Volume 12 Edisi 1

pharmaceuticals
Article
In Silico Study to Identify New Antituberculosis
Molecules from Natural Sources by Hierarchical
Virtual Screening and Molecular
Dynamics Simulations
Vinícius de S. Pinto 1 , Janay S. C. Araújo 1 , Rai C. Silva 2 , Glauber V. da Costa 3 ,
Jorddy N. Cruz 4 , Moysés F. De A. Neto 5 , Joaquín M. Campos 6 , Cleydson B. R. Santos 3, * ,
Franco H. A. Leite 1,5 and Manoelito C. S. Junior 1
1 Graduate Program in Biotechnology, State University of Feira de Santana, 44036-900 Feira de Santana, BA,
Brazil; viniciuspintto@gmail.com (V.d.S.P.); janay@hotmail.com.br (J.S.C.A.); fhenrique@uefs.br (F.H.A.L.);
mc2500@gmail.com (M.C.S.J.)
2 Graduate Program in Chemistry, Faculty of Pharmaceutical Sciences of Ribeirão Preto,
University of São Paulo, 14040-903 Ribeirão Preto, São Paulo, Brazil; raics@usp.br
3 Laboratory of Modeling and Computational Chemistry, Department of Biological and Health Sciences,
Federal University of Amapá, 68902-280 Macapá, AP, Brazil; vilhenac@hotmail.com
4 Laboratory of Preparation and Computation of Nanomaterials, Federal University of Pará, 66075-110 Belém,
PA, Brazil; jorddynevescruz@gmail.com
5 Laboratory of Molecular Modeling, State University of Feira de Santana, 44036-900 Feira de Santana, BA,
Brazil; moysesfagundes@gmail.com
6 Department of Pharmaceutical and Organic Chemistry, Faculty of Pharmacy, University of Granada,
18071 Granada, Spain; jmcampos@ugr.es
* Correspondence: breno@unifap.br; Tel.: +55-(96)-4009-2699

Received: 25 November 2018; Accepted: 27 February 2019; Published: 12 March 2019
Abstract: Tuberculosis (TB) is an infection caused by Mycobacterium tuberculosis, responsible for

1.5 million documented deaths in 2016. The increase in reported cases of M. tuberculosis resistance to
the main drugs show the need for the development of new and efficient drugs for better TB control.
Based on these facts, this work aimed to use combined in silico techniques for the discovery of
potential inhibitors to β-ketoacyl-ACP synthase (MtKasA). Initially compounds from natural sources
present in the ZINC database were selected, then filters were sequentially applied by virtual screening,
initially with pharmacophoric modeling, and later the selected compounds (based on QFIT scores)
were submitted to the DOCK 6.5 program. After recategorization of the variables (QFIT score and
GRID score), compounds ZINC35465970 and ZINC31170017 were selected. These compounds
showed great hydrophobic contributions and for each established system 100 ns of molecular
dynamics simulations were performed and the binding free energy was calculated. ZINC35465970
demonstrated a greater capacity for the KasA enzyme inhibition, with a ∆Gbind = −30.90 kcal/mol
and ZINC31170017 presented a ∆Gbind = −27.49 kcal/mol. These data can be used in other studies
that aim at the inhibition of the same biological targets through drugs with a dual action.
Keywords: virtual screening; pharmacophore model; molecular docking; β-ketoacyl-ACP

synthase; tuberculosis
Pharmaceuticals 2019, 12, 36; doi:10.3390/ph12010036 www.mdpi.com/journal/pharmaceuticals

Pharmaceuticals 2019, 12, 36 2 of 19
1. Introduction
Tuberculosis (TB) is a chronic infectious contagious disease that afflicts humanity since ancient
times. Despite being preventable and curable, tuberculosis is the ninth leading cause of death
worldwide [1]. In 2016, 10.4 million cases and 1.5 million deaths were documented [2].
TB has become a priority for the World Health Organization (WHO) since 1993 [3]. In order to
reduce its incidence and mortality, the disease has been linked to the Millennium Development Goals
(MDGs). This feature has been proposed by the United Nations, through the global plan for the fight
against tuberculosis [4].
The current treatment of tuberculosis presents some negative features, such as a lack of alternatives
for the cases of bacterial resistance to the available drugs. Multiple drug-resistant TB occurs when
there is resistance to first-line drugs (i.e., rifampicin) and at least one of the second-line injectable
drugs [5]. These mechanisms of resistance have contributed to the spread of TB.
The need for new and efficient drugs is evident for better control of TB. Treatment of the disease
should not be prolonged, and should not interfere with the administration of antiretroviral agents.
The new therapeutic approaches for the treatment of tuberculosis should act on specific targets,
without cross-resistance [6,7].
In this scenario, methodologies that can be used to optimize the identification of potential
molecules for the TB control are tests in databases [8]. Virtual screening can improve the discovery of
potential drug candidates with a reduction in cost and time [9].
Virtual strategies comprise three approaches: The first concerns ligand-based screening, the second
is based on the molecular target structure, and finally a combined approach for the first two strategies,
pertaining both in the known ligand structures and in the molecular target [10]. Ligand-based screening
can be initiated from the knowledge of at least one structure with biological activity. Structure-based
design considers the three-dimensional structure of the therapeutic target, using as main strategy the
molecular docking simulations for the selection of potential ligands with chemical, electronic and
structural characteristics that favor interactions with the molecular target orthosteric site [11].
Mycolic acids and other elements of the mycobacterial cell wall have been proposed as targets
in the mechanism of action of drugs used in the treatment of TB. The study of metabolic pathways
involved in the biosynthesis of mycolic acids has been used as important steps for the identification of
potential targets for the design of new therapeutic agents against this disease [12].
The enzyme β-ketoacyl-ACP synthase (KasA) is a validated and essential target for the survival
of mycobacteria. It is therefore an attractive study for the drug development since is based on the
Claisen acyl-ACP condensation with malonyl-ACP in the fatty acid synthase pathway in the mycolic
acid biosynthesis [13]. Moreover, it is highlighted that KasA is not present in humans, which makes it
possible to plan for more selective and safer drugs. The existence of the molecular target along with
inhibitors for the target opens up the opportunity to the use of combined virtual techniques, such as
screening by pharmacophore models and molecular docking.
After the targeting phase, the next step is the choice of the library of ligands to be explored.
Compounds from natural sources are biologically privileged structures, since they have evolved
together with proteins during their biosynthesis. Moreover, natural products have a wide and complex
chemical diversity, with some properties similar to the drugs in use [14,15]. As reactions in nature
are highly influenced by their functions, each natural product generally has a biological receptor and
therefore can potentially be a target for drugs. Natural products appear in nature to interact with a
given receptor (biomacromolecule), and therefore they could be considered as biologically-validated
chemical structures [16].
The objective of the present study is to identify molecules from natural sources through combined
virtual strategies for the control of tuberculosis through inhibition of the enzyme KasA. This work is
justified mainly by the limited number of alternatives in the treatment of tuberculosis, combined with
the high mortality rate, as well as the appearance of tuberculosis resistant cases to current medications.
Pharmaceuticals 12, 3611, x
2019, 2019,
Pharmaceuticals 3 of 19
3 of 19
2. Results and Discussion

2.1. Selected
2.1. Selected Compounds
Compounds that Exhibit
that Exhibit Biological
Biological Activity
Activity with with
TargetTarget
Thiolactomycin
Thiolactomycin (TLM;(TLM; Figure
Figure 1) is 1)
anisinhibitor
an inhibitor
withwith significant
significant efficacy
efficacy for MtKasA,
for MtKasA, but itbut
doesit does
not present
not present a considerable
a considerable inhibition
inhibition of other
of the the other
FAS FAS II condensation
II condensation enzymes,
enzymes, which
which has to
has led leda to a
greater interest in identifying other potent inhibitors for this target. TLM showed in vivo efficacy in in
greater interest in identifying other potent inhibitors for this target. TLM showed in vivo efficacy
rat models,
rat models, showing
showing goodgood pharmacokinetic
pharmacokinetic characteristics
characteristics and antibacterial
and antibacterial protection
protection [13]. [13].
Figure 1. 2D1.and
Figure 2D 3D
andstructures of thiolactomycin
3D structures (TLM).
of thiolactomycin (TLM).
2.2. Construction and Evaluation of Pharmacophore Models
2.2. Construction and Evaluation of Pharmacophore Models
Pharmacophore models are considered the most successful approach in the development of
Pharmacophore models are considered the most successful approach in the development of
new therapeutic agents, especially since the last two decades [17]. One of the advantages is that this
new therapeutic agents, especially since the last two decades [17]. One of the advantages is that this
technique allows to prioritize quickly and inexpensively molecules with high potential for interaction
technique allows to prioritize quickly and inexpensively molecules with high potential for
with the therapeutic target in large databases [18]. Ten pharmacophore models for MtKasA inhibitors
interaction with the therapeutic target in large databases [18]. Ten pharmacophore models for
were generated using six molecules of the training set (see Table 1).
MtKasA inhibitors were generated using six molecules of the training set (see Table 1).
TableFrom the models
1. Parameters generated, only
of GALAHAD™ the third
for models frommodel
MtKasAwas excluded
inhibitors. from
The the analysis,
pharmacophore because they
model
presented an energy value higher than 100.00 kcal/mol,
that presented an energy penalty is indicated by the red color. which reflects the difficulty of the molecules
to align with the model [19]. Discrepancies between energies between the remaining models can be
Models Strain Energy (kcal/mol) Hbond Mol-qry
explained mainly to steric hindrances [20]. The values for Hbond (pharmacophoric concordance)
varied from 478.102to 650.30, while Mol_QRY 7.53 650.30
(fitting of each inhibitor 112.40
to the model) showed a lower
8 7.58 613.30 114.60
variation from 112.40 to 127.7 kcal/mol.
1 8.02 619.30 127.70
For the pharmacophore
5 models 9.05generated for MtKasA, 497.20 it was122.60
possible to identify that the
KasA_002 model (see 10 Table 1) presents the best values for478.10
9.54 the GALAHAD™ 125.10 parameters: an energy
value equal to 7.534kcal/mol, Hbond equal 10.52 to 650.30 and Mol_QRY
520.90 128.40
equal to 112.40.
7 45.88 504.00 126.20
9 67.80 550.70 132.00
Table 1. Parameters of GALAHAD™ for models from MtKasA inhibitors. The pharmacophore
6 69.55 482.60 136.70
model that presented
3
an energy penalty is indicated by the655.80
496.79
red color. 133.10
Models Strain Energy (kcal/mol) Hbond Mol-qry

2 7.53 650.30
From the models generated, only the third model was excluded from the 112.40
analysis, because they
8 7.58 613.30 114.60 of the molecules
presented an energy value higher than 100.00 kcal/mol, which reflects the difficulty
1 8.02 619.30 127.70
to align with the model [19]. Discrepancies between energies between the remaining models can be
5 9.05 497.20 122.60
explained mainly to steric hindrances [20]. The values for Hbond (pharmacophoric concordance)
10 9.54 478.10 125.10
varied from 478.10 to 650.30, 4while Mol_QRY (fitting
10.52 of each inhibitor
520.90to the 128.40
model) showed a lower
variation from 112.40 to 127.77kcal/mol. 45.88 504.00 126.20
For the pharmacophore9 models generated 67.80for MtKasA, it 550.70
was possible to identify that the
132.00
KasA_002 model (see Table 1) 6 presents the best69.55
values for the GALAHAD™
482.60 parameters: an energy
136.70
value equal to 7.53 kcal/mol,3Hbond equal to 650.30
496.79 and Mol_QRY 655.80 133.10
equal to 112.40.
Small values for energy and high for Hbond and Mol_QRY are the ideal ones to select the best
model [19], because reflects a greater ease of the molecules to mold when aligned with the model,
Pharmaceuticals 2019, 11, x 4 of 19
Small values for energy and high for Hbond and Mol_QRY are the ideal ones to select the best
model [19], because reflects a greater ease of the molecules to mold when aligned with the model, a
a better pharmacophore agreement
better pharmacophore agreement andandaagood
goodfitfitofofeach
eachinhibitor
inhibitor
to to
thethe model.
model. Consequently,
Consequently, such
such facts have led to the choice of the KasA_002 model. The schematic representation
facts have led to the choice of the KasA_002 model. The schematic representation of this of this model,
model, and
and
thethe pharmacophore
pharmacophore characteristics
characteristics areare shown
shown inin Figure
Figure 2. 2.
Figure 2. Representation of the best pharmacophore model for KasA inihibitors. Pink: Hbond donor;
Figure
green: 2. Representation
Hbond of the
acceptors; cyan: best pharmacophore
hydrophobic model
centers. The forthe
size of KasA inihibitors.
spheres Pink:the
represents Hbond donor;
tolerance.
green:
The Hbond
distances areacceptors; cyan: hydrophobic
shown in angstroms (Å). centers. The size of the spheres represents the tolerance.
The distances are shown in angstroms (Å).
The spheres describe the area that should be occupied by a certain functional group with identical
characteristics to those
The spheres presented
describe thebyarea
that that
of theshould
pharmacophore pointby[21].
be occupied The selected
a certain pharmacophore
functional group with
model
identical characteristics to those presented by that of the pharmacophore point [21]. The center
has six characteristics, being two hydrogen acceptor centers (green), one hydrogen donor selected
(magenta) and fourmodel
pharmacophore hydrophobic
has sixcenters (blue). being two hydrogen acceptor centers (green), one
characteristics,
hydrogen donor center (magenta) and four hydrophobic centers (blue).
Pharmacophore-Based Virtual Screening
2.2.1.
ThePharmacophore-based
pharmacophore model Virtual Screening
displays the key features involved in the interactions within the
ligand-target complex. Thereby,
The pharmacophore this displays
model techniquethecankey
be applied
featuresininvolved
the questinofthe
compounds, which
interactions meetthe
within
theligand-target
main molecular requirements for the inhibition of a given target [22].
complex. Thereby, this technique can be applied in the quest of compounds, which
In order
meet to select
the main candidate
molecular moleculesfor
requirements forthe
theinhibition
biologicalof assays against
a given MtKasA,
target [22]. virtual screening
was performed based on the pharmacophore characteristics of
In order to select candidate molecules for the biological assays against the selected model. MtKasA,
With the total
virtual
number of compounds present in the ZINC natural-product database (142,788
screening was performed based on the pharmacophore characteristics of the selected model. molecules), 3896 were
With
aligned with the characteristics of the pharmacophore model KasA_002.
the total number of compounds present in the ZINC natural-product database (142,788 molecules),
The
3896 UNITY
were alignedplatform
with the provides a QFITofvalue
characteristics as a scoring form
the pharmacophore modelforKasA_002.
alignment in the model.
The valueThe of QFIT platform
UNITY can varyprovides
from 0 toa QFIT
100, invalue
whichas a100 is theform
scoring best for
value and presents
alignment a greater
in the model. The
complementarity with the pharmacophore [23]. The QFIT values for the
value of QFIT can vary from 0 to 100, in which 100 is the best value and presents a greater molecules submitted to
screening ranged from
complementarity with1.47the
to 66.76. The five molecules
pharmacophore [23]. Thewith
QFITthevalues
best QFIT
for value are shownsubmitted
the molecules in Table 2.to
ZINC35465970
screening ranged givesfrom
the higher QFIT value
1.47 to 66.76. formolecules
The five the alignment
with inthethe
bestbest pharmacophore
QFIT value are shown model of
in Table
KasA (QFIT = 66.76).
2. ZINC35465970 gives the higher QFIT value for the alignment in the best pharmacophore model of
However,
KasA (QFIT =the obtained results do not allow the identification of the mode of interaction, molecular
66.76).
volume,However,
as well as do
thenot quantifyresults
obtained the affinity
do energy between
not allow the the molecule and
identification ofthe
thereceptor,
mode which makes
of interaction,
necessary the accomplishment of the docking. Thus, docking assays were performed.
molecular volume, as well as do not quantify the affinity energy between the molecule and the The molecules
selected for the
receptor, whichdocking
makes followed
necessarythe the
evaluation criterion byofthe
accomplishment themean + 2 × docking
valueThus,
docking. standardassays
deviation
were
(34.73). Thereby, 186 molecules were submitted to the docking calculation.
performed. The molecules selected for the docking followed the evaluation criterion by the mean
value + 2 × standard deviation (34.73). Thereby, 186 molecules were submitted to the docking
calculation.
Pharmaceuticals
Pharmaceuticals 11,12,
2019,
2019, x 36 5 of
5 of 1919
Table 2. The five molecules with the best QFIT value.
Molecule
Table 2. The
Table fivefive
The moleculesStructure
with
molecules the the
with bestbest
QFIT value.
QFIT QFIT
value.QFIT
Molecule
Table 2.2.The Structure
five molecules with the best QFIT value.
Molecule Structure QFIT
Molecule Structure QFIT
Molecule
Molecule Structure
Structure QFIT
QFIT
ZINC35465970 66.76
ZINC35465970 66.76
ZINC35465970 66.76
ZINC35465970 66.76
ZINC35465970
ZINC35465970 66.76
66.76
ZINC15959689 62.97
ZINC15959689 62.97
ZINC15959689 62.97
ZINC15959689
ZINC15959689 62.97
62.97
ZINC15959689 62.97
ZINC16032930 62.07
ZINC16032930 62.07
ZINC16032930
ZINC16032930 62.07
62.07
ZINC16032930 62.07
ZINC16032930 62.07
ZINC31161132 59.99
ZINC31161132
ZINC31161132 59.99
59.99
ZINC31161132 59.99
ZINC31161132 59.99
ZINC31161132 59.99
ZINC72320274
ZINC72320274 59.86
59.86
ZINC72320274 59.86
ZINC72320274 59.86
ZINC72320274 59.86
ZINC72320274 59.86
2.3.2.3.
Docking-based
Docking-Based Virtual Screening
Virtual Screening
2.3.
2.3. Docking-based
Docking-based Virtual
Virtual Screening
Screening
2.3. Docking-based
Initially,
Initially, Virtual
thethe
scoring Screening
function
scoring function waswasselected.
selected.ForFor
this, we we
this, used some
used someevaluation
evaluationmetrics such
metrics as as
such
2.3.Initially,
Docking-based
the Virtual
scoring Screening
function was
Initially,
molecular
molecular the scoring
redocking,
redocking, KISSfunction
KISS score
score was selected.
selected.
calculation,
calculation, ROC
ROC
For
For this,
this,
curve
curve
we
we used
used
analysis
analysis and
some
some
and
the
evaluation
evaluation
the
metrics
metrics
enrichment
enrichment factor.
such
such
factor.
The
as
as
The
evaluation
Initially,
molecular the scoring
redocking, KISSfunction
score was selected.
calculation, ROCForcurve
this, we used some evaluation metrics suchTheas
Initially,
molecular
evaluation
of the of thethepositioning
redocking,
positioning scoring
ofKISS function
score
of the
the ligand wasorthosteric
ligand
the selected.
incalculation, ROC
in the For this,analysis
curve
orthosteric
site from we used
analysis
site
the from
DOCK’s
and
some
andthe
the
the
search
enrichment
evaluation
enrichment
DOCK's search
algorithm
factor.
metrics such
factor. Theas
algorithm
was performed
molecular
evaluation redocking,
of the KISS
positioning score
of the calculation,
ligand in ROC
the curve
orthosteric analysis
site fromandthethe enrichment
DOCK's search factor. The
algorithm
molecular
evaluation
was of
performed
through redocking,
the thethrough
values KISS
positioning
of RMSDs, score
the of the calculation,
values
RMSDh ligand
of and
RMSDs, theROC
inRMSDm. RMSDhcurve
orthosteric
The andanalysis
siteRMSDm.
RMSDs from and
evaluate thedeviation
thethe enrichment
DOCK's
The RMSDssearch factor.
algorithm
evaluate
between The
the
different
evaluation
was of the
performed positioning
through the of the ligand
values of in the orthosteric
RMSDs, RMSDh andsiteRMSDm.
from the The
DOCK's
RMSDssearch algorithm
evaluate the
evaluation of the positioning of the ligand in the orthosteric site
was performed through the values of RMSDs, RMSDh and RMSDm. The RMSDs evaluate the from the DOCK's search algorithm
was performed through the values of RMSDs, RMSDh and RMSDm. The RMSDs evaluate the
was performed through the values of RMSDs, RMSDh and RMSDm. The RMSDs evaluate the
deviation between
Pharmaceuticals 2019, 11,36
x different heavy atom pairs of hydrogen from the reference conformation 66 of
to1919
the
pose after redocking. In turn, the RMSDm is an implementation of RMSD used in AutoDock Vina
deviation
[24], which between
does not different
explicitlyheavy atom apairs
consider of hydrogen
mapping atom by fromatom;the thereference
deviationconformation
is estimated from to the the
heavy
pose atom
after pairs
redocking. of hydrogen
In turn, from
the the
RMSDm reference
is an conformation
implementation to the
of pose
RMSD after
used redocking.
in AutoDock In turn,
Vina
minimum distance between any atom of the same element in reference conformations and
the
[24],RMSDm
which does is annot implementation
explicitly of RMSD
consider used in
a mapping AutoDock atom;Vina [24], which does not explicitly
recoupled. The RMSDh performs a correction ofatom
the by
symmetry thebetween
deviation theis heavy
estimated
atomsfrom theA
[25].
consider
minimum a mapping
distance atom by atom;
between any theatom
deviationof is estimated
the same from the
element in minimum
reference distance between any
conformations and
value of RMSD < 2.0 Å is considered acceptable [26].
atom of the same
recoupled. The RMSDhelement in reference
performs a conformations
correction andsymmetry
of the recoupled. between
The RMSDh the performs a correction
The redocking showed RMSDs = 0.26Å, RMSDh = 0.26Å and RMSDm =heavy
0.14Åatoms [25]. A3).
(see Figure
of the
value symmetry
of RMSD between
< 2.0 Å isthe heavy
considered atoms [25].
acceptable A value
[26]. of RMSD < 2.0 Å is considered acceptable [26].
The methods of calculation of RMSD indicated excellent results, and the DOCK program similarly
The
The redocking
redocking showed
showed RMSDsRMSDs = 0.26Å,
0.26Å, RMSDh
RMSDh = = 0.26Å
0.26Åinand RMSDm = 0.14Å (see
(see Figure 3).
reproduced the conformation of the=crystallographic ligand and RMSDm
the active site=of0.14Å
the protein,Figure
since3). the
The
The methods
methods of calculation of RMSD indicated excellent results, and the DOCK program similarly
values foundofwere calculation
ideal as of RMSD indicated
recommended excellent[11,26].
by literature results,Theand the DOCK
second stage of program similarly
evaluation for the
reproduced
reproduced the
theconformation
conformation of
of the
the crystallographic ligand in
in the
the active
active site of
of the
the protein, since
since the
methodology of docking refers tocrystallographic
the analysis of ligand the intermolecular site
interactions protein,
after redocking. the
values
values found were ideal as recommended by literature [11,26]. The second stage of evaluation for
KISS found
score iswerethe ideal
resultasofrecommended
the ratio between by literature
the number [11,26]. The second
of hydrogen stageinofthe
bonds evaluation
redockedforligand the
the methodology
methodology of of docking
docking refers
refers to to
thetheanalysis
analysisofofthetheintermolecular
intermolecular interactions
interactions after
after redocking.
redocking.
and the number of hydrogen bonds in the crystallographic ligand (PDB = 4C6X). KISS score = 1 was
KISS
KISS score
score isis the
the result of ofthe
theratio between the
thenumber
number of hydrogen bonds in the
theredocked
redocked ligand
obtained for theresult
target. The ratio between
interaction of the hydrogen ofobserved
hydrogenwith bonds theincrystallographic ligand
ligand
and
and the
the number
number of
of hydrogen
hydrogen bonds
bonds in
in the
the crystallographic
crystallographic ligand
ligand (PDB
(PDB == 4C6X).
4C6X). KISS
KISS score
score == 11 was
was
was reproduced after redocking, and no interaction of the same nature was formed. Variations
obtained
obtained for
for the
the target.
target. The
The interaction
interaction of ofthethe
hydrogen
hydrogen observed
observed withwiththe crystallographic
the crystallographicligand was
ligand
occurred only in the hydrophobic interactions, with the His275 and Phe236 residues.
reproduced
was reproduced after redocking,
after redocking,and noand interaction of the same
no interaction of the nature
samewas formed.
nature wasVariations occurred
formed. Variations
only in the hydrophobic interactions, with the His275 and
occurred only in the hydrophobic interactions, with the His275 and Phe236 residues. Phe236 residues.
Figure 3. Result of the redocking. Crystallographic ligand in cyan and the best docking pose in
orange.
Figure 3. Result of the redocking. Crystallographic ligand in cyan and the best docking pose in orange.
Figure 3. Result of the redocking. Crystallographic ligand in cyan and the best docking pose in
orange.
The
The third
third stage
stage for analysis
for the the analysis
of theof the scoring
scoring functions
functions consisted
consisted in the evaluation
in the evaluation of the
of the Receiver
Receiver Characteristic
Operating Operating Characteristic
(ROC) curve. (ROC) curve.
This step Thisofstep
is one is one
the best of the
ways best ways
to compare theto compare the
performance
The third of
performance stage for the
scoring analysis
functions and ofclassifiers
the scoring functions
[27]. For consisted
this, a library in
wastheconstructed
evaluationwith
of the
550
of scoring functions and classifiers [27]. For this, a library was constructed with 550 decoys and six
Receiver Operating
decoys and Characteristic (ROC) curve. This step is one of the best ways to compare the
inhibitors, fivesixof inhibitors,
these withfivetwoof theseofwith
states two statesAfter
protonation. of protonation. After of
the construction thethe
construction of the
molecule bank
performance
molecule of scoring
bank (ligands functions
+ decoys),andthe
classifiers
docking [27].
tests For this,
were a library
carried out was constructed
with two scoring with 550
functions
(ligands + decoys), the docking tests were carried out with two scoring functions (Grid-Hawkins
decoys and six
(Grid-Hawkins inhibitors,
GB/SAThe five
and of these
Gridwith with
Score). two states
The curve of
graphwith protonation.
with After the construction of the
GB/SA and Grid Score). graph the ROC thethe
twoROC curve
scoring with the
functions two can
applied scoring
be
molecule
functionsbank (ligandsbe + visualized
decoys), thein docking
Figure 4. tests were carried out with two scoring functions
visualized inapplied
Figure can4.
(Grid-Hawkins GB/SA and Grid Score). The graph with the ROC curve with the two scoring
functions applied can be visualized in Figure 4.
Figure 4. ROC
Figure curves
4. ROC forfor
curves evaluation of of
evaluation Grid-Hawkins GB/SA
Grid-Hawkins and
GB/SA Grid
and Score.
Grid Score.
Figure 4. ROC curves for evaluation of Grid-Hawkins GB/SA and Grid Score.
Fromthe
From theROCROCcurve
curvegraph,
graph,it it
waswas possible
possible to to evaluate
evaluate thethe performance
performance of scalar
of scalar measures
measures of
classification, as specificity and sensitivity. The ROC curve is a graphical representation ofofthe
of classification, as specificity and sensitivity. The ROC curve is a graphical representation the
sensitivity(proportion
sensitivity (proportionofoftrue truepositives)
positives)asasaafunction
functionofofspecificity
specificity(proportion
(proportionofoffalse falsepositives).
positives).
The value of the area on the curve (AUC) provides an objective measure
The value of the area on the curve (AUC) provides an objective measure of the overall performance of the overall performance
ofofaaclassifier.
classifier.An AnAUCAUCvalue
valueequal
equalto to11(or
(or100%)
100%)indicates
indicatesthat thatthe
theactive
activeand andinactive
inactivecompounds
compounds
areperfectly
are perfectly discriminated,
discriminated, whilewhile aa value
valueofof0.5
0.5(or
(or50%)
50%) is is
understood
understood as asa random
a random performance
performance [28].
Matsubara’s studies [29] show that, in general, the accuracy of the classification
[28]. Matsubara's studies [29] show that, in general, the accuracy of the classification method can be method can be
evaluated with
evaluated with the
the following
followingscale:
scale:0.9–1:
0.9–1:excellent;
excellent;0.8–0.89:
0.8–0.89:good; 0.6–0.79:
good; 0.6–0.79: reasonable;
reasonable;0.5–0.59: poor;
0.5–0.59:
and below 0.49 corresponds to a complete
poor; and below 0.49 corresponds to a complete failure.failure.
Fromthe
From theanalysis
analysisofofthe
theROC
ROCcurve,
curve,ititwas
waspossible
possibletotoobserve
observethat thatthethescoring
scoringfunction
functionproved
proved
totobe
bethe
themost
mostefficient
efficientininthe
therecovery
recoveryof ofthe
thebioactive
bioactivecompounds
compoundsfor forMtKasA
MtKasAwas wasGrid
Grid++Hawkins
Hawkins
GB/SA, with an AUC = 0.96, which can be considered as an excellent value
GB/SA, with an AUC = 0.96, which can be considered as an excellent value for the classification. for the classification.
Oneof
One of the
the problems
problems pointed
pointedoutouttotothe AUC
the AUC value
valueis the factfact
is the thatthat
it is ait global measure,
is a global not clearly
measure, not
presenting
clearly information
presenting about about
information the early
the recognition of active
early recognition compounds
of active compounds [30,31]. On the
[30,31]. Onother hand,
the other
the enrichment
hand, factor (EF)
the enrichment quantifies
factor the proportion
(EF) quantifies of active compounds
the proportion identified when
of active compounds analyzed
identified whenin a
given proportion
analyzed in a given (nX%) of the total
proportion (nX%)setof
ofthe
ordered compounds.
total set of orderedThe enrichmentThe
compounds. factors were calculated
enrichment factors
for the functions, considering 1%, 5%, 10% and 25% of the bank that was
were calculated for the functions, considering 1%, 5%, 10% and 25% of the bank that was coupled to. coupled to. The results are
shown in Figure 5.
The results are shown in Figure 5.
Figure5.5.Enrichment
Figure Enrichmentfactor (EF)
factor forfor
(EF) scoring functions
scoring used
functions in MtKasA
used in 1, in
in MtKasA 5, 10 and
1, 5, 1025%
andof25%
the of
database.
the
database.
According to the results presented in Figure 5, it was possible to highlight that, among the two
scoring functions
According tested,
to the the Grid
results + Hawkins
presented GB/SA
in Figure function
5, it was presents
possible a 15.3 times
to highlight that,greater
amongchance
the twoof
identifying the active compounds in only 1% of the database.
scoring functions tested, the Grid + Hawkins GB/SA function presents a 15.3 times greater chance of
Considering
identifying the compounds
the active two analyzes,inthe area
only 1%onofthe
thecurve and the enrichment factor, the Grid + Hawkins
database.
GB/SA
Considering the two analyzes, the area on the curve that
scoring function was presented as the methodology and best
the classifies,
enrichment andfactor,
therefore
the presents
Grid +
the greatest
Hawkins accuracy.
GB/SA scoring function was presented as the methodology that best classifies, and
After
therefore selecting
presents thethe scoring
greatest function with greater accuracy for each target, the virtual screening
accuracy.
wasAfter
performed by docking using structures
selecting the scoring function withwith a QFIT
greater value equal
accuracy for eachto or greater
target, thethan 34.73.
virtual Out the
screening
186 performed
was molecules subjected
by docking to the docking,
using onlywith
structures 152 achieved someequal
a QFIT value affinity
to to
or MtKasA. The34.73.
greater than affinity energy
Out the
values for these molecules ranged from − 4.87 to − 67.70 kcal/mol (see Figure 6).
186 molecules subjected to the docking, only 152 achieved some affinity to MtKasA. The affinity
energy values for these molecules ranged from −4.87 to −67.70 kcal/mol (see Figure 6).
In an
anattempt
attempttotorelate thethe
relate QFIT values
QFIT obtained
values fromfrom
obtained similarity screening
similarity and the
screening affinity
and energy
the affinity
values resulting from docking calculations, a consensual analysis was performed using these
energy values resulting from docking calculations, a consensual analysis was performed using these two
variables (see Table
two variables 3). 3).
(see Table
Figure 6. Distribution of compounds according to their affinity energy

energy against
against MtKasA.
MtKasA.
Table 3. Consensus ranking of variables: relationship between the result of QFIT and Grid + Hawkins
Table 3. Consensus ranking of variables: relationship between the result of QFIT and Grid +
GB/SA for MtbKasA.
Hawkins GB/SA for MtbKasA.
MOLECULE
MOLECULE CONSENSUS
CONSENSUS
ZINC35465970
ZINC35465970 122.10
122.10
ZINC31170017
ZINC31170017 108.57
108.57
ZINC12659549
ZINC12659549 108.52
108.52
ZINC08453820
ZINC08453820 107.44
107.44
ZINC15959689
ZINC15959689 107.28
107.28
The structure with the best score for MtbKasA was ZINC35465970, which has 332.5 g/mol of
The structure with the best score for MtbKasA was ZINC35465970, which has 332.5 g/mol of
molecular mass, three Hbond acceptors, three Hbond donors, nine rotatable bonds and partition
molecular mass, three Hbond acceptors, three Hbond donors, nine rotatable bonds and partition
coefficient (logP) of 6.44. Information extracted from ZINC [32] PubChem [33] and ChemSpyder [34]
coefficient (logP) of 6.44. Information extracted from ZINC [32] PubChem [33] and ChemSpyder [34]
indicate that such a molecule has not been subjected to biological evaluation. ZINC31170017
indicate that such a molecule has not been subjected to biological evaluation. ZINC31170017 presented
presented the second best value of consensus, and the analysis of the physical-chemical properties
the second best value of consensus, and the analysis of the physical-chemical properties indicates
indicates that this molecule has 462.5 g/mol of molecular mass, nine Hbond acceptors, six Hbond
that this molecule has 462.5 g/mol of molecular mass, nine Hbond acceptors, six Hbond donors,
donors, ten rotatable bonds and 1.02 xlogP. It was not possible to identify experiments related with
ten rotatable bonds and 1.02 xlogP. It was not possible to identify experiments related with the
the biological activity for ZINC31170017. Therefore, from the analysis of the scoring function and the
biological activity for ZINC31170017. Therefore, from the analysis of the scoring function and the
similarity alignment (QFIT), molecules ZINC35465970 and ZINC31170017 were submitted for
similarity alignment (QFIT), molecules ZINC35465970 and ZINC31170017 were submitted for analysis
analysis of the intermolecular interactions.
of the intermolecular interactions.
The analysis of the intermolecular interactions is useful for identification and optimization of
The analysis of the intermolecular interactions is useful for identification and optimization of
contacts between ligands and target [35]. The results for the two molecules selected are shown in
contacts between ligands and target [35]. The results for the two molecules selected are shown
Figure 7: It is possible to point out a greater contribution of the hydrophobic interactions in both
in Figure 7: It is possible to point out a greater contribution of the hydrophobic interactions in
molecules. ZINC35465970 (Figure 7A) performs hydrophobic interactions of its aliphatic chains
both molecules. ZINC35465970 (Figure 7A) performs hydrophobic interactions of its aliphatic
attached to the dihydroxybenzene ring with the Phe403, Pro 279, Phe401, Gly402 and Gly317
chains attached to the dihydroxybenzene ring with the Phe403, Pro 279, Phe401, Gly402 and Gly317
residues. In addition to these interactions, this molecule forms aromatic interaction π-π, between the
residues. In addition to these interactions, this molecule forms aromatic interaction π-π, between the
dihydroxybenzene ring and the aromatic ring of the side chain of Phe403. It is also possible to
observe a hydrogen interaction between a hydroxyl of the molecule with the Val277 ketone.
Considering the complex formed between compound ZINC31170017 (Figure 7B) and MtbKasA,
hydrophobic interactions with Phe403 and Met212 residues can be observed. In addition, a
hydrophobic π-π interaction with Phe403 and the dihydroxy benzene ring of the molecule can be
observed, also
Pharmaceuticals 2019,present
12, 36 in the molecule with the best score, indicating that this residue may 9 of be
19
involved in the molecular recognition process. The third type of binding present is a hydrogen
interaction, occurring between the two hydroxyls of the dihydroxy benzene ring with the imidazole
dihydroxybenzene
rings of His344 andring and the
His310. aromaticinteractions
Hydrogen ring of the with
side chain of Phe403.
the same It is
residues also possible
(His310 to observe
and His344) have
aalready
hydrogen interaction between a hydroxyl
been verified in other studies [36]. of the molecule with the Val277 ketone.
Figure 7. Interactions of compounds against the active site of Mycobacterium tuberculosis KasA,
Figure
In 7. Interactions and
(A) ZINC35465970 of compounds against the active site of Mycobacterium tuberculosis KasA, In (A)
(B) ZINC31170017.
ZINC35465970 and (B) ZINC31170017.
Considering the complex formed between compound ZINC31170017 (Figure 7B) and MtbKasA,
2.4. Structural Analysis
hydrophobic of Systems
interactions with Phe403 and Met212 residues can be observed. In addition,
a hydrophobic π-π interaction with Phe403 and the dihydroxy benzene ring of the molecule can
The root mean square deviation (RMSD) of the atomic positions of protein was plotted to
be observed, also present in the molecule with the best score, indicating that this residue may be
evaluate the structural stability of the complexes (KasA-ZINC31170017 and KasA-ZINC35465970)
involved in the molecular recognition process. The third type of binding present is a hydrogen
and the binder backbone along the molecular dynamics trajectory. To plot the backbone and RMSD
interaction, occurring between the two hydroxyls of the dihydroxy benzene ring with the imidazole
of the complex graphs the Cα atoms and heavy atoms were used, respectively. The graphs plotted of
rings of His344 and His310. Hydrogen interactions with the same residues (His310 and His344) have
systems can be seen in Figure 8. Throughout the simulation time, the inhibitors remained bound to
already been verified in other studies [36].
the active site of the protein. The ZINC35465970 ligand remained in equilibrium exhibiting slight
structural
2.4. divergences.
Structural Analysis of During
Systems approximately the 20 ns of the initial simulation, ZINC31170017
underwent several conformational changes, but along the trajectory reached a balance and started to
Thesmall
exhibit root mean square
changes deviation
in KasA (RMSD) of Thus,
conformation. the atomic positions ofwere
the complexes protein was plotted
considered to evaluate
stable and not
the structural stability of the
discarded for a further analysis. complexes (KasA-ZINC31170017 and KasA-ZINC35465970) and the
binder backbone along the molecular dynamics trajectory. To plot the backbone and RMSD of the
complex graphs the Cα atoms and heavy atoms were used, respectively. The graphs plotted of systems
can be seen in Figure 8. Throughout the simulation time, the inhibitors remained bound to the active
site of the protein. The ZINC35465970 ligand remained in equilibrium exhibiting slight structural
divergences. During approximately the 20 ns of the initial simulation, ZINC31170017 underwent
several conformational changes, but along the trajectory reached a balance and started to exhibit small
changes in KasA conformation. Thus, the complexes were considered stable and not discarded for a
further analysis.
Figure 8. Evaluationof of RMSDplots

plots duringMD MD simulations. The
The KasAprotein
protein backbonehas has been
Figure 8.8. Evaluation
Figure Evaluation of RMSD
RMSD plots during
during MD simulations.
simulations. The KasA
KasA protein backbone
backbone has been
been
represented
represented in black, while the ligands graphs have been represented in different colors. (a) RMSDs
representedin inblack,
black,while
whilethe
theligands
ligandsgraphs
graphshave
havebeen represented
been in in
represented different colors.
different (a) (a)
colors. RMSDs of
RMSDs
of the
the KasA-
KasA- ZINC35465970
ZINC35465970 system
system andand
(b) (b) RMSDs
RMSDs of of the
the KasA-ZINC31170017
KasA-ZINC31170017 system.
system.
of the KasA- ZINC35465970 system and (b) RMSDs of the KasA-ZINC31170017 system.
Thisway,
This way, the mean
the root rootsquare
meanfluctuation
square (RMSF)
fluctuation (RMSF) of
of the complexes the (KasA-ZINC31170017
residues complexes residues
This way, the root mean square fluctuation (RMSF) of the complexes residues
(KasA-ZINC31170017
and KasA-ZINC35465970) andhave
KasA-ZINC35465970)
been plotted using have
the Cαbeen plotted
atoms usingthe
to analyze theprotein
Cα atoms to analyze
backbone (see
(KasA-ZINC31170017 and KasA-ZINC35465970) have been plotted using the Cα atoms to analyze
the protein
Figure 9). backbone (see Figure 9).
the protein backbone (see Figure 9).
Figure9.9.Protein
Figure Proteinbackbone
backboneRMSF
RMSFplots.
plots.
Figure 9. Protein backbone RMSF plots.
The
TheRMSF
RMSFplots
plotsrevealed
revealeddifferences
differencesin
inprotein
proteinflexibility
flexibilitythroughout
throughoutthe thetrajectory.
trajectory.The
Thelargest
largest
The
differencesRMSF plots revealed differences in protein flexibility throughout the trajectory. The
200 largest
differencesininresidue
residuefluctuations
fluctuationsoccurred
occurredin in
thethe
range
rangeof residues 60 to
of residues 6080,
to 110 to 150
80, 110 to and
150 and to 220.
200 to
differences
In Figure 10 in residue
these fluctuations
protein segments occurred
were in the
colored andrange of residues 60 to 80, 110 to 150 and 200 to
identified.
220. In Figure 10 these protein segments were colored and identified.
220. In Figure 10 these protein segments were colored and identified.
greater residue
Figure 10. Regions of protein that showed greater residue fluctuations.
fluctuations.
The
The amino
amino acids
acids 6060 to
to 80 form a loop region
region and
and two
two small
small alpha helices that are are exposed
exposed to to the
the
solvent.
solvent. The fluctuation of this backbone region may be related to structural features of loop regions
regions
that
that naturally
naturally have
have aa degree
degree ofof flexibility.
flexibility.
ZINC35465970
ZINC35465970 compound shifted from its its molecular
molecular docking
docking position
position further
further into
into the
the bonding
bonding
cavity during the simulation. In the molecular docking pose, this ligand
cavity during the simulation. In the molecular docking pose, this ligand did not show manydid not show many interactions
with the alpha
interactions helixthe
with residues
alpha andhelixtheresidues
loop comprised
and thebetween residues 200–220.
loop comprised betweenHowever,
residues after MD
200–220.
simulation
However, and afterwith
MD thesimulation
balance of ligand in the the
and with binding pocket,
balance of two hydrophobic
ligand chains ofpocket,
in the binding the complex
two
underwent
hydrophobic conformational
chains of thereorientation, thus establishing
complex underwent greater hydrophobic
conformational reorientation,interactions with the
thus establishing
residues in the range interactions
greater hydrophobic of 200 to 220.withZINC31170017
the residuesshowed interactions
in the range of 200 towith
220. few residues present
ZINC31170017 showedat
the beginning of alpha helix formed by the residues in the range that we have
interactions with few residues present at the beginning of alpha helix formed by the residues in the analyzed. In this way,
the alpha
range thathelix
we region remained freer
have analyzed. In thisto move,
way, theandalpha
consequently demonstrating
helix region remained afreer
greater
to fluctuation.
move, and
The region
consequently of the proteinathat
demonstrating showed
greater the greatest fluctuation is composed by the residues 110 to
fluctuation.
150. In theregion
The system of established
the protein with ZINC31170017
that showed this region
the greatest presented
fluctuation a more open
is composed by theconformation,
residues 110
more
to 150.exposed
In theto system
the solvent, whereaswith
established the same region of the
ZINC31170017 thisprotein
regioninpresented
the systema formed with
more open
ZINC35465970
conformation, more remained in a tightest
exposed conformation,
to the solvent, whereas in the
comparison
same regionto each other.
of the protein in the system
formed with ZINC35465970 remained in a tightest conformation, in comparison to each other.
2.4.1. Hydrogen Bonds Established between Receptor-Ligands
2.4.1.Taking
Hydrogen
the Bonds
RMSDEstablished
and RMSFbetween Receptor-Ligands
results are a visual analysis of representative MD trajectory,
the formation of hydrogen bonds during the entire
Taking the RMSD and RMSF results are a visual computational simulation time
analysis of representative MD was carried out
trajectory, the
to investigate
formation the interaction
of hydrogen bondsprofile
duringofthe
complexes. The main interactions
entire computational simulationestablished are shown
time was carried out in
to
Table 4.
investigate the interaction profile of complexes. The main interactions established are shown in
Table 4.
Table 4. Hydrogen bonds formed in complexes.
Acceptor Hydrogen Donor Donor Occupancy (%) a Average Distance (Å)

ZINC35465970
ZINC35465970_416@O25 GLN_170@HE22 GLN_170@NE2 30.46 3.17
ZINC35465970_416@O25 HIS_344@HE2 HIS_344@NE2 28.75 3.03
ZINC35465970_416@O24 LYS_339@HZ3 LYS_339@NZ 21.69 2.89
ZINC31170017
MET_212@O ZINC31170017_416@H62 ZINC31170017_416@O37 71.27 2.81
ARG_233@O ZINC31170017_416@H56 ZINC31170017_416@O22 49.69 3.01
a Occupancy is defined as the percentage of time that hydrogen bonding existed during the 100 ns simulation time.
There was a difference between the number of hydrogen bonds established by compounds in the
binding pocket. ZINC35465970 showed the highest number of hydrogen bonds formed, which can
justify its greater structural stability verified in the RMSD graph and also its greater capacity to connect
the enzyme, and consequently its greater power of inhibition verified in binding free energy value
(∆Gbind = −30.90 kcal/mol). However, ZINC31170017 established only two hydrogen bonds with
the receptor, with this presented lower conformational stability throughout the simulation and less
capacity of interaction with the protein.
2.4.2. Bind Free Energy KasA-Ligands

The free energy values and their energy components are summarized in Table 5. The results
obtained by the MM/GBSA method suggest that ZINC35465970 has a greater capacity to inhibit the
KasA enzyme, since we obtained the value of ∆Gbind = −30.90 kcal/mol, whereas the ZINC31170017
reached the value of ∆Gbind = −27.49 kcal/mol. The van der Waals interactions (∆EvdW ) were the main
responsible for maintaining the enzyme-inhibitor complexes, in the systems formed, ∆EvdW presented
values of −45.21 kcal/mol and −35.86 kcal/mol, to ZINC35465970 and ZINC31170017, respectively.
The electrostatic (∆Eele ) and non-polar (∆GNP ) contributions also favored the established systems.
For the interaction with ZINC35465970, ∆Eele = −11.93 kcal/mol and ∆GNP = −6.06 kcal/mol for
ZINC31170017 ∆Eele = −11.48 kcal/mol and ∆GNP = −5.18 kcal/mol were obtained.
Table 5. Energy contributions to the free energy binding KasA-compounds.
Compound ∆EvdW ∆Eele ∆GGB ∆GNP ∆Gbind

ZINC35465970 −45.21 −11.93 32.31 −6.06 −30.90
ZINC31170017 −35.86 −11.48 25.04 −5.18 −27.49
3. Materials and Methods
3.1. Dataset
Initially a dataset with potent KasA inhibitors (Ki < 40 µM) [36] was employed for construction
and evaluation for pharmacophore modelling (Table 6). The subset of natural products from the ZINC
database (https://zinc.docking.org) [32] with 142.788 compounds were selected for virtual screening.
All molecules were drawn on the Marvin Sketch software [37]. Then they were prepared through
calculation of the Gasteiger-Hückel charges and addition of the hydrogen atoms using the DockPrep
module implemented at Chimera 1.10.1 [38].
Pharmaceuticals
Pharmaceuticals2019,
2019, 12,xx36
11, 13
1313of
ofof19
19
19
Table
Table6.
Table 6.6.KasA
KasAinhibitors
KasA inhibitorsselected
inhibitors selectedfor
selected forconstruction
for constructionand
construction andevaluation
and evaluationof
evaluation ofpharmacophore
of pharmacophoremodels.
pharmacophore models.
models.
Molecule
Molecule
Molecule Structure
Structure
Structure Ki
Ki (µM)
(µM) Ki (µM)
1 11 0.46
0.46 0.46
2 22 0.90
0.90 0.90
3 33 1.90
1.90 1.90
4 44 7.10
7.10 7.10
5 55 16.00
16.00 16.00
6 66 34.00
34.00 34.00
3.2. Pharmacophore Models Construction

Pharmacophore models were generated through the GALAHAD module, available on the
SYBYL-X 2.0 platform [39]. For the generation of pharmacophore models, the training set (n = 6)
was aligned based on its pharmacophore characteristics to produce the respective models using default
genetic algorithm parameters (Cross-over = 1.0 and Mutation = 0.4).
The alignment process was conducted in two steps: in the first, the selected compounds were
aligned mutually in a given spatial coordinate using an advanced genetic algorithm (GA) to allow
flexibility of the molecules. In the second step, the molecular alignment was based on the overlapping
of its pharmacophore characteristics in the conformations generated by the GA of the compounds in
the Cartesian space. Thus, to achieve a hyper molecular alignment, the population size and maximum
generation of GA simulation were set at 40 and 45, respectively. The other genetic operators were
maintained as the default configuration, as established in SYBYL-X 2.0.
The pharmacophore models were evaluated based on the statistical parameters of GALAHAD [39].
These parameters include Hbond, Mol_Qry, and energy. First, models with strain energy, two orders
of magnitudes above the others, were discarded. Next, the pharmacophore models were analyzed
based on the value of Hbond and Mol_QRY, that one with best result based on the sum of these two
parameters was selected for virtual screening.
Pharmacophore-Based Virtual Screening

The model selected with the best requirements was employed to screen the natural products
in the ZINC database through UNITY 3D implemented on SYBYL X 2.0 [39]. These molecules were
evaluated by QFIT value. Molecules with QFIT = 0 were discarded.
Thus, the compounds were staggered according to the QFIT value, and the molecules were
selected for the docking-based on the cut-off point (COP) obtained by the following Equation (1):
PdC = ±2σ (1)
3.3. Docking-Based Virtual Screening

The crystallographic structure was selected in the Protein Data Bank (PDB ID 4C6X). The 3D
structure was treated with the DockPrep implemented on Chimera 1.9 [37], thereby the hydrogen
atoms were added and the Gasteiger partial atomic charges (ff12SB) were attributed to the protein
residues [40]. The PDB2PQR program [41] was employed to calculate the protonation state.
Docking was carried out in DOCK 6.5 [25]. Through molecular surface to access receiver solvent
and the negative image of the orthosteric sites were generated by the accessory program DMS [42],
SPHGEN and SPHERE_SELECTOR programs respectively [43]. Properties were calculated by the
GRID program in its standard configuration [44,45]. The following scoring functions were used:
GridScore and GridScore + Hawkings GB/SA.
The evaluation step of the scoring functions contemplated some different methodologies. First,
one was the calculation of the mean square deviation (RMSD) obtained by the two scoring functions in
relation to the repositioning of the crystallographic ligand. Then, the Key Interaction Score System
(KISS) score [46] was calculated, in which the hydrogen bonds are evaluated after the redocking
process. KISS scores were obtained using Equation (2) below [46]:
KISS score = Ir /In (2)
where, Ir represents the amount of hydrogen bonds carried out between the best posture of the coupled
ligand and In is the total of hydrogen bonds present between the crystallographic structure of the
ligand and the target.
The scoring function with RMSD ≤ 2.0Å were evaluated by ROC curve performance and AUC
value based on protocol established by our research group [47]. In order to analyze the recovery
rate of true positive inhibitors, it was necessary to generate false-positive molecules (decoys) with
DUD-E server (http://dude.docking.org) [48], using same training set employed on pharmacophore
modelling (Table 1). The data were plotted for Receiver Operating Characteristic (ROC) construction
and later Area Under the Curve (AUC) analysis. The enrichment factor was also used to complement
the evaluation of the scoring functions.
From the selection of the most sensitive scoring function in the recovery of the active molecules,
the molecules that were selected according to the QFIT scores in the alignment step in the pharmacophore
model were submitted to docking. After docking, the molecules were arranged in order of the affinity
estimated against KasA.
In order to correlate the results obtained by the docking calculations with the results of the filtering
by pharmacophore modeling, we also adopted the number-by-number classification strategy with
the application of variable recategorization, which combines the results of the strategies used [49].
The scores obtained for the QFIT value and energy values were summed up, resulting in an overall
score for each molecule. The analysis of intermolecular interactions and 2D intermolecular plots were
built up on the PoseView web 1.97.0 server [49,50].
3.4. Molecular Dynamics (MD) Simulations

The atomic charges of the ligands were obtained by performing the Restrained Electrostatic
Potential (RESP) using HF/6-31G * [51]. Then the parameters of the ligands were generated with the
Antechamber module being described by General Amber Force Field (GAFF) [52], whereas the force
field ff12SB [53] was applied to treat the proteins. To neutralize the overall partial load of the systems
sodium ions were added, soon after the complexes were solvated in a truncated octahedral water box
with periodontal conditions and water molecules described by the TIP3P model [54].
For the energy minimization calculations of the system the sander.MPI, and for the heating steps
and molecular dynamics the pmemd.CUDA of the Amber 16 package were used [55,56].
The energy of the systems was mimicked in five stages where in each one were executed
3000 cycles using the steepest descent method and 5000 cycles using the conjugate gradient algorithm.
In the first stage the hydrogen of the water molecules were minimized, and in the second stage the
water molecules and ions, soon after the hydrogen atoms of the protein and finally, all the solvent
and solute had their energy minimized. The systems were heated gradually to 300 K for 650 ps and
was performed using NVT ensemble. The collision frequency used was 3.0 ps−1 and the Langevin
thermostat was used for temperature control [57]. After warming up, 2 ns of MD simulations using
NPT ensemble were run to balance the system. Lastly, for each system 100 ns of MD simulations were
generated. The Particle Mesh Ewald method was used to calculate the electrostatic interactions [58]
and the bonds involving hydrogen atoms were restricted with the SHAKE algorithm [59].
Binding Free Energy Calculations

To estimate the binding free energy of the KasA protein with the ligands ZINC35465970 and
ZINC31170017 we used the Molecular Mechanics/Generalized Born Surface Area (MM/GBSA)
method [60,61]. For our calculations, we used 500 snapshots of the last 5 ns of MD simulation.
The free energy was estimated according to Equation (3):
∆Gbind = ∆EMM + ∆Gsolv − T∆S (3)
∆Gbind is the affinity energy resulting from the sum of the total energy in the gas phase (∆EMM ),
free energy of solvation (∆Gsolv ) and entropy (T∆S).
∆EMM is the sum of ∆Einternal (connections, angles and dihedra), ∆Eelectrostatic (electrostatic
contributions) and ∆EvdW (Van der Waals contributions), according to Equation (4):
∆EMM = ∆Einternal + ∆Eelectrostatic + ∆Evdw (4)

∆Gsolv can be obtained by solving Equation (5):
∆Gsolv = ∆GGB + ∆GSASA (5)
When the polar contributions (∆GGB ) are calculated using either the GB model or the non-polar
contributions (∆GSASA ), they were determined from the calculation of the solvent accessible surface
area (SASA).
4. Conclusions
Ligand-based and target-based techniques have been useful in identifying potential inhibitors.
In this sense, this work applied these two approaches through a pharmacophore model and molecular
docking studies. The pharmacophore modeling strategy resulted in the selection of a model with
acceptable statistical parameters. The molecules filtered by the model show the physicochemical
properties of inhibitors known for the MtbKasA. These molecules were evaluated through molecular
docking. Through these results, a consensual evaluation was made that provided the prioritization
of molecules. The two best molecules scored were analyzed for the interaction mode with the target,
which presented π-π interactions, hydrogen bonds and hydrophobic interactions. We also performed
MD simulations by which we evaluated the structural stability of the systems and observed that
certain regions of the protein showed differences in their flexibility when interacting with the inhibitors
along the simulation trajectory. We also verified by means of free energy calculations with the
MM/GBSA method that the inhibitor ZINC35465970 was that presented better ability to inhibit the
target KasA (∆Gbind = −30.90 kcal/mol) and that the interactions of the van der Waals type were the
main responsible for the formation of the systems.
This work allowed the identification of possible inhibitors of MtKasA, which is an essential phase
in the design of new drugs. It is necessary to apply in vitro and in vivo assays to confirm the biological
potential of these molecules.
Author Contributions: V.d.S.P., J.S.C.A., J.N.C., M.F.D.A.N., F.H.A.L. and M.C.S.J. conceived and designed the
experiments and analyzed the data; V.d.S.P. performed the experiments; V.d.S.P., J.S.C.A., G.V.d.C., J.M.C., J.N.C.,
M.F.D.A.N., R.C.S., C.B.R.S., F.H.A.L. and M.C.S.J. wrote—review & editing of the paper.
Funding: This research received no external funding.
Acknowledgments: The authors acknowledge the Graduate Program in Biotechnology/UEFS for academic support.
Conflicts of Interest: No conflicts of interest to this work.
References
1. Mello, F.C.Q.; Silva, D.R.; Dalcolmo, M.P. Tuberculosis: Where are we? J. Bras. Pneumol. 2018, 44, 82.
[CrossRef]
2. World Health Organization. Global Tuberculosis Report; WHO: Geneva, Switzerland, 2017.
3. Santos, D.S.; Vasconselos, I.B.; Meyer, E.; Sales, F.A.M.; Moreira, I.S.; Basso, L.A. The mode of inhibition of
Mycobacterium tuberculosis wild-type and isoniazid-resistant 2-trans-enoyl-ACP(CoA) reductase enzymes
by an inorganic complex. Anti-Infect. Agents Med. Chem. 2008, 7, 50–62. [CrossRef]
4. Moreira, T.R.; Zandonade, E.; Maciel, E.L.N. Risco de infecção tuberculosa em agentes comunitários de
saúde. Rev. Saúde Pública 2010, 44, 332–338. [CrossRef]
5. Antonova, A.V.; Gryadunov, V.A.; Zimenkov, D.V. Molecular mechanisms of drug tolerance in
Mycobacterium tuberculosis. Mol. Biol. (Mosk.) 2018, 52, 435–450. [CrossRef]
6. Ananthan, S.; Faaleolea, E.R.; Goldman, R.C.; Hobrath, J.V.; Kwong, C.D.; Laughon, B.E.; Maddry, J.A.;
Mehta, A.; Rasmussen, L.; Reynolds, R.C.; et al. High-throughput screening for inhibitors of Mycobacterium
tuberculosis H37Rv. Tuberculosis 2009, 89, 334–353. [CrossRef]
7. Ikram, A.I.; Satti, M.L.; Lalani, F.K.; Zaman, G.; Gardezi, A.H.; Ahmed, P. Evaluation of nitrate reductase assay
for early detection of multi and extensively drug resistance tuberculosis in our setup. J. Coll. Physicians Surg. Pak.
2018, 28, 22–25. [CrossRef]
8. Miranda, T.F.N.; Santos Junior, M.C. Identificação de potenciais inibidores da beta-cetoacil sintase do
Mycobacterium tuberculosis por triagem virtual. Rev. Virtual. Quim. 2016, 8, 1740–1756. [CrossRef]
9. Alves, V.M.; Braga, R.C.; Muratov, E.N.; Andrade, C.H. Quimioinformática: Uma introdução. Quím. Nova
2018, 41, 202–212. [CrossRef]
10. Kumar, V.; Krishna, S.A.; Siddiqi, M.I. Virtual screening strategies: Recent advances in the identification and
design of anti-cancer agents. Methods 2014, 71, 64–70. [CrossRef]
11. Silva, R.C.; Poiani, J.G.C.; Ramos, R.S.; Costa, J.S.; Silva, C.H.T.P.; Brasil, D.S.B.; Santos, C.B.R. Ligand- and
structure-based virtual screening from 16-(N,N-diisobutylaminomethyl)-6α-hydroxyivouacapan-7β,17β-lactone
compound with potential anti-prostate cancer activity. J. Serb. Chem. Soc. 2018, 83, 1–23. [CrossRef]
12. Bhatt, A.; Molle, V.; Besra, G.S.; Jacobs, W.R., Jr.; Kremer, L. The Mycobacterium tuberculosis FAS-II
condensing enzymes: Their role in mycolic acid biosynthesis, acid-fastness, pathogenesis and in future drug
development. Mol. Microbiol. 2007, 64, 1442–1454. [CrossRef]
13. Zhang, H.J.; Li, Z.L.; Zhu, H.L. Advances in the Research of β-Ketoacyl-ACP Synthase III (FabH) Inhibitors.
Curr. Med. Chem. 2012, 19, 1225–1237. [CrossRef]
14. Brandão, H.N.; David, J.P.; Couto, R.D.; Nascimento, J.A.P.; David, J.M. Química e farmacologia de
quimioterápicos antineoplásicos derivados de plantas. Quím. Nova 2010, 33, 1359–1369. [CrossRef]
15. Newman, D.J.; Cragg, G.M. Natural products as sources of new drugs over the 30 years from 1981 to 2010.
J. Nat. Prod. 2012, 75, 311–335. [CrossRef]
16. Cordier, C.; Morton, D.; Murrison, S.; Nelson, A.; O’leary-steele, C. Natural products as an inspiration in the
diversity-oriented synthesis of bioactive compound libraries. Nat. Prod. Rep. 2008, 25, 719–737. [CrossRef]
17. Braga, R.C.; Andrade, C.H. Assessing the Performance of 3D Pharmacophore Models in Virtual Screening:
How Good are They? Curr. Top. Med. Chem. 2013, 13, 1127–1138. [CrossRef]
18. Ferreira, R.S.; Oliva, G.; Andricopulo, A.D. Integração das técnicas de triagem virtual e triagem biológica
automatizada em alta escala: Oportunidades e desafios em P&D de fármacos. Quím. Nova 2011, 34, 1770–1778.
[CrossRef]
19. Caballero, J. 3D-QSAR (CoMFA and CoMSIA) and pharmacophore (GALAHAD) studies on the differential
inhibition of aldose reductase by flavonoid compounds. J. Mol. Graph. Model. 2010, 29, 363–371. [CrossRef]
20. Dorfman, R.J.; Smith, K.M.; Masek, B.B.; Clark, R.D. A knowledge-based approach to generating diverse but
energetically representative ensembles of ligand conformers. J. Comput. Aided Mol. Des. 2007, 22, 681–691.
[CrossRef]
21. Eyunni, S.K.; Gangapuran, M.; Redda, K.K. In-vitro antiproliferative activity of new tetrahydroisoquinolines
(THIQs) on ishikawa cells and their 3D pharmacophore models. Lett. Drug Des. Discov. 2014, 11, 428–436.
[CrossRef]
22. Rella, M.; Rushworth, C.A.; Guy, J.L.; Turner, A.J.; Langer, T.; Jackson, R.M. Structure-based pharmacophore
design and virtual screening for novel angiotensin converting enzyme 2 inhibitors. J. Chem. Inf. Model. 2006,
46, 708–716. [CrossRef]
23. Chen, Z.; Myint, Z.; Xie, Q. New QSAR prediction models derived from GPCR CB2-antagonistic triaryl
bis-sulfone analogues by a combined molecular morphological and pharmacophoric approach. SAR QSAR
Environ. Res. 2011, 22, 525–544. [CrossRef]
24. Trott, O.; Olson, A.J. Software news and update AutoDock Vina: Improving the speed and accuracy of
docking with a new scoring function, efficient optimization, and multithreading. J. Comp. Chem. 2010, 31,
455–461. [CrossRef]
25. Lang, P.T.; Kuntz, I.; Allen, W.J.; Balius, T.; Brozell, S.; Jiang, L.; Mcgee, T.D., Jr.; Moustakas, D.; Mukherjee, D.;
Zhou, Y.; et al. DOCK 6.0 Users Manual. The Official UCSF DOCK Web-Site. Fev 2015. Available online:
http://dock.compbio.ucsf.edu/DOCK_6/dock6_manual.htm (accessed on 5 February 2015).
26. Wong, C.F. Flexible ligand flexible protein docking in protein kinases systems. Biochim. Biophym. Acta 2008,
1784, 244–251. [CrossRef]
27. Nicholls, A. What do we know and when do we know it? J. Comput. Aided Mol. Des. 2008, 22, 239–255.
[CrossRef]
28. Fawcett, T. An introduction to ROC analysis. Pattern Recogn. Lett. 2006, 27, 861–874. [CrossRef]
29. Matsubara, E.T. Relações entre Ranking, Análise ROC e Calibração em Aprendizado de Máquina. Ph.D. Thesis,
USP, São Carlos, Brazil, 2008.
30. Triballeau, N.; Acher, F.; Brabet, I.; Pin, J.P.; Bertrand, H.O. Virtual screening workflow development guided
by the “receiver operating characteristic” curve approach. Application to high-throughput docking on
metabotropic glutamate receptor subtype 4. J. Med. Chem. 2008, 48, 2534–2547. [CrossRef]
31. Jain, A.N.; Nicholls, A. Recommendations for evaluation of computational methods. J. Comput. Aided Mol. Des.
2008, 22, 133–139. [CrossRef]
32. Irwin, J.J.; Sterling, T.; Mysinger, M.M.; Bolstad, E.S.; Coleman, R.G. ZINC: A Free Tool to Discover Chemistry
for Biology. J. Chem. Inf. Model. 2012, 52, 1757–1768. [CrossRef]
33. Kim, S.; Thiessen, P.A.; Bolton, E.E.; Chen, J.; Fu, G.; Gindulyte, A.; Han, L.; He, J.; He, S.; Shoemaker, B.A.; et al.
PubChem Substance and Compound databases. Nucleic Acids Res. 2016, 4, D1202–D1213. [CrossRef]
34. Pence, H.E.; Willians, A. ChemSpider: An online chemical information resource. J. Chem. Educ. 2010, 87,
1123–1124. [CrossRef]
35. Bissantz, C.; Kuhn, B.; Stahl, M. A Medicinal Chemist’s Guide to Molecular Interactions. J. Med. Chem. Persp.
2010, 53, 5061–5084. [CrossRef]
36. Schiebel, J.; Kapilashrami, K.; Fekete, A.; Bommineni, G.R.; Schaefe, R.; Mueller, M.J.; Tonge, P.J. Structural
Basis for the Recognition of Mycolic Acid Precursors by KasA, a Condensing Enzyme and Drug Target from
Mycobacterium Tuberculosis. J. Biol. Chem. 2013, 288, 34190–34204. [CrossRef]
37. Chemaxon. Marvin Sketch, version 15.4.20; ChemAxon: Budapest, Hungary, 2015.
38. Pettersen, E.F.; Goddard, T.D.; Huang, C.C.; Couch, G.S.; Greenblatt, D.M.; Meng, E.C.; Ferrin, T.E. UCSF
Chimera—A visualization system for exploratory research and analysis. J. Comput. Chem. 2004, 25, 1605–1612.
[CrossRef]
39. Tripos Inc. SYBYL-X, Discovery Software for Computational Chemistry and Molecular Modelling, Princeton, NJ,
USA, 2012.
40. Jakalian, A.; Jack, D.B.; Bayly, C.I. Fast, efficient generation of high-quality atomic charges. AM1-BCC model:
II. Parameterization and validation. J. Comput. Chem. 2002, 23, 1623–1641. [CrossRef]
41. Dolinsky, T.J.; Nielsen, J.E.; McCammon, J.A.; Baker, N.A. PDB2PQR: An automated pipeline for the
setup, execution, and analysis of Poisson-Boltzmann electrostatics calculations. Nucleic Acids Res. 2004, 32,
W665–W667. [CrossRef]
42. Ferrin, T.E.; Huang, C.C.; Jarvis, L.E.; Langridge, R. The Midas display system. J. Mol. Graf. 1988, 6, 13–27.
[CrossRef]
43. Kuntz, I.D.; Blaney, J.M.; Oatley, S.J.; Langridge, R.; Ferrin, T.E. A geometric approach to macromolecule–ligand
interactions. J. Mol. Biol. 1982, 161, 269–288. [CrossRef]
44. Shoichet, B.K.; Kuntz, I.D.; Bodian, D.L. Molecular docking using shape descriptors. J. Comput. Chem. 1992,
13, 380–397. [CrossRef]
45. Meng, E.C.; Shoichet, B.K.; Kuntz, I.D. Automated docking with grid-based energy evaluation. J. Comput. Chem.
1992, 13, 505–524. [CrossRef]
46. Udatha, D.B.R.K.; Sugaya, N.; Olsson, L.; Panagiotou, G. How well do the substrates KISS the enzyme?
Molecular docking program selection for feruloyl esterases. Sci. Rep. 2012, 2, 323. [CrossRef]
47. Cruz, J.V.; Neto, M.F.A.; Silva, L.B.; Ramos, R.d.S.; Costa, J.d.S.; Brasil, D.S.B.; Lobato, C.C.; da Costa, G.V.;
Bittencourt, J.D.A.H.M.; da Silva, C.H.T.P.; et al. Identification of novel protein kinase receptor type 2
inhibitors using pharmacophore and structure-based virtual screening. Molecules 2018, 23, 453. [CrossRef]
48. Mysinger, M.M.; Carchia, M.; Irwin, J.J.; Shoichet, B.K. Directory of Useful Decoys, Enhanced (DUD-E):
Better Ligands and Decoys for Better Benchmarking. J. Med. Chem. 2012, 55, 6582–6594. [CrossRef]
49. Feher, M. Consensus scoring for protein–ligand interactions. Drug Discov. Today 2006, 11, 421–428. [CrossRef]
50. Stierand, K.; Maaß, P.C.; Rarey, M. Molecular complexes at a glance: Automated generation of
two-dimensional complex diagrams. Bioinformatics 2006, 22, 1710–1716. [CrossRef]
51. Cornell, W.D.; Cieplak, P.; Bayly, C.I.; Kollman, P.A. Application of RESP charges to calculate conformational
energies, hydrogen bond energies, and free energies of solvation. J. Am. Chem. Soc. 1993, 115, 9620–9631.
[CrossRef]
52. Wang, J.; Wolf, R.M.; Caldwell, J.W.; Kollman, P.A.; Case, D.A. Development and testing of a general amber
force field. J. Comput. Chem. 2004, 25, 1157–1174. [CrossRef]
53. Ponder, J.W.; Case, D.A. Force fields for protein simulations. Adv. Protein Chem. 2003, 66, 27–85.
54. Jorgensen, W.L.; Chandrasekhar, J.; Madura, J.D.; Impey, R.W.; Klein, M.L. Comparison of simple potential
functions for simulating liquid water. J. Chem. Phys. 1983, 79, 926–935. [CrossRef]
55. Salomon-Ferrer, R.; Case, D.A.; Walker, R.C. An overview of the Amber biomolecular simulation package.
Wiley Interdiscip. Rev. Comput. Mol. Sci. 2013, 3, 198–210. [CrossRef]
56. Cruz, J.N.; Costa, J.F.S.; Khayat, A.S.; Kuca, K.; Barros, C.A.L.; Neto, A.M.J.C. Molecular dynamics simulation
and binding free energy studies of novel leads belonging to the benzofuran class inhibitors of Mycobacterium
tuberculosis Polyketide Synthase 13. J. Biomol. Struct. Dyn. 2018, 1–12. [CrossRef]
57. Izaguirre, J.A.; Catarello, D.P.; Wozniak, J.M.; Skeel, R.D. Langevin stabilization of molecular dynamics.
J. Chem. Phys. 2001, 114, 2090–2098. [CrossRef]
58. Darden, T.; York, D.; Pedersen, L. Particle mesh Ewald: An N·log(N) method for Ewald sums in large
systems. J. Chem. Phys. 1993, 98, 10089–10092. [CrossRef]
59. Ryckaert, J.P.; Ryckaert, J.P.; Ciccotti, G.; Berendsen, H.J.C.C. Numerical integration of the cartesian equations
of motion of a system with constraints: Molecular dynamics of n-alkanes. J. Comput. Phys. 1977, 23, 327–341.
[CrossRef]
60. Chen, F.; Liu, H.; Sun, H.; Pan, P.; Li, Y.; Li, D.; Hou, T. Assessing the performance of the MM/PBSA and
MM/GBSA methods. 6. Capability to predict protein-protein binding free energies and re-rank binding
poses generated by protein-protein docking. Phys. Chem. Chem. Phys. 2016, 18, 22129–22139. [CrossRef]
61. de Oliveira, M.S.; da Cruz, J.N.; Gomes Silva, S.; da Costa, W.A.; de Sousa, S.H.B.; Bezerra, F.W.F.; Teixeira, E.;
da Silva, N.J.N.; de Aguiar Andrade, E.H.; de Jesus Chaves Neto, A.M.; et al. Phytochemical profile,
antioxidant activity, inhibition of acetylcholinesterase and interaction mechanism of the major components
of the Piper divaricatum essential oil obtained by supercritical CO2 . J. Supercrit. Fluids 2019, 145, 74–84.
[CrossRef]
© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).
pharmaceuticals
Brief Report
Brief Report
Baricitinib: A
Baricitinib: A 2018
2018 Novel FDA-Approved Small
Novel FDA-Approved Small
Molecule Inhibiting Janus Kinases
Annie Mayence
Annie Mayence 11 and
and Jean Jacques Vanden
Jean Jacques Vanden Eynde
Eynde 2,2,**
11 Haute
Haute Ecole Provincialedede
Ecole Provinciale Hainaut-Condorcet,
Hainaut-Condorcet, 73307330 Saint-Ghislain,
Saint-Ghislain, Belgium;Belgium;
annie.mayence@alumini.umons.ac.be
annie.mayence@alumini.umons.ac.be
22
Formerly
Formerly head ofthe
head of theDepartment
Department of Organic
of Organic Chemistry
Chemistry (FS), University
(FS), University of Mons-UMONS,
of Mons-UMONS, 7000
7000 Mons,
Belgium
Mons, Belgium
Correspondence: jean-jacques.vandeneynde@ex.umons.ac.be;
** Correspondence: jean-jacques.vandeneynde@ex.umons.ac.be; Tel.: +32-2355-8161
Tel.: +32-2355-8161

Received:1818February
Received: February 2019;
2019; Accepted:
Accepted: 10 March
10 March 2019; 2019; Published:
Published: 122019
12 March March 2019
Abstract: In
Abstract: In 2018,
2018, Baricitinib
Baricitinib was
was approved
approved by by the
the Food
Food and
and Drig
Drig Administration
Administration (FDA)
(FDA) for
for the
the
treatment of
treatment ofrheumatoid
rheumatoidarthritis.
arthritis.Baricitinib
Baricitinibexerts
exerts
itsits action
action by by targeting
targeting Janus
Janus kinases
kinases (JAK).
(JAK). In
In this
this study, we describe the necessary steps for preparing the drug using two alternative
study, we describe the necessary steps for preparing the drug using two alternative routes. routes.
Keywords: approved
Keywords: approveddrugs; baricitinib;
drugs; FDA; Janus
baricitinib; FDA; kinases;
Janus protein
kinases;kinase inhibitor;
protein kinaserheumatoid
inhibitor;
arthritis
rheumatoid arthritis
1. Introduction
1. Introduction
In
In May
May2001,
2001,Food
Foodand Drig
and Administration
Drig (FDA)
Administration approved
(FDA) the first
approved thekinase
first inhibitor [1], imatinib
kinase inhibitor [1],
(1, Figure(1,
imatinib Gleevec1;® Gleevec
1; Figure from Novartis, Basel, Switzerland),
® from Novartis, for the treatment
Basel, Switzerland), of chronicofmyeloid
for the treatment chronic
leukemia. [2] Since[2]
myeloid leukemia. that year,
Since more
that than
year, 45 than
more kinase
45 inhibitors have been
kinase inhibitors havemarketed, most of
been marketed, them
most of
in
themoncology.
in oncology.
N N
H N
N N
H N
1 N
Structure of
Figure1.1.Structure
Figure ofimatinib
imatinib1.1.
In
In 2018,
2018, 59
59 novel
novel drugs
drugs have
have been
been approved
approved by by FDA.
FDA. Among
Among them,
them, 4040 can
can be
be considered
considered asas
small molecules, 16 are derived from amino acids, and 3 from nucleic acids. Interestingly,
small molecules, 16 are derived from amino acids, and 3 from nucleic acids. Interestingly, almost 25% almost
25% of approved
of the the approved small
small molecules
molecules acted
acted as as kinase
kinase inhibitorsand
inhibitors andpredominantly
predominantlyas as protein
protein kinase
kinase
inhibitors.
inhibitors. The purpose of this review is to provide a series of specific information on one
The purpose of this review is to provide a series of specific information on one of
of those
those
inhibitors:
inhibitors: baricitinib (LY3009104, formerly developed by Incyte Co as INCB028050 and subject of
baricitinib (LY3009104, formerly developed by Incyte Co as INCB028050 and subject of aa
license agreement with Eli Lilly and Co in 2009).
license agreement with Eli Lilly and Co in 2009).
2. Baricitinib
2. Baricitinib
2.1. Names and Structure
2.1. Names and Structure
Baricitinib (2, Figure 2) is the active ingredient of Olumiant® , commercialized by Eli Lilly and Co.
Its IUPAC name is: 2-[1-(ethanesulfonyl)-3-(4-{7H-pyrrolo[2,3-d]pyrimidin-4-yl}-1H-pyrazol-1-yl)aze
tidin-3-yl] acetonitrile, CAS 1187594-09-7.
Pharmaceuticals 2019, 12, x; doi: www.mdpi.com/journal/pharmaceuticals

interactions (hydrogen bonds) with amino acid residues in the hinge region of the ATP binding site
[5]. Studies started during a screening of approximately 400,000 compounds from a Pfizer library in
order to discover an inhibitor of JAK3. This enabled the identification of 9-(7H-Pyrrolo[2,3-
d]pyrimidin-4-yl)-2,3,4,4a,9,9a-hexahydro-1H-carbazole (5) as a lead, which was reported by
Flanagan et al.
Pharmaceuticals [10].
2019, Improvements of the properties of 5, among which its metabolic stability [10]2 led
12, 37 of 7
to the identification and development of commercialized drugs 2-4.
Figure 2. 2.
Figure Structure
Structureof
ofJanus kinases(JAK)
Janus kinases (JAK)inhibitors
inhibitors
2–5.2–5.
2.2. Uses
Table 1. IC50 values for the inhibition of JAK1, JAK2, JAK3, and tyrosine kinases 2 (TYK2)
by 2, 3,
After and 4, following
a rejection in Aprilthe methodology
2017, ofmg
baricitinib (2 Clarck et al.has
tablets) [5].been approved on May 31, 2018 for
treatment of rheumatoid arthritis. [3] Noticeably, it had been approved, for the same purpose, in the
IC50 Values, Enzyme Assay (nM)
Compound
European Union (EU) in February 2017. [4]JAK1 JAK2 JAK3 TYK2
2 5.9 5.7 >400 53
2.3. Targets
3 3.3 2.8 323 19
4
Janus kinases (JAK) are tyrosine 3.2 (TYK)
kinases 4.1 that play
1.6 a crucial
34 role in cell signaling [5,6].
They can be divided into four families: JAK1, JAK2, JAK3, and TYK2, and constitute interesting
2.4. In Vitro Studies,
therapeutic Rodent
targets [5,7]. Models,
The and inhibitor
first JAK Clinical Trials
approved by FDA (November 2011) was ruxolitinib
(3) [8]. Through its selective inhibition of JAK1
Several proinflammatory cytokines are involved and JAK2in(Table 1), it is clinically
the pathogenesis used for intermediate
of rheumatoid arthritis.
or high-risk
Mention canmyelofibrosis.
be made of interleukin (IL) 6, IL-15, IL-17, IL-23, interferon-α/β, interferon-γ, and
granulocyte-macrophage colony as stimulating factors. [10] As elegantly depicted by Furumoto and
Table 1. IC50 values for the inhibition of JAK1, JAK2, JAK3, and tyrosine kinases 2 (TYK2) by 2, 3, and
Gadina [11], such activity is critically linked to JAK signaling pathways and the signal transducer
4, following the methodology of Clarck et al. [5].
IC50 Values, Enzyme Assay (nM)

Compound
JAK1 JAK2 JAK3 TYK2
2 5.9 5.7 >400 53
3 3.3 2.8 323 19
4 3.2 4.1 1.6 34
Baricitinib (2) is also a selective and reversible inhibitor of JAK1 and JAK2 with less affinity
for JAK3 and TYK2 (Table 1). Interestingly, tofacitinib (4, Figure 2), another FDA-approved kinase
(November 2012) used to treat rheumatoid arthritis [9], is even more selective (Table 1).
Selectivity of inhibitors within the Janus kinases has been tentatively correlated to specific
interactions (hydrogen bonds) with amino acid residues in the hinge region of the ATP binding
site [5]. Studies started during a screening of approximately 400,000 compounds from a
Pfizer library in order to discover an inhibitor of JAK3. This enabled the identification of
9-(7H-Pyrrolo[2,3-d]pyrimidin-4-yl)-2,3,4,4a,9,9a-hexahydro-1H-carbazole (5) as a lead, which was
reported by Flanagan et al. [10]. Improvements of the properties of 5, among which its metabolic
stability [10] led to the identification and development of commercialized drugs 2-4.
2.4. In Vitro Studies, Rodent Models, and Clinical Trials

Several proinflammatory cytokines are involved in the pathogenesis of rheumatoid arthritis.
Mention can be made of interleukin (IL) 6, IL-15, IL-17, IL-23, interferon-α/β, interferon-γ, and
granulocyte-macrophage colony as stimulating factors. [10] As elegantly depicted by Furumoto and
Gadina [11], such activity is critically linked to JAK signaling pathways and the signal transducer and
activator of transcription (STAT) signaling pathways. Therefore, targeting those pathways represented
and still represents a challenging field of research [11,12].
In an in-depth preclinical study performed by Incyte Co, Fridman et al. [13], they reported that
the action of baricitinib in peripheral blood mononuclear cells (PBMCs) could prevent the production
of pathogenic and proinflammatory cytokines. That production was not altered by structural analogs
that did not inhibit JAK1 and JAK 2. Baricitinib has been administered orally (1, 3, and 10 mg/kg/day)
or by constant infusion in several rodent models (rats and mice). Clinical, histologic, radiographic,
and hematologic data demonstrated the efficacy and safety of the drug, thus justifying clinical trials.
Following the NIH [14], baricitinib has been the subject of 30 completed clinical trials, namely
17 phase 1 studies, 7 phase 2 studies, and 6 phase 3 studies; 20 other trials are ongoing or scheduled.
Historically, the first trial was a phase 2 study launched on May 15, 2009. It was entitled “INCB028050
Compared to Background Therapy in Patients with Active Rheumatoid Arthritis (RA) with Inadequate
Response to Disease Modifying Anti-Rheumatic Drugs” (NCT00902486) and was summarized as
follows: “This was a randomized, double blind, placebo controlled, dose ranging, parallel group
study. Participants who had active rheumatoid arthritis (RA) who had inadequate response to any
disease modifying anti-rheumatic drug (DMARD) therapy including biologics were enrolled. Screening
evaluations were performed within approximately 28 days of randomization. The duration of the study
was [six] months with the primary endpoint assessed at [three] months. Eligible participants were
randomly assigned to one of three doses (4, 7, or 10 mg QD) of INCB028050 (Baricitinib) or placebo.”
The first phase 3 studies were initiated at the end of 2012 under the titles “A Study in Moderate to
Severe Rheumatoid Arthritis (RA-BEAM)” (NCT01710358; first posted October 19, 2012), “A Study in
Participants with Moderate to Severe Rheumatoid Arthritis (RA-BEGIN)” (NCT01711359; first posted
October 22, 2012), “A Moderate to Severe Rheumatoid Arthritis Study (RA-BEACON)” (NCT01721044;
first posted November 2, 2012), and “A Study in Moderate to Severe Rheumatoid Arthritis Participants
(RA-BUILD)” (NCT01721057; first posted November 2, 2012). All results confirmed the high efficiency
of baricitinib and underlined a limited incidence of side effects such as a decrease in hemoglobin and
an increase in LDL, HDL, creatinine, and creatine phosphokinase. Further details on clinical trials can
be found in references [14–19].
2.5. Syntheses
There are essentially two routes for the preparation of baricitinib 2. As depicted in Scheme 1,
they can be distinguished by introducing central pyrazole ring in the molecule. In the original
procedure [20,21], the pyrazole ring was linked to the pyrrolo[2,3-d]pyrimidine system (to afford 6)
and then coupled to the azetidine moiety 7 to give the intermediate 8. In an alternative route [22,23],
the bound between the pyrazole and the azetidine was formed (to yield 10) before reaction with the
fused system 9.
Thus (Scheme 2), 4-chloro-7H- pyrrolo[2,3-d]pyrimidine was protected on position 7 by reaction
with 2-(trimethylsilyl)ethoxymethyl chloride. The protected fused system was then coupled with
4-pyrazoleboronic acid pinacol ester 12 by a Suzuli-Miyaura reaction, giving 6. Parallelly, 7 was
obtained from 1-Boc-3-azetidinone 13 and diethyl cyanomethylphosphonate. Reaction between 6 and 7
in the presence of DBU afforded the ester 8. Subsequent hydrolysis, decarboxylation, sulfonation, and
finally deprotection of the pyrrolopyrimidine moiety yielded the targeted derivative 2. In a variant [21],
also used to prepare deuterated samples of 2 [24], the azetidine derivative 7 has been deprotected and
sulfonated before coupling with 6.
In a more recent patent [22], the sulfonated azetidine 14 (Scheme 3) was prepared from
azetidine-3-ol by a sequence including a sulfonation, an oxidation, and introduction of the
cyanomethylene moiety. Interestingly, there is no need to protect any position in that sequence.
Additionally, let us emphasize that the oxidation step could be performed both in batch or under
flow conditions. [22,25]. Then, 14 was reacted with 4-pyrazoleboronic acid pinacol ester 12 to yield 10.
The bound between the azetidinylpyrazole group and the pyrrolo[2,3-d]pyrimidine system was then
created through a Suzuki-Miyaura reaction involving 7-Boc-4-chloro-7H-pyrrolo[2,3-d]pyrimidine 9 or
even the unprotected 4-chloro-7H-pyrrolo[2,3-d]pyrimidine.
Pharmaceuticals
Pharmaceuticals 2019,
2019, 12,
12, xx 44 of
of 77
Scheme 1.
1.1.Key
Scheme
Scheme Keysteps
Key stepsin
steps intwo
in twoalternate
two routes
alternate routes
alternate yielding
routesyielding baricitinib
yieldingbaricitinib
baricitinib2. 2.
2.
Scheme 2.
2. Preparation
Scheme
Scheme 2. Preparationof
Preparation ofofbaricitinib
baricitinib222from
baricitinib from aa 4-pyrazolyl-7H-pyrrolo[2,3-d]pyrimidine.
from 4-pyrazolyl-7H-pyrrolo[2,3-d]pyrimidine.
4-pyrazolyl-7H-pyrrolo[2,3-d]pyrimidine.
moiety. Interestingly, there is no need to protect any position in that sequence. Additionally, let us
emphasize that the oxidation step could be performed both in batch or under flow conditions. [22,25].
Then, 14 was reacted with 4-pyrazoleboronic acid pinacol ester 12 to yield 10. The bound between
the azetidinylpyrazole group and the pyrrolo[2,3-d]pyrimidine system was then created through a
Suzuki-Miyaura
Pharmaceuticals reaction
2019, 12, 37 involving 7-Boc-4-chloro-7H-pyrrolo[2,3-d]pyrimidine 9 or even 5 the
of 7
unprotected 4-chloro-7H-pyrrolo[2,3-d]pyrimidine.
Scheme 3. Preparation
Scheme ofof
3. Preparation baricitinib
baricitinib22from
from aa 4-chloro-7H-pyrrolo[2,3-d]pyrimidine.
4-chloro-7H-pyrrolo[2,3-d]pyrimidine.
3. Perspectives
3. Perspectives
As expected,
As expected, success
success of
of the
the 7H-pyrrolo[2,3-b]pyrimidine scaffold in
7H-pyrrolo[2,3-b]pyrimidine scaffold in the
the development
development of of drugs
drugs
for treatment
for of various
treatment of various diseases
diseases and
and essentially
essentially rheumatoid
rheumatoid arthritis has initiated
arthritis has initiated many
many researches
researches
on structurally
on structurally related
related analogs. Among them,
analogs. Among them, the
the 1H-pyrrolo[2,3-b]pyridine skeleton emerged
1H-pyrrolo[2,3-b]pyridine skeleton emerged as as aa
moiety of particular interest, as indicated by the number of recent publications and patents in which
moiety of particular interest, as indicated by the number of recent publications and patents in which it
has
it been
has described
been described[26–31].
[26–31].
Author Contributions: Both authors contributed equally to this manuscript.
Conflicts of
Conflicts Interest: The
of Interest: The authors
authors declare
declare no
no conflict
conflict of
of interest.
interest.
References
1.
1. Wu, P.;
Wu, P.; Nielsen,
Nielsen, T.E.;
T.E.; Clausen,
Clausen, M.H.
M.H. FDA-Approved
FDA-Approved Small-molecules Kinase Inhibitors.
Small-molecules Kinase Trends Pharmacol.
Inhibitors. Trends Pharmacol.
Sci. 2015,
Sci. 36, 422–439.
2015, 36, 422–439. doi:10.1016/j.tips.2015.04.005.
[CrossRef]
2.
2. CenterWatch. 2001
CenterWatch. 2001 FDA
FDA Approved
Approved Drugs. Drugs. Available
Availableonline:
online:https://www.centerwatch.com/drug-
https://www.centerwatch.com/drug-
information/fda-approved-drugs/year/2001 (accessed
information/fda-approved-drugs/year/2001 (accessed on 5 March on 5 March 2019).
2019).
3.
3. Smolen, J.S.;
Smolen, J.S.; Genovese,
Genovese, M.C.;
M.C.; Takeuchi,
Takeuchi, T.;T.; Hyslop,
Hyslop, D.L.;
D.L.; Macias,
Macias, W.L.;
W.L.; Rooney,
Rooney, T.;
T.; Chen,
Chen, L.;
L.; Dickson,
Dickson, C.L.;
C.L.;
Riddle Camp,
Riddle Camp,J.;J.;Cardillo,
Cardillo,T.E.; et al.
T.E.; et Safety Profile
al. Safety of Baricitinib
Profile in Patients
of Baricitinib with Active
in Patients withRheumatoid Arthritis
Active Rheumatoid
with over 2with
Arthritis Years Median
over 2 Time Median J.Time
in Treatment.
Years Rheumatol. 2019, 46, 7–18.J. [CrossRef]
in Treatment. Rheumatol. 2019. 46, 7–18.
4. Drugbank. Baricitinb.
doi:10.3899/jrheum.171361.Available online: https://www.drugbank.ca/drugs/DB11817 (accessed on 5 March 2019).
5.
4. Clarck, J.D.;Baricitinb.
Drugbank. Flanagan, Available
M.E.; Telliez, J.-B. https://www.drugbank.ca/drugs/DB11817
online: Discovery and Development of Janus Kinase (JAK) Inhibitors
(accessed for
on 5 March
Inflammatory Diseases. J. Med. Chem. 2014, 57, 5023–5038. [CrossRef] [PubMed]
2019).
6. Villarino, A.V.; Kanno, Y.; O’Shea, J.J. Mechanisms and Consequences of Jak–STAT Signaling in the Immune
System. Nat. Immunol. 2017, 18, 374–384. [CrossRef]
7. Schwartz, D.M.; Kanno, Y.; Villarino, A.; Ward, M.; Gadina, M.; O’Shea, J.J. JAK inhibition as a therapeutic
strategy for immune and inflammatory diseases. Nat. Rev. Drug Discov. 2017, 16, 843–862. [CrossRef]
[PubMed]
8. Drugbank. Ruxolitinib. Available online: https://www.drugbank.ca/drugs/DB08877 (accessed on 5 March 2019).
9. Drugbank. Tofacitinib. Available online: https://www.drugbank.ca/drugs/DB08895 (accessed on 5 March 2019).
10. Flanagan, M.E.; Blumenkopf, T.A.; Brissette, W.H.; Brown, M.F.; Casavant, J.M.; Shang-Poa, C.; Doty, J.L.;
Elliott, E.A.; Fisher, M.B.; Hines, M.; et al. Discovery of CP-690,550: A Potent and Selective Janus Kinase
(JAK) Inhibitor for the Treatment of Autoimmune Diseases and Organ Transplant Rejection. J. Med. Chem.
2010, 53, 8468–8484. [CrossRef] [PubMed]
11. Furumoto, Y.; Gadina, M. The Arrival of JAK Inhibitors: Advancing the Treatment of Immune and
Hematologic Disorders. BioDrugs 2013, 27, 431–438. [CrossRef] [PubMed]
12. Schwartz, D.M.; Bonelli, M.; Gadina, M.; O’Shea, J.J. Type I/II Cytokines, JAKs, and New Strategies for
Treating Autoimmune Diseases. Nat. Rev. Rheumatol. 2016, 12, 25–36. [CrossRef] [PubMed]
13. Fridman, J.S.; Scherle, P.A.; Collins, R.; Burn, T.C.; Li, Y.; Li, J.; Covington, M.B.; Thomas, B.; Collier, P.;
Favata, M.F.; et al. Selective Inhibition of JAK1 and JAK2 is Efficacious in Rodent Models of Arthritis:
Preclinical Characterization of INCB028050. J. Immunol. 2010, 184, 5298–5307. [CrossRef]
14. NIH US National Library of Medicine. ClinicalTrials.gov. Baricitinib. Available online: https://www.
clinicaltrials.gov/ct2/results?term=baricitinib&age_v=&gndr=&type=&rslt=&Search=Apply (accessed on 5
March 2019).
15. Jiang, J.-J.J.; Wang, X.-Y.; Zhang, Y.; Jin, Y.; Lin, J. Advances in the Inhibitors of Janus Kinases. Med. Chem.
2014, 4, 540–548. [CrossRef]
16. Yamaoka, K. Janus Kinase Inhibitors for Rheumatoid Arthritis. Curr. Opin. Chem. Biol. 2016, 32, 29–33.
[CrossRef]
17. Tanaka, Y.; Emoto, K.; Cai, Z.; Aoki, T.; Schlichting, D.E.; Rooney, T.P.; Macias, W. Efficacy and Safety of
Baricitinib in Japanese Patients with Active Rheumatoid Arthritis Receiving Background Methotrexate
Therapy: A 12-week, Double-blind, Randomized Placebo-controlled Study. J. Rheumatol. 2016, 43, 3.
[CrossRef]
18. Keystone, E.C.; Taylor, P.C.; Drescher, E.; Schlichting, D.E.; Beattie, S.D.; Berclaz, P.-Y.; Lee, C.H.;
Fidelus-Gort, R.K.; Luchi, M.E.; Rooney, T.P.; et al. Safety and Efficacy of Baricitinib at 24 Weeks in Patients
with Rheumatoid Arthritis Who Have Had an Inadequate Response to Methotrexate. Ann. Rheum. Dis. 2015,
74, 333–340. [CrossRef] [PubMed]
19. Kuryia, B.; Cohen, M.D.; Keystone, E.C. Baricitinib in Rheumatoid Arthritis: Evidence-to-date and Clinical
Potential. Ther. Adv. Musculoskelet. Dis. 2017, 9, 37–44. [CrossRef]
20. Rodgers, J.; Shepard, S.; Maduskuie, T.; Wang, H.; Falahatpisheh, N.; Rafalski, M.; Arvanitis, A.; Storace, L.;
Jalluri, R.; Fridman, J.; et al. Heteroaryl Substituted Pyrrolo[2,3-b]pyridines and Pyrrolo[2,3-b]pyrimidines
as Janus Kinase Inhibitors. US20070135461A1, 14 June 2007.
21. Rodgers, J.D.; Shepard, S.; Li, Y.-L.; Zhou, J.; Liu, P.; Meloni, D.; Xia, M. Azetidine and Cyclobutane
Derivatives as JAK Inhibitors. U.S. Patent US20090233903A1, 17 September 2009.
22. Kobierski, M.E.; Kopach, M.E.; Martinelli, J.R.; Varie, D.L.; Wilson, T.M. Processes and Intermediates for the
Preparation of {1(Ethylsulfonyl)-3-[4-(7H-pyrrolo[2,3-d]pyrimidin-4-yl)-1H-pyrazol-1yl]azetidin-3-yl}acetonitrile.
WO 2016/ 205487, 22 December 2016.
23. Xu, J.; Cai, J.; Chen, J.; Zong, X.; Wu, X.; Ji, M.; Wang, P. An Efficient Synthesis of Baricitinib. J. Chem. Res.
2016, 40, 205–208. [CrossRef]
24. Tung, R.D. Deuterated Baricitinib. U.S. Patent US20180221374A1, 9 August 2018.
25. Hughes, D.L. Applications of Flow Chemistry in Drug Development: Highlights of Recent Patent Literature.
Org. Process Res. Dev. 2018, 22, 13–20. [CrossRef]
26. Rodgers, J.D.; Wang, H.; Combs, A.P.; Sparks, R.B. Pyrrolo[2,3-b]pyridin-4-yl-amines and Pyrrolo[2,3-b]
pyrimidin-5-yl-amines as Janus Kinase Inhibitors. U.S. Patent US9879010B2, 30 January 2018.
27. Hamaguchi, H.; Amano, Y.; Moritomo, A.; Shirakami, S.; Nakajima, Y.; Nakai, K.; Nomura, N.; Ito, M.;
Higashi, Y.; Inoue, T. Discovery and Structural Characterization of Peficitinib (ASP015K) as a Novel and
Potent JAK Inhibitor. Bioorg. Med. Chem. 2018, 26, 4971–4983. [CrossRef] [PubMed]
28. Crocetti, L.; Giovannoni, M.P.; Schepetkin, I.A.; Quinn, M.T.; Khlebnikov, A.I.; Cantini, N.; Guerrini, G.;
Iacovone, A.; Teodori, E.; Vergelli, C. 1H-pyrrolo[2,3-b]pyridine: A New Scaffold for Human Neutrophil
Elastase (HNE) Inhibitors. Bioorg. Med. Chem. 2018, 26, 5583–5585. [CrossRef] [PubMed]
29. Tang, Q.; Duan, Y.; Wang, L.; Wang, M.; Ouyang, Y.; Wang, C.; Mei, H.; Tang, S.; Xiong, Y.;
Zheng, P.; et al. Synthesis and Antiproliferative Activity of Pyrrolo[2,3-b]pyridine Derivatives Bearing
the 1,8-Naphthyridin-2-one Moiety. Eur. J. Med. Chem. 2018, 143, 266–275. [CrossRef] [PubMed]
30. Wang, W.; Xu, S.; Duan, Y.; Liu, X.; Li, X.; Wang, C.; Zhao, B.; Zheng, P.; Zhu, W. Synthesis and Bioevaluation
and Docking Study of 1H-pyrrolo[2,3-b]pyridine Derivatives Bearing Aromatic Hydrazone Moiety as c-Met
Inhibitors. Eur. J. Med. Chem. 2018, 145, 315–327. [CrossRef] [PubMed]
31. Günther, M.; Laux, J.; Laufer, S. Synthesis and Structure-activity-relationship of
3,4-Diaryl-1H-pyrrolo[2,3-b]pyridines as Irreversible Inhibitors of Mutant EGFR-L858R/T790M.
Eur. J. Pharm. Sci. 2019, 128, 91–96. [CrossRef] [PubMed]
pharmaceuticals
Review
Potential of the Other Genetic Information Coded by
the Viral RNA Genomes as Antiviral Target
Alfredo Berzal-Herranz * , Cristina Romero-López, Beatriz Berzal-Herranz and
Sara Ramos-Lorente
Instituto de Parasitología y Biomedicina López-Neyra, (IPBLN-CSIC); Av. del Conocimiento 17, PTS Granada,
Armilla, 18016 Granada, Spain; cristina_romero@ipb.csic.es (C.R.-L.); bbh@ipb.csic.es (B.B.-H.);
seramos@correo.ugr.es (S.R.-L.)
* Correspondence: aberzalh@ipb.csic.es; Tel.: +34-958181648

Received: 26 February 2019; Accepted: 10 March 2019; Published: 13 March 2019
Abstract: In addition to the protein coding information, viral RNA genomes code functional
information in structurally conserved units termed functional RNA domains. These RNA domains
play essential roles in the viral cycle (e.g., replication and translation). Understanding the molecular
mechanisms behind their function is essential to understanding the viral infective cycle. Further,
interfering with the function of the genomic RNA domains offers a potential means of developing
antiviral strategies. Aptamers are good candidates for targeting structural RNA domains. Besides its
potential as therapeutics, aptamers also provide an excellent tool for investigating the functionality
of RNA domains in viral genomes. This review briefly summarizes the work carried out in our
laboratory aimed at the structural and functional characterization of the hepatitis C virus (HCV)
genomic RNA domains. It also describes the efforts we carried out for the development of antiviral
aptamers targeting specific genomic domains of the HCV and the human immunodeficiency virus
type-1 (HIV-1).
Keywords: Viral RNA genome; Functional RNA domains; RNA–RNA interactions; Antivirals; RNA
aptamers; RNA structure/function; RNA tools
1. Introduction
The biological function of RNA has, for long time, been considered to be restricted to its role in
the transmission of genetic information from DNA to proteins. The discovery in the early 19800 s that
the RNA is capable of catalyzing chemical reactions [1,2] led to the reconsideration of this dogma
and aroused the interest in unraveling the molecular mechanisms of the catalytic activity of RNA
molecules and, later, in deciphering the unknown functional roles of RNA within cells. Currently, it is
widely accepted that RNA plays a central role in many biological processes in all living organisms.
An increasing number of functional RNA molecules, so-called “noncoding RNAs”, has been identified.
Although they do not encode proteins, they encode information that is also essential for life. In fact,
errors in RNA metabolism or the loss of function of RNA molecules are the cause of pathological
processes of sanitary importance. Therefore, determining the function of certain RNA molecules has
made it possible to clarify the molecular mechanisms of different biological processes. The genomes
of RNA viruses constitute a particular class of functional RNA molecules. Viral RNA genomes are
compact entities that carry all the information that the virus requires to complete the infectious cycle.
Besides acting as a replication and translation template, viral RNA genomes play several essential
functions for the completion of the viral cycle and their regulation. To achieve all these functions, they
have developed an information coding system that complements and overlaps the protein coding
one. This information coding system uses discrete RNA units that fold into their specific structures

for storing information. These structural units are, in fact, functional genomic RNA domains and are
highly conserved among the viral population. They can be grouped in complex folded RNA regions
that interact among them to achieve the proper functioning. Thus, the structure of the entire genome
has to be preserved for the efficient viral propagation. This structural conservation contrasts with
the great sequence variability of the viral RNA genomes, which helps the viral populations to adapt
to novel molecular and cellular contexts and to escape host defenses. It also contributes toward the
development of resistance to antiviral drugs. The genome variability is, therefore, limited by the
preservation of the genome structure. The precise equilibrium between the sequence variability and
structure preservation determines the success of the RNA viral populations.
Deciphering the mechanisms that underlie each of the functions performed by functional genomic
RNA domains and understanding the molecular strategies used by RNA genomes to contain all
the information necessary for each of them are scientific problems of great relevance and interest.
Elucidating these mechanisms would provide information of enormous importance to address the
control of infections caused by RNA viruses. The current information indicates that structural RNA
domains play out their different biological roles (e.g., in replication, translation, or encapsidation)
by directly recruiting viral and/or cellular factors and/or by forming high-order structures via the
establishment of long-range RNA–RNA interaction networks (for a review, see Reference [3]).
The essentiality of these genomic RNA domains for the virus makes them excellent candidates
as targets for the development of antiviral strategies. Drugs aimed at interfering with their function
either by blocking their interactions with viral or cellular factors or by impeding their proper folding
may seriously compromise the viral viability. Therefore, they should be considered promising antiviral
agents. Aptamers specifically recognize and efficiently bind to structural elements within a target
molecule, being among the most promising molecules for interfering with the function of structural
domains of viral RNA genomes.
Members of the family Flaviviridae bear a positive single-stranded RNA genome. They are
responsible of worldwide public health problems. The family comprises three genera: Flavivirus, that
includes important human pathogens like Dengue virus (DENV), West Nile virus (WNV), yellow
fever virus (YFV), or Zika virus (ZIKV), among many others; Pestivirus, represented by the virus
responsible of the Bovine viral diarrhea (BVDV) or the classic swine fever (CSFV); and Hepacivirus,
with the hepatitis C virus (HCV) as the most significant member. Their genome contains multiple
highly conserved structural RNA domains, which store essential information for the completion of the
viral cycle. This review focuses on the HCV genome, paying special attention to the work carried out
in our laboratory aimed at the structural and functional characterization of genomic RNA domains.
The second part of the review summarizes the main strategies we have developed seeking aptamers
targeted against specific genomic RNA domains of the HCV and HIV genomes.
2. HCV Genomic Functional RNA Domains

HCV is likely the member of the family Flaviviridae that has attracted major efforts to understand
the molecular mechanisms that govern its infective cycle. Its genome is a 9600-nt-long single-stranded
RNA molecule, which codes for a single open reading frame (ORF) flanked by highly conserved
untranslated regions (UTRs). Although the UTRs comprise numerous conserved structural/functional
RNA elements that play essential roles for the consecution of the viral cycle, these RNA elements
are also distributed throughout the coding region (Figure 1). The 50 UTR is a highly conserved and
complexly folded region that is mainly occupied by an internal ribosome entry site (IRES) that spans
around 30 nt within the viral capsid coding region [4,5]. The IRES directs the recruitment of the 40S
ribosomal subunit in the absence of a cap structure and initiates the viral protein synthesis [6,7]. It is
folded into four structural domains that comprise a set of highly conserved subdomains, each with
essential roles in the ribosome recruitment and translation initiation. In addition, 50 UTR structural
domains are essential for viral replication and infectivity [6–12]. The initiation of replication takes
places at the 30 UTR [8,10,11]. It is a highly conserved 200–250-nt-long region, which contains several

functional RNA elements grouped into highly conserved domains required for efficient viral replication.
0
The 3 UTR also plays an important role in the viral translation, regulating the IRES activity [13,14].
domains required for efficient viral replication. The 3′ UTR also plays an important role in the viral
The use ofregulating
translation, complementary the IRESexperimental
activity [13,14].approaches (bioinformatics tools, secondary structure
mapping,The and genetic strategies) has
use of complementary experimental providedapproaches
a good overview of the HCV
(bioinformatics tools, genome
secondaryfolding
structure[15–23].
It hasmapping,
allowedand the genetic
identification
strategies) has provided a good overview of the HCV genome folding [15–23]. viral
of up to 20 highly conserved structural elements among different
isolates throughout
It has allowed the theidentification
ORF of the HCV of up genome (Figure
to 20 highly 1). This
conserved high structural
structural elementsconservation
among differentwithin a
genomeviralwith
isolates throughout
a high rate of the ORF of variability
sequence the HCV genome (Figure
indicates that1).each
Thisstructural
high structural
unit conservation
codes important
within afor
information genome with a high
the efficiency ratevirus
of the of sequence
infective variability
cycle. Inindicates
contrast that each
to the structural
genomic unitinformation
UTRs, codes
important information for the efficiency of the virus infective cycle. In contrast
about the structure and function of the ORF structural elements is scarce. Among them, the so-called to the genomic UTRs,
information about the structure and function of the ORF structural elements is scarce. Among them,
CRE, cis-replication element, is probably the one that has attracted more attention (Figure 1). The CRE
the so-called CRE, cis-replication element, is probably the one that has attracted more attention
is defined by three stable stem-loops known as 5BSL3.1, 5BSL3.2, and 5BSL3.3 located at the 30 end of
(Figure 1). The CRE is defined by three stable stem-loops known as 5BSL3.1, 5BSL3.2, and 5BSL3.3
the protein
located coding
at the 3′ region
end of the[24,25]. The
protein central
coding domain,
region 5BSL3.2,
[24,25]. The centralis absolutely indispensable
domain, 5BSL3.2, for HCV
is absolutely
propagation, acting as an essential element for viral replication. Further, it has been
indispensable for HCV propagation, acting as an essential element for viral replication. Further, it shown that CRE is
a negative
has been regulator
shown that of the
CREHCV IRES-dependent
is a negative regulator of translation [26].
the HCV IRES-dependent translation [26].
Figure
Figure 1. The1.hepatitis
The hepatitis C virus
C virus (HCV) RNA (HCV)
genome.RNA genome.
Upper panel:Upper panel:representation
A schematic A schematicof the
representation
genetic organizationofofthe
thegenetic
viral genome. The 50ofand
organization the30viral
UTRs genome.
flankingThe
the5′single
and 3′ UTRs
ORF areflanking
depicted by a
blackthe single
line. ORF proteins
The viral are depicted by afunctions
and their black line. The viral The
are indicated. proteins and their
translational functions
start and stopare
codons
indicated. The translational start and stop codons are marked by arrows. The
are marked by arrows. The numbering corresponds to the codon positions in the ORF according to numbering
corresponds
the HCV to the genotype
Con1 isolate, codon positions in the
1b. Lower ORF A
panel: according
structuraltoconservation
the HCV Con1 isolate,
map of thegenotype
HCV RNA is
represented by gray boxes denoting structurally conserved regions among different viralby
1b. Lower panel: A structural conservation map of the HCV RNA is represented gray The
isolates.
boxes denoting structurally conserved regions among different viral isolates. The
predicted secondary structures of conserved elements in the ORF of the viral RNA genome are shown predicted
secondary structures of conserved elements in the ORF of the viral RNA genome are shown
at the bottom. The color code and labels at the bottom indicate the position where each stem-loop is
at the bottom. The color code and labels at the bottom indicate the position where each stem-
located. The figure is adapted from Reference [27].
loop is located. The figure is adapted from Reference [27].
3. Long-Range Genomic RNA–RNA Interactions
3. Long-Range Genomic RNA–RNA Interactions
Specific structural elements of the 50 and 30 ends of the HCV genome are involved in the
Specific structural elements of the 5′ and 3′ ends of the HCV genome are involved in the viral
viral replication and translation. This implies the existence of a communication between the two
replication and translation. This implies the existence of a communication between the two genomic
genomic ends. It was commonly accepted that the acquisition of a genomic circular conformation was
ends. It was commonly accepted that the acquisition of a genomic circular conformation was
mediated by the
mediated recruitment
by the recruitmentatatboth
bothends
ends of
of the cellularand
the cellular andviral
viral proteins,
proteins, as itaswas
it was demonstrated
demonstrated
for the recruitment of IRES-activity stimulating proteins by the genomic HCV 3 0 UTR [13,28–30].
for the recruitment of IRES-activity stimulating proteins by the genomic HCV 3′ UTR [13,28–30]. We
We questioned whether the conserved structural RNA elements located in the two ends of the HCV
genome were directly involved in the formation of the circular conformation. We first tested the
questioned
ability of whether the
two independent conserved
RNA structural
fragments comprising the 50 or
RNA elements located
the 3in0 genomic
the two ends
endsoftothe
bindHCV
with each
genome were directly involved in the formation of the circular conformation. We first tested the
other. Interestingly both RNA genomic ends form a stable complex in in vitro assays [31]. Further
ability of two independent RNA fragments comprising the 5′ or the 3′ genomic ends to bind with
characterization allowed us to demonstrate that this interaction takes place between the two essential
each other. Interestingly both RNA genomic ends form a stable complex in in vitro assays [31].
structural RNA elements
Further characterization IIIdallowed
within us
domain III of the
to demonstrate IRES
that this and the 5BSL3.2
interaction within
takes place the CRE
between (Figure 2).
the two
It responds to a kissing ALIL (apical loop–internal loop) interaction. This result
essential structural RNA elements IIId within domain III of the IRES and the 5BSL3.2 within the CRE represented the
(Figure of
first evidence 2).a Itlong-range
responds to a kissing ALIL
RNA–RNA (apical involving
interaction loop–internal theloop) interaction.ends,
two genomic This supporting
result a
represented the circularization
protein-independent first evidence ofofa long-range
the HCV RNA RNA–RNA
genome. interaction involving characterization
The functional the two genomic of this
ends, supporting a protein-independent circularization of the HCV RNA genome. The functional
interaction allowed us to conclude that it actually occurs in vivo, being the CRE a negative regulator
characterization of this interaction allowed us to conclude that it actually occurs in vivo, being the
of the HCV
CRE aIRES activity
negative [26] of
regulator (Figure
the HCV2).IRES
Theactivity
use of [26]
CRE(Figure
mutants depleted
2). The of specific
use of CRE mutants structural
depleted units
demonstrated that this translation regulation activity is mainly restricted to the central
of specific structural units demonstrated that this translation regulation activity is mainly restricted structural unit
5BSL3.2to[26].
the central structural unit 5BSL3.2 [26].
A schematic representation of the 0 and 30 ends of the HCV genome depicting the highly
Figure 2.
Figure 2. A schematic representation of 5the 5′ and 3′ ends of the HCV genome depicting the
highly
conserved conserved
structural RNA structural RNAThe
domains: domains:
defined Thefunctional
defined functional
RNA–RNA RNA–RNA interactions
interactions are indicated
are indicated with broken gray lines. The translational start and stop codons are indicated
with broken gray lines. The translational start and stop codons are indicated by arrows. The figure is
adaptedbyfrom
arrows. The figure
Reference [32].is adapted from Reference [32].
The 5BSL3.2
The 5BSL3.2 domain
domain is also
is also involved in
involved in an
an Apical
Apicalloop–Apical
loop–Apical looploop
(ALAL) long-distance
(ALAL) long-distance
interaction with the highly conserved 0 3′SLII domain included in the 03′X tail region at the 3′0 UTR
interaction with the highly conserved 3 SLII domain included in the 3 X tail region at the 3 UTR [33,34]
[33,34] (Figure 2). This interaction promotes viral replication. A third functional long-range
(Figure interaction
2). This interaction promotes
has been described, whichviral replication.
involves the internalA third
loop offunctional
the 5BSL3.2long-range
domain and interaction
the
has been described,
upstream which
Alt region involves
[17,35,36] the 2).
(Figure internal loop of the 5BSL3.2 domain and the upstream Alt
A functional
region [17,35,36] (FigureRNA–RNA
2). interaction seems to take place between domain VI, at the 5′ end of the
protein coding
A functional region, and interaction
RNA–RNA the highly conserved
seems to linker
takesequence between domains
place between domainI VI,andat the 50 the
II within end of the
5′ UTR. This interaction has been proposed to participate in the regulation of both the viral replication
protein coding region, and the highly conserved linker sequence between domains I and II within the 50
and translation processes [20,37–40]. Interestingly, the interacting sequence within the interdomain
UTR. This interaction has been proposed to participate in the regulation of both the viral replication and
region overlaps with the interaction site of the liver-specific micro RNA miR-122, which has been shown
translation
to beprocesses
involved in[20,37–40]. Interestingly,
viral replication the interacting
and translation [41,42]. It hassequence
been proposedwithinthatthe
the interdomain
interaction of region
overlaps with the
miR-122 wouldinteraction site of the liver-specific
induce a conformational switch of the micro RNA
5′ genomic endmiR-122,
encompassingwhich thehas
IRES,been
whichshown to
wouldin
be involved beviral
responsible for the regulation
replication of the activity
and translation of this
[41,42]. It region
has been[39]. proposed that the interaction of
miR-122 would induce a conformational switch of the 50 genomic end encompassing the IRES, which
would be responsible for the regulation of the activity of this region [39].
All together, these results probe the existence of a network of long-range RNA–RNA interactions.
This network involves essential structural genomic RNA domains of at least the IRES, CRE, and the
genomic 30 UTR X-tail region and governs the regulation of the essential viral processes (reviewed
in References [27,43]; Figure 2), in which the CRE 5BSL3.2 domain seems to be the conductor of the
orchestra directing the regulatory interactions of all other structural elements.
All together, these results probe the existence of a network of long-range RNA–RNA
interactions. This network involves essential structural genomic RNA domains of at least the IRES,
CRE, and the genomic 3′ UTR X-tail region and governs the regulation of the essential viral processes
(reviewed in References [27,43]; Figure 2), in which the CRE 5BSL3.2 domain seems to be the
conductor of the orchestra directing the regulatory interactions of all other structural elements.
To understand the molecular mechanism by which long-distance interactions may regulate viral
To understand the molecular mechanism by which long-distance interactions may regulate viral
processes, a detailed structural analysis at the nucleotide level of the HCV genome was carried out.
processes, a detailed structural analysis at the nucleotide level of the HCV genome was carried out.
For this
For purpose, a combination
this purpose, of chemical
a combination probing
of chemical methods,
probing SHAPE
methods, (Selective
SHAPE 20 -hydroxyl
(Selective acylation
2′-hydroxyl
and primer extension)-based technology analysis and accessibility using oligonucleotide
acylation and primer extension)-based technology analysis and accessibility using oligonucleotide microarrays,
was used. The results
microarrays, was used.of this analysis
The results allowed
of this usallowed
analysis to conclude that interactions
us to conclude between
that interactions elements
between
elements
of both ends of of both ends genomes
the viral of the viral genomestopromote
promote to each
each other other
their their structural
structural tuning tuning
at the at the
secondary
secondary
and tertiary and which
levels, tertiaryundoubtedly
levels, whichleads
undoubtedly leads modulation
to the mutual to the mutual modulation
of their of their
functionality [32,44]
functionality [32,44] (Figure 3). Each one tunes the structural conformation of essential
(Figure 3). Each one tunes the structural conformation of essential domains of the other end of the domains of
the other end of the genome. These structural modifications correlate well with the functional
genome. These structural modifications correlate well with the functional regulation observed. Thus,
regulation observed. Thus, modulation of the translation efficiency at the 5′ genomic end by the 3′
modulation of the translation efficiency at the 50 genomic end by the 30 end is achieved by promoting
end is achieved by promoting the conformational fine-tuning of the IRES-essential domains III and
the conformational fine-tuning of the IRES-essential domains III and IV, while the 50 genomic end
IV, while the 5′ genomic end promotes significant structural changes at the 3′ genomic end, mainly
promotes
at thesignificant structural
3′X tail region. changes
The latest the 30 genomic
couldataccount end, in
for variations mainly at the 30 X tail
the translational andregion. The latest
replicational
couldefficiencies.
account for variations in the translational and replicational efficiencies.
Figure 3. The
Figure conformational
3. The tuning
conformational of long-distance
tuning functional
of long-distance RNA
functional RNAdomains:
domains: AAsummary
summaryof the
comparative structural analysis of the 5 0 and 30 HCV genomic ends in the presence and absence of
of the comparative structural analysis of the 5′ and 3′ HCV genomic ends in the presence
each and
other end. Only
absence nucleotides
of each other end.that
Onlyshow a differential
nucleotides chemical
that show reactivitychemical
a differential are indicated with red
reactivity
figures.
are indicated with red figures. The solid figures indicate increases in reactivity. The empty in
The solid figures indicate increases in reactivity. The empty figures indicate decreases
reactivity.
figures4indicate
indicatesdecreases
the DMS in
(dimethyl
reactivity.  indicates
sulfate) the #
reactivity. indicates
DMS the NMIA
(dimethyl (N-methyl-isatoic
sulfate) reactivity.
 indicates
anhydride) reactivity.
the NMIA (N-methyl-isatoic anhydride) reactivity.
Besides intramolecular
Besides intramoleculargenomic
genomicRNA–RNA interactions,intermolecular
RNA–RNA interactions, intermolecular onesones
havehave
also also
been been
described.
described. Thus,Thus,
thethe dimerizationofofthe
dimerization theHCV
HCV genome
genome isisinitiated
initiatedatat
a highly
a highlyconserved palindromic
conserved palindromic
sequence
sequence namednamedDLSDLS (DimerizationLinkage
(Dimerization Linkage Sequence)
Sequence) located
located at at
thethe 30 SLII
3′SLII element of the
element of 3′X 30 X tail
thetail
region [45,46] (Figure 2). The 30 X tail region folds into two alternative conformations [46,47]: a three
stem-loop (30 SLI-30 SLII-30 SLIII) [48] and a two stem-loop (30 SLI-30 SLII’) conformers, being the latest
dimerization competent one, which exposes the DLS in the apical portion of the stem-loop SLII’.
A functional analysis of the 30 X tail in the presence of the IRES and/or the CRE allowed us to
demonstrate that genomic dimerization is modulated by long-distance elements [49]. This confirms
the functionality of the described long-distance RNA–RNA interactions involving the CRE-30 X tail and
IRES-30 X tail.
All together, these data demonstrate the existence of a dynamic network of RNA–RNA contacts
that tune the structural conformation of essential distal RNA elements, resulting in the regulation of
the different stages of the viral cycle. Our data support that the 5BSL3.2 domain operates at the core of
this regulatory system.
4. Functional Genomic RNA Domains as Therapeutic Targets

Interfering with the function of viral genomic structural domains either by modifying their folding
or by impeding the interactions they are involved in offers a potential means to compromise their
viral viability and in the consequence of fighting viral infections. Much effort has been devoted to the
development of nucleic acid-based strategies aimed to interfere with the functioning of viral RNA
genomes. Several types of RNA molecules have been shown to be efficient antiviral agents, and some
of them have even entered clinical trials [50–52]. Out of the different inhibitor RNAs, aptamers have
been postulated as promising tools for targeting structural RNA elements.
4.1. Aptamers
Aptamers are short single-stranded DNA or RNA oligonucleotides that recognize and bind
efficiently and specifically to a target molecule [53,54]. Aptamers have been isolated against a
wide range of molecules varying in nature, size, and complexity: from ions to full cells including
small molecules (such as amino acids, nucleotides, antibiotics, or metabolites), peptides, proteins,
nucleic acids, macromolecular aggregates, virus, cell organelles, or tissues (for examples, see
References [55–59]). In all cases, aptamers are isolated following a common experimental in vitro
strategy known as SELEX (Systematic Evolution of Ligands by EXponential enrichment) [54]. It consists
in iterative selection cycles that comprise the following steps: binding to the target molecule,
partitioning the binders, and molecularly amplifying the binders. The SELEX strategy allows the
identification of efficient binder molecules from a large pool of variants, usually 1012 –1015 . The winner
molecules are known as aptamers. Since it is an entirely in vitro process, the selection conditions are
controlled by the researcher and can be modified along the process.
Herein, we summarize the attempts performed at our laboratory to identify aptamers with a
potential antiviral activity targeting highly conserved structural RNA elements within viral genomes,
paying special attention to the work we have performed to develop RNA inhibitors targeting the HCV
RNA genome.
4.1.1. Aptamers Targeting HCV IRES

We developed an innovative SELEX strategy that was applied to a theoretical population of
more than 1 × 1015 variant RNA molecules with lengths of 84 nt. The original population resulted
from the randomization of 25 consecutive nucleotides linked at the 30 end of the catalytically active
hammerhead ribozyme targeted against position 363 of the HCV IRES. The resulting population
was subjected to a two-sequential-selection-steps procedure [60]. The first selection step selected for
the binding to the complexly folded HCV IRES, the aptamers’ selection; the second step selected
for cleavage of the viral RNA by the ribozyme domain of the previously selected binders. After six
rounds of selection, the heterogeneity of the population was analyzed and the resulting aptamers
were classified within seven groups defined by a consensus sequence motif ranging from 5 to 9 nt in
length. Each consensus motif is complementary to a specific sequence within the IRES (Figure 4a).
This indicates that the complementary sequence partners are required for the binding of the two
molecules. An in vitro analysis of the inhibition of the IRES activity indicated that the majority of the
selected inhibitory RNAs significantly reduced the IRES-dependent translation, yielding inhibitions
up to 90% [61] (Figure 4b).
A further in vivo analysis in the Huh-7 cell culture (human hepatoma cell line) of the most efficient
in vitro inhibitors yielded efficient antiviral activities showing inhibitions up to 60% and 70% of the
viral translation and replication, respectively [61–64] (Figure 5).
selected inhibitory RNAs significantly reduced the IRES-dependent translation, yielding inhibitions
up to 90% [61] (Figure 4b).
(a) (b)
Figure 4. The selection of RNA inhibitors targeting the HCV internal ribosome entry site
(IRES): (a) The identification of the aptamer targets within the IRES. The targets determined
by the complementary sequence to the consensus motif that define each aptamer’s group
are indicated using a colour code on the schematic representation of the secondary structure
of the IRES. (b) (a)The inhibition of the IRES activity by a collection (b) of selected aptamers: The
IRES-dependent translation was measured in vitro as the Fluc activity. The values are the
Figure
Figuremean 4.ofThe
4. The at selection
selection
least three of independent
of RNA RNA inhibitors
inhibitors targeting
targeting the
the HCV
experiments andHCV internal
internal ribosome
ribosome
are referred entry
to the entry
site sitethe
(IRES):
activity in (a) The
(IRES): (a)
identification The
absence ofofan identification
theRNAaptamer of the
targets
inhibitor. aptamer
within
Figure targets
the IRES.
adapted within
fromThe the IRES. The targets determined
targets determined by the complementary
[61].
by the complementary
sequence to the consensussequence
motif that to define
the consensus motif that
each aptamer’s define
group are each aptamer’s
indicated usinggroup
a colour code
areAindicated using a colour
further in representation
vivo analysis incode on the schematic
the secondary representation
Huh-7 cellstructure
culture (human of the secondary
hepatoma structure
on the schematic of the of the IRES. (b) Thecell line)
inhibition ofofthe
themost
IRES
of the in
efficient IRES.
vitro(b) The inhibition
inhibitors yieldedofaptamers:
the IRES
efficient activityactivities
antiviral by a collection
showingof inhibitions
selected aptamers:
up to 60%The and 70%
activity by a collection of selected The IRES-dependent translation was measured in vitro as
of IRES-dependent
the viral translationtranslation
and was measured
replication, in vitro
respectively as the(Figure
[61–64] Fluc activity.
5). The values are the
the Fluc activity. The values are the mean of at least three independent experiments and are referred to
mean of at least three independent experiments and are referred to the activity in the
the activity in the absence of an RNA inhibitor. Figure adapted from [61].
absence of an RNA inhibitor. Figure adapted from [61].
A further in vivo analysis in the Huh-7 cell culture (human hepatoma cell line) of the most
efficient in vitro inhibitors yielded efficient antiviral activities showing inhibitions up to 60% and 70%
of the viral translation and replication, respectively [61–64] (Figure 5).
(a) (b)
FigureFigure 5. The
5. The antiviral
antiviral activity
activity of selected
of selected RNARNA inhibitors
inhibitors in ainhuman
a human hepatoma
hepatoma cellcell line.
line. (a) The
(a) The
inhibition inhibition
of HCV of HCV IRES-dependent
IRES-dependent translation
translation in Huh-7 cells byin Huh-7
three RNAcells by three
inhibitors: TheRNAcells were
co-transfected with a mix containing the transcripts IRES-Fluc and cap-RLuc mRNAs and an excess of
a specific RNA inhibitor. The IRES translation efficiency was measured as Fluc versus RLuc activity
and referred to the value (a)obtained in the control reaction without an RNA (b) inhibitor. The activity values
areFigure
the mean of three independent experiments. (b) The inhibition of
5. The antiviral activity of selected RNA inhibitors in a human hepatoma viral replication in a Huh-7
cell line.
cell line-based HCV sub-genomic replication system (Huh-7 NS3-3 0 ): The cells supporting the stable
(a) The inhibition of HCV IRES-dependent translation in Huh-7 cells by three RNA
replication of an HCV sub-genomic replicon were transfected with individual RNA inhibitors and
incubated for 20 h. The viral replication was measured as the viral RNA amount by qRT-PCR from a
total cell RNA extraction. The data are the mean of three independent experiments.
4.1.2. Anti-HCV CRE Aptamers

The highly conserved genomic CRE element codes essential information for the regulation of
the viral processes. It has been shown to be an essential partner in the viral replication [24,25], and
it is involved in the translation regulation [26,65]. CRE is, therefore, an excellent candidate to be an
efficient anti-HCV therapeutic target. To assess this hypothesis, we applied a SELEX procedure to
isolate aptamers that could interfere with the functioning of the CRE element, which is defined by the
information encoded by each of its structural domains (5BSL3.1-5BSL3.2 and 5BSL3.3).
For this assay, a starting population of RNA molecules resulting from the randomization of
30 contiguous nucleotides (yielding a theoretical heterogeneity of more than 1 × 1018 variants) was
synthesized and introduced in a standard SELEX procedure. With this strategy, we searched for binders
to a 194-nt-long viral genomic RNA fragment containing the CRE structural element [66] (Figure 6a).
After nine rounds of selection, a collection of RNA aptamers could be classified in five different groups
defined by a consensus sequence, which was complementary to a specific domain within the viral
target RNA (Figure 6a). These complementary sequence motifs are likely involved in the binding
of the two molecules. Interestingly, the majority of the selected aptamers contained more than one
consensus sequence and potentially more than one site of binding in the target molecule [66,67]. To test
the potential of the CRE as anti-HCV target, a collection of 44 aptamers was assayed for their capacity
to inhibit HCV replication. For this purpose, the sub-genomic viral replication system based on the
Huh-7 cell lines mentioned above (see Figure 5) was used. The results of this analysis showed than a
significant number of tested aptamers efficiently reduced the viral RNA levels (Figure 6b). It is worth
noting that those aptamers that yield the highest inhibition levels contained the consensus sequences
of groups 2 and 3, which target the apical and internal loops of the 5BSL3.2 domain, respectively
(Figure 6). Further characterization of the inhibitory effect of aptamers P6-89, P6-96, and P6-103
competed out the recruitment of the viral polymerase by the 5BSL3.2 region [67]. This result was also
confirmed with the group 2 aptamers P-58 and P-78 [68]. Further, the P7-49 that is also an efficient
HCV replication
Pharmaceuticals 2019,inhibitor
12, x seems to bind to the 5BSL3.4 rather than 5BSL3.2 domain. Our data9suggests
of 14
for the 5BSL3.4, which bears the translational stop codon, to have new unknown roles in the viral cycle.
(a) (b)
Figure6. 6.RNA
Figure RNA aptamers
aptamers selected
selected against
against the HCVthe cis-replication
HCV cis-replication element(a)(CRE):
element (CRE): Sequence (a) and
Sequence and secondary structure of the 194-nt-long viral RNA fragment used
secondary structure of the 194-nt-long viral RNA fragment used as the target for the selection of as the target
for the selection
anti-HCV aptamers.ofTheanti-HCV aptamers.structural
highly conserved The highly conserved
domains structural
of the CRE elementdomains
(5BSL3.1, of5BSL3.2,
the
CRE5BSL3.3)
and element and(5BSL3.1,
the 5BSL3.45BSL3.2, and the
that includes 5BSL3.3) and the
translational stop5BSL3.4
codon, are that includes
indicated. Thethe
target
translational stop codon, are indicated. The target sequences complementary
sequences complementary to the consensus sequence that define the five groups of selected aptamers to the
consensus
are indicatedsequence
by coloredthat define the
sequences. five inhibition
(b) The groups ofofselected aptamersHuh-7
HCV replication: are indicated by
cells supporting
colored sequences. (b) The inhibition of HCV replication: Huh-7 cells supporting
the stable replication of a sub-genomic HCV replicon were transfected with independent aptamers. the stable
replication
Viral of a were
RNA levels sub-genomic
quantifiedHCVfromreplicon were extracted
the total RNA transfected18 hwith independentusing
post-transfection aptamers.
qRT-PCR.
Viral
The barRNA
chartlevels were
indicates quantified
the HCV RNAfrom thereferred
levels total RNA extracted
to those obtained18 hinpost-transfection using
the absence of a non-related
qRT-PCR.
RNA TheRNA80.
molecule, bar chart
Theindicates
values arethethe HCV
mean ofRNA levels
at least fourreferred
independentto those obtained
experiments in The
[67]. the red
absence of a non-related RNA molecule, RNA80. The values are the mean of
arrows and boxes indicate the aptamers for which the mechanism of action was further characterized.at least four
independent experiments [67]. The red arrows and boxes indicate the aptamers for which
Taken together, of
the mechanism theaction
results offurther
was this study support the potential use of the HCV CRE element as a
characterized.
therapeutic target. More importantly, they provide evidences that aptamers are also useful molecular
Taken together, the results of this study support the potential use of the HCV CRE element as a
therapeutic target. More importantly, they provide evidences that aptamers are also useful molecular
tools for understanding the function of RNA elements and the molecular mechanisms underlying the
RNA-mediated regulation in viral genomes.
4.1.3. Aptamers Targeting the Structural Elements of the 5′ UTR of the HIV-1 RNA Genome
tools for understanding the function of RNA elements and the molecular mechanisms underlying the
RNA-mediated regulation in viral genomes.
4.1.3. Aptamers Targeting the Structural Elements of the 50 UTR of the HIV-1 RNA Genome
Reports describing the successful targeting of different functional RNA elements in several viral
RNA genomes by inhibitor RNA molecules can be found in the literature. Besides HCV, we have also
targeted RNA aptamers against the 50 UTR of the HIV-1 RNA genome. For this purpose, we developed
an innovative approach in which we combined a standard SELEX strategy with an in silico analysis of
the selected aptamers, allowing the design of more efficient ones [69].
The SELEX procedure was applied to a starting population of more than 1 × 1015 variants of
64-nt-long RNA molecules. They were binding-challenged against a HIV-1 genomic RNA fragment
comprising the first 308 nt of its 50 UTR. After 11 selection rounds, we observed that the specific
8-nt-long sequence 50 -GGCAAGGA-30 appeared in high frequency in the RNA population reaching
89.2% from the total analyzed sequences after cycle 14 [69]. Interestingly, this consensus motif is
complementary to the sequence exposed in the apical loop of the highly conserved polyA domain
of the genomic HIV-1 RNA, indicating that it is the target of the selected aptamers (Figure 7). An in
silico structural analysis of the selected RNA molecules containing the consensus octamer sequence
reveled the existence of a consensus 16-nt-long structural RNA element, which folded into a stem-loop
motif that exposes the octamer sequence in the loop. This 8-nt-long loop was closed by four base
pairs of variable sequence (Figure 8a). This result prompted us to hypothesize that the 16-nt structural
element was responsible of the aptamer binding capacity and, therefore, its potential inhibitory activity.
To validate this hypothesis, an in silico-designed new 16-nt-long RNA aptamer was synthesized.
This aptamer This
synthesized. was aptamer
named RNApt16
was named andRNApt16
consistedand
in the consensus
consisted octamer
in the sequence
consensus exposed
octamer in a
sequence
loop closed by a 4-bp-long stem, which the sequence of allows the highest thermodynamic
exposed in a loop closed by a 4-bp-long stem, which the sequence of allows the highest stability of
the folded molecule
thermodynamic (Figure
stability 8a). folded molecule (Figure 8a).
of the
Figure7.7. Aptamers
Figure Aptamersselected
selected against
against thethe genomic
genomic 50 UTR:
HIVHIV 5′ UTR: A schematic
A schematic representation
representation of the
0
of the secondary
secondary structurestructure
of the 5 ofUTRtheof
5′ the
UTR of the
viral RNAviral RNA genome.
genome. The well-characterized
The well-characterized functional
structural
functionalelements are elements
structural depicted. areThedepicted.
sequence The
and secondary
sequence structure of the polyA
and secondary domain
structure is
of the
enlarged in the box
polyA domain with bluein
is enlarged letters.
the box Thewith
octamer
blueconsensus sequence
letters. The of consensus
octamer the selectedsequence
aptamers of
is
indicated in red.
the selected aptamers is indicated in red.
AA comparison
comparison study
study ofof the
the inhibition
inhibition of
of HIV
HIV particles
particles production
production inin the
the cell
cell culture
culture byby the
the
RNApt16
RNApt16 and the two most represented selected aptamers XIV22 and XIV26 demonstrated that the
and the two most represented selected aptamers XIV22 and XIV26 demonstrated that the
three
three of
of them
them efficiently
efficiently inhibited
inhibited viral
viral particle
particle production
production reaching
reaching aa maximum
maximum inhibition
inhibition value
value ofof
85%
85% for
for the
the RNApt16
RNApt16 [69]
[69] (Figure
(Figure 8b).
8b). Thus,
Thus, RNApt16
RNApt16 is is the
the shortest
shortest RNA
RNA aptamer
aptamer soso far
far described
described
that
that efficiently
efficiently interferes
interfereswith
withits
itscorresponding
correspondingtarget.
target.
the selected aptamers is indicated in red.
A comparison study of the inhibition of HIV particles production in the cell culture by the
RNApt16 and the two most represented selected aptamers XIV22 and XIV26 demonstrated that the
three of them efficiently inhibited viral particle production reaching a maximum inhibition value of
85% for the
Pharmaceuticals RNApt16
2019, 12, 38 [69] (Figure 8b). Thus, RNApt16 is the shortest RNA aptamer so far described 10 of 14
that efficiently interferes with its corresponding target.
(a) (b)
Figure
Figure 8. The
8. The in silico-designed
in silico-designed RNApt16:
RNApt16: (a) The
(a) The structural
structural analysis
analysis of representative
of the the representative
examples
examples selected
of 64-nt-long of 64-nt-long selected
aptamers showsaptamers shows free
the minimum the minimum free energy
energy isoform. isoform. The
The probabilities of every
probabilities
nucleotide of every
to actually holdnucleotide to actually
the structural hold the
conformation structural
shown conformation
are represented by a shown are (red,
colour code
represented
highest; by alowest).
dark blue, colour The
codepink,
(red, shadowed
highest; dark
boxblue, lowest).
indicates The pink, shadowed
the consensus 16-nt-long box
structural
indicates the consensus 16-nt-long structural domain. The sequence and
domain. The sequence and secondary structure of the consensus motif and the designed secondary structure
RNApt16 are
of the consensus motif and the designed RNApt16 are indicated at the bottom. R, purine; Y,
indicated at the bottom. R, purine; Y, pirimidine; N, any ribonucleotide, (b) HIV-1 inhibition assays:
pirimidine; N, any ribonucleotide, (b) HIV-1 inhibition assays: The inhibition of HIV-1
The inhibition of HIV-1 particles production measured as the p24 antigen production by the XIV22,
particles production measured as the p24 antigen production by the XIV22, XIV26, and
XIV26, and RNApt16 aptamers (compared to a negative control). The data represent the mean of three
RNApt16 aptamers (compared to a negative control). The data represent the mean of three
independent assays. ** The significant differences as compared to the control (p < 0.01). The figure is
adapted from Reference [69].
5. Conclusions
Viral RNA genomes have developed an information storing system beyond the protein coding
one, which stores essential information for the completion of the viral cycle. This system uses structural
RNA elements to perform essential functions. The high conservation of these elements among different
viral isolates and their important roles in the viral cycle make them potentially good therapeutic targets
for new antiviral strategies, which has been confirmed by the reported examples of interfering with
their activity by the use of RNA aptamers or other RNA inhibitors. The approach of RNA aptamers
targeting highly conserved structural genomic RNA elements offers a potential means of developing
new and efficient antiviral strategies. In addition, aptamers are excellent molecular tools to study the
functionality of structural RNA domains in viral RNA genomes.

Conflicts of Interest: The authors declare no conflict of interest.
References
1. Kruger, K.; Grabowski, P.J.; Zaug, A.J.; Sands, J.; Gottschling, D.E.; Cech, T.R. Self-splicing RNA: Autoexcision
and autocyclization of the ribosomal RNA intervening sequence of Tetrahymena. Cell 1982, 31, 147–157.
[CrossRef]
2. Guerrier-Takada, C.; Gardiner, K.; Marsh, T.; Pace, N.; Altman, S. The RNA moiety of ribonuclease P is the
catalytic subunit of the enzyme. Cell 1983, 35, 849–857. [CrossRef]
3. Romero-López, C.; Berzal-Herranz, A. Unmasking the information encoded as structural motifs of viral
RNA genomes: A potential antiviral target. Rev. Med. Virol. 2013, 23, 340–354. [CrossRef] [PubMed]
4. Reynolds, J.E.; Kaminski, A.; Kettinen, H.J.; Grace, K.; Clarke, B.E.; Carroll, A.R.; Rowlands, D.J.; Jackson, R.J.
Unique features of internal initiation of hepatitis C virus RNA translation. EMBO J. 1995, 14, 6010–6020.
[CrossRef] [PubMed]
5. Wang, T.H.; Rijnbrand, R.C.; Lemon, S.M. Core protein-coding sequence, but not core protein, modulates the
efficiency of cap-independent translation directed by the internal ribosome entry site of hepatitis C virus.
J. Virol. 2000, 74, 11347–11358. [CrossRef] [PubMed]
6. Tsukiyama-Kohara, K.; Iizuka, N.; Kohara, M.; Nomoto, A. Internal ribosome entry site within hepatitis C
virus RNA. J. Virol. 1992, 66, 1476–1483. [PubMed]
7. Wang, C.; Sarnow, P.; Siddiqui, A. Translation of human hepatitis C virus RNA in cultured cells is mediated
by an internal ribosome-binding mechanism. J. Virol. 1993, 67, 3338–3344.
8. Kolykhalov, A.A.; Mihalik, K.; Feinstone, S.M.; Rice, C.M. Hepatitis C virus-encoded enzymatic activities
and conserved RNA elements in the 3’ nontranslated region are essential for virus replication in vivo. J. Virol.
2000, 74, 2046–2051. [CrossRef]
9. Friebe, P.; Lohmann, V.; Krieger, N.; Bartenschlager, R. Sequences in the 5’ nontranslated region of hepatitis
C virus required for RNA replication. J. Virol. 2001, 75, 12047–12057. [CrossRef]
10. Friebe, P.; Bartenschlager, R. Genetic analysis of sequences in the 3’ nontranslated region of hepatitis C virus
that are important for RNA replication. J. Virol. 2002, 76, 5326–5338. [CrossRef]
11. Yi, M.; Lemon, S.M. 3’ nontranslated RNA signals required for replication of hepatitis C virus RNA. J. Virol.
2003, 77, 3557–3568. [CrossRef] [PubMed]
12. Yi, M.; Lemon, S.M. Structure-function analysis of the 3’ stem-loop of hepatitis C virus genomic RNA and its
role in viral RNA replication. RNA 2003, 9, 331–345. [CrossRef]
13. McCaffrey, A.P.; Ohashi, K.; Meuse, L.; Shen, S.; Lancaster, A.M.; Lukavsky, P.J.; Sarnow, P.; Kay, M.A.
Determinants of hepatitis C translational initiation in vitro, in cultured cells and mice. Mol. Ther. 2002,
14. Bung, C.; Bochkaeva, Z.; Terenin, I.; Zinovkin, R.; Shatsky, I.N.; Niepmann, M. Influence of the hepatitis C
virus 3’-untranslated region on IRES-dependent and cap-dependent translation initiation. FEBS Lett. 2010,
15. Chu, D.; Ren, S.; Hu, S.; Wang, W.G.; Subramanian, A.; Contreras, D.; Kanagavel, V.; Chung, E.; Ko, J.;
Amirtham Jacob Appadorai, R.S.; et al. Systematic analysis of enhancer and critical cis-acting RNA elements
in the protein-encoding region of the hepatitis C virus genome. J. Virol. 2013, 87, 5678–5696. [CrossRef]
16. Davis, M.; Sagan, S.M.; Pezacki, J.P.; Evans, D.J.; Simmonds, P. Bioinformatic and physical characterizations of
genome-scale ordered RNA structure in mammalian RNA viruses. J. Virol. 2008, 82, 11824–11836. [CrossRef]
[PubMed]
17. Diviney, S.; Tuplin, A.; Struthers, M.; Armstrong, V.; Elliott, R.M.; Simmonds, P.; Evans, D.J. A hepatitis C
virus cis-acting replication element forms a long-range RNA-RNA interaction with upstream RNA sequences
in NS5B. J. Virol. 2008, 82, 9008–9022. [CrossRef] [PubMed]
18. Fricke, M.; Dunnes, N.; Zayas, M.; Bartenschlager, R.; Niepmann, M.; Marz, M. Conserved RNA secondary
structures and long-range interactions in hepatitis C viruses. RNA 2015, 21, 1219–1232. [CrossRef]
19. Mauger, D.M.; Golden, M.; Yamane, D.; Williford, S.; Lemon, S.M.; Martin, D.P.; Weeks, K.M. Functionally
conserved architecture of hepatitis C virus RNA genomes. Proc. Natl. Acad. Sci. USA 2015, 112, 3692–3697.
[CrossRef] [PubMed]
20. McMullan, L.K.; Grakoui, A.; Evans, M.J.; Mihalik, K.; Puig, M.; Branch, A.D.; Feinstone, S.M.; Rice, C.M.
Evidence for a functional RNA element in the hepatitis C virus core gene. Proc. Natl. Acad. Sci. USA 2007,
21. Pirakitikulr, N.; Kohlway, A.; Lindenbach, B.D.; Pyle, A.M. The Coding Region of the HCV Genome Contains
a Network of Regulatory RNA Structures. Mol. Cell 2016, 62, 111–120. [CrossRef] [PubMed]
22. Tuplin, A.; Wood, J.; Evans, D.J.; Patel, A.H.; Simmonds, P. Thermodynamic and phylogenetic prediction
of RNA secondary structures in the coding region of hepatitis C virus. RNA 2002, 8, 824–841. [CrossRef]
[PubMed]
23. Tuplin, A.; Evans, D.J.; Simmonds, P. Detailed mapping of RNA secondary structures in core and
NS5B-encoding region sequences of hepatitis C virus by RNase cleavage and novel bioinformatic prediction
methods. J. Gen. Virol. 2004, 85, 3037–3047. [CrossRef]
24. Lee, H.; Shin, H.; Wimmer, E.; Paul, A.V. cis-acting RNA signals in the NS5B C-terminal coding sequence of
the hepatitis C virus genome. J. Virol. 2004, 78, 10865–10877. [CrossRef] [PubMed]
25. You, S.; Stump, D.D.; Branch, A.D.; Rice, C.M. A cis-acting replication element in the sequence encoding the
NS5B RNA-dependent RNA polymerase is required for hepatitis C virus RNA replication. J. Virol. 2004,
78, 1352–1366. [CrossRef]
26. Romero-López, C.; Berzal-Herranz, A. The functional RNA domain 5BSL3.2 within the NS5B coding sequence
influences hepatitis C virus IRES-mediated translation. Cell Mol. Life Sci. 2012, 69, 103–113. [CrossRef]
[PubMed]
27. Romero-López, C.; Berzal-Herranz, A. The 5BSL3.2 Functional RNA Domain Connects Distant Regions in
the Hepatitis C Virus Genome. Front. Microbiol. 2017, 8, 2093. [CrossRef]
28. Bradrick, S.S.; Walters, R.W.; Gromeier, M. The hepatitis C virus 3’-untranslated region or a poly(A) tract
promote efficient translation subsequent to the initiation phase. Nucleic Acids Res. 2006, 34, 1293–1303.
[CrossRef] [PubMed]
29. Song, Y.; Friebe, P.; Tzima, E.; Junemann, C.; Bartenschlager, R.; Niepmann, M. The hepatitis C virus RNA
3’-untranslated region strongly enhances translation directed by the internal ribosome entry site. J. Virol.
2006, 80, 11579–11588. [CrossRef] [PubMed]
30. Lourenco, S.; Costa, F.; Debarges, B.; Andrieu, T.; Cahour, A. Hepatitis C virus internal ribosome entry
site-mediated translation is stimulated by cis-acting RNA elements and trans-acting viral factors. FEBS J.
2008, 275, 4179–4197. [CrossRef]
31. Romero-López, C.; Berzal-Herranz, A. A long-range RNA-RNA interaction between the 5’ and 3’ ends of the
HCV genome. RNA 2009, 15, 1740–1752. [CrossRef] [PubMed]
32. Romero-López, C.; Barroso-Deljesus, A.; García-Sacristán, A.; Briones, C.; Berzal-Herranz, A. End-to-end
crosstalk within the hepatitis C virus genome mediates the conformational switch of the 3’X-tail region.
Nucleic Acids Res. 2014, 42, 567–582. [CrossRef] [PubMed]
33. Friebe, P.; Boudet, J.; Simorre, J.P.; Bartenschlager, R. Kissing-loop interaction in the 3’ end of the hepatitis C
virus genome essential for RNA replication. J. Virol. 2005, 79, 380–392. [CrossRef] [PubMed]
34. Cantero-Camacho, A.; Fan, L.; Wang, Y.X.; Gallego, J. Three-dimensional structure of the 3’X-tail of hepatitis
C virus RNA in monomeric and dimeric states. RNA 2017, 23, 1465–1476. [CrossRef] [PubMed]
35. Tuplin, A.; Struthers, M.; Simmonds, P.; Evans, D.J. A twist in the tail: SHAPE mapping of long-range
interactions and structural rearrangements of RNA elements involved in HCV replication. Nucleic Acids Res.
2012, 40, 6908–6921. [CrossRef] [PubMed]
36. Shetty, S.; Stefanovic, S.; Mihailescu, M.R. Hepatitis C virus RNA: Molecular switches mediated by long-range
RNA-RNA interactions? Nucleic Acids Res. 2013, 41, 2526–2540. [CrossRef]
37. Kim, Y.K.; Lee, S.H.; Kim, C.S.; Seol, S.K.; Jang, S.K. Long-range RNA-RNA interaction between the 5’
nontranslated region and the core-coding sequences of hepatitis C virus modulates the IRES- dependent
translation. RNA 2003, 9, 599–606. [CrossRef]
38. Beguiristain, N.; Robertson, H.D.; Gomez, J. RNase III cleavage demonstrates a long range RNA: RNA duplex
element flanking the hepatitis C virus internal ribosome entry site. Nucleic Acids Res. 2005, 33, 5250–5261.
[CrossRef]
39. Diaz-Toledano, R.; Ariza-Mateos, A.; Birk, A.; Martinez-Garcia, B.; Gomez, J. In vitro characterization of a
miR-122-sensitive double-helical switch element in the 5’ region of hepatitis C virus RNA. Nucleic Acids Res.
2009, 37, 5498–5510. [CrossRef]
40. Ariza-Mateos, A.; Diaz-Toledano, R.; Block, T.M.; Prieto-Vega, S.; Birk, A.; Gomez, J. Geneticin Stabilizes the
Open Conformation of the 5’ Region of Hepatitis C Virus RNA and Inhibits Viral Replication. Antimicrob.
Agents Chemother. 2016, 60, 925–935. [CrossRef]
41. Jopling, C.L.; Yi, M.; Lancaster, A.M.; Lemon, S.M.; Sarnow, P. Modulation of hepatitis C virus RNA
abundance by a liver-specific MicroRNA. Science 2005, 309, 1577–1581. [CrossRef]
42. Henke, J.I.; Goergen, D.; Zheng, J.; Song, Y.; Schuttler, C.G.; Fehr, C.; Junemann, C.; Niepmann, M.
microRNA-122 stimulates translation of hepatitis C virus RNA. EMBO J. 2008, 27, 3300–3310. [CrossRef]
[PubMed]
43. Romero-López, C.; Berzal-Herranz, A. Structure-function relationship in viral RNA genomes. The case of
HCV. World J. Med. Genet. 2014, 4, 6–18. [CrossRef]
44. Romero-López, C.; Barroso-Deljesus, A.; García-Sacristán, A.; Briones, C.; Berzal-Herranz, A. The folding of
the hepatitis C virus internal ribosome entry site depends on the 3’-end of the viral genome. Nucleic Acids Res.
2012, 40, 11697–11713. [CrossRef] [PubMed]
45. Cristofari, G.; Ivanyi-Nagy, R.; Gabus, C.; Boulant, S.; Lavergne, J.P.; Penin, F.; Darlix, J.L. The hepatitis C
virus Core protein is a potent nucleic acid chaperone that directs dimerization of the viral (+) strand RNA
in vitro. Nucleic Acids Res. 2004, 32, 2623–2631. [CrossRef] [PubMed]
46. Ivanyi-Nagy, R.; Kanevsky, I.; Gabus, C.; Lavergne, J.P.; Ficheux, D.; Penin, F.; Fosse, P.; Darlix, J.L. Analysis
of hepatitis C virus RNA dimerization and core-RNA interactions. Nucleic Acids Res. 2006, 34, 2618–2633.
[CrossRef] [PubMed]
47. Blight, K.J.; Rice, C.M. Secondary structure determination of the conserved 98-base sequence at the 3’
terminus of hepatitis C virus genome RNA. J. Virol. 1997, 71, 7345–7352.
48. Tanaka, T.; Kato, N.; Cho, M.J.; Sugiyama, K.; Shimotohno, K. Structure of the 3’ terminus of the hepatitis C
virus genome. J. Virol. 1996, 70, 3307–3312.
49. Romero-López, C.; Barroso-delJesus, A.; Berzal-Herranz, A. The chaperone-like activity of the hepatitis C
virus IRES and CRE elements regulates genome dimerization. Sci Rep. 2017, 7. [CrossRef]
50. Reyes-Darias, J.A.; Sanchez-Luque, F.J.; Berzal-Herranz, A. Inhibition of HIV-1 replication by RNA-based
strategies. Curr. HIV Res. 2008, 6, 500–514. [CrossRef]
51. Poolsup, S.; Kim, C.Y. Therapeutic applications of synthetic nucleic acid aptamers. Curr. Opin. Biotechnol.
2017, 48, 180–186. [CrossRef] [PubMed]
52. Zuckerman, J.E.; Davis, M.E. Clinical experiences with systemically administered siRNA-based therapeutics
in cancer. Nat. Rev. Drug Discov. 2015, 14, 843–856. [CrossRef] [PubMed]
53. Ellington, A.D.; Szostak, J.W. In vitro selection of RNA molecules that bind specific ligands. Nature 1990,
54. Tuerk, C.; Gold, L. Systematic evolution of ligands by exponential enrichment: RNA ligands to bacteriophage
T4 DNA polymerase. Science 1990, 249, 505–510. [CrossRef] [PubMed]
55. Wilson, D.S.; Szostak, J.W. In vitro selection of functional nucleic acids. Annu. Rev. Biochem. 1999, 68, 611–647.
[CrossRef] [PubMed]
56. Rajendran, M.; Ellington, A.D. Selection of fluorescent aptamer beacons that light up in the presence of zinc.
Anal. Bioanal. Chem. 2008, 390, 1067–1075. [CrossRef] [PubMed]
57. Raddatz, M.S.; Dolf, A.; Endl, E.; Knolle, P.; Famulok, M.; Mayer, G. Enrichment of cell-targeting and
population-specific aptamers by fluorescence-activated cell sorting. Angew. Chem. Int. Ed. Engl. 2008,
47, 5190–5193. [CrossRef]
58. Marton, S.; Reyes-Darias, J.A.; Sanchez-Luque, F.J.; Romero-Lopez, C.; Berzal-Herranz, A. In vitro and ex
vivo selection procedures for identifying potentially therapeutic DNA and RNA molecules. Molecules 2010,
15, 4610–4638. [CrossRef]
59. Kumar, P.K. Monitoring Intact Viruses Using Aptamers. Biosensors (Basel) 2016, 6, 40. [CrossRef]
60. Romero-López, C.; Barroso-delJesus, A.; Puerta-Fernández, E.; Berzal-Herranz, A. Interfering with hepatitis
C virus IRES activity using RNA molecules identified by a novel in vitro selection method. Biol. Chem. 2005,
61. Romero-López, C.; Díaz-González, R.; Berzal-Herranz, A. Inhibition of hepatitis C virus internal ribosome
entry site-mediated translation by an RNA targeting the conserved IIIf domain. Cell Mol. Life Sci. 2007,
62. Romero-López, C.; Díaz-González, R.; Barroso-delJesus, A.; Berzal-Herranz, A. Inhibition of hepatitis C virus
replication and internal ribosome entry site-dependent translation by an RNA molecule. J. Gen. Virol. 2009,
63. Romero-López, C.; Berzal-Herranz, B.; Gómez, J.; Berzal-Herranz, A. An engineered inhibitor RNA that
efficiently interferes with hepatitis C virus translation and replication. Antiviral Res. 2012, 94, 131–138.
[CrossRef] [PubMed]
64. Romero-López, C.; Lahlali, T.; Berzal-Herranz, B.; Berzal-Herranz, A. Development of Optimized Inhibitor
RNAs Allowing Multisite-Targeting of the HCV Genome. Molecules 2017, 22, 861. [CrossRef] [PubMed]
65. Romero-López, C.; Ríos-Marco, P.; Berzal-Herranz, B.; Berzal-Herranz, A. The HCV genome domains 5BSL3.1
and 5BSL3.3 act as managers of translation. Sci. Rep. 2018, 8, 16101. [CrossRef] [PubMed]
66. Marton, S.; Berzal-Herranz, B.; Garmendia, E.; Cueto, F.J.; Berzal-Herranz, A. Anti-HCV RNA Aptamers
Targeting the Genomic cis-Acting Replication Element. Pharmaceuticals (Basel) 2012, 5, 49–60. [CrossRef]
[PubMed]
67. Fernández-Sanlés, A.; Berzal-Herranz, B.; González-Matamala, R.; Ríos-Marco, P.; Romero-López, C.;
Berzal-Herranz, A. RNA Aptamers as Molecular Tools to Study the Functionality of the Hepatitis C Virus
CRE Region. Molecules 2015, 20, 16030–16047. [CrossRef]
68. Marton, S.; Romero-Lopez, C.; Berzal-Herranz, A. RNA aptamer-mediated interference of HCV replication
by targeting the CRE-5BSL3.2 domain. J. Viral. Hepat. 2013, 20, 103–112. [CrossRef]
69. Sánchez-Luque, F.J.; Stich, M.; Manrubia, S.; Briones, C.; Berzal-Herranz, A. Efficient HIV-1 inhibition by
a 16 nt-long RNA aptamer designed by combining in vitro selection and in silico optimisation strategies.
Sci. Rep. 2014, 4, 6242. [CrossRef]
pharmaceuticals
Communication
Development of a Microwave-assisted
Chemoselective Synthesis of Oxime-linked Sugar
Linkers and Trivalent Glycoclusters
Katherine McReynolds * , Dustin Dimas † , Grace Floyd ‡ and Kara Zeman
Department of Chemistry, California State University, Sacramento, 6000 J Street, Sacramento, CA 95819-6057,
USA; Dustin.Dimas-1@ou.edu (D.D.); gracepfloyd@gmail.com (G.F.); karazeman@csus.edu (K.Z.)
* Correspondence: kdmcr@csus.edu; Tel.: +01-916-278-6551
† Current address: Department of Chemistry and Biochemistry, The University of Oklahoma, 101 Stephenson
Parkway, SLSRC, Room 1000, Norman, OK 73019-5251, USA.
‡ Current address: Biocare Medical, 60 Berry Dr., Pacheco, CA 94553, USA.

Abstract: A rapid, high-yielding microwave-mediated synthetic procedure was developed and

optimized using a model system of monovalent sugar linkers, with the ultimate goal of using
this method for the synthesis of multivalent glycoclusters. The reaction occurs between the
aldehyde/ketone on the sugars and an aminooxy moiety on the linker/trivalent core molecules used
in this study, yielding acid-stable oxime linkages in the products and was carried out using equimolar
quantities of reactants under mild aqueous conditions. Because the reaction is chemoselective, sugars
can be incorporated without the use of protecting groups and the reactions can be completed in
as little as 30 min in the microwave. As an added advantage, in the synthesis of the trivalent
glycoclusters, the fully substituted trivalent molecules were the major products produced in excellent
yields. These results illustrate the potential of this rapid oxime-forming microwave-mediated reaction
in the synthesis of larger, more complex glycoconjugates and glycoclusters for use in a wide variety
of biomedical applications.
Keywords: Microwave reactions; chemoselective; oxime; aminooxy; glycoclusters; multivalent
1. Introduction
Multivalent glycoconjugates such as glycoclusters, glycodendrimers, glyconanoparticles and
glycan microarrays have found a wide range of applications in biochemistry and medicine from
studying protein-carbohydrate interactions to developing potential therapeutic agents [1–4]. It has
long been recognized in nature that proteins and their binding partners are often displayed in a
multivalent fashion. It is also well known for carbohydrate-containing molecules that their binding
interactions with their biological partners are much stronger when present in multiple copies rather
than in a 1:1 ratio. This is known as the multivalent or cluster glycoside effect [2,3,5,6].
Given the complexity and size of multivalent glycoconjugates, as well as the requirement for
multiple simultaneous reactions to append the sugars, it is critical that a synthetic strategy be
developed to create these molecules in an efficient and fully functionalized manner. Commonly,
carbohydrate chemistry requires multiple protection/deprotection steps due to the similar reactivity
of the hydroxyl groups present. This can lead to time-consuming reaction pathways with low yielding
final products. Some studies have sought to simplify the attachment of the carbohydrates to the
multivalent scaffold. We previously showed that we could use amide coupling between the carboxylic
acid-containing unprotected sugar, sialic acid and a poly(amidoamine) (PAMAM) dendrimer core to
create a fully substituted 2nd generation 16-mer glycodendrimer that showed µM activity against

Pharmaceuticals 2019,
Pharmaceuticals 2019, 12,
12, 39
x 22 of 14
of 14
dendrimer core to create a fully substituted 2nd generation 16-mer glycodendrimer that showed µM
HIV-1
activity[7].
againstHowever,
HIV-1[7]. thisHowever,
strategy is only
this applicable
strategy is onlytoapplicable
sugars with eitherwith
to sugars an existing
either anamine existing or
carboxylic acid group or would require the introduction of these groups,
amine or carboxylic acid group or would require the introduction of these groups, leading to further leading to further steps.
Another approach
steps. Another uses click
approach chemistry,
uses however, this
click chemistry, still requires
however, functionalization
this still of the carbohydrate
requires functionalization of the
moiety with either
carbohydrate moiety anwith
alkyne or an
either an alkyne
azide group, requiring
or an azide group, a requiring
multi-stepa synthetic
multi-stepprocess
synthetic involving
process
protecting group chemistry
involving protecting group to install those
chemistry groups
to install [2,8–10].
those groups[2,8–10].
As an
analternative
alternativestrategystrategy to either
to either amide amide or chemistry,
or click click chemistry,
we soughtwe asought
streamlineda streamlined
synthetic
synthetic
process that process
wouldthat allow would
us toallow us to use unprotected
use unprotected carbohydrates carbohydrates
in the coupling in the
stepcoupling step to
to a multivalent
ascaffold
multivalent
molecule scaffold molecule
to create thetodesired
create the desired
fully fully substituted
substituted glycoclusters.glycoclusters.
This involved This ainvolved
two-part a
two-part simplification
simplification strategy. strategy. First, a chemoselective
First, a chemoselective oxime-forming
oxime-forming reactionreaction was selected [11,12].
was selected[11,12]. Oxime
Oxime
linkageslinkages
are known aretoknown
be bothto acid-be and
bothglycosidase-stable,
acid- and glycosidase-stable,
making them making
more robustthem thanmore robust
glycosidic
than glycosidic
linkages linkagesapplications[13–15].
for biological for biological applications [13–15]. reaction
The oximation The oximation
occurs reaction
under mild, occurs under
aqueous
mild, aqueous
conditions between conditions between an aminooxy-containing
an aminooxy-containing molecule, linker
molecule, here a monovalent here aormonovalent
a multivalent linker
core
or a multivalent core [16] and an unprotected reducing aldose (hemiacetal)
[16] and an unprotected reducing aldose (hemiacetal) or ketose (hemiketal) sugar[17,18]. Oxime or ketose (hemiketal)
sugar [17,18].
linkages Oxime
in sugars canlinkages
exist inin ansugars can exist
equilibrium in an equilibrium
mixture containing two mixture
formscontaining two formsthe
in protic solutions, in
protic solutions,
ring opened oxime, thewith
ringE/Z opened
isomers oxime, with
(major E/Z isomers
products) and the(major products)
ring-closed and thecomprised
glycoside, ring-closed of
the α- and comprised
glycoside, β-anomers (minor of the α- and β-anomers
products, (minor The
Figure 1)[11,19]. products,
secondFigure 1) [11,19].
simplification to the Thesynthetic
second
simplification
process involved to thethe synthetic process involved thereaction
use of a microwave-mediated use of afor microwave-mediated
the formation of thereaction for the
oxime linkage.
formation of the oxime linkage. This was done for multiple reasons. First,
This was done for multiple reasons. First, to shorten the reaction times from several hours down to shorten the reaction timesto
from several
minutes hours down Next,
in duration[20]. to minutes in duration
to synthesize the[20].
desiredNext, to synthesize the
glycoconjugates in desired
good yieldsglycoconjugates
and for the
in good yieldsglycoclusters,
multivalent and for the multivalent
to ensure glycoclusters,
that simultaneous to ensure that simultaneous
reactions of each reactions
sugar with of eachthe
sugar with the complementary
complementary aminooxy group aminooxy
on the group on themolecule
linker/core linker/corecouldmolecule could be
be achieved. achieved.
Finally, Finally,
to simplify
to
thesimplify
process theandprocess
ensureand ensure reproducibility
reproducibility of the reactionof thethrough
reactionthe through
use of the use of programmed
programmed methods.
methods. These characteristics
These characteristics of microwave-mediated
of microwave-mediated reactions reactions
make themmake them particularly
particularly attractive
attractive for thefor
the synthesis
synthesis of carbohydrate-containing
of carbohydrate-containing molecules,
molecules, which
which cancanbebesensitive
sensitivetotolong
longand
and harsh
harsh reaction
conditions
conditions[21].[21]. Additionally,
Additionally,there thereareare only
only aa few
few reports
reports of microwave-assisted reactions for the
purpose
purpose ofof synthesizing
synthesizingmultivalent
multivalentglycoconjugates [22–30]. Most
glycoconjugates[22–30]. Mostofofthese
thesereports
reports focus
focus ononthethe
useuseof
microwave
of microwave irradiation
irradiation to mediate
to mediateclick chemistry reactions.
click chemistry There is
reactions. only is
There one reported
only use of alternate
one reported use of
microwave-mediated
alternate microwave-mediated reactionsreactions
to form multivalent oxime-linked
to form multivalent glycopeptoids
oxime-linked [30]. In this In
glycopeptoids[30]. paper,
this
Carrasco and coworkers
paper, Carrasco reported thereported
and coworkers microscalethe microwave-assisted synthesis of a glycopeptoid
microscale microwave-assisted synthesis using of a
aglycopeptoid
50 to 100-fold excess
using a 50oftosugar.
100-fold excess of sugar.
1. Reaction of a typical
Figure 1. typical reducing
reducing aldose
aldose sugar
sugar with
with an
an aminooxy-containing
aminooxy-containing compound in a
protic
protic solvent such as water results in the initial formation of the ring-opened
solvent such as water results in the initial formation of the ring-openedE/Z
E/Z oximes, which over
time in aqueous solution, will equilibrate with the ring closed glycosides.
equilibrate with the ring closed glycosides.
Other
Other traditional,
traditional,non-microwave-mediated
non-microwave-mediated studies similarly
studies show the
similarly use ofthe
show multiple
use of equivalents
multiple
of sugar relative to oxime forming partners. For simple monovalent systems,
equivalents of sugar relative to oxime forming partners. For simple monovalent systems,2–3 equivalents of aldose
2–3
sugar were reacted
equivalents of aldose with aminooxy-containing
sugar were reacted with linkers/peptides,
aminooxy-containing bothlinkers/peptides,
in the presence/absence of
both in the
aniline for 1–48 h at 25 ◦ C or 60 ◦ C [31]. Here, Jensen and coworkers reported yields for the GlcNAc
presence/absence of aniline for 1–48 h at 25 °C or 60 °C[31]. Here, Jensen and coworkers reported
oxime ◦ C., respectively, for the non-aniline
yields linker
for theas GlcNAc
7 and 72% for a 1-h
oxime reaction
linker conducted
as 7 and at 25
72% for or 60reaction
a 1-h conducted at 25 or 60 °C.,
catalyzed reaction,
respectively, for theillustrating
non-aniline that elevatedreaction,
catalyzed temperatures alone that
illustrating improved reaction
elevated rates. For
temperatures the
alone
same GlcNAc reaction conducted with 0.1 M aniline included, the yields were 20 and 80% for a 1 or
6 h reaction time at 25 ◦ C, respectively [31]. For the ketose sialic acid sugars, Szabo and coworkers
reported the synthesis of sialic acid and tetrasialic acid conjugated to one side of a di-aminooxy
linker [32]. The reactions were carried out at 37 ◦ C for 22 h using 19 equivalents of the linker to
the sugar to avoid crosslinking. These reactions resulted in yields of 33 and 50%, respectively, for
sialic acid and tetrasialic acid-linked monovalent conjugates. Finally, for an example of a traditional
multivalent oxime-forming reaction, Renaudet et al. reported yields of 66, 65 and 74% for the syntheses
of tetravalent oxime-linked glycopeptoids using the anomeric aminooxy-sugars α-Fuc, β-Fuc and
β-Gal and a tetra-aldehyde-bearing cyclic peptide. The reactions were conducted at 37 ◦ C for 2 h and
utilized 2 equivalents of sugar per reactive site [33].
Here we report the development of an efficient microwave-mediated method used to synthesize
both mono- and multivalent oxime-linked linkers and glycoclusters, respectively. Our method uses
equimolar ratios of sugar:aminooxy-linker/core and can be completed in as little as 30 min of total
reaction time, such that many reactions can be completed in the space of a day and precious/rare
glycans used sparingly. This microwave procedure is also simple to set up and operate, making it
possible for the rapid production of a wide variety of glycoconjugates by junior researchers/technicians
in the lab. Reaction condition uniformity can also be maintained through the use of a programmed
method, thereby increasing method consistency and minimizing trial-to-trial variation. Through our
microwave-mediated procedure, we have been able to demonstrate both the preparative production
of sugar-linkers in a single step, such that they can be isolated and utilized in further synthetic
transformations or the facile synthesis of novel glycoclusters in excellent yields that can then be readily
incorporated into biological studies. Finally, we have also illustrated that we can create large quantities
of the sugar-linker conjugates using our microwave-mediated conditions in good yields without
the addition of the common catalyst, aniline [31,34]. Removal of the aniline catalyst and using the
microwave to shorten the reaction time both lend themselves to making the reaction greener overall.
2. Results
In this paper we outline the successful combination of chemoselectivity with a
microwave-mediated reaction for purposes of synthesizing a series of oxime-linked monovalent
sugar-linker molecules and three trivalent glycoclusters. We first evaluated seven common aldose
mono-, di- and tri-saccharides (N-acetyl glucosamine, cellobiose, gentiobiose, lactose, maltose,
maltotriose and melibiose, 1–7, Scheme 1) for the preparation of the monovalent sugar linkers.
This chemistry was undertaken to determine the optimum microwave reaction conditions necessary for
the reaction in equimolar quantities of sugar to aminooxy-linker. The study had a goal of minimizing
the use of excess reactants, particularly if expensive/difficult to create sugars were to be used, which
would also serve to simplify the purification process. Traditionally, aniline is used as a catalyst in
oxime-forming reactions because it yields significant increases to the reaction rate [31,34]. Therefore,
in our development of the microwave-mediated method, we evaluated whether or not aniline was
required to improve the yields for oxime formation or whether it could be omitted to make the
reaction greener and easier to purify, without significantly sacrificing the reaction yield. The resultant
sugar-linker molecules are useful intermediates in the development of multivalent glycoconjugates.
The trivalent glycoclusters synthesized in this study provide proof of concept for the synthesis of higher
order glycoclusters efficiently and excellent isolated yields via a chemoselective microwave synthesis.
Pharmaceuticals 2019, 12, 39
x 44 of
of 14
14
Scheme 1. Microwave synthesis of sugar linker conjugates 9–15. (a) 0.1 M NH4OAc, pH 4.5, 25% of
1. Microwave
Scheme 1.
Scheme Microwave synthesis
synthesis of
of sugar
sugar linker
linker conjugates 9–15. (a) 0.1
conjugates 9–15. 0.1 M
M NH
NH44OAc,
OAc, pH
pH 4.5,
4.5, 25%
25% of
400 W, 50 °C, 30 min. Optional: 0.1 .
400 W, 50 ◦°C,
400 C, 30 min. Optional:
Optional: 0.1
0.1..
For the trivalent

trivalent glycocluster
glycocluster synthesis,
synthesis, we employed both
we employed
employed both anan aldose
aldose reducing
reducing disaccharide,
disaccharide,
For the trivalent glycocluster synthesis, we both aldose reducing disaccharide,
cellobiose
cellobiose (2,(2, Scheme
(2, Scheme
Scheme 2) 2) and
2) and two
and two ketose
two ketose sugars,
ketose sugars, sialic
sugars, sialic acid
sialic acid (N-acetyl
acid (N-acetyl neuraminic
(N-acetyl neuraminic
neuraminic acid, acid, Neu5Ac,
acid, Neu5Ac,
Neu5Ac, 18,18,
18,
cellobiose
Scheme
Scheme 3)
3) and
3) and
and the
the α-28-linked dimer
→8-linked dimer
dimer of
of sialic
of sialic
acid (disialic acid) (19, Scheme 3)[35]. These sugars
Scheme the α-2
α-28-linked sialic acid
acid (disialic
(disialic acid)
acid) (19,
(19, Scheme
Scheme 3) [35]. These
3)[35]. These sugars
sugars
were chosen
were
chosen to to illustrate that the microwave reaction worked efficiently for for both
both types
types of sugars and
were chosen to illustrate that the microwave reaction worked efficiently for both types of sugars and
that the
that
the glycosidic
glycosidic
bonds
bonds
present
present
in
in
22and
and
1919
would
would
bebestable
stable
totothe microwave
the microwave
heating
heating
conditions
conditions
at
that the glycosidic bonds present in 2 and 19 would be stable to the microwave heating conditions at
a pH of 4.5. Sialic acid-containing glycans are widely found in nature and are important markers in
aatpH
a pH
of of
4.5.4.5. Sialic
Sialic acid-containing
acid-containing glycans
glycans areare widely
widely found
found in nature
in nature andandareare important
important markers
markers in
disease states such as cancer, influenza and meningococcal meningitis[36,37]. The produced
in disease
disease states
states suchasascancer,
such cancer, influenza
influenza and meningococcal
meningococcal meningitis [36,37]. The
meningitis[36,37]. The produced
produced
glycoclusters contain the acid/glycosidase stable oxime linkage, which can help ensure the integrity
glycoclusters contain
glycoclusters contain the theacid/glycosidase
acid/glycosidase stable
stableoxime
oximelinkage,
linkage,which
whichcancanhelp
helpensure
ensure thethe
integrity of
integrity
of the molecules if they are ultimately used in biological applications[13–15]. It is also worth noting
thethe
of molecules
molecules if they are are
if they ultimately used
ultimately in biological
used applications
in biological [13–15]. It is Italso
applications[13–15]. worth
is also noting
worth that
noting
that longer oligosaccharides may be necessary given that the reducing end sugar will exist as a
longer oligosaccharides may be necessary given that the reducing end sugar
that longer oligosaccharides may be necessary given that the reducing end sugar will exist as a will exist as a mixture
mixture of the native closed ring conformation and the open ring oxime, which may impact the
of the native
mixture of theclosed
nativering conformation
closed and the open
ring conformation and theringopen
oxime,ringwhich
oxime, may impact
which maythe resultant
impact the
resultant biological activity.
biologicalbiological
resultant activity. activity.
Scheme 2. Microwave synthesis of trivalent cellobiose glycocluster (17, 94% yield). (a) 0.1 M NH4OAc,
Scheme 2.
Scheme Microwave synthesis
2. Microwave synthesis of
of trivalent
trivalent cellobiose
cellobiose glycocluster
glycocluster (17,
(17, 94%
94% yield). (a) 0.1
yield). (a) 0.1 M
M NH
NH444OAc,
OAc,
pH 4.5,
pH 4.5, 0.1 M
M aniline, 25%
25% of 400
400 W,30 30 min.
pH 4.5, 0.1
0.1 M aniline,
aniline, 25% of
of 400 W,
W, 30 min.
min.
Scheme 3. Microwave synthesis of trivalent sialic acid (20, 82%) and disialic acid (Sia(α-2 →8)Sia, 21,
(Sia(α-28)Sia,
Scheme 3. Microwave synthesis of trivalent sialic acid (20, 82%) and disialic acid (Sia(α-28)Sia, 21,
88%) glycoclusters. (a) 0.1 M NH4OAc, pH 4.5, 0.1 M aniline, 25% of 400 W, 30–90
30-90 min.
88%) glycoclusters. (a) 0.1 M NH44OAc, pH 4.5, 0.1 M aniline, 25% of 400 W, 30-90 min.
2.1. Medium Scale Microwave-mediated Synthesis of Monovalent Sugar-Linkers

To begin the medium scale (≤0.250 mmol) synthesis of the monovalent sugar linker molecules, a
bifunctional, hydrophilic aminooxy-Boc-protected amine linker was used (8, Scheme 1) [16]. This was
combined 1:1 with any of the seven off-the-shelf aldose mono-, di- and tri-saccharides (1–7) in 0.1 M
ammonium acetate, pH 4.5, in the presence or absence of 0.1 M aniline. The reactions were carried
out in a CEM MARS 5 laboratory-grade microwave. Many different combinations of power level and
time were attempted, with the optimum combination found to be 30 min at 25% of 400 W, with a 50 ◦ C
maximum temperature. Upon completion of the reaction, the solution was freeze dried then purified
by flash chromatography on silica gel in a 6:4:0.5 mixture of chloroform:methanol:water. Examination
of the pooled fractions by 1 H NMR showed no evidence of degradation of either the products or
unreacted starting materials (See Supplementary Materials for details). This initial set of reactions,
using the aniline catalyst, gave rise to 73–93% yields of the Boc-protected sugar linker products (9–15,
Table 1), while the same reactions conducted without the aniline catalyst resulted in yields ranging
from 60–68%, a decrease of 9–28%, depending on the sugar used.
Table 1. Summary of sugar-linker 50 ◦ C traditional and microwave syntheses. Medium scale (≤0.250
mmol) reactions were run in the presence or absence of 0.1 M aniline (final concentration), while the
large scale (≥0.800 mmol) reactions were all run in the presence of 0.1 M aniline as a catalyst. The ∆%
yield column compares the yields of the medium scale aniline catalyzed reaction with the uncatalyzed
reaction of the same scale. For the large-scale reactions, the comparison is between the medium and
large-scale aniline-catalyzed reactions.
Sugar Additive % Yield % Yield

Medium Scale (≤0.250 mmol)
Non-Microwave Conditions:
Cellobiose 0.1 M Aniline 65
N/A 56 −9
Microwave-Mediated Conditions:
GlcNAc 0.1 M Aniline 78
N/A 68 −10
Cellobiose 0.1 M Aniline 76
N/A 63 −13
Gentiobiose 0.1 M Aniline 73
N/A 60 −13
Lactose 0.1 M Aniline 93
N/A 65 −28
Maltose 0.1 M Aniline 74
N/A 65 −9
Maltotriose 0.1 M Aniline 92
N/A 68 −24
Melibiose 0.1 M Aniline 80
N/A 68 −12
Large Scale (≥0.800 mmol)
GlcNAc 0.1 M Aniline 79 1
Cellobiose 0.1 M Aniline 75 −1
Gentiobiose 0.1 M Aniline 65 −8
Lactose 0.1 M Aniline 63 −30
Maltose 0.1 M Aniline 62 −12
Maltotriose 0.1 M Aniline 64 −28
Melibiose 0.1 M Aniline 60 −20
For further comparison, two traditional, non-microwave-mediated reactions to yield the

Boc-protected cellobiosyl-linker (10) were carried out at 50 ◦ C for 30 min in the presence or
absence of the aniline catalyst. Cellobiose was chosen for the sugar, as it represents a typical
aldose disaccharide. For the aniline-catalyzed reaction, a 65% yield of 10 resulted, while for the
non-catalyzed reaction a yield of 56% of 10 was observed (Table 1). Comparing these results against 10
synthesized in the microwave, modest yield improvements were noted in the microwave mediated
reactions. The microwave aniline-catalyzed reaction gave rise to an 11% higher yield and the
non-aniline catalyzed microwave reaction resulted in a 7% higher yield. Overall the average %
yield increase for all 7 aldoses for the microwave versus traditional reactions in the presence of aniline
was 16% and 9% for the non-aniline catalyzed reactions. Interestingly, the average magnitude of
the difference between aniline versus non-aniline catalyzed reactions was greater (~15%) for the
microwave-mediated reactions than for the traditional heated reactions (9%). These results indicate
that modest improvements of yield can be gained by using microwave-mediated conditions, using
equimolar quantities of sugar and linker. In addition, if desired, it was found that the aniline catalyst
could be left out of the reaction mixture to simplify product purification and make the reaction greener,
all without unreasonable yield losses.
2.2. Large Scale Microwave-mediated Synthesis of Monovalent Sugar-Linkers

To further evaluate reaction scalability, the microwave reaction was then carried out on the same
seven aldose sugars at higher quantities (≥0.800 mmol) using identical microwave reaction conditions
as described above. All of the reactions included the 0.1 M aniline catalyst to maximize the yields.
Here it was found that the isolated yields in the large-scale reactions ranged from 60–79%, a decrease
of 1–30% compared to the medium scale aniline-catalyzed reactions, again depending on the sugar
incorporated (Table 1). We noted that the decreases in the yields for the medium scale reactions in the
absence of aniline were similar to the decreases seen in the larger scale microwave-mediated reactions
in the presence of the aniline catalyst. This means that while the microwave can be an excellent tool for
shortening the reaction times for these reactions, a balance must be struck between reaction scale and
reaction time savings. For our purposes, it made sense to significantly scale up the reactions, given
that the sugars used were all commercially available and the linker could be produced efficiently in
large quantities as well [16]. These new monovalent sugar-linker molecules are useful intermediates
that can be utilized in the synthesis of further glycoconjugates. Once the Boc group is removed,
the resultant amine can be used in amide coupling reactions to attach the sugar linker to whatever
carboxyl-containing scaffold/surface is desired.
2.3. Multivalent Glycocluster Microwave-mediated Synthesis

Based on the results for the model monovalent sugar linkers, we moved into the
microwave-mediated synthesis of multivalent glycoclusters. With multivalent scaffolds, in addition
to the desired fully functionalized product (here the trisubstituted glycocluster), under-substituted
products are possible (un-, mono- and disubstituted glycoclusters). It was hypothesized that by
utilizing our best reaction conditions developed for the equimolar system described above (25%
of 400 W power, 30 min, 0.1 M aniline), that the production of under-substituted products would
be limited. The synthesis of three novel trivalent glycoclusters was undertaken beginning with
the optimized conditions and included one aldose disaccharide, cellobiose (2, Scheme 2), as well
as a ketose monosaccharide sialic acid (18, Scheme 3) and a ketose disaccharide, α-2→8-disialic
acid (19, Scheme 3), with a previously synthesized trivalent aminooxy-terminated hydrophilic core
(16) [16]. Here, cellobiose was chosen as a representative aldose disaccharide and both sialic acid and
α-2→8-disialic acid [35] were chosen to represent more hindered, less reactive ketose substrates to
show the utility of this method for these interesting, biologically important sugars.
Beginning with a 3:1 ratio of cellobiose to the trivalent core (2 and 16, respectively), the 30-min
reaction time was sufficient to produce only the desired trivalent product, 17, in 94% yield following
purification via size exclusion chromatography (SEC, Scheme 2). This reaction was carried out on a
10-fold lower scale than the monovalent, aniline-catalyzed reaction, which is one possible reason why
the yields were higher. No under-substituted products (un-, mono- or disubstituted), sugar degradation
or unreacted starting materials were observed for this reaction by 1 H NMR upon purification.
For the ketoses, the sialic acid (18) reaction with the trivalent core (16, Scheme 3), a 3:1 ratio
of sugar to core was utilized. The reaction was carried out as described above for 30 min and after
purification by SEC, an 82% yield of the desired trivalent product (20) was achieved. However, unlike
the reaction to produce 17, where no under-substituted products were produced, the disubstituted
byproduct was isolated from a separate peak from the SEC purification and identified by 1 H NMR.
Carrying this method forward, the disaccharide ketose, disialic acid, 19, was reacted in a 3:1 sugar to
core (16) ratio under the same conditions as used for sialic acid. However, it was noted that 30 min was
not sufficient to achieve a good yield for the desired trivalent product, 21. This is likely due to steric
issues, so two additional 30-min cycles were carried out under the same conditions for a total of 90 min
of microwave reaction time. After purification by SEC, an 88% yield of the desired trivalent product
(21) was achieved. Similar to the sialic acid reaction, the disubstituted byproduct was isolated from
a separate peak after SEC purification and was identified by 1 H NMR. No other under-substituted
products or other byproducts were noted from the pooled column fractions for any of the observed
peaks, showing again that the microwave-mediated reaction conditions are mild enough to use for
these more expensive sugars.
3. Discussion
In conclusion, the development of an efficient microwave-mediated oxime forming reaction
between equimolar quantities of unprotected aldose or ketose sugars and either a monovalent
or trivalent aminooxy-containing linker or core was undertaken with the aims to create a facile
and reproducible method that resulted in decreased reaction times, minimization of the amounts
of sugars/catalysts used to reduce costs and make the reactions greener, while still creating the
desired glycoconjugates efficiently and in good to excellent yields. We began our studies with the
formation of model monovalent oxime-linked sugar linker molecules at two different preparative
reaction scales (medium and large). This was done to determine the optimum microwave-mediated
conditions necessary to achieve the best yields in the smallest amount of time without requiring an
excess of either the sugar or linker molecules. Once this was accomplished, the microwave reaction
conditions were then applied to more complex multivalent systems, such that biologically relevant
glycoclusters could be prepared in their desired fully substituted forms in a matter of minutes in a
single chemoselective step.
To begin with the monovalent oxime-linked sugar linkers, first a medium scale reaction
(≤0.250 mmol) was tested both in the presence and absence of 0.1 M aniline as a catalyst and equimolar
quantities of sugar and linker. The aniline-based reaction conditions were superior to the reactions
without aniline, which was not unexpected, however, if greener reaction conditions are sought or
simplified purification procedures are desired, the reactions still work well without the catalyst when
the reaction is carried out using microwave conditions. When the reaction was scaled up to a more
preparative scale (≥0.800 mmol) in the presence of 0.1 M aniline, slightly lower yields were obtained,
however, the yields were considered to be acceptable, given the ease of setup and the short reaction
time needed to produce larger quantities of simple glycoconjugates.
The real advantage of using a microwave-mediated reaction in the formation of oxime-linked
glycoconjugates was realized when the method was applied to one disaccharide aldose and two
ketose sugars in a multivalent reaction. The complete substitution of multivalent glycoclusters can be
difficult to achieve, as multiple, simultaneous reactions are required between the individual sugars
and the reactive moieties on the multivalent scaffold. However, in this study using our optimized
microwave-mediated reaction conditions developed with the monovalent sugar linkers, it was found
that the tri-cellobiosyl product (17) was synthesized in a 94% yield, with no under-substitution products
observed, while the tri-sialic acid (20) and trivalent di-sialic acid (21) were the predominant products
formed in 82 and 88% yields, respectively. For the latter two reactions, the only other observed minor
product in each case was the disubstituted glycocluster.
This newly developed microwave method allows for the efficient production of monovalent or
multivalent glycoconjugates. It offers ease of set-up, consistent trial-to-trial reaction control through the
use of programmable methods, short reaction times and good yields of the desired products, all while
utilizing equimolar quantities of the reactant oxime-forming partners. This method, when applied to
larger, more complex/hindered oligosaccharides gives rise to primarily the desired fully substituted
products, with minimal to no production of undesired under-substituted products. This is valuable,
particularly when one is working with sugars that are rare or expensive to produce/purchase.
4.1. General Methods

Unless otherwise noted all chemicals were purchased from commercial sources and used without
further purification/treatment. All microwave reactions were carried out in a CEM MARS 5 microwave.
All reaction solutions were freeze dried upon reaction completion prior to purification. Size exclusion
chromatography (SEC) separations were conducted on either a BioRad BioLogic DuoFlow 10 system
or a Pharmacia LC 500 system, using a BioRad 2.5 × 120 cm column packed with BioGel P-10 in
0.03 M NH4 HCO3 . 3.5 mL fractions were collected, and the absorbance measured at 214 nm and 225
nm. 1 H and 13 C spectra (internal methanol standard) in D2 O were collected on a Bruker Avance III
500 MHz spectrometer. 1 H NMR integration data for 17, 20 and 21 were normalized to 1/3 of the total
molecule. Mass spectrometry data were obtained at the Campus Chemical Instrument Center (CCIC)
Mass Spectrometry and Proteomics Facility at The Ohio State University (OSU).
4.2. Synthesis of Monovalent Sugar-Linkers
4.2.1. General Procedure for the Medium-Scale (0.187–0.250 mmol) Synthesis of Sugar-Linkers
Without Aniline-microwave or Traditional Heating

Compounds 9–15 were synthesized using 1 equivalent of the aminooxy linker (Compound 8) [16],
prepared as a 100 mg/mL solution in methanol. The appropriate volume of this solution was
transferred to a flask and evaporated under reduced pressure, then freeze-dried to get an accurate
mass of the oil. Next, 1 equivalent of: N-acetylglucosamine, cellobiose, gentiobiose, lactose, maltose,
maltotriose or melibiose was separately added to each flask containing Compound 8 [16]. These were
each dissolved in 3.0 mL of 0.1 M ammonium acetate (NH4 OAc) at a pH of 4.5. The reactions were
conducted either stirring at 50 ◦ C in an oil bath (traditional) or at 400 W in a microwave (CEM MARS 5)
at 25% power with a 2 min ramp to temperature and a hold time of 30 min at a maximum temperature
of 50 ◦ C. After the reaction was complete, the solutions were freeze-dried. The products were then
purified by flash chromatography in 6:4:0.5 CHCl3 :MeOH:H2 O, yielding off-white amorphous solids.
Tert-butyl N-[3-(2-{(E/Z)-[2-acetamido-2-deoxy-D-glucopyranosyl]oxime}ethoxy)propyl] carbamate
(Compound 9): 54.3 mg (0.232 mmol) of Compound 8 plus 53.2 mg (0.241 mmol) of Compound 1
were utilized, resulting in 68.9 mg (67.9%) of an off white solid (Compound 9). 1 H NMR (500 MHz,
D2 O): 1 H NMR (500 MHz, D2 O): δ 7.49 (d, J = 6.2 Hz, 0.7H, E isomer), 6.83 (d, J = 6.6 Hz, 0.2H, Z
isomer), 5.10, (t, J = 6.7 Hz, 0.2H, Z isomer), 4.67 (t, J = 6.8 Hz, 0.7H, E isomer), 4.32 (d, J = 9.8 Hz, 0.1 H,
closed ring), 4.30-4.17 (m, overlapping, 2H), 4.17-4.03 (m, overlapping, 1H), 3.39-3.47 (m, overlapping,
9.2H), 3.41 (d, J = 3.4 Hz, 0.1H, closed ring), 3.10 (t, J = 6.3 Hz, 2H), 2.02 (s, 3H), 1.71 (p, J = 6.3, 13.0 Hz,
2H), 1.40 (s, 9H). 13 C NMR (125 MHz, D2 O with internal MeOH standard): δ 174.21, 171.20, 149.99,
148.78, 81.03, 72.79, 71.50, 71.07, 70.22, 69.43, 68.65, 68.60, 63.02, 52.07, 49.03, 37.20, 28.84, 23.42, 22.38,
22.14, 22.03. HRMS ESI+: Calc. for C18 H36 N3 O9 (M + H)+ : 438.2416. Found: 438.2455.
Tert-butyl N-[3-(2-{(E/Z)-[β-D-glucopyranosyl-(1→4)-D-glucopyranosyl]oxime}ethoxy)propyl]
carbamate (Compound 10-microwave): 43.7 mg (0.187 mmol) of Compound 8 plus 63.9 mg (0.187
mmol) of Compound 2 were utilized, resulting in 64.7 mg (63.1%) of an off white solid (Compound
10). 1 H NMR (500 MHz, D2 O): δ 7.69 (d, J = 5.5 Hz, 0.6H, E isomer), 7.00 (d, J = 5.5 Hz, 0.1H, Z
isomer), 4.99 (dd, J = 4.1, 5.4 Hz, 0.1H, Z isomer), 4.58 (dd, J = 5.7, 6.8 Hz, 0.7H, E isomer), 4.56-4.49 (m,
overlapping, 0.9H), 4.30 (d, J = 9.2 Hz, 0.2H), 4.28-4.21 (m, overlapping, 1.5H), 4.09 (t, J = 3.7 Hz, 0.1H,
Z isomer), 3.98 (dd, J = 1.8, 6.9 Hz, 0.7H, E isomer), 3.96-3.39 (m, overlapping, 14.8 H), 3.46-3.29 (m,
overlapping, 1H), 3.17-3.11 (m, overlapping, 2H), 1.75 (p, J = 6.5, 12.9 Hz, 2H), 1.43 (s, 9H). 13 C NMR
(125 MHz, D2 O with internal MeOH standard): δ 158.39, 153.22, 152.06, 102.69, 90.25, 81.08, 80.77,
78.64, 78.38, 76.16, 75.97, 75.76, 75.45, 73.51. 73.35, 73.28, 72.81, 71.51, 71.34, 70.59, 69.63, 69.51, 69.52,
68.82, 68.66, 68.53, 68.49, 66.46, 62.27, 62.09, 60.76, 60.66, 60.33, 49.06, 37.29, 28.94, 27.88. HRMS ESI+:
Calc. for C22 H43 N2 O14 (M + H)+ : 559.2709. Found: 559.2724.
Cellobiose (Compound 10-oil bath, traditional): 55.3 mg (0.236 mmol) of Compound 8 plus 73.3
mg (0.214 mmol) of Compound 2 were utilized, resulting in 67.3 mg (56.1%) of an off white solid
(Compound 10).
Tert-butyl N-[3-(2-{(E/Z)-(β-D-glucopyranosyl-(1→6)-D-glucopyranosyl)oxime}ethoxy)propyl]
carbamate (Compound 11): 46.6 mg (0.199 mmol) of Compound 8 plus 70.8 mg (0.207 mmol) of
Compound 3 were utilized, resulting in 65.8 mg (60.3%) of a fluffy white solid (Compound 11). 1 H
NMR (500 MHz, D2 O): δ 7.49 (d, J = 6.5 Hz, 0.7H, E isomer), 6.84 (d, J = 6.3 Hz, 0.1H, Z isomer), 4.42 (d,
J = 9.1 Hz, 1H), 4.39 (d, J = 3.7 Hz, 0.7H), 4.39 (t, J = 3.7 Hz, 2H), 3.88-3.36 (m, overlapping, 15H), 3.05
(t, J = 6.5 Hz, 2H), 1.67 (p, J = 6.5, 13.1 Hz, 2H), 1.34 (s, 9H). 13 C NMR (125 MHz, D2 O with internal
MeOH standard): δ 157.48, 151.76, 150.70, 102.10, 89.60, 80.15, 75.94, 75.37, 75.22, 75.16, 74.90, 72.60,
72.47, 72.36, 72.30, 71.94, 70.72, 70.14, 69.80, 69.53, 69.48, 69.07, 69.01, 68.94, 68.66, 68.52, 67.90, 67.74,
67.63, 65.61, 59.99, 48.17, 36.35, 28.02, 27.00, 22.53. HRMS ESI+: Calc. for C22 H43 N2 O14 (M + H)+ :
559.2715. Found: 559.2723.
Tert-butyl N-[3-(2-{(E/Z)-(β-D-galactopyranosyl-(1→4)-D-glucopyranosyl)oxime}ethoxy)propyl]
Compound 4 were utilized, resulting in 68.1 mg (64.6%) of an off white solid (Compound 12). 1 H
NMR (500 MHz, D2 O): δ 7.70 (d, J = 5.5 Hz, 0.7H, E isomer), 7.00 (d, J = 5.4 Hz, 0.1H, Z isomer), 4.59 (d,
J = 6.0 Hz, 0.7H), 4.50 (d, J = 7.8 Hz, 0.8H), 4.24 (t, J = 4.4 Hz, 2H), 3.98-3.54 (m, overlapping, 16H), 3.14
(t, J = 6.4 Hz, 2H), 1.76 (p, J = 6.5, 13.0 Hz, 2H), 1.43 (s, 9H). 13 C NMR (125 MHz, D2 O with internal
MeOH standard): δ 158.43, 153.38, 152.23, 103.70, 103.26, 103.13, 90.30, 81.08, 80.86, 78.50, 78.35, 76.23,
75.56, 75.43, 75.27, 73.58, 73.32, 72.85, 72.76, 71.52, 71.39, 71.26, 71.18, 70.55, 69.60, 69.48, 68.89, 68.79,
68.71, 68.54, 66.75, 62.31, 62.11, 61.26, 61.09, 61.00, 60.39, 49.12, 37.29, 28.99, 27.95. HRMS ESI+: Calc.
for C22 H43 N2 O14 (M + H)+ : 559.2715. Found: 559.2727.
Tert-butyl N-[3-(2-{(E/Z)-(α-D-glucopyranosyl-(1→4)-D-glucopyranosyl)oxime}ethoxy)propyl]
Compound 5 were utilized, resulting in 79.5 mg (65.3%) of a white powdery solid (Compound 13).
1 H NMR (500 MHz, D O): δ 7.51 (d, J = 6.1 Hz, 0.7H, E isomer), 6.88 (d, J = 5.5 Hz, 0.2H, Z isomer),
2
5.02 (d, J = 4.0 Hz, 1H), 4.44 (t, J = 5.5 Hz, 0.7H), 4.16 (t, J = 1.7 Hz, 2H), 4.92-3.35 (m, overlapping,
16H), 3.04 (t, J = 6.6 Hz, 2H), 1.67 (p, J = 6.5, 13.1 Hz, 2H), 1.35 (s, 9H). 13 C NMR (125 MHz, D2 O with
internal MeOH standard): δ 158.38, 153.53, 151.85, 129.60, 128.03, 125.53, 100.73, 100.59, 99.83, 90.27,
81.61, 81.04, 80.34, 77.36, 76.87, 75.94, 73.52, 73.27, 73.12, 73.03, 72.94, 72.86, 72.64, 72.42, 72.29, 71.84,
71.45, 69.59, 69.54, 69.49, 69.43, 68.83, 68.68, 68.66, 68.48, 65.86, 62.48, 62.26, 60.98, 60.64, 60.54, 49.05,
37.27, 28.91, 27.89. HRMS ESI+: Calc. for C22 H43 N2 O14 (M + H)+ : 559.2715. Found: 559.2713.
Tert-butyl N-[3-(2-{(E/Z)-(α-D-glucopyranosyl-(1→4)-α-D-glucopyranosyl-(1→4)-D-glucopyran
osyl)oxime} ethoxy)propyl] carbamate (Compound 14): 53.4 mg (0.228 mmol) of Compound 8 plus
115.5 mg (0.229 mmol) of Compound 6 were utilized, resulting in 112.2 mg (68.3%) of an off white
amorphous solid (Compound 14). 1 H NMR (500 MHz, D2 O): δ 7.62 (d, J = 6.1 Hz, 0.7 H, E isomer),
6.96 (d, J = 5.5 Hz, 0.1 H, Z isomer), 5.59 (d, J = 3.9 Hz, 1 H), 5.36 (d, J = 4.0 Hz, 0.8 H), 4.54 (t, J = 5.6
Hz, 0.7 H), 4.29 (t, J = 3.7 Hz, 2 H), 4.04-3.61 (m, overlapping, 21H), 3.58 (t, J = 3.5 Hz, 1 H), 3.15 (t,
J = 6.3 Hz, 2H), 1.77 (p, J = 6.5, 13.0 Hz, 2H), 1.45 (s, 9H). 13 C NMR (125 MHz, D2 O with internal
MeOH standard): δ 158.51, 153.68, 151.96, 100.53, 100.06, 81.84, 81.14, 80.54, 77.46, 77.11, 76.04, 73.69,
73.65, 73.52, 73.20, 72.99, 72.94, 72.51, 72.44, 72.08, 71.77, 71.62, 71.51, 71.30, 69.72, 69.62, 69.52, 68.96,
68.79, 68.60, 65.99, 62.57, 62.36, 61.03, 60.77, 60.65, 49.15, 37.37, 29.04, 27.99. HRMS ESI+: Calc. for
C28 H53 N2 O19 (M + H)+ : 721.3244. Found: 721.3244.
Tert-butyl N-[3-(2-{(E/Z)-(α-D-galactopyranosyl-(1→6)-α-D-glucopyranosyl-(1→4)-D-glucopyrano
syl)oxime}ethoxy)propyl] carbamate (Compound 15): 46.1 mg (0.197 mmol) of Compound 8 plus
70.9 mg (0.197 mmol) of Compound 7 were utilized, resulting in 73.0 mg (67.6%) of an off white
amorphous solid (Compound 15). 1 H NMR (500 MHz, D2 O): δ 7.60 (d, J = 6.5 Hz, 0.7H, E isomer),
6.95 (d, J = 9.4 Hz, 0.1H, Z isomer), 4.99 (d, J = 3.6 Hz, 2H), 4.44 (d, J = 6.8 Hz, 0.7H), 4.26 (d, J = 6.8 Hz,
2H), 4.00-3.51 (m, overlapping, 16H), 1.75 (p, J = 6.6, 13.1 Hz, 2H), 1.44 (s, 9H). 13 C NMR (125 MHz,
D2 O with internal MeOH standard): δ 158.44, 152.85, 151.76, 98.58, 81.11, 75.97, 73.27, 72.92, 71.17,
71.12, 71.04, 70.61, 70.49, 70.36, 69.99, 69.77, 69.52, 69.48, 68.94, 68.88, 68.84, 68.72, 68.70, 66.67, 61.37,
61.31, 57.66, 49.12, 37.31, 28.98, 27.96, 22.39, 17.04. HRMS ESI+: Calc. for C22 H42 N2 NaO14 (M + Na)+ :
581.2535. Found: 581.2556.
With Aniline
Compounds 9–15 were synthesized using 1 equivalent of the aminooxy linker (Compound 8) [16],
mass of the oil. Next, 1 equivalent of: N-acetylglucosamine, cellobiose, gentiobiose, lactose, maltose,
maltotriose or melibiose was separately added to each flask containing Compound 8. These were each
dissolved in 3.0 mL of 0.1 M ammonium acetate (NH4 OAc) at a pH of 4.5. Next, 27.3 µL of aniline
(0.1 M final concentration) were added to each flask and the pH checked to ensure that it remained at
4.5. The reactions were conducted either stirring at 50 ◦ C in an oil bath (traditional) or at 400 W in a
microwave (CEM MARS 5) at 25% power with a 2 min ramp to temperature and a hold time of 30 min
at a maximum temperature of 50 ◦ C. After the reaction was complete, the solutions were freeze-dried.
The products were then purified by flash chromatography in 6:4:0.5 CHCl3 :MeOH:H2 O, yielding off
white amorphous solids.
N-acetylglucosamine (Compound 9): 55.3 mg (0.240 mmol) of Compound 8 plus 52.2 mg (0.240
mmol) of Compound 1 were utilized, resulting in 81.4 mg (77.6%) of an off white solid (Compound 9).
Cellobiose (Compound 10-microwave): 53.1 mg (0.227 mmol) of Compound 8 plus 77.7 mg (0.227
mmol) of Compound 2 were utilized, resulting in 96.3 mg (76.1%) of an off white solid (Compound 10).
Cellobiose (Compound 10-oil bath, traditional): 50.4 mg (0.215 mmol) of Compound 8 plus 74.2
mg (0.217 mmol) of Compound 2 were utilized, resulting in 77.8 mg (64.8%) of an off white solid
(Compound 10).
Gentiobiose (Compound 11): 52.0 mg (0.222 mmol) of Compound 8 plus 80.1 mg (0.222 mmol) of
Compound 3 were utilized, resulting in 90.6 mg (73.1%) of a fluffy white solid (Compound 11).
Lactose (Compound 12): 56.5 mg (0.241 mmol) of Compound 8 plus 82.6 mg (0.241 mmol) of
Compound 4 were utilized, resulting in 124.7 mg (92.7%) of an off white solid (Compound 12).
Maltose (Compound 13): 43.6 mg (0.186 mmol) of Compound 8 plus 67.0 mg (0.186 mmol) of
Compound 5 were utilized, resulting in 76.4 mg (73.6%) of an off white amorphous solid (Compound 13).
Maltotriose (Compound 14): 48.1 mg (0.206 mmol) of Compound 8 plus 103.7 mg (0.206 mmol) of
Compound 6 were utilized, resulting in 137.0 mg (92.3%) of an off white amorphous solid (Compound 14).
Melibiose (Compound 15): 47.7 mg (0.204 mmol) of Compound 8 plus 73.4 mg (0.204 mmol) of Compound
7 were utilized, resulting in 91.2 mg (80.1%) of an off white amorphous solid (Compound 15).
4.2.2. General Procedure for the Large-scale (≥0.800 mmol) Synthesis of Sugar-linkers:
Compounds 9–15 were synthesized using 1 equivalent of the aminooxy linker (Compound 8), [16]
mass. Next, 1 equivalent of: N-acetylglucosamine, cellobiose, gentiobiose, lactose, maltose, maltotriose
or melibiose was separately added to each flask containing Compound 8. These were dissolved in 5.0
mL of 0.1 M ammonium acetate (NH4 OAc) at a pH of 4.5, followed by 45.6 µL of aniline (0.1 M final
concentration). The pH was checked after aniline addition to confirm it was still 4.5. The reactions were
conducted at 400 W in a microwave (CEM MARS 5) at 25% power with a 2 min ramp to temperature
and a hold time of 30 min at a maximum temperature of 50 ◦ C. After the reaction was complete, the
solutions were freeze-dried. The products were then purified by flash chromatography in 6:4:0.5
CHCl3 :MeOH:H2 O followed by dialysis in 100 molecular weight cutoff (MWCO) tubing against water
(for gentiobiose, maltose and maltotriose only), yielding white to off-white solids.
N-acetylglucosamine (Compound 9): 213.5 mg (0.912 mmol) of Compound 8 plus 201.6 mg (0.912
mmol) of Compound 1 were utilized, resulting in 315.8 mg (79.2%) of a white amorphous solid
(Compound 9).
Cellobiose (Compound 10): 186.8 mg (0.798 mmol) of Compound 8 plus 273.0 mg (0.798 mmol) of
Compound 2 were utilized, resulting in 333.6 mg (74.9%) of a white solid (Compound 10).
Gentiobiose (Compound 11): 205.1 mg (0.876 mmol) of Compound 8 plus 299.8 mg (0.876 mmol) of
Compound 3 were utilized, resulting in 319.4 mg (65.3%) of an off-white solid (Compound 11).
Lactose (Compound 12): 196.8 mg (0.841 mmol) of Compound 8 plus 303.0 mg (0.841 mmol) of
Maltose (Compound 13): 203.2 mg (0.868 mmol) of Compound 8 plus 312.9 mg (0.868 mmol) of
Compound 5 were utilized, resulting in 299.8 mg (61.9%) of a white powdery solid (Compound 13).
Maltotriose (Compound 14): 208.0 mg (0.889 mmol) of Compound 8 plus 448.4 mg (0.889 mmol) of
Melibiose (Compound 15): 193.8 mg (0.828 mmol) of Compound 8 plus 298.4 mg (0.828 mmol) of
Compound 7 were utilized, resulting in 275.9 mg (59.7%) of an off-white powdery solid (Compound 15).
4.3. Synthesis of Trivalent Glycoclusters

(Cellobiose)3 -Glycocluster, Compound 17: Compound 16 [16] was transferred to a round-bottomed
flask as a 10 mg/mL solution in methanol and then evaporated under reduced pressure to give 9.9 mg
(0.0183 mmol) of Compound 16 as an oil. Next, 3 equivalents of Compound 2 (18.8 mg, 0.0549 mmol)
were added to the reaction flask. The solutes were dissolved in 1.5 mL of 0.1 M NH4 OAc buffer, pH
4.5, plus 13.7 µL aniline (0.1 M final concentration) as a catalyst. The pH was confirmed to be 4.5 after
aniline addition. The reaction was conducted at 400 W in a microwave (CEM MARS 5) at 25% power
for 30 min. After the reaction was complete, the solution was freeze-dried then purified by SEC as
described in general methods, yielding Compound 17 (26 mg, 93.9%) as an off white solid. 1 H NMR
(500 MHz, D2 O): δ 7.67 (d, J = 5.5 Hz, 0.5 H, E isomer), 7.00 (d, J = 5.5 Hz, 0.1H, Z isomer), 4.61–4.51 (m,
1.4H), 4.31–4.08 (m, 2H), 3.99–3.89 (m, 4H), 3.87-3.66 (m, 7H), 3.65–3.50 (m, 1H), 3.49–3.39 (m, 3H), 2.70
(br s, 2H), 2.53 (app t, 2H). 13 C NMR (125 MHz, D2 O with internal MeOH standard): δ 174.39, 152.17,
102.83, 90.39, 78.72, 78.48, 76.30, 76.10, 75.87, 75.81, 75.58, 73.63, 73.59, 73.47, 72.92, 71.42, 69.76, 69.70,
69.63, 69.02, 68.81, 67.09, 66.97, 62.37, 60.89, 60.77, 52.70, 36.27. MALDI-MS: Calc for C57 H106 N7 O39 (M
+ H)+ : 1512.652. Found: 1512.687.
(Sialic Acid)3 -Glycocluster, Compound 20: Compound 16 [16] was transferred to a pear shaped flask
as a 10 mg/mL solution in methanol, then evaporated, yielding 10.8 mg (0.02 mmol) of the trivalent
core. Next, 3 equivalents of sialic acid (Compound 18, Nacalai Tesque, 18.6 mg, 0.06 mmol) were
weighed into the flask. The solutes were then dissolved in 1.5 mL of 0.1 M NH4 OAc, pH 4.5, plus
13.7 µL aniline (0.1 M final concentration) as a catalyst. The pH was confirmed to be 4.5 after aniline
addition. The reaction was conducted at 400 W in a microwave (CEM MARS 5) at 25% power for 30
min. After the reaction was complete, the solution was freeze-dried then purified by SEC as described
in general methods, yielding Compound 20 (23 mg, 81.6%) of an off white solid. 1 H NMR (500 MHz,
D2 O): δ 4.41 (m, 1H), 4.28 (app t, 1.5H), 4.15 (app t, 0.5H), 4.01-3.91 (m, 2H), 3.85-3.73 (m, 6H), 3.65-3.56
(m, 3H), 3.47-3.45 (app d, 1H), 3.39-3.36 (m, 2H), 2.82-2.70 (m, 1.5H), 2.58 (app t, 2H), 2.47 (d, J = 6.8 Hz,
0.5H), 2.08 (s, 3H). 13 C NMR (125 MHz, D2 O with internal MeOH standard): δ 175.10, 175.07, 174.50,
170.44, 170.05, 157.22, 156.15, 73.33, 72.65, 70.72, 69.50, 69.45, 69.02, 68.90, 67.90, 67.86, 66.71, 66.55,
65.97, 63.35, 53.94, 53.56, 53.49, 53.39, 35.90, 35.83, 34.70, 34.64, 30.94, 22.05, 22.01. MALDI-MS: Calc. for
C54 H96 N10 O33 (M + H)+ : 1413.62134. Found: 1413.799.
(Disialic Acid)3 -Glycocluster, Compound 21: Compound 16 [16] was transferred to a pear shaped
flask as a 10 mg/mL solution in methanol, then evaporated, yielding 5.9 mg (0.011 mmol) of the
trivalent core as an oil. Next, the α-2→8 linked dimer of sialic acid (Compound 19, disialic acid, 21.1
mg, 0.033 mmol) [35] was added. The solutes were then dissolved in 1.5 mL of 0.1 M NH4 OAc, pH 4.5,
plus 13.7 µL of aniline (0.1 M final concentration) as a catalyst. The pH was confirmed to be 4.5 after
aniline addition. The reaction was conducted at 400 W in a microwave (CEM MARS 5) at 25% power
for 90 min with a temperature maximum set at 50 ◦ C. After the reaction was complete, the solution
was freeze-dried then purified by FPLC as described in general methods, yielding Compound 21 (22
mg, 87.6%) as an off white solid. 1 H NMR (500 MHz, D2 O): δ 4.46-4.40 (m, 1H), 4.29 (app t, 1.5H), 4.15
(app t, 0.5H), 3.99-3.91 (m, 4H), 3.89-3.74 (m, 9.5H), 3.70-3.61 (m, 4H), 3.57-3.53 (m, 3H), 3.26 (br s, 2H),
2.83-2.67 (m, 2.5H), 2.57 (t, J = 5.9 Hz, 2H), 2.47 (m, 0.5H), 2.09 (s, 3H), 2.03 (s, 3H), 1.78 (t, J = 12.2 Hz,
1H). 13 C NMR (125 MHz, D2 O with internal MeOH standard): δ 175.20, 174.89, 174.36, 173.47, 101.84,
74.30, 74.24, 72.71, 72.51, 71.80, 71.73, 68.96, 68.77, 68.21, 68.05, 67.75, 67.70, 67.49, 67.45, 66.57, 66.28,
65.79, 62.66, 61.14, 53.72, 53.40, 53.36, 53.31, 51.65, 40.05, 35.79, 35.75, 35.68, 34.42, 30.82, 21.97, 21.95.
MALDI-MS: Calc. for C87 H147 N13 O57 (M + H)+ : 2286.90755. Found: 2286.791.
Supplementary Materials: The following are available online at http://www.mdpi.com/1424-8247/12/1/39/s1,

Figures S1–S20: 1 H (odd) and 13 C (even) NMR spectra, Table S1: Summary of the E/Z ratios and % ring open
oxime for all products.
Author Contributions: Conceptualization, K.M.; methodology, K.M., D.D., G.F.; validation, K.M., D.D., G.F.,
K.Z.; formal analysis, K.M., D.D., G.F., K.Z.; investigation, K.M., D.D., G.F., K.Z.; resources, K.M.; data curation,
K.M.; writing—original draft preparation, K.M., D.D., G.F.; writing—review and editing, K.M., D.D., G.F., K.Z.;
visualization, K.M., D.D., G.F., K.Z.; supervision, K.M.; project administration, K.M.; funding acquisition, K.M.
Funding: This work was financially supported by a Research Development grant from California State University
Program for Research and Education in Biotechnology (CSUPERB).
Acknowledgments: The authors would like to thank Lucia Gwarada and Juan Gonzalez for their technical
assistance. Mass spectrometry was carried out at Campus Chemical Instrument Center (CCIC) Mass Spectrometry
and Proteomics Facility at The Ohio State University (OSU) (NIH grants: P30CA016058, 1S10RR025660-01A1,
1S10OD018507).
Conflicts of Interest: The authors declare no conflict of interest. The funders had no role in the design of the
study; in the collection, analyses or interpretation of data; in the writing of the manuscript or in the decision to
publish the results.
References
1. Kottari, N.; Chabre, Y.M.; Sharma, R.; Roy, R. Applications of glyconanoparticles as “sweet” glycobiological
therapeutics and diagnostics. Adv. Polym. Sci. 2013, 254, 297–342.
2. Mellet, C.O.; Mendez-Ardoy, A.; Ferenandez, J.M.G. Click multivalent glycomaterials: Glycoclusters,
glycodendrimers, glycopolymers, hybrid glycomaterials and glycosurfaces. In Click Chemistry in Glycoscience:
New Developments and Strategies, 1st ed.; Witczak, Z.J., Bielski, R., Eds.; John Wiley and Sons, Inc.: Hoboken,
NJ, USA, 2013; pp. 143–182.
3. Adak, A.K.; Yu, C.-C.; Lin, C.-C. Synthesis and applications of glyconanoparticles, glycodendrimers, and
glycoclusters in biological systems. In Glycochemical Synthesis: Strategies and Applications; Hung, S.-C.,
Zulueta, M.M.L., Eds.; John Wiley and Sons, Inc.: Hoboken, NJ, USA, 2016; pp. 425–454.
4. Jebali, A.; Nayeri, E.K.; Roohana, S.; Aghaei, S.; Ghaffari, M.; Daliri, K.; Fuente, G. Nano-carbohydrates:
Synthesis and application in genetics, biotechnology, and medicine. Adv. Colloid Interface 2017, 240, 1–14.
[CrossRef] [PubMed]
5. Ramella, D.; Polito, L.; Mazzini, S.; Ronchi, S.; Scaglioni, L.; Marelli, M.; Lay, L. A strategy for multivalent
presentation of carba analogues from N. meningitidis a capsular polysaccharide. Eur. J. Org. Chem. 2014, 2014,
5915–5924. [CrossRef]
6. Lee, Y.-C.; Lee, R.T. Carbohydrate-protein interactions: Basis of glycobiology. Acc. Chem. Res. 1995, 28,
7. Clayton, R.; Hardman, J.; LaBranche, C.C.; McReynolds, K.D. Evaluation of the synthesis of sialic
acid-PAMAM glycodendrimers without the use of sugar protecting groups, and the anti-HIV-1 properties of
these compounds. Bioconj. Chem. 2011, 22, 2186–2197. [CrossRef] [PubMed]
8. Cecioni, S.; Praly, J.-P.; Matthews, S.E.; Wimmerova, M.; Imberty, A.; Vidal, S. Rational design and synthesis
of optimized glycoclusters for multivalent lectin-carbohydrate interactions: Influence of the linker arm.
Chem. Eur. J. 2012, 18, 6250–6263. [CrossRef]
9. Yang, Z.-L.; Zeng, X.-F.; Liu, H.-P.; Yu, Q.; Meng, X.; Yan, Z.-L.; Fan, Z.-C.; Xiao, S.S.; Yang, Y.; Yu, P. Synthesis
of multivalent difluorinated zanamivir analogs as potent antiviral inhibitors. Tetrahedron Lett. 2016, 57,
2579–2582. [CrossRef]
10. Bagul, R.S.; Hosseini, M.; Shiao, T.C.; Saadeh, N.K.; Roy, R. Heterolayered hybrid dendrimers with optimized
sugar head groups for enhancing carbohydrate-protein interactions. Polym. Chem. 2017, 8, 5354–5366.
[CrossRef]
11. Villadson, K.; Martos-Maldonado, M.C.; Jensen, K.J.; Thygesen, M.B. Chemoselective reactions for the
synthesis of glycoconjugates from unprotected carbohydrates. ChemBioChem 2017, 18, 574–612. [CrossRef]
12. Kolmel, D.K.; Kool, E.T. Oximes and hydrazones in bioconjugation: Mechanism and catalysis. Chem. Rev.
2017, 117, 10358–10376. [CrossRef]
13. Kalia, J.; Raines, R.T. Hydrolytic stability of hydrazones and oximes. Angew. Chem. Int. Ed. Engl. 2008, 47,
7523–7526. [CrossRef]
14. Gudmundsdottir, A.V.; Paul, C.E.; Nitz, M. Stability studies of hydrazide and hydroxylamine-based
glycoconjugates in aqueous solution. Carbohydr. Res. 2009, 344, 278–284. [CrossRef]
15. Iqbal, A.; Chibli, H.; Hamilton, C.J. Stability of aminooxy glycosides to glycosidase catalyzed hydrolysis.
Carbohydr. Res. 2013, 377, 1–3. [CrossRef]
16. McReynolds, K.D.; Dimas, D.; Le, H. Synthesis of hydrophilic aminooxy linkers and multivalent cores for
chemoselective aldehyde/ketone conjugation. Tetrahedron Lett. 2014, 55, 2270–2273. [CrossRef]
17. Chen, N.; Xie, J. N-O linkage in carbohydrates and glycoconjugates. Org. Biomol. Chem. 2016, 14, 11028–11047.
[CrossRef]
18. Pifferi, C.; Daskhan, G.C.; Fiore, M.; Shiao, T.C.; Roy, R. Aminooxylated carbohydrates: Synthesis and
applications. Chem. Rev. 2017, 117, 9839–9873. [CrossRef]
19. Thygesen, M.B.; Sauer, J.; Jensen, K.J. Chemoselective capture of glycans for analysis on gold nanoparticles:
Carbohydrate oxime tautomers provide functional recognition by proteins. Chem. Eur. J. 2009, 15, 1649–1660.
[CrossRef]
20. Rathi, A.K.; Gawande, M.B.; Zboril, R.; Varma, R.S. Microwave-assisted synthesis-catalytic applications in
aqueous media. Coord. Chem. Rev. 2015, 291, 68–94. [CrossRef]
21. Corsaro, A.; Pistaria, V.; Chiacchio, M.A.; Romeo, G. A journey into recent microwave-assisted carbohydrate
chemistry. In Microwaves in Organic Synthesis, 3rd ed.; de la Hoz, A., Loupy, A., Eds.; Wiley-VCH Verlag
GmbH and Co.: Weinheim, Germany, 2012; pp. 961–1011.
22. Joosten, J.A.F.; Tholen, N.T.H.; El Maate, F.A.; Brouwer, A.J.; van Esse, G.W.; Rijkers, D.T.S.;
Liskamp, R.M.J.; Pieters, R.J. High-yielding microwave-assisted synthesis of triazole-linked glycodendrimers
by copper-catalyzed [3+2] cycloaddition. Eur. J. Org. Chem. 2005, 2005, 3182–3185. [CrossRef]
23. Brun, M.A.; Disney, M.D.; Seeberger, P.H. Miniaturization of microwave assisted carbohydrate
functionalization to create oligosaccharide microarrays. ChemBioChem 2006, 7, 421–424. [CrossRef]
24. Cecioni, S.; Faure, S.; Darbost, U.; Bonnamour, I.; Parrot-Lopez, H.; Roy, O.; Taillefumier, C.; Wimmerova, M.;
Praly, J.-P.; Imberty, A.; et al. Selectivity among two lectins: Probing the effect of topology, multivalency and
flexibility of “clicked” multivalent glycoclusters. Chem. Eur. J. 2011, 17, 2146–2159. [CrossRef]
25. Cagnoni, A.J.; Varela, O.; Gouin, S.G.; Kovensky, J.; Uhrig, M.L. Synthesis of multivalent glycoclusters from
1-thio-β-D-galactose and their inhibitory activity agains the β-galactosidase from E. coli. J. Org. Chem. 2011,
76, 3064–3077. [CrossRef]
26. Cecioni, S.; Matthews, S.E.; Blanchard, H.; Praly, J.-P.; Imberty, A.; Vidal, S. Synthesis of lactosylated
glycoclusters and inhibition studies with plant and human lectins. Carbohydr. Res. 2012, 356, 132–141.
[CrossRef]
27. Cagnoni, A.J.; Kovensky, J.; Uhrig, M.L. Design and synthesis of hydrolytically stable multivalent ligands
bearing thiodigalactoside analogues for peanut lectin and human galectin-3 binding. J. Org. Chem. 2014, 79,
6456–6467. [CrossRef]
28. Ding, F.; Ji, L.; William, R.; Chai, H.; Liu, X.-W. Design and synthesis of multivalent neoglycoconjugates by
click conjugations. Beilstein J. Org. Chem. 2014, 10, 1325–1332. [CrossRef]
29. Chuang, Y.-J.; Zhou, X.; Pan, Z.; Turchi, C. A convenient method for synthesis of glyconanoparticles for
calorimetric measuring carbohydrate-protein interaction. Biochem. Biophys. Res. Commun. 2009, 389, 22–27.
[CrossRef]
30. Seo, J.; Michaelian, N.; Owens, S.C.; Dashner, S.T.; Wong, A.J.; Barron, A.E.; Carrasco, M.R. Chemoselective
and microwave-assisted synthesis of glycopeptoids. Org. Lett. 2009, 11, 5210–5213. [CrossRef]
31. Thygesen, M.B.; Munch, H.; Sauer, J.; Cló, E.; Jorgensen, M.R.; Hindsgaul, O.; Jensen, K.J. Nucleophilic
catalysis of carbohydrate oxime formation by anilines. J. Org. Chem. 2010, 75, 1752–1755. [CrossRef]
32. Ray, G.J.; Siekman, J.; Scheinecker, R.; Zhang, Z.; Gerasimov, M.V.; Szabo, C.M.; Kosma, P. Reaction of
oxidized polysialic acid and a diaminooxy linker: Characterization and process optimization using nuclear
magnetic resonance spectroscopy. Bioconj. Chem. 2016, 27, 2071–2080. [CrossRef]
33. Berthet, N.; Thomas, B.; Bossu, I.; DuFour, E.; Gillon, E.; Garcia, J.; Spinelli, N.; Imberty, A.; Dumy, P.;
Renaudet, O. High affinity glycodendrimers for the lectin LecB from Pseudomonas aeruginosa. Bioconj. Chem.
2013, 24, 1598–1611. [CrossRef]
34. Dirksen, A.; Dawson, P.E. Rapid oxime and hydrazone ligations with aromatic aldehydes for biomolecular
labeling. Bioconj. Chem. 2008, 19, 2543–2548. [CrossRef]
35. Patane, J.; Trapani, V.; Villavert, J.; McReynolds, K.D. Preparative production of colominic acid oligomers via
a facile microwave hydrolysis. Carbohydr. Res. 2009, 344, 820–824. [CrossRef]
36. Sato, C.; Kitajima, K. Disialic, oligosialic and polysialic acids: Distribution, functions and related disease. J.
Biochem. 2013, 154, 115–136. [CrossRef]
37. Bhide, G.P.; Colley, K.J. Sialylation of N-glycans: Mechanism, cellular compartmentalization and function.
Histochem. Cell Biol. 2017, 147, 149–174. [CrossRef]
pharmaceuticals
Article
Physical Stability and Viscoelastic Properties of
Co-Amorphous Ezetimibe/Simvastatin System
Justyna Knapik-Kowalczuk 1, * , Krzysztof Chmiel 1 , Karolina Jurkiewicz 1 ,
Natália T. Correia 2 , Wiesław Sawicki 3 and Marian Paluch 1
1 Institute of Physics, University of Silesia, SMCEBI, 75 Pułku Piechoty 1a, 41-500 Chorzów, Poland;
krzysztof.chmiel@smcebi.edu.pl (K.C.); karolina.jurkiewicz@us.edu.pl (K.J.);
marian.paluch@us.edu.pl (M.P.)
2 Univ Lille, CNRS, UMR 8207, UMET, Unité Matériaux et Transformations, F59000 Villeneuve d’Ascq, France;
natalia.correia@univ-lille.fr
3 Department of Physical Chemistry, Medical University of Gdansk, 84-416 Gdansk, Poland;
wsawicki@gumed.edu.pl
* Correspondence: justyna.knapik-kowalczuk@us.edu.pl; Tel.: +48-32-3497629

Received: 5 March 2019; Accepted: 16 March 2019; Published: 19 March 2019
Abstract: The purpose of this paper is to examine the physical stability as well as viscoelastic
properties of the binary amorphous ezetimibe–simvastatin system. According to our knowledge,
this is the first time that such an amorphous composition is prepared and investigated. The tendency
toward re-crystallization of the amorphous ezetimibe–simvastatin system, at both standard storage
and elevated temperature conditions, have been studied by means of X-ray diffraction (XRD).
Our investigations have revealed that simvastatin remarkably improves the physical stability of
ezetimibe, despite the fact that it works as a plasticizer. Pure amorphous ezetimibe, when stored
at room temperature, begins to re-crystallize after 14 days after amorphization. On the other
hand, the ezetimibe-simvastatin binary mixture (at the same storage conditions) is physically
stable for at least 1 year. However, the devitrification of the binary amorphous composition was
observed at elevated temperature conditions (T = 373 K). Therefore, we used a third compound to
hinder the re-crystallization. Finally, both the physical stability as well as viscoelastic properties
of the ternary systems containing different concentrations of the latter component have been
thoroughly investigated.
Keywords: ezetimibe; simvastatin; co-amorphous; melt viscosity; solubility enhancement
1. Introduction
Cardiovascular diseases (CVDs) are currently the leading cause of death worldwide for both
men and women [1,2]. Based on data from the Centers for Disease Control and Prevention
(CDC), about 610,000 people in the United States die each year from these diseases. That is
roughly the entire population of Washington, D.C., Vermont, or Wyoming. The cornerstone in
the prevention of CVDs is lowering high blood cholesterol level [3]. Currently, the first-choice
medications for reducing the low-density lipoprotein cholesterol (LDL-C) are statins. They act
by inhibiting 3-hydroxy-3-methylglutaryl coenzyme A (HMG-CoA) reductase, thereby blocking
cholesterol synthesis.
Among all statins, atorvastatin and simvastatin are considered as the most potent [4,5].
This explains why they can be found on the list of TOP 5 most prescribed drugs [6]. It has to be pointed
out that an increase in statins dose offers unfortunately limited LDL-C lowering effect (saturation
effect) while at the same time it creates an increased incidence of side effects. Thus, novel active

pharmaceutical ingredients (APIs) that might reduce LDL-C levels when co-administered with a statin
are of great interest. Ezetimibe is one of such compounds, i.e., a lipid-lowering agent of a new type.
In contrast to statins, which suppress the cholesterol synthesis, ezetimibe inhibits intestinal and biliary
cholesterol absorption by selectively blocking the Niemann–Pick C1- like 1 (NPC1L1) protein. It has
been proven that co-administration of this API with simvastatin is much more effective in reducing
mean plasma concentrations of LDL-C than ezetimibe, simvastatin, and also other statins alone [7,8].
As H. E. Bays et al. indicated, the similar reduction in plasma LDL-C level can be achieved with
the co-administration of ezetimibe 10 mg plus simvastatin 10 mg as with simvastatin 80 mg alone.
In addition, the fixed-dose combination therapy of ezetimibe and simvastatin is well-tolerated by
patients [9]. The efficacy and safety of the combined use of both aforementioned APIs have resulted
in the appearance of such a combo-product on the market. It is available for oral use as tablets
containing 10 mg of ezetimibe and 10, 20, 40, or 80 mg of simvastatin. It is worth mentioning that
the commercial forms of ezetimibe and simvastatin exhibit low oral bioavailability, attributed to its
poor water solubility. Ezetimibe, which has water solubility is equal to 8.46 ng/L, is characterized by
a bioavailability of 35% [10,11]. Simvastatin’s solubility and bioavailability are equal to 1.45 mg/L
and 5%, respectively [12]. Since the permeability of both these drugs is high, they are classified to the
second class of the Biopharmaceutics Classification System (BCS).
There are several methods that can be employed to increase the solubility and consequently also
the bioavailability of APIs from BCS class II [13–15]. One of them is based on the preparation of the
amorphous form of a chosen compound [16–20]. In the literature, one can find a lot of examples
indicating that amorphization remarkably improves the water solubility and bioavailability of poorly
soluble drugs [21–25]. It has to be noted that amorphous APIs have unfortunately one disadvantage,
which blocks their widespread use [26–30]. Mainly, these APIs are physically unstable. It means that
during the time of storage or manufacturing they tend to revert back to their crystalline form [31–35].
In this paper, the binary amorphous composition of ezetimibe and simvastatin in the weight
ratio of 1:1 and ternary amorphous compositions containing ezetimibe 1:1 simvastatin and 5, 20, 40
and 60 wt. % of co-polymer Kollidon VA64 were prepared and thoroughly investigated. Thermal
properties of both the pure components and the obtained systems were investigated by means of
differential scanning calorimetry (DSC). Because the main limitation of widespread use of amorphous
pharmaceuticals is physical instability, we thoroughly examined the tendency toward re-crystallization
of the investigated system at both standard storage and elevated temperature conditions. The physical
stability studies of ezetimibe–simvastatin system were performed by means of X-ray diffraction
technique (XRD). Finally, the oscillatory shear rheology was performed to measure the viscoelastic
properties of all investigated binary and ternary systems. Based on the obtained rheological data,
we plotted the temperature dependences of complex viscosity (η*(T)) of all investigated systems,
which can be used for quick determination of the appropriate concentration, temperature, and viscosity
for the chosen production method of the ezetimibe–simvastatin system.
2.1. Thermal Properties of Ezetimibe, Simvastatin And Its Binary System

Figure 1 shows the DSC thermograms obtained during heating of the crystalline ezetimibe
(EZB), crystalline simvastatin (SVT), and their physical mixture (EZB/SVT). As can be seen, the DSC
traces of pure components reveal only one endothermic peak corresponding to the samples melting.
The melting onset of EZB is equal to 436 K, while SVT’s Tm = 412 K. Both these values are in good
agreement with the literature data [36,37]. For the physical mixture of EZB/SVT, two thermal effects
are observed on the DSC heating curve. The first with the onset at 391 K is related to solidus (eutectic)
temperature. The second thermal event with the maximum of the peak at a temperature equal to 406 K
and a characteristic tail shape corresponds to the liquidus.
Figure
Figure 1. Differential scanning
1. Differential scanning calorimetry
calorimetry (DSC)
(DSC) thermograms
thermograms of the pure
of the pure crystalline
crystalline ezetimibe
ezetimibe (EZB;
(EZB;
blue line), pure crystalline simvastatin (SVT; red line), and EZB/SVT physical mixture (green line)
blue line), pure crystalline simvastatin (SVT; red line), and EZB/SVT physical mixture (green line)
obtained
obtained during
during heating
heating at
at 10
10 K/min.
K/min.
It is worth emphasizing that the observed decrease in the melting points of both EZB and SVT
It is worth emphasizing that the observed decrease in the melting points of both EZB and SVT
after mixing these two components might be beneficial from the production of the amorphous APIs
after mixing these two components might be beneficial from the production of the amorphous APIs
point of view. Currently, for the manufacturing of amorphous pharmaceuticals, the hot melt extrusion
point of view. Currently, for the manufacturing of amorphous pharmaceuticals, the hot melt
(HME) method is increasingly employed. During this production process, the crystalline API or
extrusion (HME) method is increasingly employed. During this production process, the crystalline
a drug-polymer composition containing crystalline API is exposed to an elevated temperature in order
API or a drug-polymer composition containing crystalline API is exposed to an elevated temperature
to melt the system. The lower the extrusion temperature, the safer production process, due to the
in order to melt the system. The lower the extrusion temperature, the safer production process, due
decreased probability that the sample will undergo thermal decomposition.
to the decreased probability that the sample will undergo thermal decomposition.
Figure 2 presents the second heating run of EZB, SVT, and EZB/SVT, which were performed
Figure 2 presents the second heating run of EZB, SVT, and EZB/SVT, which were performed
immediately after melting and quenching the samples in DSC with a rate equal to 10 K/min. As can
immediately after melting and quenching the samples in DSC with a rate equal to 10 K/min. As can
be seen, the DSC thermograms of freshly melt-quenched samples reveal only one thermal event with
be seen, the DSC thermograms of freshly melt-quenched samples reveal only one thermal event with
a characteristic step-like behavior corresponding to the glass transition. The Tg midpoints of EZB, SVT,
a characteristic step-like behavior corresponding to the glass transition. The Tg midpoints of EZB,
and EZB/SVT are equal to 337 K, 305 K, and 323 K, respectively.
SVT, and EZB/SVT are equal to 337 K, 305 K, and 323 K, respectively.
Figure
Figure 2. DSC thermograms
2. DSC thermograms of
of quench
quench cooled
cooled pure
pure EZB
EZB(blue
(blue line),
line), pure
pureSVT
SVT(red
(redline),
line),and
andEZB/SVT
EZB/SVT
composition
composition (green
(green line).
line).
Recently, Martinez-Jimenez C. et al. showed that neat amorphous SVT reveals no tendency
toward re-crystallization for at least one year when stored at room temperature conditions [38].
Amorphous EZB, however, quickly reverts to its crystalline form when stored at similar conditions.
Recently, Martinez-Jimenez C. et al. showed that neat amorphous SVT reveals no tendency toward
re-crystallization for at least one year when stored at room temperature conditions [38]. Amorphous
EZB, however, quickly reverts to its crystalline form when stored at similar conditions. The first
The first sign of re-crystallization of amorphous EZB was detected after only 14 days from
sign of re-crystallization of amorphous EZB was detected after only 14 days from vitrification by the
vitrification by the melt-quench method [37]. In light of these facts, the physical stability
melt-quench method [37]. In light of these facts, the physical stability improvement of EZB is needed.
improvement of EZB is needed. By far, the most common approach to suppress the re-crystallization
By far, the most common approach to suppress the re-crystallization tendency of amorphous APIs is
tendency of amorphous APIs is to slow down its molecular mobility (i.e., increase their Tg). This is
to slow down its molecular mobility (i.e., increase their Tg ). This is because the molecular dynamic
because the molecular dynamic is believed to be a key factor governing the physical stability of
is believed to be a key factor governing the physical stability of amorphous materials. In EZB/SVT
amorphous materials. In EZB/SVT case the opposite situation is observed—mainly the molecular
case the opposite situation is observed—mainly the molecular mobility of physically-unstable EZB
mobility of physically-unstable EZB accelerates after the incorporation of SVT. Consequently, one
accelerates after the incorporation of SVT. Consequently, one would speculate that SVT might decrease
would speculate that SVT might decrease the physical stability of an amorphous EZB.
the physical stability of an amorphous EZB.
2.2.
2.2. Physical
Physical Stability
Stability Studies
Studies of
of the
the Amorphous Ezb/Svt System
Amorphous Ezb/Svt System Stored
stored at
at Both
Both Supercooled
Supercooled Liquid
Liquid and
and
Glassy State
Glassy State
In
In order
order to
toassess
assesshow
howthe
theSVT
SVTchanges
changes the physical
the physicalstability of the
stability amorphous
of the amorphous form of EZB,
form the
of EZB,
amorphous EZB/SVT mixture was subjected into two independent XRD experiments.
the amorphous EZB/SVT mixture was subjected into two independent XRD experiments. The idea of The idea of the
first one one
the first waswas
to check the mixture’s
to check physical
the mixture’s stability
physical at standard
stability storage
at standard conditions,
storage i.e., at
conditions, 298
i.e., at K and
298 K
RH
and=RH 60%.= During the second
60%. During the experiment, the sample
second experiment, thewas stored
sample at elevated
was stored attemperature conditions
elevated temperature
(T = 373 K).(TThe
conditions representative
= 373 XRD patterns
K). The representative XRD obtained
patternsduring these
obtained experiments
during are presented,
these experiments are
together with the diffractograms of the crystalline EZB, SVT, and EZB/SVT, in Figure
presented, together with the diffractograms of the crystalline EZB, SVT, and EZB/SVT, in Figure 3a–c. It is worth
3a–c.
mentioning that the XRD
It is worth mentioning thatpatterns
the XRDofpatterns
pure crystalline APIs are APIs
of pure crystalline in agreement with previous
are in agreement results
with previous
[39,38]. The samples measured immediately after quenching, as expected, were fully amorphous.
results [38,39]. The samples measured immediately after quenching, as expected, were fully amorphous. The
-sharp
The Bragg’s
-sharp peaks
Bragg’s are not
peaks are visible on their
not visible XRDXRD
on their patterns.
patterns.
Figure 3.
Figure XRD patterns
3. XRD patterns of
of (a)
(a) crystalline
crystalline systems
systems stored
stored at
at 298
298 K
K (red
(red line—pure
line—pure SVT;
SVT; blue line—pure
blue line—pure
EZB; green line—EZB/SVT physical mixture), (b) amorphous EZB/SVT system
EZB; green line—EZB/SVT physical mixture), (b) amorphous EZB/SVT system stored at 298 stored at 298 K,
K, (c)
(c) amorphous EZB/SVT system stored
amorphous EZB/SVT system stored at 373 K. at 373 K.
The amorphous
The amorphous form
form of of EZB/SVT composition stored
EZB/SVT composition stored at
at standard
standard storage
storage conditions
conditions waswas
measured by means of XRD at the beginning once daily for a week, later once a week
measured by means of XRD at the beginning once daily for a week, later once a week for a month, for a month,
and then
and then once
once aa month
month until
until 11 year.
year. As
As can
can be
be clearly
clearly seen
seen even
even after
after 11 year,
year,the
theexamined
examinedEZB/SVT
EZB/SVT
system did not reveal any sign of re-crystallization. Comparing this
system did not reveal any sign of re-crystallization. Comparing this result with theresult with the previously
previously
published
published by byKnapik
Knapiketetal.,
al.,data
dataforfor pure
pure EZB,
EZB, oneone
cancan conclude
conclude that that
SVT SVT impressively
impressively stabilizes
stabilizes EZB.
EZB. It should be emphasized that herein the improvement of physical stability was
It should be emphasized that herein the improvement of physical stability was achieved despite achieved despite
acceleration the
acceleration EZB’s molecular
the EZB’s molecular mobility.
mobility. The
The similar
similar situation
situation was previously observed
was previously observed only
only inin two
two
cases: Nifedipine + nimodipine and simvastatin + nifedipine [39,40]. Such a stabilization effect was
explained by the existence of some specific interactions between both drugs. Accordingly, one can
expect that, in the EZB/SVT system, the source of the observed stabilization is analogous to the
aforementioned cases.
cases: Nifedipine + nimodipine and simvastatin + nifedipine [39,40]. Such a stabilization effect was
explained by the existence of some specific interactions between both drugs. Accordingly, one can
expect that, in the EZB/SVT system, the source of the observed stabilization is analogous to the
aforementioned cases.
During the second XRD experiment, in which the sample was exposed to the elevated temperature
conditions (T = 373 K), the XRD patterns were collected every 30 min for 30 h. The representative
diffraction patterns are presented in Figure 3c. At this temperature, the first appearance of sharp
Bragg’s peaks was noticed after only 210 min. This result indicates that even if the amorphous
EZB/SVT system reveals high physical stability when stored below the glass transition temperature,
it tends to re-crystallize at a supercooled liquid state. Such a situation is acceptable if for sample
manufacturing one would choose a method which does not require sample heating (ex. Spray Drying).
It, however, might not be sufficient in cases of production methods involving thermal processing,
such as hot melt extrusion or 3D printing.
2.3. Impact of KVA 64 Polymer on the Thermal Properties as Well as Physical Stability of the EZB/SVT System
Kollidon VA64 (KVA 64) is a vinylpyrrolidone—a vinyl acetate copolymer. It is amorphous in
nature, with a glass transition temperature equal to 378 K and degradation temperature ~500 K [40].
This polymer is well soluble in water and has good processability, and therefore it is commonly used as
an excipient for manufacturing of amorphous pharmaceutical products using either solvent (ex. Spray
Drying (SD)) or melting (ex. Hot Melt Extrusion (HME)) methods. In such formulations, usually,
KVA64 plays a dual role. On the one hand it increases the dissolution rate of the APIs, and on the other
hand, it well stabilizes them.
In this section, the impact of KVA64 on the thermal properties and physical stability of
a co-amorphous composition containing EZB and SVT is described. Figure 4a presents DSC
thermograms of the EZB/SVT physical mixture, neat (as received) KVA 64 polymer, as well as four
ternary physical mixtures of EZB, SVT, and KVA in the concentrations: (i) EZB/SVT + 5wt. % KVA,
(ii) EZB/SVT + 20wt. % KVA, (iii) EZB/SVT + 40wt. % KVA and (iv) EZB/SVT + 60wt. % KVA. As it
has been shown in the first section, the DSC curve of the EZB/SVT crystalline system reveals two
thermal events. The first one associated with the eutectic, and the second one reflecting the liquidus.
DSC thermogram of the first run of KVA64 reveals only one endothermic event visible in the vicinity of
320–380 K that corresponds to water evaporation. This broad endothermic peak covers the polymer’s
glass transition that can be easily noticed at a second DSC heating run (see the black curve in Figure 4b),
or (ii) first DSC heating run obtained after sample heating at 373 K for 15 min (data not shown). In the
DSC traces of ternary systems (i) and (ii), three thermal endotherms might be noted. First, a very
broad thermal event reflecting water evaporation can be found in the vicinity of 320–350 K. A second,
with an onset in the neighborhood of 390 K, is connected with the eutectic transition. The third one is
visible around 410 K and possesses a characteristic tail shape associated with the liquids phase. On the
thermograms of systems (iii) and (iv) the latter endothermic peak becomes invisible, indicating that
the ternary eutectic EZB/SVT/KVA was obtained.
Melts of the examined ternary systems have been quenched in DSC with a rate equal to 10 K/min
to obtain the amorphous compositions. The glassy samples were subsequently reheated up to T = 440 K
while obtained thermograms and are presented in Figure 4b. As can be seen, the amorphous mixtures
containing EZB, SVT, and KVA are characterized by a single glass transition event which shifts toward
a higher temperature with increasing co-polymer content. The presence of a single Tg in binary or
ternary compositions usually indicates that these systems are homogenous [41]. If mixtures are not
or are only partially miscible, the DSC trace of an amorphous system should reveal multiple Tg s.
These can be due to the presence of various drug or polymer rich domains [29]. The values obtained
by means of DSC technique, Te , Tl , and Tg , for all examined samples are presented in Figure 4 and are
collected in Table 1.
Figure 4.4.DSCDSC
Figure thermograms
thermograms of (a) crystalline
of a) crystalline and (b) amorphous
and b) amorphous EZB/SVT
EZB/SVT (green lines),(green lines),
EZB/SVT +
EZB/SVT + 5wt. % (Kollidon VA64 (KVA) (blue lines), EZB/SVT + 20wt. % KVA
5wt. % (Kollidon VA64 (KVA) (blue lines), EZB/SVT + 20wt. % KVA (red lines), EZB/SVT + 40wt. % (red lines),
EZB/SVT
KVA (yellow+ 40wt. %EZB/SVT
lines), KVA (yellow lines),
+ 60wt. % EZB/SVT + 60wt.
KVA (orange % and
lines) KVAKVA
(orange lines)
(black and KVA (black lines).
lines).
Mixing of the multi-components can either be ideal and non-ideal [41–43]. The deviation from the
Mixing of the multi-components can either be ideal and non-ideal [41–43]. The deviation from
ideal mixing in amorphous binary, ternary, or quaternary systems might be examined by comparing
the ideal mixing in amorphous binary, ternary, or quaternary systems might be examined by
the experimental glass transition temperature with the T predicted from the Couchman-Karasz (CK)
comparing the experimental glass transition temperatureg with the Tg predicted from the Couchman-
equation given as follows [44,45]:
Karasz (CK) equation given as follows [44,45]:
∑ ∆C pi xi Tgi
Tg = ∑ ∆𝐶 𝑥 𝑇 (1)
∑ ∆C pi xi
𝑇 = (1)
∑∆𝐶 𝑥
where xi is a mass fraction of component i, ∆Cpi refers to the change in heat capacity of component i
where
betweenxi is
itsaliquid-like
mass fractionand of component
glassy i, ΔC
states, and refers toathe
Tgipi denotes change
glass in heat
transition capacity of
temperature ofcomponent
component ii.
between its liquid-like
The similarity and glassy
in predicted states, and Tgi determined
and experimentally denotes a glass transition
Tg values temperature
strongly suggestsofideal
component
mixing,
i.while
The positive
similarity andin negative
predicted and experimentally
deviation indicates thatdetermined
the mixing is Tg non-ideal
values strongly
[41]. Assuggests ideal
can be seen in
mixing, while positive and negative deviation indicates that the mixing is non-ideal
Figure 5 and Table 1, the predicted and experimentally obtained Tg values are in very good agreement [41]. As can be
seen in Figure
indicating that 5there
and is
Table
good1,molecular
the predicted and experimentally
miscibility between all threeobtained Tg values
components, are in
as well as very good
an absence
agreement indicating that
of the intermolecular there is between
interactions good molecular
them. miscibility between all three components, as well
as an absence of the intermolecular interactions between them.
Table 1. Comparison of Te , Tl , Tg exp , and Tg pred values of pure KVA, the binary mixture of EZB/SVT,
as well as the ternary system of EZB/SVT/KVA containing 5, 20, 40 and 60 wt. % of KVA.
System Te [K] Tl [K] Tg exp [K] Tg pred [K]

EZB/SVT 391 406 323 323
EZB/SVT + 5wt.% KVA 390 406 324 324
EZB/SVT + 20wt.% KVA 389 401 332 332
EZB/SVT + 40wt.% KVA 388 - 345 342
EZB/SVT + 60wt.% KVA 388 - 355 353
KVA - - 378 -
Figure 5. Variation of glass transition temperature in the ternary phase diagram of EZB/SVT/KVA.
Figure 5. Variation of glass transition temperature in the ternary phase diagram of EZB/SVT/KVA.
To determineTable how the polymeric

1. Comparison additive
of Te, Tl, Tg exp , and Tg predinflux
values ofeffects
pure KVA, thethephysical stability
binary mixture of EZB/SVT
of EZB/SVT, as system,
well as the ternary
the ternary amorphous systemsystem of EZB/SVT/KVA
containing containing
the lowest 5, 20, 40 and
amount 60 wt.
of the % of KVA.
co-polymer has been investigated
by means of XRD. The short-termSystem XRD experiment e T [K] Tof
l [K]EZB/SVT
T
g exp [K] T + 5wt.
g pred [K] % KVA was performed at
EZB/SVT 391 406 323 323
elevated temperature conditions (T +=5wt.%
EZB/SVT 373KVA K). During390 these studies,
406 324 the mixture’s XRD patterns were
324
collected every 30 min for 60 h.EZB/SVT The obtained
+ 20wt.% KVA results 389 are401presented 332 in Figure
332 6. The XRD patterns of both
EZB/SVT + 40wt.% KVA 388 - 345 342
freshly vitrified as well as stored at elevated
EZB/SVT + 60wt.% KVA temperature
388 - conditions
355 for
353 60 h EZB/SVT + 5wt. % KVA
KVA - - 378 -
mixture reveal a lack of Bragg’s peaks. This indicates that the sample remained amorphous throughout
the duration of Tothe experiment.
determine Comparing
how the polymeric thisinflux
additive result with
effects thethe datastability
physical presented in Figure
of EZB/SVT system,3c, one can
conclude that KVA significantly
the ternary amorphous system improves
containing thethephysical stability
lowest amount of theof the amorphous
co-polymer EZB/SVT system at
has been investigated
Pharmaceuticals
by 2019,
means
elevated temperature 12,ofx XRD. The short-term
conditions (T = 373 XRD experiment of EZB/SVT + 5wt. % KVA was performed at
K). 8 of 15
elevated temperature conditions (T = 373 K). During these studies, the mixture’s XRD patterns were
collected every 30 min for 60 h. The obtained results are presented in Figure 6. The XRD patterns of
both freshly vitrified as well as stored at elevated temperature conditions for 60 h EZB/SVT + 5wt. %
KVA mixture reveal a lack of Bragg’s peaks. This indicates that the sample remained amorphous
throughout the duration of the experiment. Comparing this result with the data presented in Figure
3c, one can conclude that KVA significantly improves the physical stability of the amorphous
EZB/SVT system at elevated temperature conditions (T = 373 K).
Figure 6. XRD diffraction patterns of ternary amorphous EZB/SVT + 5wt. % KVA measured as
Figure 6. XRD diffraction patterns of ternary amorphous EZB/SVT + 5wt. % KVA measured as a
a function of storage time at elevated temperature conditions (T = 373 K).
function of storage time at elevated temperature conditions (T = 373 K).
2.4. The Impact of KVA on the Complex Viscosity of the EZB/SVT System
2.4. The Impact of KVA on the Complex Viscosity of the EZB/SVT System
The experiments presented in the previous section proved that even a very small amount of
KVAThe(5wt.experiments
%) is able topresented in the suppress
very effectively previous the
section proved
tendency that even
toward a very smallofamount
re-crystallization of
EZB/SVT
KVA (5wt. %) is able to very effectively suppress the tendency toward re-crystallization of EZB/SVT
composition at elevated temperature conditions. It is worth repeating that the physical stability
improvement of this binary amorphous drug-drug system is needed only if the chosen production
method will involve thermal processing. The best representatives of such methods are hot melt
extrusion (HME) or 3D printing (3DP). This kind of manufacturing processes typically includes three
stages: (i) Heating and softening of a physical mixture containing API and thermoplastic polymer;
composition at elevated temperature conditions. It is worth repeating that the physical stability
improvement of this binary amorphous drug-drug system is needed only if the chosen production
method will involve thermal processing. The best representatives of such methods are hot melt
extrusion (HME) or 3D printing (3DP). This kind of manufacturing processes typically includes three
stages: (i) Heating and softening of a physical mixture containing API and thermoplastic polymer;
(ii) pressurization of molten mass through a die, and (iii) sample solidification, during which the
final dosage form is formulated [46–51]. To obtain the desired shape of a final product, samples
besides high physical stability needs to be characterized by suitable melt viscosity. For example,
the small scale extruder required the sample’s melt viscosity at a range of 800–10,000 Pa·s [52]. Thus,
in this section, the viscosity of EZB/SVT and its mixtures with KVA have been also investigated.
Both the EZB/SVT binary physical mixture as well as the ternary mixtures containing EZB/SVT and
KVA polymer have been placed between the fixtures of rheometer and heated up to 413 K. At this
particular temperature, the samples were melted and then the gap was set. Next, the material’s
complex viscosity was examined in the wide frequency (0.016–15.916 Hz) and temperature (413–353 K)
Pharmaceuticals 2019, 12, x
range. The representative 9 of 15
complex viscosity curves as a function of frequency are presented in Figure 7.
Figure 7. Temperature sweep data showing complex viscosity as a function of frequency of (a) the
Figure 7. Temperature sweep data showing complex viscosity as a function of frequency of (a) the
binary amorphous EZB/SVT system, (b) the ternary amorphous system of EZB/SVT + 5wt.% KVA,
binary amorphous EZB/SVT system, (b) the ternary amorphous system of EZB/SVT + 5wt.% KVA, (c)
(c) the ternary amorphous system of EZB/SVT + 20wt.% KVA, (d) the ternary amorphous system of
the ternary amorphous system of EZB/SVT + 20wt.% KVA, (d) the ternary amorphous system of
EZB/SVT + 40wt.% KVA, (e) the ternary amorphous system of EZB/SVT + 60wt.% KVA and f) the
EZB/SVT + 40wt.% KVA, (e) the ternary amorphous system of EZB/SVT + 60wt.% KVA and f) the neat
neat KVA polymer.
KVA polymer.
As can be seen with decreasing the temperature, the samples become more viscous and the
Figure
viscosity 8 shows
plateau the temperature
is less pronounced dependences of viscosity
and shifts toward values from the
lower frequencies. plateau
In this case,region for all
the complex
investigated materials, i.e. pure KVA, the binary amorphous system of EZB/SVT, and four
viscosity is low, i.e., the sample is more elastic at higher frequencies and increases with decreasingternary
amorphous systems containing
frequency culminating EZB/SVTviscosity
in a consistent and 5, 20,plateau.
40, and 60 wt.lower
The % of KVA. In ordercontent,
the polymer to parameterize
the less
the temperature dependence of complex viscosity, we employed the Vogel-Fulcher-Tamman
viscous the sample is at constant T, and the more pronounced is the viscosity plateau on the |η*|(f). (VFT)
equation that is defined as follows [53–55]:
𝐷𝑇
𝑙𝑜𝑔  𝑇 = 𝑙𝑜𝑔  + (2)
𝑇−𝑇
where T is temperature and η∞, D, T0 are parameters obtained by fitting Equation 2 to experimentally
measured viscosity data. All fitting parameters are collected in Table 2.
The shadowed area indicates the generally accepted ‘rule-of-thumb viscosity range’ for small-scale
extrusion [52]. The |η*|(f) marked by the dashed lines refer to the temperatures higher or equal to the
liquidus temperature. As can be seen, EZB/SVT, EZB/SVT + 5wt. % KVA and EZB/SVT + 20wt. % of
KVA possess suitable for small scale extruder viscosity value at the temperatures below the liquidus
temperature. Therefore, extrusion of such products is impossible. Suitable melting viscosity at T equals
or higher than the liquidus temperature can be reached when 40 or higher wt. % of the polymer is
employed (see Figure 7d–f).
Figure 8 shows the temperature dependences of viscosity values from the plateau region for all
investigated materials, i.e., pure KVA, the binary amorphous system of EZB/SVT, and four ternary
amorphous systems containing EZB/SVT and 5, 20, 40, and 60 wt. % of KVA. In order to parameterize
the temperature dependence of complex viscosity, we employed the Vogel-Fulcher-Tamman (VFT)
equation that is defined as follows [53–55]:
DT0
log10 η ( T ) = log10 η∞ + (2)
T − T0
where T is temperature and η∞ , D, T0 are parameters obtained by fitting Equation (2) to experimentally
measured viscosity data. All fitting parameters are collected in Table 2.
Table 2. Comparison of the VFT fitting parameters for neat KVA, the binary mixture of EZB/SVT, and
ternary systems of EZB/SVT/KVA containing 5, 20, 40 and 60wt. % of the polymer.
System Log10 η∞ B = DT0 T0

EZB/SVT 4.67 ± 0.17 1148.34 ± 60.19 295.12 ± 1.86
EZB/SVT + 5wt.% KVA −4.55 ± 0.09 1303.53 ± 37.27 290.77 ± 1.22
EZB/SVT + 20wt.% KVA −6.19 ± 2.67 2411.64 ± 148.98 269.49 ± 3.66
EZB/SVT + 40wt.% KVA −5.59 ± 0.22 2515.29 ± 126.57 273.42 ± 3.04
EZB/SVT + 60wt.% KVA −7.15 ± 0.4 3816.61 ± 278.45 253.35 ± 5.37
KVA −4.13 ± 0.83 2981.70 ± 565.27 278.26 ± 13.87
Based on Figure 8, where the temperature dependencies of |η*| of all investigated materials are
summarized, it is possible to quickly determine both the KVA polymer content and the temperature
range, which are suitable to produce physically-stable amorphous EZB/SVT system by means of
the melting methods. For example, in the case of small-scale extrusion, where the recommended
material’s viscosity range between 800–10,000 Pa·s, the lowest available concentration is equal to
40wt. %. Choosing this particular concentration of the excipient, the minimum temperature needed for
sample melting is equal to 398 K. It is worth highlighting that for higher polymer concentrations, higher
T conditions should be selected (ex. 60wt% of KVA – T ≥ 403 K). To better visualize it, we marked in
Figure 8 the viscosity area which is appropriate for small scale extrusion and the liquidus temperatures
for each examined concentration. Finally, in the shadowed area one can find suitable for extrusion
EZB/SVT/KVA compositions and temperatures.
According to the experimental results presented above, in order to produce the EZB/SVT system
by means of small-scale extruder, a minimum of 40wt. % of KVA co-polymer should be added.
Such a ternary system should be melted at T ≥ 389 K. To make sure that T = 389 K is indeed high
enough to fully amorphize the sample, as well as to investigate whether such a system will be physically
stable at both standard storage and elevated temperature conditions, we investigated a quenched
from T = 398 K EZB/SVT + 40wt. % KVA sample by means of XRD. The representative XRD patterns
obtained during this experiment are presented in Figure 9. As can be seen, the diffraction pattern of the
freshly prepared sample (quenched from T = 398 K) do not reveal sharp Bragg’s peaks, indicating that
T = 389 K is high enough to fully melt the sample. In the next step of the XRD experiment, the vitrified
sample has been heated up to 373 K and stored for 30 h. After each 30 min, the XRD pattern has been
registered. As data presented in Figure 9 proves, the investigated EZB/SVT/KVA system was for all
range, which
Pharmaceuticals 2019,are
12,suitable
40 to produce physically-stable amorphous EZB/SVT system by means of10
the
of 15
melting methods. For example, in the case of small-scale extrusion, where the recommended
material’s viscosity range between 800–10,000 Pa·s, the lowest available concentration is equal to
that time%.fully
40wt. amorphous.
Choosing The lastconcentration
this particular stage of the XRDof themeasurement consisted temperature
excipient, the minimum in keeping the cooled
needed
for sample
down melting
to T = 298 is equalcomposition
K ternary to 398 K. It isfor
worth highlighting
a period that forThe
of 3 months. higher polymer
sample’s concentrations,
diffraction patterns
higher
were T conditions
collected one a dayshould
for abeweek,
selected
next(ex.
once60wt% of KVA
a week for a–month,
T ≥ 403andK). To better
finally visualize
once a monthit, we
for 3
marked
months. in Figure
Data 8 thein
presented viscosity
Figure 9area which is
indicates appropriate
that for small
the examined scale
ternary extrusion
system and the liquidus
is physically stable for
temperatures
at least 3 monthsfor each
after examined concentration.
amorphization and annealing Finally,
at Tin= the
373 shadowed
K for 30 h.area one can find suitable
for extrusion EZB/SVT/KVA compositions and temperatures.
Figure
Figure8. 8.Comparison
Comparisonofofthe thetemperature
temperature dependences
dependences of ofcomplex
complexviscosity
viscosity ofof KVA
KVA (black
(black stars),
stars),
EZB/SVT (green pentagons), EZB/SVT + 5wt. % of KVA (blue squares), EZB/SVT
EZB/SVT (green pentagons), EZB/SVT + 5wt. % of KVA (blue squares), EZB/SVT + 20wt. % of KVA + 20wt. % of KVA
(red circles),
(red EZB/SVT
circles), EZB/SVT++40wt. 40wt.% %ofof KVA
KVA(yellow
(yellow triangles) and EZB/SVT
triangles) and EZB/SVT+ +60wt.60wt.%% of of
KVA KVA (orange
(orange
hexagons).
hexagons). The
Thesolid lines
solid correspond
lines correspond totothethe
VFT
VFTfits, dashed
fits, dashed horizontal
horizontallines
linesdenote
denotethe
theupper
upperandandthe
lower
the range
lower of viscosity
range suitablesuitable
of viscosity for small
forscale
smallextrusion, and dashed
scale extrusion, verticalvertical
and dashed lines represent liquidius
lines represent
temperature of the examined
liquidius temperature of the systems,
examinedwhile shadowed
systems, area marks
while shadowed the
area suitable
marks for extrusion
the suitable region of
for extrusion
temperature and viscosity
region of temperature andofviscosity
the EZB/SVT/KVA ternary system.
of the EZB/SVT/KVA ternary system.
According to the experimental results presented above, in order to produce the EZB/SVT system
by means of small-scale extruder, a minimum of 40wt. % of KVA co-polymer should be added. Such
a ternary system should be melted at T ≥ 389 K. To make sure that T = 389 K is indeed high enough
to fully amorphize the sample, as well as to investigate whether such a system will be physically
stable at both standard storage and elevated temperature conditions, we investigated a quenched
from T = 398 K EZB/SVT + 40wt. % KVA sample by means of XRD. The representative XRD patterns
obtained during this experiment are presented in Figure 9. As can be seen, the diffraction pattern of
the freshly prepared sample (quenched from T = 398 K) do not reveal sharp Bragg’s peaks, indicating
that T = 389 K is high enough to fully melt the sample. In the next step of the XRD experiment, the
vitrified sample has been heated up to 373 K and stored for 30 h. After each 30 min, the XRD pattern
has been registered. As data presented in Figure 9 proves, the investigated EZB/SVT/KVA system
was for all that time fully amorphous. The last stage of the XRD measurement consisted in keeping
the cooled down to T = 298 K ternary composition for a period of 3 months. The sample’s diffraction
patterns were collected one a day for a week, next once a week for a month, and finally once a month
for 3 months. Data presented in Figure 9 indicates that the examined ternary system is physically
Figure
stable for 9.
Figure XRD
at least 3diffraction
9. XRD diffraction
months patterns
patterns
after of ternary
of ternaryamorphous
amorphization amorphous
and EZB/SVT
EZB/SVT
annealing +K
at T+=40wt.
373 40wt.
% KVA
for 30%h.KVA measured
measured as a
function ofofstorage
as a function time.
storage Yellow
time. XRD XRD
Yellow diffractograms denote the
diffractograms first step
denote of thestep
the first experiment, during
of the experiment,
duringwhich
whichthethe
sample waswas
sample stored at elevated
stored temperature
at elevated conditions
temperature (T = (T
conditions 373= K),
373while the brown
K), while the brown
diffraction
diffraction pattern
pattern corresponds
corresponds toto thesecond
the secondstage
stage of
of the
theexperiment
experiment when
whenthethe
sample was was
sample stored at
stored at
room temperature.
room temperature.
3.1. Materials
Ezetimibe drug (EZB) of purity greater than 99% and molecular mass Mw = 409.4 g/mol was
purchased from Polpharma (Starogard Gdański, Poland) and used as received. This pharmaceutical
is described chemically as ((3R,4S)-1-(4-fluorophenyl)-3-[(3S)-3-(4- fluorophenyl)-3-hydroxypropyl] -
4-(4-hydroxyphenyl)azetid in-2-one), with its chemical structure presented in blue inset of Figure 1.
3.1. Materials
Ezetimibe drug (EZB) of purity greater than 99% and molecular mass Mw = 409.4 g/mol was
purchased from Polpharma (Starogard Gdański, Poland) and used as received. This pharmaceutical
is described chemically as ((3R,4S)-1-(4-fluorophenyl)-3-[(3S)-3-(4-fluorophenyl)-3-hydroxypropyl]-4-
(4-hydroxyphenyl)azetid in-2-one), with its chemical structure presented in blue inset of Figure 1.
Simvastatin drug (SVT) of purity greater than 98% and molecular mass Mw = 418.6 g/mol was
purchase from ACOFARMA and used as received. This pharmaceutical is described chemically
as Butanoic acid, 2,2-dimethyl-(1S,3R,7S,8S,8aR)-1,2,3,7,8,8a-hexahydro-3,7-dimethyl-8-[2-[(2R,4R)
-tetrahydro-4-hydroxy-6-oxo-2H-pyran-2-yl]ethyl]-1-naphthalenyl ester, with its chemical structure
presented in the red inset of Figure 1. Kollidon VA64 polymer (KVA) of molecular mass
Mw = 45,000–47,000 g/mol was purchased from BASF SE (Germany) and used as received.
3.2. Preparation of Binary and Ternary Systems

Both binary amorphous mixtures of EZB and SVT, as well as ternary amorphous systems of
EZB/SVT and KVA with different content of the polymer (5, 20, 40, 60wt.% KVA), were prepared by
the quench cooling technique. To obtain the homogenous systems prior to the quenching, the physical
mixtures were obtained by gentle mixing the appropriate number of samples in a mortal for 15–20 min.
3.3. Differential Scanning Calorimetry (DSC)

Thermodynamic properties of EZB, SVT, KVA and their binary and ternary systems were
examined using a Mettler–Toledo DSC 1 STARe System. The measuring device was equipped with
an HSS8 ceramic sensor having 120 thermocouples. The instrument was calibrated for temperature
and enthalpy using indium and zinc standards. Crystallization and melting points were determined
as the onset of the peak, whereas the glass transition temperature as the midpoint of the heat capacity
increment. The samples were measured in an aluminum crucible (40 µL). All measurements were
carried out with a heating rate equal to 10 K/min.
3.4. X-ray Diffraction (XRD)

X-ray diffraction experiments were performed for samples stored at both ambient and
elevated temperature conditions on a Rigaku-Denki D/MAX RAPID II-R diffractometer (Rigaku
Corporation, Tokyo, Japan) with a rotating anode Ag KR tube (λ = 0.5608 Å), an incident beam
(002) graphite monochromator, and an image plate in the Debye−Scherrer geometry. The pixel
size was 100 µm × 100 µm. Measurements were performed on sample-filled and empty capillaries,
to allow subtraction of the background intensity. The beam width at the sample was 0.1 mm.
The two-dimensional diffraction patterns were converted into one-dimensional intensity data versus
the scattering angle.
3.5. Rheological Studies

The viscoelastic properties of neat KVA, the binary system of EZB and SVT, as well as ternary
systems containing EZB1:1SVT + different KVA content (5wt%, 20wt.%, 40wt.% and 60wt.%) were
measured by means of ARES G2 Rheometer. The instrument was equipped with aluminum parallel
plates geometry (diameter = 25 mm). Prior to the oscillation, temperature sweep rheological
experiments the samples were placed between the fixtures of rheometer and heated up to 413 K
(or T = 458 K for neat KVA). At this particular temperature, the examined materials were melted (or
liquefied in the case of neat polymer) and then the gap was set at ~1 mm. Next, the material’s complex
viscosity was examined in the wide temperature range and frequencies from 0.016–15.916 Hz.
4. Conclusions
In this article, the binary amorphous composition of ezetimibe and simvastatin was investigated by
means of differential scanning calorimetry, X-ray diffraction, and oscillatory shear rheology. We found
that the simple melt-quenching method can be applied to obtain a homogeneous binary system of
these two pharmaceuticals. Thermal analysis of the pure crystalline components, as well as physical
mixture of ezetimibe and simvastatin, indicated that co-administration of these two APIs leads to
melting point depression. The long-term stability studies performed by XRD at standard storage
conditions (i.e., T = 298 K and RH = 60%) have proven that simvastatin, despite the fact that it acts as
a plasticizer, remarkably improves the physical stability of amorphous ezetimibe drugs. Unfortunately,
the stabilization effect is not enough for the case when the sample is planned to be produced at
elevated temperature—at elevated temperature conditions (T = 373 K), the re-crystallization of the
sample was registered. To overcome this problem, the third component, co-polymer Kollidon VA 64,
was proposed as a mixture stabilizer. The physical stability studies performed at elevated temperature
conditions indicated that even a small amount of the polymeric additive (5 wt. %) may significantly
suppress tendency towards the devitrification of the ezetimibe–simvastatin composition. The first
sign of devitrification of the binary amorphous composition noted after 3h and 30 min, while the
ternary system containing 5 wt. % of the polymer was physically stable for at least 60 h (when stored
at T = 373 K).
Finally, the effect of Kollidon VA64 on the viscoelastic properties of the EZB/SVT system in their
ternary amorphous compositions was investigated using oscillatory shear rheology. Plots containing
temperature dependencies of complex viscosity of all examined systems—constructed based on the
rheological data—allowed us to estimate the most suitable, for small scale extrusion, temperature
conditions, as well as the polymer concentrations. In order to obtain suitable melt viscosity of the
ezetimibe–simvastatin composition, at least 40 wt. % of the polymer is needed. For this particular case,
the lowest available temperature is equal to 398 K.
In summary, the binary amorphous ezetimibe–simvastatin system is a very promising candidate
for new formulations intended for combined therapy. Apart from the medical and economic benefits
of using these APIs together in amorphous forms, such a binary composition provides high physical
stability that is essential for further commercial application. For the production of this binary system by
melting methods, such as hot melt extrusion or 3D printing, the addition of Kollidon VA64 is required.
Author Contributions: Conceptualization and methodology, J.K.-K. and M.P.; investigation and validation,
J.K.-K., K.J. and K.C.; data analysis, J.K.-K.; resources, N.T.C. and W.S.; writing—original draft preparation, J.K.-K.;
writing—review and editing, M.P., W.S. and N.T.C; visualization, J.K-K.; supervision, M.P.
Funding: This research was founded by the National Science Centre, Poland [Project No. 2015/16/W/NZ7/00404
(SYMFONIA 3)] and polonium.
Acknowledgments: This work was supported by the National Science Centre, Poland [Project No.
2015/16/W/NZ7/00404 (SYMFONIA 3)] and polonium.
References
1. Bangalore, S.; Kamalakkannan, G.; Parkar, S.; Messerli, F.H. Fixed-Dose Combinations Improve Medication
Compliance: A Meta-Analysis. Am. J. Med. 2007, 120, 713–719. [CrossRef] [PubMed]
2. Zaragoza, C.; Gomez-Guerrero, C.; Martin-Ventura, J.L.; Blanco-Colio, L.; Tarin, C.; Mas, S.; Ortiz, A.; Egido, J.
Animal Models of Cardiovascular Diseases. J. Biomed. Biotechnol. 2011, 2011, 497841. [CrossRef] [PubMed]
3. Kastelein, J.J.P.; Akdim, F.; Stroes, E.S.G.; Zwinderman, A.H.; Bots, M.L.; Stalenhoef, A.F.H.; Visseren, F.L.J.;
Sijbrands, E.J.G.; Trip, M.D.; Stein, E.A.; et al. Simvastatin with or without Ezetimibe in Familial
Hypercholesterolemia. N. Engl. J. Med. 2008, 358, 1431–1443. [CrossRef]
4. DiPiro, J.T.; Talbert, R.L.; Yee, G.C.; Matzke, G.R.; Wells, B.G.; Posey, L.M. Pharmacotherapy: A Pathophysiologic
Approach, 7th ed.; The McGraw-Hill Companies: New York, NY, USA, 2008; ISBN 0071643257.
5. Al-Husein, B.A.A. Molecular Mechanisms Regulating Simvastatin-Mediated Inhibition of Prostate Cancer

Cellular Functions In Vitro AND Tumor Growth In Vivo. Ph.D. Thesis, University of Georgia, Athens, GA,
USA, 2013.
6. Fuentes, A.; Pineda, M.; Venkata, K. Comprehension of Top 200 Prescribed Drugs in the US as a Resource for
Pharmacy Teaching, Training and Practice. Pharmacy 2018, 6, 43. [CrossRef]
7. Davidson, M.H.; McGarry, T.; Bettis, R.; Melani, L.; Lipka, L.J.; LeBeaut, A.P.; Suresh, R.; Sun, S.; Veltri, E.P.
Ezetimibe coadministered with simvastatin in patients with primary hypercholesterolemia. J. Am. Coll.
Cardiol. 2002, 40, 2125–2134. [CrossRef]
8. Ballantyne, C.M.; Abate, N.; Yuan, Z.; King, T.R.; Palmisano, J. Dose-comparison study of the combination of
ezetimibe and simvastatin (Vytorin) versus atorvastatin in patients with hypercholesterolemia: The Vytorin
Versus Atorvastatin (VYVA) Study. Am. Heart J. 2005, 149, 464–473. [CrossRef]
9. Baigent, C.; Landray, M.J.; Reith, C.; Emberson, J.; Wheeler, D.C.; Tomson, C.; Wanner, C.; Krane, V.; Cass, A.;
Craig, J.; et al. The eff ects of lowering LDL cholesterol with simvastatin plus ezetimibe in patients with
chronic kidney disease (Study of Heart and Renal Protection): A randomised placebo-controlled trial. Lancet
2011, 377, 2181–2192. [CrossRef]
10. Kumari Nanam, P.; Thadkala, K.; Sailu, C.; Aukunuru, J. Investigation of various practical techniques to
enhance dissolution of ezetimibe from oral tablets: A comparative study. J. Young Pharm. 2014, 6, 8–14.
[CrossRef]
11. Bandyopadhyay, S.; Katare, O.P.; Singh, B. Optimized self nano-emulsifying systems of ezetimibe with
enhanced bioavailability potential using long chain and medium chain triglycerides. Colloids Surf. B
Biointerfaces 2012, 100, 50–61. [CrossRef] [PubMed]
12. Rao, M.; Mandage, Y.; Thanki, K.; Bhise, S. Dissolution improvement of simvastatin by surface solid
dispersion technology. Dissolut. Technol. 2010, 17, 27–34. [CrossRef]
13. Kaminski, K.; Adrjanowicz, K.; Wojnarowska, Z.; Grzybowska, K.; Hawelek, L.; Paluch, M.; Zakowiecki, D.;
Mazgalski, J. Molecular Dynamics of the Cryomilled Base and Hydrochloride Ziprasidones by Means of
Dielectric Spectroscopy. J. Pharm. Sci. 2011, 100, 2642–2657. [CrossRef] [PubMed]
14. Cao, Y.; Teng, J.; Selbo, J.; Cao, Y.; Teng, J.; Selbo, J. Amorphous Solid Dispersion of Epigallocatechin Gallate
for Enhanced Physical Stability and Controlled Release. Pharmaceuticals 2017, 10, 88. [CrossRef] [PubMed]
15. Kalepu, S.; Nekkanti, V. Insoluble drug delivery strategies: Review of recent advances and business prospects.
Acta Pharm. Sin. B 2016, 5, 442–453. [CrossRef]
16. Hancock, B.C.; Parks, M. What is the True Solubility Advantage for Amorphous Pharmaceuticals? Pharm.
Res. 2000, 17, 397–404. [CrossRef] [PubMed]
17. Craig, D.Q.; Royall, P.G.; Kett, V.L.; Hopton, M.L. The relevance of the amorphous state to pharmaceutical
dosage forms: Glassy drugs and freeze dried systems. Int. J. Pharm. 1999, 179, 179–207. [CrossRef]
18. Wojnarowska, Z.; Grzybowska, K.; Adrjanowicz, K.; Kaminski, K.; Paluch, M.; Hawelek, L.; Wrzalik, R.;
Dulski, M.; Sawicki, W.; Mazgalski, J.; et al. Study of the amorphous glibenclamide drug: Analysis of the
molecular dynamics of quenched and cryomilled material. Mol. Pharm. 2010, 7, 1692–1707. [CrossRef]
[PubMed]
19. Adrjanowicz, K.; Wojnarowska, Z.; Wlodarczyk, P.; Kaminski, K.; Paluch, M.; Mazgalski, J. Molecular
mobility in liquid and glassy states of Telmisartan (TEL) studied by Broadband Dielectric Spectroscopy.
20. Grzybowska, K.; Paluch, M.; Grzybowski, A.; Wojnarowska, Z.; Hawelek, L.; Kolodziejczyk, K.; Ngai, K.L.
Molecular dynamics and physical stability of amorphous anti-inflammatory drug: Celecoxib. J. Phys. Chem.
B 2010, 114, 12792–12801. [CrossRef] [PubMed]
21. Forster, A.; Hempenstall, J.; Rades, T. Characterization of glass solutions of poorly water-soluble drugs
produced by melt extrusion with hydrophilic amorphous polymers. J. Pharm. Pharmacol. 2001, 53, 303–315.
[CrossRef] [PubMed]
22. Amidon, G.L.; Lennernäs, H.; Shah, V.P.; Crison, J.R. A Theoretical Basis for a Biopharmaceutic Drug
Classification: The Correlation of in Vitro Drug Product Dissolution and in Vivo Bioavailability. Pharm. Res.
1995, 12, 413–420. [CrossRef]
23. Rani, T.N.; Muzeeb, Y.I.; Rani, P.S.; Pradesh, A. Solubility Enhancement of Poorly Soluble Drug Ezetimibe By
Solid Dispersion Technique. J. Adv. Pharm. Educ. Res. 2013, 4, 75–81.
24. Murdande, S.B.; Pikal, M.J.; Shanker, R.M.; Bogner, R.H. Solubility advantage of amorphous pharmaceuticals:
I. A thermodynamic analysis. J. Pharm. Sci. 2010, 99, 1254–1264. [CrossRef] [PubMed]
25. Williams, H.; Trevaskis, N.; Charman, S.; Shanker, R.; Charman, W.; Pouton, C.; Porter, C. Strategies to
address low drug solubility in discovery and development. Pharmacol. Rev. 2013, 65, 315–499. [CrossRef]
[PubMed]
26. Descamps, M.; Willart, J.-F. Scaling laws and size effects for amorphous crystallization kinetics: Constraints
imposed by nucleation and growth specificities. Int. J. Pharm. 2018, 542, 186–195. [CrossRef] [PubMed]
27. Descamps, M.; Dudognon, E. Crystallization from the Amorphous State: Nucleation-Growth Decoupling,
Polymorphism Interplay, and the Role of Interfaces. Pharm. Assoc. J. Pharm. Sci. 2014, 103, 2615–2628.
[CrossRef]
28. Knapik-Kowalczuk, J.; Wojnarowska, Z.; Rams-Baron, M.; Jurkiewicz, K.; Cielecka-Piontek, J.; Ngai, K.L.;
Paluch, M. Atorvastatin as a promising crystallization inhibitor of amorphous probucol. Dielectric studies at
ambient and elevated pressure. Mol. Pharm. 2017, 14, 2670–2680. [CrossRef]
29. Chmiel, K.; Knapik-Kowalczuk, J.; Jurkiewicz, K.; Sawicki, W.; Jachowicz, R.; Paluch, M. A New Method To
Identify Physically Stable Concentration of Amorphous Solid Dispersions (I): Case of Flutamide + Kollidon
VA64. Mol. Pharm. 2017, 14, 3370–3380. [CrossRef]
30. Löbmann, K.; Laitinen, R.; Grohganz, H.; Gordon, K.C.; Strachan, C.; Rades, T. Coamorphous drug systems:
Enhanced physical stability and dissolution rate of indomethacin and naproxen. Mol. Pharm. 2011, 8,
1919–1928. [CrossRef]
31. Shi, Q.; Moinuddin, S.M.; Cai, T. Advances in coamorphous drug delivery systems. Acta Pharm. Sin. B 2019,
9, 19–35. [CrossRef]
32. Laitinen, R.; Löbmann, K.; Strachan, C.J.; Grohganz, H.; Rades, T. Emerging trends in the stabilization of
amorphous drugs. Int. J. Pharm. 2013, 453, 65–79. [CrossRef]
33. Knapik, J.; Wojnarowska, Z.; Grzybowska, K.; Jurkiewicz, K.; Stankiewicz, A.; Paluch, M. Stabilization of
the Amorphous Ezetimibe Drug by Confining Its Dimension. Mol. Pharm. 2016, 13, 1308–1316. [CrossRef]
[PubMed]
34. Knapik, J.; Wojnarowska, Z.; Grzybowska, K.; Jurkiewicz, K.; Tajber, L.; Paluch, M. Molecular dynamics
and physical stability of coamorphous ezetimib and indapamide mixtures. Mol. Pharm. 2015, 12, 3610–3619.
[CrossRef]
35. Szafraniec, J.; Antosik, A.; Knapik-Kowalczuk, J.; Kurek, M.; Syrek, K.; Chmiel, K.; Paluch, M.; Jachowicz, R.
Planetary ball milling and supercritical fluid technology as a way to enhance dissolution of bicalutamide.
Int. J. Pharm. 2017, 30, 470–479. [CrossRef]
36. Yan, J.; Yin, Q.; Jiang, C.; Gong, J.; Zhang, M.; Wang, Y.; Hou, B.; Hao, H. Solution thermodynamics of
simvastatin in pure solvents and binary solvent mixtures. Fluid Phase Equilib. 2015, 406, 77–90. [CrossRef]
37. Knapik, J.; Wojnarowska, Z.; Grzybowska, K.; Hawelek, L.; Sawicki, W.; Wlodarski, K.; Markowski, J.;
Paluch, M. Physical stability of the amorphous anticholesterol agent (Ezetimibe): The role of molecular
mobility. Mol. Pharm. 2014, 11, 4280–4290. [CrossRef]
38. Martínez-Jiménez, C.; Cruz-Angeles, J.; Videa, M.; Martínez, L.; Martínez-Jiménez, C.; Cruz-Angeles, J.;
Videa, M.; Martínez, L.M. Co-Amorphous Simvastatin-Nifedipine with Enhanced Solubility for Possible Use
in Combination Therapy of Hypertension and Hypercholesterolemia. Molecules 2018, 23, 2161. [CrossRef]
[PubMed]
39. Knapik-Kowalczuk, J.; Tu, W.; Chmiel, K.; Rams-Baron, M.; Paluch, M. Co-Stabilization of Amorphous
Pharmaceuticals - The Case of Nifedipine and Nimodipine. Mol. Pharm. 2018, 15, 2455–2465. [CrossRef]
[PubMed]
40. Shah, N.; Sandhu, H.; Choi, D.S.; Chokshi, H.; Malick, A.W. Amorphous Solid Dispersions Theory and Practice;
Springer International Publishing: Berlin/Heidelberg, Germany, 2014; ISBN 9781493915989.
41. Prasad, D.; Chauhan, H.; Atef, E. Amorphous stabilization and dissolution enhancement of amorphous
ternary solid dispersions: Combination of polymers showing drug-polymer interaction for synergistic effects.
J. Pharm. Sci. 2014, 103, 3511–3523. [CrossRef]
42. Lu, Q.; Zografi, G. Phase behaviour of binary and ternary amourphous mixtures containing indomethacin,
citric acid and PVP. Pharm. Res. 1998, 15, 1202–1206. [CrossRef] [PubMed]
43. Janssens, S.; Nagels, S.; Armas, H.N. de; D’Autry, W.; Van Schepdael, A.; Van den Mooter, G. Formulation
and characterization of ternary solid dispersions made up of Itraconazole and two excipients, TPGS 1000
and PVPVA 64, that were selected based on a supersaturation screening study. Eur. J. Pharm. Biopharm. 2008,
69, 158–166. [CrossRef]
44. Couchman, P.R. Compositional Variation of Glass-Transition Temperatures. 2. Application of the
Thermodynamic Theory to Compatible Polymer Blends. Macromolecules 1978, 11, 1156–1161. [CrossRef]
45. Couchman, P.R.; Karasz, F.E. A Classical Thermodynamic Discussion of the Effect of Composition on
Glass-Transition Temperatures. Macromolecules 1978, 11, 117–119. [CrossRef]
46. Repka, M.A.; Battu, S.K.; Upadhye, S.B.; Thumma, S.; Crowley, M.M.; Zhang, F.; Martin, C.; McGinity, J.W.
Pharmaceutical applications of hot-melt extrusion: Part II. Drug Dev. Ind. Pharm. 2007, 33, 1043–1057.
[CrossRef] [PubMed]
47. Sarode, A.L.; Sandhu, H.; Shah, N.; Malick, W.; Zia, H. Hot melt extrusion (HME) for amorphous solid
dispersions: Predictive tools for processing and impact of drug-polymer interactions on supersaturation.
48. Gupta, S.S.; Solanki, N.; Serajuddin, A.T.M. Investigation of Thermal and Viscoelastic Properties of Polymers
Relevant to Hot Melt Extrusion, IV: AffinisolTM HPMC HME Polymers. AAPS PharmSciTech 2016, 17, 148–157.
[CrossRef] [PubMed]
49. Bochmann, E.S.; Üstüner, E.E.; Gryczke, A.; Wagner, K.G. Predicting melt rheology for hot-melt extrusion by
means of a simple Tg-measurement. Eur. J. Pharm. Biopharm. 2017, 119, 47–55. [CrossRef] [PubMed]
50. Bochmann, E.S.; Steffens, K.E.; Gryczke, A.; Wagner, K.G. Numerical simulation of hot-melt extrusion
processes for amorphous solid dispersions using model-based melt viscosity. Eur. J. Pharm. Biopharm. 2018,
51. Solanki, N.; Gupta, S.S.; Serajuddin, A.T.M. Rheological analysis of itraconazole-polymer mixtures to
determine optimal melt extrusion temperature for development of amorphous solid dispersion. Eur. J.
Pharm. Sci. 2018, 111, 482–491. [CrossRef]
52. Aho, J.; Edinger, M.; Botker, J.; Baldursdottir, S.; Rantanen, J. Oscillatory Shear Rheology in Examining the
Drug-Polymer Interactions Relevant in Hot Melt Extrusion. J. Pharm. Sci. 2016, 105, 160–167. [CrossRef]
53. Vogel, H. Das Temperaturabhangigkeitgesetz der Viskosität von Flüssigkeiten. J. Phys. Z. 1921, 22, 645–646.
54. Fulcher, G.S. Analysis of Recent Measurements of the Viscosity of Glasses. J. Am. Ceram. Soc. 1925, 8, 339–355.
[CrossRef]
55. Tammann, G.; Hesse, W. Die Abhängigkeit der Viscosität von der Temperatur bie unterkühlten Flüssigkeiten.
Z. Anorg. Allg. Chem. 1926, 156, 245–257. [CrossRef]
pharmaceuticals
Review
BACE-1 and γ-Secretase as Therapeutic Targets for
Alzheimer’s Disease
Miguel A. Maia 1 and Emília Sousa 1,2, *
1 Laboratory of Organic and Pharmaceutical Chemistry, Department of Chemical Sciences,
Faculty of Pharmacy, University of Porto, Rua Jorge Viterbo Ferreira, 228, 4050-313 Porto, Portugal;
miguelmaia2@gmail.com
2 Interdisciplinary Centre of Marine and Environmental Research (CIIMAR), Terminal de Cruzeiros do Porto
de Leixões, Av. General Norton de Matos s/n, 4450-208 Matosinhos, Portugal
* Correspondence: esousa@ff.up.pt; Tel.: +351-220428689

Abstract: Alzheimer’s disease (AD) is a growing global health concern with a massive impact on
affected individuals and society. Despite the considerable advances achieved in the understanding
of AD pathogenesis, researchers have not been successful in fully identifying the mechanisms
involved in disease progression. The amyloid hypothesis, currently the prevalent theory for AD,
defends the deposition of β-amyloid protein (Aβ) aggregates as the trigger of a series of events
leading to neuronal dysfunction and dementia. Hence, several research and development (R&D)
programs have been led by the pharmaceutical industry in an effort to discover effective and safety
anti-amyloid agents as disease modifying agents for AD. Among 19 drug candidates identified in the
AD pipeline, nine have their mechanism of action centered in the activity of β or γ-secretase proteases,
covering almost 50% of the identified agents. These drug candidates must fulfill the general rigid
prerequisites for a drug aimed for central nervous system (CNS) penetration and selectivity toward
different aspartyl proteases. This review presents the classes of γ-secretase and beta-site APP cleaving
enzyme 1 (BACE-1) inhibitors under development, highlighting their structure-activity relationship,
among other physical-chemistry aspects important for the successful development of new anti-AD
pharmacological agents.
Keywords: Alzheimer’s disease; amyloid hypothesis; γ-secretase; BACE-1
1. Introduction
1.1. Alzheimer’s Disease–Epidemiology

AD is recognized by the World Health Organization (WHO) as a global public health priority [1].
Notwithstanding the advances in the understanding of AD pathogenesis since Alois Alzheimer
reported the first case in 1907 [2], there are still a lot of uncertainties regarding the mechanisms involved
in disease progression. AD is the most prevalent cause of dementia-acquired progressive cognitive
impairment sufficient to impact on activities of daily living, being a major cause of dependence,
disability and mortality. The WHO estimated that in 2010, 35.6 million worldwide were living with
dementia [3]. This figure is projected to almost double every 20 years, reaching 65.7 million by 2030 and
115.4 million by 2050. Europe, with an estimated 10 million cases of dementia in 2010 and acquiring
progressively an old population structure [4], has a projected increase to 14 million cases in 2030 [1].
In the United States of America (USA), the total annual payments for health care, long-term care
and hospice care for people with AD or other dementias are projected to increase from $259 billion in
2017 to more than $1.1 trillion in 2050 [5]. These numbers show the dramatic impact that AD and the
other dementias will have in the health care systems in the future.

Pharmaceuticals 2018, 11, 2 of 31
in 2017
in 2017 to
to more
more than
than $1.1
$1.1 trillion
trillion in
in 2050
2050 [5].
[5]. These
These numbers
numbers show
show the
the dramatic
dramatic impact
impact that
that AD
AD and
and
the other dementias
Pharmaceuticals will
2019, 12,have
41 in the health care systems in the future.
the other dementias will have in the health care systems in the future. 2 of 31
1.2. Pathology
1.2. Pathology
1.2. Pathology
The main characteristics of Alzheimer’s pathology are the presence of amyloid plaques and
The mainThe characteristics
main characteristics of Alzheimer’s pathology are theare presence of amyloid plaques and and
neurofibrillary
neurofibrillary tangles
tangles (aggregates of
(aggregates ofofhyperphosphorylated
Alzheimer’s
hyperphosphorylated pathology
tau
the presence
tau protein)
protein) [5]. In
[5].
of amyloid
In addition,
addition, plaques
neuropil
neuropil
treads,neurofibrillary
treads, dystrophic neurites,
dystrophic
tangles
neurites, (aggregates
associated
associated
of hyperphosphorylated
astrogliosis,
astrogliosis, and microglial
and
tau protein)
microglial activation
activation [5]. In addition,
frequently
frequently coexist. The
coexist.
neuropil
The
treads,consequences
downstream dystrophic neurites, associated astrogliosis,
of these pathological processes include and microglial activationwith
neurodegeneration frequently
synapticcoexist.
downstream
The consequences
downstream of these pathological
consequences of these processes
pathological include neurodegeneration
processes include with synaptic with
neurodegeneration
and neuronal
and neuronal lossloss leading
leading to to macroscopic
macroscopic atrophy
atrophy [6]. [6]. Research
Research suggests
suggests thatthat the
the brain
brain changes
changes
synaptic
associated with and
AD neuronal
may begin loss20 leading
or more toyears
macroscopic
before atrophy [6].
symptoms Research
appear [7,8]. suggests
When the that the brain
initial
associated with
changes AD may begin 20 or more years before symptoms appear [7,8]. When the initial
changes occur,associated
the brain with AD may begin
compensates 20 or enabling
for them, more years before symptoms
individuals to continue appear to [7,8].
functionWhen the
changes occur,
initial the brain
changes occur, compensates
the brain for them, for
compensates enabling
them, individuals
enabling to continue
individuals to to function
continue to function
normally. As
normally. As neuronal
neuronal damagedamage increases,
increases, the
the brain
brain cancan nono longer
longer compensate
compensate for for the
the changes
changes and and
individuals show subtle cognitive decline. Later, neuronal damage is so significant that individuals and
normally. As neuronal damage increases, the brain can no longer compensate for the changes
individuals show subtle
individuals
cognitive decline.
show decline,
subtle cognitive
Later, neuronal damage is so significant that individuals
show obvious
show obvious cognitive
cognitive decline, includingdecline.
including symptoms
symptoms
Later, neuronal
such
such as memory
as memorydamage loss is
loss orso
or
significant
confusion
confusion as to
as
that
to timeindividuals
time or
or
place show
[5]. obvious cognitive decline, including symptoms such as memory loss or confusion as to time or
place [5].
place [5].
1.3. Current Pharmacology and Drug Development
1.3. Current Pharmacology
1.3. Current and Drug
Pharmacology andDevelopment
Drug Development
Currently, five drugs are approved for the treatment of AD (Table 1). These include four
Currently, five drugs
Currently, are approved
five drugs for thefortreatment
are approved the treatmentof ADof(Table 1). These
AD (Table includeinclude
1). These four four
acetylcholinesterase inhibitors
acetylcholinesterase inhibitors (AChEi)
(AChEi) -- tacrine
tacrine (approved
(approved in in 1993),
1993), donepezil
donepezil (approved
(approved in in 1996),
1996),
acetylcholinesterase inhibitors (AChEi) - tacrine (approved in 1993), donepezil (approved in 1996),
rivastigmine (1998),
rivastigmine (1998), and
and galantamine
galantamine (approved
(approved in in 2001)
2001) [9].
[9]. This
This class
class ofof drugs
drugs actact byby blocking
blocking the the
rivastigmine (1998), and galantamine (approved in 2001) [9]. This class of drugs act by blocking
process that downregulate the neurotransmitter acetylcholine (ACh), a key signaling agent for nerve
processthethat downregulate
process the neurotransmitter
that downregulate acetylcholine
the neurotransmitter (ACh), a key(ACh),
acetylcholine signalinga keyagent for nerve
signaling agent for
cells communication.
cells communication. Although Although tacrine tacrine has
has been
been the the first
first AChEi
AChEi approved
approved by by Food
Food and and DrugDrug
nerve cells communication. Although tacrine has been the first AChEi approved by Food and Drug
Administration (FDA),
Administration (FDA), it it is
is currently
currently not not used
used due
due to to its
its hepatotoxicity [10]. [10]. Donepezil,
Donepezil, rivastigmine,
rivastigmine,
Administration (FDA), it is currently not used duehepatotoxicity
to its hepatotoxicity [10]. Donepezil, rivastigmine,
and galantamine belong to the same therapeutic class, but present different pharmacodynamic (PD)
and galantamine
and galantaminebelongbelong
to the sameto thetherapeutic
same therapeuticclass, but present
class, different
but present pharmacodynamic
different pharmacodynamic (PD) (PD)
and pharmacokinetic
and pharmacokinetic (PK) (PK) profiles:
profiles: donepezil
donepezil is is aa noncompetitive
noncompetitive reversible
reversible AChEi;
AChEi; galantamine
galantamine is is
and pharmacokinetic (PK) profiles: donepezil is a noncompetitive reversible AChEi; galantamine is
a selective reversible AChEi and a positive allosteric modulator of nicotinic ACh receptors; and
a selective reversible
a selective AChEiAChEi
reversible and a and positive allosteric
a positive modulator
allosteric of nicotinic
modulator ACh receptors;
of nicotinic ACh receptors; and and
rivastigmine acts
rivastigmine acts asas an
an inhibitor
inhibitor of of both
both acetylcholinesterase
acetylcholinesterase (AChE) (AChE) and and butyrylcholinesterase
butyrylcholinesterase
rivastigmine acts as an inhibitor of both acetylcholinesterase (AChE) and butyrylcholinesterase
(BuChE) [11].
(BuChE) [11].
(BuChE) [11].
Table 1. Approved drugs for the treatment of Alzheimer’s disease.
Table 1.Table
Approved drugs for
1. Approved the for
drugs treatment of Alzheimer’s
the treatment disease.disease.
of Alzheimer’s
Name Structure
Name Name Structure
Structure
Tacrine
Tacrine Tacrine
Donepezil
DonepezilDonepezil
Rivastigmine
Rivastigmine
Rivastigmine
Galantamine
Galantamine
Galantamine
MemantineMemantine
Memantine

An alternative therapy to AChEi is the administration of memantine, a non-competitive

An alternative
antagonist therapy to AChEi is
of N-methyl-D-aspartate the administration
(NMDA) of memantine,
receptor, approved a non-competitive
in 2004. antagonist
This drug regulates the
of N-methyl-D-aspartate
activity of glutamate in the brain. Attachment of the neurotransmitter glutamate to the cell surface of
(NMDA) receptor, approved in 2004. This drug regulates the activity of
glutamate in the brain. Attachment of the neurotransmitter glutamate to the cell surface
NMDA receptors allows calcium to enter the cell. This is an important event in cell signaling, as well of NMDA
receptors allows
as in learning andcalcium
memory. to However,
enter the cell.
in AD This is an important
patients, event inreleased
excess glutamate cell signaling, as well as
from damaged in
cells
learning and memory. However, in AD patients, excess glutamate released from damaged
leads to overexposure to calcium and accelerates cell damage. Memantine disrupts this chain of cells leads
to overexposure
events by blockingto calcium
the NMDAand receptors
accelerates cell[12].
[11], damage. Memantine
In Table disruptsthe
1 are comprised thiscurrent
chain of events
drugs by
used
blocking the NMDA
in the therapy of AD. receptors [11,12]. In Table 1 are comprised the current drugs used in the therapy
of AD.
1.4. Disease-Modifying Therapies and Drug Development
1.4. Disease-Modifying Therapies and Drug Development
The drugs currently available for the treatment of AD only relieve symptoms, not being able to
The drugs currently available for the treatment of AD only relieve symptoms, not being able
impact in the progression of the disease. There is an urgent need for the development of disease
to impact in the progression of the disease. There is an urgent need for the development of disease
modifying therapies or treatments (DMT’s) able to prevent, delay, or slow the progression and target
modifying therapies or treatments (DMT’s) able to prevent, delay, or slow the progression and target
the primary pathophysiology mechanisms of AD. It has been estimated that the overall frequency of
the primary pathophysiology mechanisms of AD. It has been estimated that the overall frequency of
the disease would be decreased by 40% if the onset of the disease could be delayed by 5 years [13].
the disease would be decreased by 40% if the onset of the disease could be delayed by 5 years [13].
Unfortunately, drug development for AD has proven to be very difficult. The high cost of drug
Unfortunately, drug development for AD has proven to be very difficult. The high cost of drug
development, the relatively long time needed to observe whether an investigational treatment affects
development, the relatively long time needed to observe whether an investigational treatment affects
disease progression and the needed capacity of the drug to cross the blood-brain barrier (BBB) are
disease progression and the needed capacity of the drug to cross the blood-brain barrier (BBB) are some
some factors that contribute to the absence of new approved drugs for AD. Even considering the
factors that contribute to the absence of new approved drugs for AD. Even considering the compounds
compounds that reach clinical trials, many of them fail to prove efficacy or present unacceptable
that reach clinical trials, many of them fail to prove efficacy or present unacceptable toxicity, both small
toxicity, both small molecules and immunotherapies [9]. Considering the decade between 2002 and
molecules and immunotherapies [9]. Considering the decade between 2002 and 2012, 244 compounds
2012, 244 compounds were assessed in 413 trials for AD. Of the agents which advanced to phase III,
were assessed in 413 trials for AD. Of the agents which advanced to phase III, only one was advanced
only one was advanced to FDA and approved for marketing (memantine). This represents an overall
to FDA and approved for marketing (memantine). This represents an overall success rate for approval
success rate for approval of 0.4%, which is among the lowest for any therapeutic area [9].
of 0.4%, which is among the lowest for any therapeutic area [9].
However, besides the mentioned difficulties in the development of new therapies for AD, there
However, besides the mentioned difficulties in the development of new therapies for AD, there
are currently large research programs for the drug development of new anti-AD agents. In 2018, there
are currently large research programs for the drug development of new anti-AD agents. In 2018, there
were 112 agents in the AD pipeline, of which 23 in phase I, 63 in phase II and 26 in phase III. Across
were 112 agents in the AD pipeline, of which 23 in phase I, 63 in phase II and 26 in phase III. Across all
all stages, 63% are DMTs, 34% symptomatic agents for neuropsychiatric and behavioral changes, and
stages, 63% are DMTs, 34% symptomatic agents for neuropsychiatric and behavioral changes, and 3%
3% have undisclosed mechanism of action (MoA) [14]
have undisclosed mechanism of action (MoA) [14]
Figure 1 summarizes the MoA of the agents in clinical trials in 2018. Twenty five percent of the
Figure 1 summarizes the MoA of the agents in clinical trials in 2018. Twenty five percent of the
agents have an amyloid-related MoA (12% immunotherapy and 14% small molecules), 14% anti-tau,
agents have an amyloid-related MoA (12% immunotherapy and 14% small molecules), 14% anti-tau,
22% neurotransmitter-based (symptomatic treatment agents), 18% neuroprotection/ anti-
22% neurotransmitter-based (symptomatic treatment agents), 18% neuroprotection/ anti-inflammatory,
inflammatory, 18% metabolic and 3% with other mechanisms [14]
18% metabolic and 3% with other mechanisms [14].
Others
3%
Anti-amyloid (small
Neurotransmitter molecules &
based immunotherapy)
22% 25%
Neuroprotection /
anti-inflammatory
18% Anti-tau
Metabolic 14%
18%
Figure 1. Mechanism of action of AD agents in clinical trials in 2018. Adapted from [14].
Figure 1. Mechanism of action of AD agents in clinical trials in 2018. Adapted from [14].
As depicted in Figure 1, Aβ appears as the main therapeutic target for AD, with several
As depicted inindustries
pharma/biopharma Figure 1,trying
AΆ appears as agents
to develop the main therapeutic targetorfor
(as immunotherapies AD,
small with several
molecules) able
pharma/biopharma industries trying to develop agents (as immunotherapies or small
to decrease Aβ accumulation/aggregation and to show a positive effect on AD pathology. molecules) able
to decrease AΆ accumulation/aggregation and to show a positive effect on AD pathology.

2. The Amyloid
Pharmaceuticals Hypothesis
2019, 12, 41 of AD 4 of 31
The amyloid hypothesis, the current prevalent theory of AD pathogenesis, suggests that the
accumulation
2.
2. The AmyloidofHypothesis
The Amyloid pathological
Hypothesis of forms of AΆ, due to an increase of production and/or decreased
of AD
AD
clearance, is the primary pathological process in AD [3,15]. The accumulation of AΆ peptides leads
The amyloid
amyloid hypothesis,
The oligomerization
hypothesis, the
the current
current prevalent
prevalent theory
theory of AD
AD pathogenesis, suggests
suggests that
that the
to their and formation of AΆ plaques. Theseofplaques pathogenesis,
generate an anti-inflammatory the
accumulation
accumulation of pathological
of pathological forms
formsof Aβ, due to
of AΆ, and an increase
duedisrupting of
to an increaseproduction and/or
ofkinase
production decreased clearance,
and/or decreased
response causing oxidative stress in neurons normal and phosphatase activity,
is the primary
clearance, is thepathological
primary process inprocess
pathological AD [3,15].
in AD The accumulation
[3,15]. The of Aβ peptides
accumulation of AΆ leads to leads
peptides their
resulting in hyperphosphorylation of tau protein and subsequent neurofibrillary tangle formation.
oligomerization
to their and
oligomerizationformation of Aβ
and formation plaques.
of AΆ These
plaques. plaques
These generate
plaques an anti-inflammatory
generate resulting response
an anti-inflammatory
This cascade of events leads to abnormal signaling and synaptic impairment, ultimately in
causing
response oxidative
causing stress in neurons
oxidative stress inand disrupting
neurons and normal kinase
disrupting normaland phosphatase
kinase and activity, resulting
phosphatase activity,
neuronal dead and dementia in the AD patient [16]. This is schematized in Figure 2.
in hyperphosphorylation
resulting of tau protein
in hyperphosphorylation andprotein
of tau subsequent
and neurofibrillary tangle formation.
subsequent neurofibrillary tangleThis cascade
formation.
of events leads to abnormal signaling and synaptic impairment, resulting ultimately
This cascade of events leads to abnormal signaling and synaptic impairment, resulting ultimately in neuronal deadin
and dementia in the AD patient [16]. This is schematized in Figure
neuronal dead and dementia in the AD patient [16]. This is schematized in Figure 2. 2.
Figure 2. Cascade of events according the amyloid hypothesis. Adapted from [16].
AΆ is produced by a two-step sequential cleavage of amyloid precursor protein (APP): first beta-
2. Cascade
Figure enzyme
site APP-cleaving of events cleaves
(BACE-1) according the to
APP amyloid hypothesis.
solubleAdapted
hypothesis.
generate from
from [16].
APPΆ (sAPPΆ) [16].
and a 99 amino
acid fragment (C99), which then suffers several cleavage events by ·-secretase to produce peptides
Aβ is produced
AΆ is producedby bya atwo-step
two-step sequential cleavage of amyloid precursor protein (APP): first
of different lengths from 38 to 43sequential
amino acids,cleavage
beingofAΆ40
amyloidand precursor
AΆ42 the protein (APP):
main products first
andbeta-
both
beta-site
site APP-cleaving enzyme
APP-cleaving (BACE-1) cleaves toAPP to generate soluble(sAPPΆ)
APPβ (sAPPβ) and a
playing a key roleenzyme (BACE-1)
in the aggregation cleaves APPplaques
of neuritic generate soluble
[15], [17] APPΆ
(Figure 3). The longer and a 99 amino
peptide AΆ42
99 amino
acid acid fragment
fragment (C99), (C99), which then sufferscleavage
several events
cleavage byevents by γ-secretase to peptides
produce
has been shown to bewhich
the mostthen suffers
prone to several
aggregation, with an increased·-secretase
AΆ42:AΆ40 to produce
ratio observed in
peptides
of of differentfrom
different lengths from 38 to 43 amino acids, being
andAβ40
AΆ42andthe Aβ42
main the main products
the familiallengths 38 to
form of the disease 43 amino
[17,18]. acids, being AΆ40 products and both
and both key
playing playing a key role in the aggregation of neuritic plaques [15,17] (Figure 3).peptide
The longer
The aactivity
role
ofin the aggregation
Ά-cleavage of neuritic
is considered plaques
to determine [15],
the[17]
total(Figure
amount 3).of
TheAΆlonger
production, andAΆ42
the
peptide
has Aβ42 has been shown to be the most prone to aggregation, with an increased Aβ42:Aβ40 ratio
efficiency of the successive ·-cleavage impacts the production ratio of toxic AΆ42 to total AΆ [15]. in
been shown to be the most prone to aggregation, with an increased AΆ42:AΆ40 ratio observed
observed
the familialin form
the familial form of [17,18].
of the disease the disease [17,18].
The activity of Ά-cleavage is considered to determine the total amount of AΆ production, and the
efficiency of the successive ·-cleavage impacts the production ratio of toxic AΆ42 to total AΆ [15].
Figure 3. Scheme of the production of Aβ by the two step sequential cleavage of APP by β-secretase
and γ-secretase. Adapted from [15].
Figure 3. Scheme of the production of AΆ by the two step sequential cleavage of APP by Ά-secretase
andactivity
The ·-secretase. Adapted from
of β-cleavage [15].
is considered to determine the total amount of Aβ production, and the
efficiency of the successive γ-cleavage impacts the production ratio of toxic Aβ42 to total Aβ [15].
Theamyloid
Figure
The amyloid
3. Scheme hypothesis isisstrongly
of the production
hypothesis strongly supported
of AΆ supported
by bygenetic
the two step
by geneticfindings.
sequential findings. All
cleavage of APP
All the
the known
byknown familial
Ά-secretase
familial
and ·-secretase.
Alzheimer’s diseaseAdapted
(fAD) from [15].
mutations are involved in either AΆ increase generation or
Alzheimer’s disease (fAD) mutations are involved in either Aβ increase generation or processing and processing and

The amyloid hypothesis is strongly supported by genetic findings. All the known familial
Alzheimer’s disease (fAD) mutations are involved in either AΆ increase generation or processing and

result in relative overproduction of toxic forms of Aβ, namely Aβ42 [3]. On the other hand, it was
reported that a mutation in the APP gene (A673T), identified from a set of whole-genome sequence
data of 1795 Iceland people, results in a lifelong decrease in APP cleavage by β-secretase conferring a
reduced clinical risk of AD and age-related decline [19,20].
Besides the fact that occurrence of AD and genetic mutations associated with Aβ formation
supports the amyloid hypothesis, some observations remains without explanation, with histological
studies being controversial in the correlation between the formation of Aβ plaques and cognitive
decline [21].
Karran et al. [22] defend the prevention of Aβ deposition in patients that are in risk of developing
AD as a robust test of the amyloid hypothesis. If individuals are at risk of developing AD but Aβ
deposition has not yet occurred, then a significant reduction of Aβ42 production or a decrease in the
longer amyloid-β/shorter amyloid-β ratio would be expected to prevent the onset of disease within a
normal lifetime [22]. Although there are further questions concerning the exact neuronal toxicity of
Aβ, the amyloid hypothesis is still broadly accepted as the general pathological cascade of events in
AD [22,23].
3. Amyloid Targeting Strategies

Following the amyloid hypothesis, where a high brain Aβ level is observed as an important factor
in AD pathogenesis, pharmacological intervention to reduce its production or improve the clearance
has become a logical approach for AD therapy development.
Considering the 112 agents in the AD pipeline in 2018 [14], there are 29 with a mechanism of
action amyloid related, 13 corresponding to immunotherapeutic agents (Table 2) and 16 to small
molecules (Table 3) involved in phase I, II and III clinical trials.
Table 2. Immunotherapeutic agents (amyloid related) in the AD pipeline (2018).
Phase
Agent Mechanism of Action Sponsor
(Clinical Trial(s))
I, III Aducanumab Monoclonal antibody Biogen
Albumin +
III Polyclonal antibody Grifols
immunoglobulin
I, II, III Crenezumab Monoclonal antibody Roche/Genentech
II, III Gantenerumab Monoclonal antibody Roche
Washington University, Eli Lilly,
Roche, NIA, Alzheimer’s
III Solanezumab Monoclonal antibody
Association, ATRI (Alzheimer’s
Therapeutic Research Institute)
II, III CAD106 Amyloid vaccine Novartis, Amgen, NIA,
University of Colorado, Denver,
II Sargramostim (GM-CSF) Immunostimulator
The Dana Foundation
II BAN2401 Monoclonal antibody Eisai
II UB-311 Monoclonal antibody United Neuroscience
I, II LY3002813 Monoclonal antibody Eli Lilly and Company
I LY3303560 Monoclonal antibody Eli Lilly and Company
I Lu AF20513 Polyclonal antibody H. Lundbeck A/S
I KHK6640 Monoclonal antibody Kyowa Hakko Kirin Pharma
Table 3. Small molecules (amyloid related) in AD pipeline (2018).
Phase
Agent Mechanism of Action Sponsor
Clinical Trial(s)
NeuroGenetic
I NGP 555 GSM
Pharmaceuticals
Phosphatidylinositol
II ID1201 3-kinase/Akt pathway II Dong Pharmaceutical Co
activation
II Nilotinib Tyrosine kinase inhibitor Georgetown University
III CNP520 (γ-secretase modulator) Alzheimer’s Association
ALZT-OP1a (cromolyn)+
III BACE1 inhibitor AZTherapies
ALZT-OP1b (ibuprofen)
Sodium Oligo-mannurarate
III Increases amyloid clearance Shanghai Green Valley
(GV-971)
RAGE1) (Receptor for advanced
III TTP488 (Azeliragon) glycation end products) vTv Therapeutics
antagonist
II, III JNJ-54861911 BACE1 inhibitor Janssen
II, III E2609 (Elenbecestat) BACE1 inhibitor Eisai, Biogen
II LY3202626 BACE1 inhibitor Eli Lilly
Adrenergic uptake inhibitor,
II Atomoxetine Emory University, NIA
SNRI
II AZD0530 (Saracatinib) Kinase inhibitor Yale University, ATRI,
Sigma-2 receptor
II CT1812 Cognition Therapeutics
competitive inhibitor2)
Selective inhibitor of APP
II Posiphen QR Pharma, ADCS
production
II Valacyclovir Antiviral agent 4) Umea University
III AZD3293 (LY3314814) BACE1 inhibitor AstraZeneca, Eli Lilly
With regard to small molecules, different targets have been identified for activity modulation
hoping to get a marked impact in the amyloid cascade and in disease progression. Among
16 anti-amyloid agents identified in the AD pipeline, six have their mechanism of action centered in
the activity of β or γ-secretase proteases, covering almost 40% of the identified agents.
In fact, in the last years several pharmaceutical industries and other research centers have led
research programs to discovered potent compounds for the modulation or inhibition of these two
targets [16,17,24–26]. The discovery and optimization activities which lead to the development of these
compounds will be herein detailed.
3.1. Gamma Secretase Inhibitors and Modulators

Gamma secretase (GS) (Figure 4) is an aspartyl protease composed by a complex of four different
membrane proteins: presenilin (PS), presenilin enhancer 2 (Pen- 2), nicastrin (Nct), and anterior
pharynx-defective 1 (Aph-1) [27]. PS is the catalytic component of γ-secretase. In humans, PS is
encoded by the PSEN1 (PS-1) gene on chromosome 14 or the PSEN2 (PS-2) gene on chromosome 1,
and mutations in both genes have been found to cause fAD [28]. The products of these genes (PS-1
and PS-2) are nine transmembrane domain (TMD) proteins that form the catalytic subunit of GS [17].
GS cleaves several type-I transmembrane proteins (over 90 reported substrates), being APP and Notch
the best characterized substrates [29].
The activity of GS on the substrate APP occurs after the cleavage performed by β-secretase
(BACE-1). Then, GS perform a series of cleavages within the transmembranar domain of the remaining
fragment (C99), termed epsilon (ε), zeta (ζ), and gamma (γ) cleavages, allowing the generation of Aβ
peptides of different lengths (Figure 5).
Notch the best characterized substrates [29].
and mutations in both genes have been found to cause fAD [28]. The products of these genes (PS-1
and PS-2) are nine transmembrane domain (TMD) proteins that form the catalytic subunit of GS [17].
GS cleaves several
Pharmaceuticals 2019, 12,type-I
41 transmembrane proteins (over 90 reported substrates), being APP 7and
of 31
Notch the best characterized substrates [29].
Figure 4. Crystal structure of human ·-secretase. Adapted by permission from Xiao-chen Bai et al.
Nature [30], Copyright 2015.
The activity of GS on the substrate APP occurs after the cleavage performed by Ά-secretase
(BACE-1). Then, GS perform a series of cleavages within the transmembranar domain of the
remaining fragment (C99), termed epsilon (Ή), zeta (Ί), and gamma (·) cleavages, allowing the
Figure 4.
Figure 4. Crystal
Crystal structure
structureof
ofhuman
human·-secretase.
γ-secretase.Adapted
Adaptedbyby
permission from
permission Xiao-chen
from BaiBai
Xiao-chen et al.
et al.
generation of AΆ peptides of different lengths (Figure 5).
Nature [30], Copyright 2015. Nature [30], Copyright 2015.
The activity of GS on the substrate APP occurs after the cleavage performed by Ά-secretase
(BACE-1). Then, GS perform a series of cleavages within the transmembranar domain of the
remaining fragment (C99), termed epsilon (Ή), zeta (Ί), and gamma (·) cleavages, allowing the
generation of AΆ peptides of different lengths (Figure 5).
Figure 5. Cleavage
Figure steps
5. Cleavage of C99
steps by GS.
of C99 Adapted
by GS. withwith
Adapted permission fromfrom
permission [17].[17].
Copyright 20162016
Copyright American
Chemical Society. American Chemical Society
The ε cleavage (1) releases the APP intracellular domain (AICD) and produces Aβ49 or
The Ή cleavage (1) releases the APP intracellular domain (AICD) and produces AΆ49 or AΆ48
Aβ48 [28,31]. Then, the carboxypeptidase cleavages ζ (2) and γ (3 and 4) progressively trims these
[28,31]. Then, the carboxypeptidase cleavages Ί (2) and · (3 and 4) progressively trims these longer
longer Aβ forms in both Aβ40 and Aβ42 [31,32]. The successive cleavage events performed by GS
AΆ forms in both AΆ40 and AΆ42 [31], [32]. The successive cleavage events performed by GS consists
consists Figure
in four5.cycles
Cleavage steps of C99
to generate by GS.
Aβ40 Adapted with
(49-46-43-40) permission
and from [17]. Copyright
Aβ38 (48-45-42-38). Further2016
cleavage will
in four cycles to generate AΆ40 (49-46-43-40) and AΆ38 (48-45-42-38). Further cleavage will
American Chemical
subsequently generate the shorter isoforms Aβ39and Aβ37 [33]. Society
subsequently generate the shorter isoforms AΆ39and AΆ37 [33].
Many fAD-causing mutations in PS have been found to decrease the catalytic activity of GS [34,35]
Many fAD-causing(1) mutations in PS have been found to decrease theand
catalytic activity of GS [34],
with the Ήmost
The cleavage
pronouncedreleases
effectthe
on APP intracellular
the fourth cleavagedomain (AICD)
cycle [36,37]. produces
This loss AΆ49
of function or AΆ48
contributes
[35] with
[28,31]. the most pronounced
Then, the Aβ42:Aβ40
carboxypeptidase effect on
cleavagesthe fourth cleavage cycle [36,37]. This loss of function
to the increased ratio observed inΊthe
(2) familial
and · (3form
and of
4) the
progressively
disease [38].trims
Aβ42these longer
is consider
contributes
AΆ forms intoboth
the AΆ40
increased
and AΆ42:AΆ40
AΆ42 [31], ratioThe
[32]. observed in the
successive familial
cleavage form performed
events of the diseaseby [38].
GS AΆ42
consists
the most toxic Aβ isoform due to its high propensity to form fibrillary and non-fibrillary aggregates.
in four cycles to generate AΆ40 (49-46-43-40) and AΆ38 (48-45-42-38). Further cleavage will
On the other hand, shorter Aβ peptides are speculated to be less toxic or even neuroprotective [33].
subsequently generate the shorter isoforms AΆ39and AΆ37 [33].
3.1.1.
Many Inhibitors
fAD-causing
γ-Secretase mutations in PS have been found to decrease the catalytic activity of GS [34],
[35] with the most pronounced effect on the fourth cleavage cycle [36,37]. This loss of function
Based on the genetic evidence of the role of GS and Aβ formation on AD, the pharmaceutical
contributes to the increased AΆ42:AΆ40 ratio observed in the familial form of the disease [38]. AΆ42
industry made efforts in developing compounds able to inhibit this protease and, consequently, reduce
the amount of Aβ peptide formation. Two key γ-secretase inhibitors (GSIs), semagacestat (1) from
Eli Lilly & Co and later avagacestat (2) from Bristol-Myers Squibb (Figure 6) advanced in late-stage
clinical trials for AD (phase III and phase II, respectively) [39,40].
Based on the genetic evidence of the role of GS and AΆ formation on AD, the pharmaceutical
industry made efforts in developing compounds able to inhibit this protease and, consequently,
reduce the amount of AΆ peptide formation. Two key ·-secretase inhibitors (GSIs), semagacestat (1)
from Eli Lilly2019,
Pharmaceuticals & Co and later avagacestat (2) from Bristol-Myers Squibb (Figure 6) advanced in8late-
12, 41 of 31
stage clinical trials for AD (phase III and phase II, respectively) [39,40].
&O
2 6 22
1
1+
)

) )
)
1 1
2

Figure 6. ·-Secretase
γ-Secretase inhibitors (1–2) advanced in late-stage clinical trials.
Unfortunately, due to GS high promiscuity in terms of substrates, several side effects were
observed throughout the clinical trials with these compounds [25]. The main responsible pointed for
these adverse effects was Notch, a single-pass type I transmembrane receptor [41]. Notch is processed processed
in a similar way as C99, where the Notch intracellular domain (NICD) is released after the Ή-cleavage ε-cleavage
and translocate to the nucleus to regulate gene expression and to mediate important intercellular
communication functions
communication functions as cellular differentiation. Additionally,
Additionally, Notch
Notch pathway
pathway is is involved in
neurogenesis, neuritic
neurogenesis, neuritic growth
growthand andlong-term
long-termmemorymemory[41].
[41].The
Theinhibitory
inhibitory effect
effect ofof GSIs
GSIs blocks
blocks thethe
Ή-
ε-cleavage
cleavage ofof affectingthe
γ-secretaseaffecting
·-secretase theNotch
Notchsignaling,
signaling,leading
leadingtotoseveral
severaladverse
adverseevents
eventsas asskin
skincancer,
cancer,
decreased lymphocyte count and/or memoryloss.
and/or memory loss. These
Theseadverse
adverse effects
effects were
were observed
observed in in phase III
clinical trial of semagacestat (1) [40].
subsequent drug-development
Consequently, subsequent drug-development programsprograms have have aimed
aimed to to achieve
achieve a greater
separation between APP and Notch inhibition. Bristol-Myers Squibb
separation between APP and Notch inhibition. Bristol-Myers Squibb developed avagacestat developed avagacestat (2)
(2) and
and reported
reported it as having
it as having a 137-fold
a 137-fold selectivity
selectivity for APP forover
APPNotch
over Notch in cell culture,
in cell culture, and to robustly
and to robustly reduce
reduce cerebrospinal
cerebrospinal fluid (CSF)
fluid (CSF) Aβ levels
AΆ levels without
without causing
causing Notch-related
Notch-related toxicity
toxicity in in ratsand
rats anddogs
dogs[42].
[42].
However, aaphase
However, phaseII II
clinical trial
clinical performed
trial performed withwith
avagacestat demonstrated
avagacestat a side-effect
demonstrated profile similar
a side-effect profile
to the one
similar found
to the onewith
found semagacestat (1) [39].(1)
with semagacestat These
[39]. results were attributed
These results to an overestimation
were attributed of the
to an overestimation
selectivity of avagacestat between Notch and APP, with other researchers indicating
of the selectivity of avagacestat between Notch and APP, with other researchers indicating an only an only 3-fold
selectivity
3-fold for cleavage
selectivity of an APP
for cleavage of ansubstrate compared
APP substrate with a with
compared Notch substrate
a Notch [43]. [43].
substrate
The disappointing
disappointingresults obtained
results obtainedwith with
these two
theseGSI led to
two GSIdiscontinuation of the development
led to discontinuation of the
development of GSIs as therapeutic strategy against AD. Additionally, other options assafer
of GSIs as therapeutic strategy against AD. Additionally, other options as GS modulators arose as GS
disease modifying
modulators arose as therapies for AD.
safer disease modifying therapies for AD.
3.1.2. Gamma
3.1.2. Gamma secretase
secretase Modulators
Modulators
A γ-secretase
A ·-secretase modulator
modulator (GSM)
(GSM)isisdefined
definedasasaacompound
compoundthatthatchanges
changesthe relative
the proportion
relative of
proportion
thethe
of AβAΆ
isoforms generated
isoforms while
generated maintaining
while maintainingthethe
rate at which
rate APP
at which APPis processed [33].
is processed [33].
Nonsteroidal Anti-inflammatory Drug Derived GSM’s

The first generation of GSMs was discovered from an epidemiological study documenting a
reduced prevalence of AD among users of nonsteroidal anti-inflammatory drugs (NSAIDs) [44]. Then,
in 2001 Weggen et al. [44] reported that NSAIDs (Figure 7) were able to decrease the Aβ42 peptide
accompanied by an increase in the Aβ38 isoform, indicating that NSAIDs modulate γ-secretase activity
without significantly perturbing other APP processing pathways or Notch cleavage [44].
The
The first
first generation
generation of of GSMs
GSMs waswas discovered
discovered from
from anan epidemiological
epidemiological study
study documenting
documenting aa
reduced
reduced prevalence
prevalence of of AD
AD among
among users
users of
of nonsteroidal
nonsteroidal anti-inflammatory
anti-inflammatory drugs
drugs (NSAIDs)
(NSAIDs) [44].
[44].
Then, in 2001 Weggen et al. [44] reported that NSAIDs (Figure 7) were able to decrease
Then, in 2001 Weggen et al. [44] reported that NSAIDs (Figure 7) were able to decrease the AΆ42 the AΆ42
peptide
peptide accompanied
accompanied by by an
an increase
increase in
in the
the AΆ38
AΆ38 isoform,
isoform, indicating
indicating that
that NSAIDs
NSAIDs modulate
modulate ·- ·-
secretase
secretaseactivity
activitywithout
withoutsignificantly
significantlyperturbing
perturbing other
otherAPP
APPprocessing
processingpathways
pathwaysor orNotch
Notchcleavage
cleavage
[44].
[44].
33 44 55
Ibuprofen
Ibuprofen Indomethacin
Indomethacin Sulindac
Sulindac sulfide
sulfide
AΆ
AΆ4242IC
IC5050== 200-300
200-300 μM
μM AΆ
AΆ4242IC
IC5050== 25-50
25-50 μM
μM AΆ
AΆ42 IC50 = 34
42 IC50 = 34 μM
μM
Figure
Figure 7.
7.NSAIDs
7. NSAIDs
NSAIDs GSMs
GSMs 3–5
GSMs and
3–53–5 respective
andand AΆ42
respective
respective half-maximal
Aβ42
AΆ42 inhibitory
half-maximal
half-maximal concentration
inhibitory
inhibitory (IC5050))values
(IC
concentration
concentration (IC50 )
values
[44].
values
[44]. [44].
This
This change
This change
change in in
in the
the cleavage
cleavage pattern
pattern could
could be be explained
explained by by (1)
1) aaa decrease
1) decrease in
decrease in the
in the probability
the probability
probability of of
of
releasing
releasing longer
releasing longer
longer AΆ Aβ
AΆ fromfrom
from the the enzyme-substrate
the enzyme-substrate
enzyme-substrate complex complex (defined
complex (defined
(defined as as dissociation
as dissociation
dissociation constant, constant,
constant, Ύκ Ύdd)))of
ofGS,
of GS,
GS,
or
or (2)
or 2)
2) ananincrease
an increasein
increase ininthe
thecleavage
the cleavageactivity
cleavage activity(defined
activity (definedby
(defined bythe
by thecatalytic
the catalyticconstant,
catalytic constant,ΎΎcat
constant, κcatcat ) of
)) of
of GS.
GS.GS. Using
Using
Using C99
C99 C99asas
as aa
a substrate,
substrate, Chavez-Gutierrez
substrate, Chavez-Gutierrez
Chavez-Gutierrez et et
et al.al. [36]
al. [36] showed
[36] showed
showed that that the Aβ40/Aβ43
that the AΆ40/AΆ43
AΆ40/AΆ43 and and Aβ38/Aβ42
and AΆ38/AΆ42
AΆ38/AΆ42 ratios ratios
ratios werewere
were
decreased
decreased by
decreased byallallthe
all thePS
the PSmutations
PS mutationsstudied,
mutations studied,suggesting
studied, suggesting
suggesting anandeterioration
an deterioration
deterioration of of
the
of the four
the four
fourcleavage
cleavage
cleavage cycle of GS
cycle
cycle of
of
(γ’
GS cleavage).
(·’ cleavage).Additionally,
Additionally, using Aβ42
using as
AΆ42a substrate,
as a Okochi
substrate, et al.
Okochi
GS (·’ cleavage). Additionally, using AΆ42 as a substrate, Okochi et al. [35] measured the kinetic [35]
et measured
al. [35] the
measured kinetic theconstants
kinetic
for the γ cleavage
constants
constants for the and
for the reportedand
·· cleavage
cleavage thatreported
and GSMs decreased
reported that
that GSMsκd and
GSMs increasedΎΎdκ
decreased
decreased and
and
d cat , while
increased
increasedPS mutations
ΎΎcat
cat,, whilecaused
while PS
PS
the opposite
mutations
mutations effect.
caused
caused theIn
the sum, these
opposite
opposite effect.
effect. results
In
Insum,
sum, suggest
these that the
theseresults
results modulation
suggest
suggest that theeffect
thatthe modulation
modulation of GSMs could
effect
effect of be an
ofGSMs
GSMs
effective
could
could be approach
be anan effective
effective toapproach
reverse the
approach toeffect
to reverse
reverseof PS
themutations
the effect
effect of
of PS
PSbymutations
restoring by
mutations the normal ratios
by restoring
restoring the of Aβ peptides
the normal
normal ratios
ratios of of
formed,
AΆ
AΆ peptides and formed,
peptides in this way
formed, andmodifying
and in
in this
this way
way the disease pathology
modifying
modifying the
the disease
disease and progression
pathology
pathology and[17].
and progression
progression [17]. [17].
Though,
Though,
Though, and and as
and as expected,
as expected,
expected, the the use
the use of
use of NSAID
of NSAID as
NSAID as GSMs
as GSMs present
GSMs present
present some some
some issues issues
issues in in terms
in terms
terms of of
of
gastrointestinal
gastrointestinal and
gastrointestinal andrenal renal toxicity
renaltoxicity
toxicitydue due
dueto to its
toits activity
itsactivity against
activityagainst cyclooxygenase
againstcyclooxygenase
cyclooxygenase11(COX-1), 1 (COX-1), compromising
(COX-1),compromising
compromising
its
its use
its use
use asasaaalong-term
as long-termtherapeutic
long-term therapeuticsolution
therapeutic solution[45].
solution [45].Fortunately,
[45]. Fortunately,
Fortunately, COX-1
COX-1
COX-1 inhibition
inhibition
inhibition activity
activity
activity was
was was shown
shownshown to
to
to
be be independent
independent of of γ-secretase
·-secretase modulation
modulation activity.
activity. For
For instance,
instance,
be independent of ·-secretase modulation activity. For instance, flurbiprofen is a COX-1 inhibitor flurbiprofen
flurbiprofen isis aa COX-1
COX-1 inhibitor
inhibitor
administered
administeredin
administered inthe
in theclinic
the clinicasas
clinic a racemate.
as aa racemate.
racemate. However,
However,
However, the R-enantiomer,
the
the R-enantiomer, tarenflurbil
R-enantiomer, tarenflurbil(6) (Figure
tarenflurbil (6) 8),
(6) (Figure showed
(Figure 8),
8),
reduced
showed activity in
showed reduced
reduced termsin
activity
activity ofterms
in COX-1of
terms ofinhibition
COX-1 while maintaining
COX-1 inhibition
inhibition while its actionits
while maintaining
maintaining asaction
its a GSMas
action as[45].
aa GSM
GSM [45]. [45].
66
Figure
Figure 8.
8. Tarenflurbil
Tarenflurbil (R-flurbiprofen)
(R-flurbiprofen) (6).
(R-flurbiprofen) (6).
Tarenflurbil
Tarenflurbil (6,
Tarenflurbil Flurizan™,
(6, Flurizan™, Myriad
Flurizan™, Myriad Genetics
Myriad Genetics
Genetics & & Laboratories)
& Laboratories) reached phase
Laboratories) reached
reached phase III
phase III clinical
III clinical trials.
clinical trials.
trials.
Unfortunately,
Unfortunately, results
Unfortunately, results showed
results showed
showed nono difference
no difference between
difference between and
between 666 and
and the placebo,
the placebo, the
placebo, the
the failure being
failure being attributed
being attributed
attributed
to
to the insufficient PD properties of Flurizan™, namely its inadequate capacity
to the insufficient PD properties of Flurizan™, namely its inadequate capacity to penetrate the
the insufficient PD properties of Flurizan™, namely its inadequate capacity to to penetrate
penetrate the
the brain
brain
brain
and and
and engage engage
engage its its
its target target
target at at
at dosesdoses sufficient
doses sufficient
sufficient to to
to yieldyield
yield an an effect.
an effect. Actually,
effect. Actually,
Actually, poor poor
poor CNS CNS penetration
CNS penetration
penetration of of
of tarenflurbil (6) was previously reported in preclinical studies in rodents, with a CSF/plasma ratio
of
1.3%. The unsatisfactory results obtained led to the discontinuation of tarenflurbil (6) clinical
development [46].
Later, two NSAID carboxylic acid derivatives developed by Chiesi (CHF5074, 7) and FORUM
Pharmaceuticals (EVP-0015962, 8) (Figure 9) were also tested in clinical trials.
tarenflurbil (6) was previously reported in preclinical studies in rodents, with a CSF/plasma ratio of
1.3%. The unsatisfactory results obtained led to the discontinuation of tarenflurbil (6) clinical
development [46].
Later, two NSAID carboxylic acid derivatives developed by Chiesi (CHF5074, 7) and FORUM
Pharmaceuticals (EVP-0015962, 8) (Figure 9) were also tested in clinical trials.
7 7 8 8
CHF5074 (Chiesi)
CHF5074 (Chiesi) EVP-00150962 (FORUM Pharmaceuticals)
EVP-00150962 (FORUM Pharmaceuticals)
AΆ42 IC50=40 μM
Aβ42 IC50=40 µM AΆ42 IC50= 67 μM
Aβ42 IC50= 67 µM
Figure 9.
Figure 9. CHF5074
CHF5074 (7)
(7) and
and EVP-0015962
EVP-0015962 (8).
(8).
Chiesi Farmaceutici
Chiesi Farmaceutici developed
developed CHF5074
CHF5074(7)(7)based
basedon on
the the
scaffold of tarenflurbil
scaffold (6). The
of tarenflurbil (6).
replacement
The of theofR-methyl
replacement substituent
the R-methyl by a cyclopropyl
substituent group group
by a cyclopropyl led to aledcomplete removalremoval
to a complete of COX
inhibition
of (at 100 μM
COX inhibition (at for
100COX-1
µM forand 300 μM
COX-1 andfor
300COX-2).
µM for AΆ42 production
COX-2). inhibition inhibition
Aβ42 production were improved
were
with the addition of chlorine substituents on the terminal phenyl ring, allowing
improved with the addition of chlorine substituents on the terminal phenyl ring, allowing a 7-fold comparing
a 7-fold
with tarenflurbil
comparing (6) [47] (Figure
with tarenflurbil 10).
(6) [47] The carboxylic
(Figure acid function
10). The carboxylic was proposed
acid function to interact
was proposed with a
to interact
lysine residue of APP located close to the membrane interface, with the lipophilic substituents
with a lysine residue of APP located close to the membrane interface, with the lipophilic substituents serving
as membrane
serving anchorsanchors
as membrane [48]. [48].
Figure
Figure 10. Improvement activity
10. Improvement activity and
and selectivity
selectivity of
of CHF5074
CHF5074 (7)
(7) against
against tarenflurbil
tarenflurbil (6).
(6).
Despite
Despite the
the increase
increase in
in potency
potency and
and selectivity
selectivity accomplished
accomplished withwith CHF5074
CHF5074 (7),(7), no
no sufficient
sufficient
improvements were achieved in terms of PK parameters. CHF5074 (7) failed
improvements were achieved in terms of PK parameters. CHF5074 (7) failed to demonstrate efficacy to demonstrate
efficacy
due to itsdue todruglike
poor its poor properties
druglike properties and extremely
and extremely poor CNS penetration
poor CNS penetration (brainratio
(brain to plasma to plasma
= 0.03–
ratio = 0.03–0.05)
0.05) [49]. [49].
EVP-0015962
EVP-0015962(8)(8)was developed
was by FORUM
developed Pharmaceutical
by FORUM (ex Envivo)
Pharmaceutical (exthrough
Envivo)thethrough
introduction
the
of additional substituents on the biphenylacetic acid core of R-flurbiprofen. This compound
introduction of additional substituents on the biphenylacetic acid core of R-flurbiprofen. This showed
improved
compoundinshowed
vitro potency (Aβ42
improved IC50 =
in vitro 67 µM)(AΆ42
potency with reproducible
IC50 = 67 μM)animal efficacy, with
with reproducible a reduction
animal of
efficacy,
22% and 38% with oral doses of 10 and 30 mg/kg, respectively [33]. However, as general
with a reduction of 22% and 38% with oral doses of 10 and 30 mg/kg, respectively [33]. However, as characteristic
of NSAIDcharacteristic
general derived GSMs,ofit presents
NSAID suboptimal properties
derived GSMs, as high lipophilicity,
it presents suboptimal which could result
properties as highin
low free fraction and poor solubility [17,33]. Additionally, the presence of the carboxylic
lipophilicity, which could result in low free fraction and poor solubility [17,33]. Additionally, the moiety can
lead to the formation of reactive metabolites such as acyl glucuronides, compounds chemically reactive

leading to covalent binding with macromolecules and toxicity [49,50]. EVP-0015962 (8) move on to
a phase II clinical trial in 2012 but results were not publicly reported. The latest update information
in the platform clinicaltrials.org was in January 2014. Additionally, EVP- 0015962 (8) is not part of
1
presence
presence of of the
the carboxylic
carboxylic moiety
moiety cancan lead
lead to
to the
the formation
formation of of reactive
reactive metabolites
metabolites such
such as
as acyl
acyl
glucuronides,
glucuronides, compounds
compounds chemically reactive leading to covalent binding with macromolecules
chemically reactive leading to covalent binding with macromolecules and
and
toxicity
toxicity [49],
[49], [50].
[50]. EVP-0015962
EVP-0015962 (8) (8) move
move onon to
to aa phase
phase IIII clinical
clinical trial
trial in
in 2012
2012 but
but results
results were
were not
not
publicly
publiclyreported.
reported.The Thelatest
latestupdate
updateinformation
informationin inthe
theplatform
platformclinicaltrials.org
clinicaltrials.orgwaswasininJanuary
January2014.
2014.
Additionally,
Additionally, EVP- 0015962 (8) is not part of the AD pipeline in 2018, leading to the assumptionthat
the AD pipelineEVP- in 0015962
2018, (8)
leadingis not
to part
the of the
assumption AD pipeline
that the in 2018,
clinical leading
developmentto the assumption
of this compoundthat
the
theclinical
was development
developmentof
discontinued.
clinical ofthis
thiscompound
compoundwas wasdiscontinued.
discontinued.
Non-NSAID
Non-NSAID Derived
Non-NSAIDDerived GSM’s
DerivedGSM’s
GSM’s
One
One of
Oneof the
ofthe firstGSM
thefirst
first GSMseries
GSM seriesnot
series notpresenting
not presentingaaacarboxylic
presenting carboxylicacid
carboxylic acidmoiety
acid moiety(non-NSAID)
moiety (non-NSAID)was
(non-NSAID) was developed
wasdeveloped
developed
by
by Neurogenetics
byNeurogenetics
Neurogeneticsinin 2004,
in2004, leading
2004,leading to
leadingto the
tothe discovery
thediscovery
discoveryofof NGP555
ofNGP555 (Figure
NGP555(Figure 11) [33].
(Figure 11) [33].
11
++ 11
11 11
66 ))
11
11
Figure
Figure11.
11.Structure
Structureof
ofNGP555
NGP555(11).
(11).
This
Thisclass
This classof
class ofcompounds
of compoundsshare
compounds shareaa
share ascaffold
scaffoldconsisting
scaffold consistingof
consisting offour
of fourconsecutives
four consecutiveslinked
consecutives linked(hetero)aromatic
linked (hetero)aromatic
(hetero)aromatic
rings
rings identified
rings identified as
identified as A,
as A,B, C
B, C
A, B, and
C and D
and D which
D which focus
which focus on
focus on aryl-
onaryl- or
aryl-or heteroarylimidazoles
orheteroarylimidazoles
heteroarylimidazoles with with an
ananilinothiazole.
with an anilinothiazole.
anilinothiazole.
Figure
Figure 12
12 represents
represents thethe basic
basic scaffold
scaffold and and
the the structure-activity
structure-activity relationship
relationship
Figure 12 represents the basic scaffold and the structure-activity relationship established. established.
established. The
The addition
The
addition
of of
a methyl
addition a
of ormethyl or
a halogen
a methyl a halogen
or aathalogen at the
the 4-position 4-position of
of the imidazole
at the 4-position the imidazole ring
ring does not
of the imidazole does
ringhave not
doesanot have a
significant significant
impact on
have a significant
impact
impacton
potency, potency,
while
on while
whilethe
the addition
potency, theofaddition of
ofaaCF
a CF3 substituent
addition CF33substituent in
inring
in ring B, ortho
substituent B,
B,ortho
to the
ring to
tothe
thiazole,
ortho thiazole,
leads
the leads
leadsto
to a decrease
thiazole, aa
toin
decrease
activity.
decreaseA in activity.
activity.A
inpyridine Apyridine
or pyrimidine
pyridine or
orpyrimidine
ring at B ring
pyrimidine ring
ring at
atBBring
increases increases
ringpotency,
increasesaspotency, as
well as the
potency, aswell as
asthe
addition
well ofaddition
the an ortho
addition
of an
methylortho methyl
substituent
of an ortho substituent
in the aniline
methyl substituent in the aniline
in [51]. [51].
the aniline [51].
Structure-activity relationship(SAR)
Figure 12.Structure-activity
Figure (SAR) ofnon-NSAID
non-NSAID GSMs four
four rings scaffold.
scaffold.
Figure12.
12. Structure-activityrelationship
relationship (SAR)of
of non-NSAIDGSMs
GSMs fourrings
rings scaffold.
Compound NGP555
Compound (11) from Neurogenetics (Figure 11) showed a decrease in CSF Aβ 42 between
Compound NGP555NGP555 (11) (11) from
from Neurogenetics
Neurogenetics (Figure (Figure 11)11) showed
showed aa decrease
decrease inin CSF
CSF AΆAΆ4242
20–40%
between and an increase of the shorter forms in rodent studies. Additionally, it demonstrated protection
between20–40%
20–40%and andan
anincrease
increaseof ofthe
theshorter
shorterforms
formsin inrodent
rodentstudies.
studies.Additionally,
Additionally,ititdemonstrated
demonstrated
from cognitive
protection decline in decline
two independent mouse studies using different memory memory
and learning
protection from cognitive decline in two independent mouse studies using
from cognitive in two independent mouse studies using different
different memory and and
tasks [51].
learning NGP 555 (11) entered in phase I clinical trials in 2015. According to a press release
release
learningtasks
tasks[51].
[51].NGP
NGP555 555(11)
(11)entered
enteredin inphase
phaseIIclinical
clinicaltrials
trialsinin2015.
2015.According
Accordingto toaapress
press release
from Neurogenetics
from Neurogenetics on January
January 2017, NGP555 NGP555 showed to to be safesafe and well-tolerated
well-tolerated in healthy
healthy
from Neurogenetics on on January 2017, 2017, NGP555 showed showed to be be safe and
and well-tolerated in in healthy
volunteers
volunteers [52]. Detailed results and future clinical studies with this compound have not been
volunteers [52].
[52]. Detailed
Detailed results
results and
and future
future clinical
clinical studies
studies with
with this
this compound
compound havehave notnot been
been
disclosed yet.
disclosed yet.
disclosed yet.
Based on
Based this A-DA-Dscaffold, Eisai Pharmaceuticals developed a series of patented diarylcinnamide
Based on on this
this A-D scaffold,
scaffold, Eisai
Eisai Pharmaceuticals
Pharmaceuticals developed
developed aa series
series ofof patented
patented
derivatives (12–15,
diarylcinnamide Figure 13) [53], where the aminothiazole group present in Neurogenetics series was
diarylcinnamide derivatives (12–15, Figure 13) [53], where the aminothiazole group present
derivatives (12–15, Figure 13) [53], where the aminothiazole group present in in
replaced by anseries
Neurogenetics α, β-unsaturated
was replaced amide
by anor΅,aΆ-unsaturated
piperidone. amide or a piperidone.
Neurogenetics series was replaced by an ΅, Ά-unsaturated amide or a piperidone.

12
12 13
13
14
14 15
15
Figure 13. Examples of Eisai cinnamides.
Beyond the common A-D scaffold, the molecules 12–15 developed shared an hydrogen bond
Beyond the
Beyond the common
common A-DA-D scaffold,
scaffold, the molecules
molecules 12–15
12–15 developed
developed shared
shared an hydrogen bondbond
donor (as a α, β-unsaturated amide or thea piperidone) suggesting the importanceanofhydrogen
a hydrogen bond
donor (as
donor (as a ΅,
΅, Ά-unsaturated
Ά-unsaturated amide
amide oror aa piperidone)
piperidone) suggesting
suggesting the
the importance
importance ofof a hydrogen bond
bond
donor ina this region [33]. The work around this cinnamide series from Eisai leadatohydrogen
the discovery of
donor
donor in this region
in this region [33]. The
[33]. The work
work around this cinnamide series from Eisai lead to the discovery of
the clinical compounds E2012 (16)around
and E2212this (17)
cinnamide
(Figure series from Eisai lead to the discovery of
14) [33].
the clinical compounds E2012 (16) and E2212 (17) (Figure
the clinical compounds E2012 (16) and E2212 (17) (Figure 14) [33].14) [33].
16
16 17
17
E2012
E2012 E2212 (predicted structure)
E2212 (predicted structure)
Figure 14. Structures of E2012 (16) and E2212 (17, predicted structure).
Figure 14.14.
Figure Structures of E2012
Structures (16)(16)
of E2012 and E2212
and (17,(17,
E2212 predicted structure).
predicted structure).
E2012
E2012
E2012 (16)
(16)
(16) decreased
decreased
decreased levels
levels
levels of of
of AΆ40
AΆ40Aβ40 and
and and AΆ42
AΆ42 Aβ42in rat
in ratrat
in CSF,
CSF, CSF,brain
brain
brain and
andand plasma
plasma
plasma in in
in vivo
vivovivoin in
in aa dose
dose
a dose
dependent
dependent manner.
manner. The reported
The
dependent manner. The reported IC50 values reportedIC 50 ICvalues
50
of
values E2012
of (16)
E2012 for
(16)AΆ40
for and
Aβ40 AΆ42
and
of E2012 (16) for AΆ40 and AΆ42 were 330 and 92 nM, were
Aβ42 330
were and 33092 nM,92
and
respectively
nM, respectively
respectively [54]. In
[54]. In[54].
rat CSF,
rat CSF, E2012
In ratE2012 (16) significantly
CSF, E2012
(16) significantly
(16) significantly decreased
decreased AΆ42 Aβ42
decreased
AΆ42 levelslevels
levels by 16.6%
by 16.6% and 47.2%
by 16.6%
and 47.2%
and 47.2%at
at
doses
doses of
at doses10
of 10 of and
and 30
1030andmg/kg,
30 mg/kg,
mg/kg, respectively.
respectively.
respectively. It was
It was also also
It was revealed that
also revealed
revealed E2012
that E2012 (16)
that(16)
E2012 reduced AΆ40
(16) reduced
reduced AΆ40 and andAβ40 AΆ42
AΆ42and
and increased shorter AΆ peptides, such as AΆ37 and AΆ38,
and increased shorter AΆ peptides, such as AΆ37 and AΆ38, without changing total amount of AΆ of
Aβ42 and increased shorter Aβ peptides, such as Aβ37 and without
Aβ38, changing
without total
changing amount
total of
amount AΆ
peptides
peptides [55]. E2012
Aβ peptides
[55]. E2012 (16) was
[55]. E2012
(16) was the
(16)the
was first non-carboxylic
thenon-carboxylic
first first non-carboxylic acid to
acid to enter
acid in clinical
to enter
enter in clinical trialstrials
in clinical
trials in 2006
in 2006
in 2006and and
and it
it
showed
showed to
it showed be efficacious
to betoefficacious
be efficacious in reduce
in reduce
in reduce plasma
plasma plasma levels
levels of
levels AΆ42
of Aβ42
of AΆ42 of ~ 50%
of ~of50% ~ 50%in a phase
in ainphase
a phase I clinical
I clinical
I clinical trial
trial [56].
trial [56].
[56].
However,
However, lenticular
lenticular opacity
opacity waswasobserved
observed in
in a ahigh-dose
high-dose
However, lenticular opacity was observed in a high-dose group of a 13-week preclinical group
group of
of aa 13-week
13-week preclinical
preclinical safety
safety study
study in
in rats, running
in rats, running in
running in parallelparallel to the phase I study leading to the suspension
parallel to the phase I study leading to the suspension of the clinical study. of the clinical study. Follow-
study. Follow-up
Follow-
up studies
studies upup toto the
the highest
highest dose
dose tolerated
tolerated in in monkeys
monkeys for for E2012
E2012
up studies up to the highest dose tolerated in monkeys for E2012 (16) did not show ocular toxicity (16)
(16) diddid
notnotshowshow ocular
ocular toxicity
toxicity [57].
[57].
[57]. However,
However,
However, EisaiEisai
Eisai decided
decided
decided to develop
to develop
to develop theirtheir improved
improved
their improved E2212E2212
E2212 compound
compound
compound (17),
(17), instead
instead
(17), instead [17].[17].
[17].
E2212
E2212
E2212 (17)
(17) entered
(17) entered
entered in in
in aa phase
phase
a phase II clinical
clinical
I clinical trial
trial
trial in in
in 2010
2010
2010 (clinicaltrials.gov
(clinicaltrials.gov
(clinicaltrials.gov Identifier:
Identifier:
Identifier: NCT01221259).
NCT01221259).
NCT01221259).
It demonstrated
It demonstrated to have
to have a similar
a similarpharmacological
pharmacological profile
profile
It demonstrated to have a similar pharmacological profile as E2012 (16) and a better safety as E2012
as E2012 (16) and
(16) anda better
a bettersafety
safetyprofile,
profile,
profile,
with
with
with no
nono clinically
clinically
clinically significant
significant
significant ophthalmologic
ophthalmologic
ophthalmologic findings
findings
findings [58].
[58].[58].The
TheThe PD
PDPD response
response
response measured
measured
measured in
in inplasma
plasma
plasma
increased
increased
increased with
withwith thethe
the dose
dose
dose and
and and was
was
was shown
shown
shown toto
to perform
perform
perform 54%
54%54% AΆ42
Aβ42
AΆ42 reduction
reduction
reduction at at
at the
the the250
250250 mg
mg dose.
mgdose. NoNo
dose.No
further development
furtherdevelopment
further development has has beenbeenreported
been reported
reported to dateto date
to date
for E2212for (17)
for E2212
E2212 (17) and
by Eisai
(17) by the
by Eisai
Eisai and the
platform
and the platform
clinicaltrials.gov
platform
clinicaltrials.gov
does not present
clinicaltrials.gov does
doesnew not present
not studies.
present The new studies.
newpredicted
studies. The The predicted
structure
predicted structure
of E2212 of
(17),of
structure E2212
possessing
E2212 (17),(17), possessing
four aromaticfour
possessing four
rings,
aromatic rings, a high molecular weight (480 g/mol)
aromatic rings, a high molecular weight (480 g/mol) and a high cLogP (5.5) [17] could have and a high cLogP (5.5) [17] could have
conditioned its
conditioned its further
further development
development due due to to its
its poor
poor drug-like
drug-like properties.
properties.

a high molecular
Pharmaceuticals weight
2018, 11, (480 g/mol) and a high cLogP (5.5) [17] could have conditioned its further 13 of 31
development due to its poor drug-like properties.
Other
Otherpharmaceutical
pharmaceuticalcompanies,
companies,namely
namelySchering,
Schering,Roche,
Roche,AstraZeneca,
AstraZeneca,Bristol-Myers
Bristol-MyersSquibb,
Squibb,
Amgen, between others, had their own programs for the development of
Amgen, between others, had their own programs for the development of non-NSAID derived non-NSAID derived
GSM GSM[59].
[59]. The scaffold
The scaffold of theoffour
the consecutive
four consecutive
linkedlinked rings firstly
rings firstly reported
reported by Neurogenetics
by Neurogenetics can bein
can be found found
most
in most
of the of the non-NSAID
non-NSAID compounds
compounds developed developed
[33]. [33]. they
Consequently, Consequently,
share the samethey share the
suboptimal same
drug-like
suboptimal drug-like properties as the first non-NSAID derived GSM, leading to a
properties as the first non-NSAID derived GSM, leading to a lack of compounds reaching clinicallack of compounds
reaching clinical
evaluation. evaluation.
The lack The lackabout
of information of information about
the structural the structural
characteristics ofcharacteristics
γ-secretase hasofbeen
·-secretase
pointed
has been pointed as an important factor for the difficulty in developing potent GSM
as an important factor for the difficulty in developing potent GSM compounds [33]. The recently compounds [33].
The recently solved structure of gamma-secretase bound to the C99 fragment
solved structure of gamma-secretase bound to the C99 fragment could help the development of potentcould help the
development
and selective of
GSMpotent and selective
compounds [60]. GSM compounds [60].
3.2.
3.2.BACE-1
BACE-1Inhibitors
Inhibitors
BACE-1
BACE-1isis aa type-1
type-1 membrane-anchored
membrane-anchoredaspartyl
aspartylprotease
proteaseresponsible
responsiblefor
forthe
thefirst
firststep
stepof
of the
the
proteolysis of APP, identified in 1999 [61]. BACE-1 cleaves APP in the luminal surface of the
proteolysis of APP, identified in 1999 [61]. BACE-1 cleaves APP in the luminal surface of the plasma plasma
membrane
membrane andandreleases
releases the
thesoluble
solubleectodomain
ectodomain of
ofAPP,
APP, leaving
leaving C99
C99 (AΆ
(Aβplus
plusAICD)
AICD)in inthe
themembrane
membrane
to
to be subsequently cleaved by GS to generate Aβ peptides of different lengths as previouslydescribed
be subsequently cleaved by GS to generate AΆ peptides of different lengths as previously described
(Figure
(Figure 15)
15) [61,62].
[61,62].
Figure 15.
Figure 15. Processing steps
steps of
of APP
APPby
by(A)
(A)BACE-1
BACE-1and
and(B)
(B)GS.
GS.Adapted
Adaptedwith
withpermission
permissionfrom [17].
from [17].
Copyright
Copyright 2016 American Chemical 2016 American Chemical Society
Society.
APPmutations
APP mutationsclosecloseto
to the
the Ά-cleavage
β-cleavagesitesitethat
thatincrease
increasethe the efficiency
efficiency of of Ά-cleavage
β-cleavageand andresult
result
inoverproduction
in overproductionofof AΆAβ peptides
peptides strongly
strongly influence
influence the ofrisk
the risk ADof ADOn
[52]. [52]. On the
the other and,other and, a
a mutation
mutation adjacent to the β-cleavage site that reduces the formation of amyloidogenic
adjacent to the Ά-cleavage site that reduces the formation of amyloidogenic peptides has a strong peptides has a
strong protective
protective effect against
effect against AD [19].
AD [19]. Considering
Considering this this genetic
genetic information,
information, thethe inhibitionofofAPP
inhibition APP
proteolysis by
proteolysis by BACE-1
BACE-1totolower
lower thethe
concentration
concentrationof neurotoxic Aβ peptides
of neurotoxic becamebecame
AΆ peptides a rational strategy
a rational
for clinical
strategy for intervention.
clinical intervention.
BACE-1protease
BACE-1 proteaseisischaracterized
characterizedby byaalarge
largecatalytic
catalyticdomain
domainwhich whichisismarked
markedby bythe
thecentrally
centrally
locatedcatalytic
located catalyticaspartates
aspartatesAsp32
Asp32and andAsp228.
Asp228.FreeFreeBACE-1
BACE-1features
featuresaaflap-open
flap-open conformation
conformationthatthat
is energetically stable due to the multiple hydrogen bonds in the flap region of
is energetically stable due to the multiple hydrogen bonds in the flap region of the enzyme. When a the enzyme. When
substrate is bound, BACE-1 assumes a flap-closed or a flap-open conformation, depending on the
characteristics of the substrate [16], [63].
The catalytic domain of BACE-1 contain eight pockets consisted of different amino acid residues
(Table 4).

a substrate is bound, BACE-1 assumes a flap-closed or a flap-open conformation, depending on the

characteristics of the substrate [16,63].
The catalytic domain of BACE-1 contain eight pockets consisted of different amino acid residues
(Table 4).
Table 4. Pockets and their amino acid residues on the catalytic domain of BACE-1 [64].
Pocket Amino Acid Residues Pocket Amino Acid Residues

S1 Leu30, Asp32, Tyr71, Leu119, Gln73, Phe108 S10 Val31, Tyr71, Thr72, Asp228
S2 Asn233, Arg235, Ser325 S20 Ser35, Val69, Pro70, Tyr198
S3 Leu133, Ile110, Ser229 S30 Arg128, Tyr198
S4 Gln73, Thr232, Arg307 S40 Glu125, Ile126, Tyr197, Tyr198
3.2.1. BACE-1 Inhibitor Development

Initially, the development of BACE-1 inhibitors appeared to be a relatively simple approach. First,
the development of successful clinical aspartic proteases inhibitors for other therapeutic areas, namely
human immunodeficiency virus (HIV) and hypertension, had established an important knowledge for
the development of others aspartic protease inhibitors [20]. Second, the first crystal structure of this
secretase, elucidated in 2000, provided powerful information for the structure-based drug design of
BACE-1 inhibitors [65]. However, progress has been difficulted by combination of properties needed
for being efficacious: compounds must fulfill the general rigid prerequisites for a drug aimed for CNS
penetration and at the same time be compatible with the large and hydrophobic catalytic pocket of
BACE-1. Moreover, selectivity toward different aspartyl proteases have been an additional attrition
factor in BACE-1 drug discovery.
Nevertheless, despite the many challenges in the design of selective and effective BACE-1
inhibitors, several pharmaceutical industries have made impressive efforts for the improvement
of BACE-1 inhibitors, with several compounds currently under clinical development.
Peptidomimetics, compounds mimicking the sequences of BACE-1 substrates were the first
BACE-1 inhibitors developed and showed potent activity in vitro [15]. However, these large
compounds suffer from poor PK properties, such as low bioavailability and low penetrance across the
BBB, leading to the unsuccess of their development. Consequently, the design of BACE-1 inhibitors
has focused on small molecules with nonpeptidic nature with improved PK properties and BBB
penetration. Herein, a general overview of the structural evolution of BACE-1 inhibitors with a focus
on the medicinal chemistry aspects of drug development programs will be provided.
Acyl-Guanidine-Based Inhibitors
A series of acyl guanidine-based BACE-1 inhibitors were discovered by high-throughput screening
(HTS) by Wyeth (acquired by Pfizer in 2009) [63] represented by compounds 18–21 in Figure 16.
An X-ray crystal structure of compound 18 complexed with the catalytic domain of BACE-1 revealed
that the acyl guanidine moiety forms four key hydrogen interactions with the catalytic aspartic acids
Asp32 and Asp228 [63]. A structural change in BACE-1 upon binding with the compound was
observed, in which the flap region adopts an “open conformation”, due to stabilized interactions
between Tyr71 and the π-system of the diarylpyrrole ligand 18 [63].
Additionally, it was observed that the two aryl groups extend into the S1 and S20 pockets and
the para position of the P1 phenyl group projects directly to an unoccupied S3 subsite, allowing the
addition of substituents in the P1 phenyl and potentially increasing binding affinity. In contrast to the
S1-S3 pockets, S20 provides access to more polar/charged groups, allowing to explore analogues of
compound 18 able to form hydrogen-bonds in this region [63].
compounds suffer from poor PK properties, such as low bioavailability and low penetrance across
the BBB, leading to the unsuccess of their development. Consequently, the design of BACE-1
inhibitors has focused on small molecules with nonpeptidic nature with improved PK properties and
BBB penetration. Herein, a general overview of the structural evolution of BACE-1 inhibitors with a
focus on the medicinal chemistry aspects of drug development programs will be provided.

18 19
Acyl-guanidine-Based Inhibitors
IC50 (BACE-1) = 3700 nM IC50 (BACE-1) = 110 nM
A series of acyl guanidine-based BACE-1 inhibitors were discovered by high-throughput
screening (HTS) by Wyeth (acquired by Pfizer in 2009) [63] represented by compounds 18–21 in
Figure 16. An X-ray crystal structure of compound 18 complexed with the catalytic domain of BACE-
1 revealed that the acyl guanidine moiety forms four key hydrogen interactions with the catalytic
aspartic acids Asp32 and Asp228 [63]. A structural change in BACE-1 upon binding with the
compound was observed, in which the flap region adopts an “open conformation”, due to stabilized
interactions between Tyr71 and the Δ-system of the diarylpyrrole ligand 18 [63].
Additionally, it was observed that the two aryl groups extend into the S1 and S2’ pockets and
the para position of the 20 P1 phenyl group projects directly to an unoccupied 21 S3 subsite, allowing the
IC50 (BACE-1) = 5.0 nM IC 50 (BACE-1) = 9.5 nM
addition of substituents in the P1 phenyl and potentially increasing binding affinity. In contrast to the
S1-S3 pockets,
Figure S2’
Figure provides
16.16.
Structuresaccess
and
Structures to more
activity
and polar/charged
of of
activity acyl groups,
guanidine-based
acyl BACE-1
guanidine-based allowing
BACE-1 to explore
inhibitors 18–21. analogues of
18–21.
inhibitors
compound 18 able to form hydrogen-bonds in this region [63].

According
According with
with the
the information
information substitutions
substitutions were
were made
made toto compound
compound 18 18 in
in order
ordertotoimprove
improve
potency.
potency.Compound
Compound19, containingaapara
19,containing parapropyloxyphenyl
propyloxyphenylmoiety
moietyin inthe
theunsubstituted
unsubstitutedarylarylring
ringand
and
aapropyl alcohol in the third guanidine nitrogen, allowed an improvement of approximately
propyl alcohol in the third guanidine nitrogen, allowed an improvement of approximately 30-fold 30-fold
in
in potency
potency comparing
comparing with with compound
compound 18. 18. The
The 2-chloro
2-chloro group
group inin PP22 does
does not
not appear
appear to
tocontribute
contribute
significantly
significantlyto
topotency
potency(Figure
(Figure17)
17)[63].
[63].
SRWHQF\ SRWHQF\
DGGLWLRQDO K\GURJHQ
LQWHUDFWLRQVLQ ERQGLQJ
6SRFNHW
§SRWHQF\
Figure 17. Optimization of 18 to 19.

Bristol-Myers Squibb (BMS) also worked in an acyl guanidine series (Figure 18). Starting with hit
Bristol-Myers Squibb (BMS) also worked in an acyl guanidine series (Figure 18). Starting with
compound 22 it was developed compound 20 with a good potency against BACE-1 (IC50 = 5.0 nM)
hit compound 22 it was developed compound 20 with a good potency against BACE-1 (IC50 = 5.0 nM)
and inactive against other aspartyl proteases tested, namely cathepsin D (CatD), cathepsin E (CatE)
and inactive against other aspartyl proteases tested, namely cathepsin D (CatD), cathepsin E (CatE)
and pepsin (IC50 > 100.000 nM) [66].
and pepsin (IC50 > 100.000 nM) [66].
The optimized compound 20 was evaluated in rats in order to access its effect on Aβ40 levels in
plasma, brain and CSF. Although a marked and dose-dependent reduction was observed in plasma
1+
Aβ40 (about 80%) no significant reduction in brain and in CSF was achieved (<20%). This lack of
efficacy on brain and CSF was ascribed to P-glycoprotein (P-gp)1efflux.1 Further improvement
2 was
+
necessary to maximize its PK properties [66].
6 2 1
+
1 &O 1
2 0H
22 20
BACE-1 hit compound BACE-1 optimized compound

IC50 (Cathepesin D) = 14000 nM IC50 (Cathepesin D) = > 100 000 nM
Bristol-Myers Squibb (BMS) also worked in an acyl guanidine series (Figure 18). Starting with
hit compound 22 it was developed compound 20 with a good potency against BACE-1 (IC50 = 5.0 nM)
and inactive against
Pharmaceuticals other aspartyl proteases tested, namely cathepsin D (CatD), cathepsin E (CatE)
2019, 12, 41 16 of 31
and pepsin (IC50 > 100.000 nM) [66].
1+
1 1 2
+
6 2 1
+
1 &O 1
2 0H
22 20
BACE-1 hit compound BACE-1 optimized compound
IC 50 (Cathepesin D) = 14000 nM IC 50 (Cathepesin D) = > 100 000 nM
The optimized compound 20 was evaluated in rats in order to access its effect on AΆ40 levels in
plasma, brain and (Cathepesin
IC50 CSF. E) = 17000
Although nM and dose-dependent
a marked IC50 (Cathepesin
reduction wasE)observed
= > 100 000innM
plasma
IC50 (Pepsin) = 11000 nM IC50 (Pepsin) = > 100 000 nM
AΆ40 (about 80%) no significant reduction in brain and in CSF was achieved (< 20%). This lack of
efficacy on brain Figure
and CSF
Figure 18. was
18.Acyl ascribed to P-glycoprotein
Acylguanidine-based
guanidine-based BACE-1 (P-gp)
BACE-1inhibitors
inhibitors efflux. Further
developed
developed by BMSimprovement
byBMS [66].
[66]. was
necessary to maximize its PK properties [66].
Array
Array BioPharma
BioPharma together
together with
with Genentech
Genentech developed
developed a series
a series of of chromane-based
chromane-based spirocyclic
spirocyclic
acyl
acyl guanidine-derived
guanidine-derived BACE-1
BACE-1 inhibitorsleading
inhibitors leadingtotocompound
compound21 (Figure19).
21(Figure 19).Although
Althoughititshowed
showed a

good selectivity to BACE-1
a good BACE-1 and andwaswasable
abletotoreduce
reduceCSFCSFAβ40
AΆ40levels from
levels from53%53%to to
63%63%in rat andand
in rat cyno,
respectively,
cyno, this this
respectively, compound
compound also also
showed a high
showed efflux
a high ratioratio
efflux by P-gp [67].[67].
by P-gp
21
IC50 = 9.5 nM
CSF AΆ40 reduction (Cyno): 63%
CatD /BACE-1 affinity: 1/88
Figure 19.19.
Figure Chromane-based spirocyclic
Chromane-based acyl
spirocyclic guanidine-derived
acyl BACE-1
guanidine-derived inhibitor
BACE-1 21 21
inhibitor develop by by
develop Array
Array
BioPharma together with Genentech [67].
BioPharma together with Genentech [67].
2-Aminopyridine-Based
2-Aminopyridine-Based Inhibitors
Inhibitors
AA key
key advance
advance in the
in the development
development of small
of small molecule
molecule BACE-1
BACE-1 inhibitors
inhibitors was was the discovery
the discovery of 2- of
amino heterocycles that, comparing with the initial acyl guanidine compounds, generally enable a a
2-amino heterocycles that, comparing with the initial acyl guanidine compounds, generally enable
better
better physico-chemical
physico-chemical profile,
profile, improved
improved brain
brain penetration
penetration and
and inin vivo
vivo efficacy
efficacy [68].
[68].
Wyeth developed a series of pyrrolyl 2-aminopyridines, as an extension
Wyeth developed a series of pyrrolyl 2-aminopyridines, as an extension of their of their work
work onon acyl
acyl
guanidine inhibitors represented in Figure 16. Acyl guanidine inhibitors are polar compounds as as
guanidine inhibitors represented in Figure 16. Acyl guanidine inhibitors are polar compounds
suggested
suggested byby
thethe high
high total
total polar
polar surface
surface area(TPSA)
area (TPSA)ƿ80 ≈80 particularly
particularly duedueto to
thethe acyl
acyl guanidine
guanidine
moiety, which was associated to the poor BBB permeation (<5%) observed with
moiety, which was associated to the poor BBB permeation (<5%) observed with this class of this class of compounds.
The bioisosteric
compounds. replacementreplacement
The bioisosteric of the acyl guanidine
of the acylmoiety in compound
guanidine by an aminopyridine
moiety in18compound 18 by an
(compound 23) was performed to improve compound permeability, maintaining
aminopyridine (compound 23) was performed to improve compound permeability, maintaining the the hydrogen-bonding
interactions with interactions
hydrogen-bonding the aspartic with
acidsthe
in the catalytic
aspartic site
acids inoftheBACE-1 [69]
catalytic site(Figure 20). [69] (Figure 20).
of BACE-1
&\FOL]DWLRQ
18 23
Acyl guanidine-based compound Amynopiridine-based compound
IC50 (BACE-1) = 3.7 μM IC50 (BACE-1) = 5.2 μM
guanidine inhibitors represented in Figure 16. Acyl guanidine inhibitors are polar compounds as
suggested by the high total polar surface area (TPSA) ƿ80 particularly due to the acyl guanidine
moiety, which was associated to the poor BBB permeation (<5%) observed with this class of
compounds. The bioisosteric replacement of the acyl guanidine moiety in compound 18 by an
aminopyridine (compound
Pharmaceuticals 2019, 12, 41 23) was performed to improve compound permeability, maintaining17the of 31
hydrogen-bonding interactions with the aspartic acids in the catalytic site of BACE-1 [69] (Figure 20).
&\FOL]DWLRQ
18 23
Acyl guanidine-based compound Amynopiridine-based compound
IC50 (BACE-1) = 3.7 μM IC50 (BACE-1) = 5.2 μM
TPSA = 86.4 TPSA = 43.8
Figure 20. Bioisosteric replacement of the acyl guanidine moiety of compound 18.
Figure 20. Bioisosteric replacement of the acyl guanidine moiety of compound 18.
The similarity of the hydrogen interaction between these two compounds was confirmed by X-ray
The similarity of the hydrogen interaction between these two compounds was confirmed by X-
studies which demonstrate the practically total sobreposition of the two moieties interacting with the
ray studies which demonstrate the practically total sobreposition of the two moieties interacting with
aspartic acids Asp228 and Asp32 of the catalytic site of BACE-1 [69].
the aspartic acids Asp228 and Asp32 of the catalytic site of BACE-1 [69].
The modulation
Pharmaceuticals
2018, 11, of the TPSA parameter enhanced the permeability, with compound 23 showing a31
17ofof31
17
The modulation of the TPSA parameter enhanced the permeability, with compound 23 showing
good central drug exposure with a brain to plasma ratio of 1.7, compared with 0.04 achieved with the
a good central drug exposure with a brain to plasma ratio of 1.7, compared with 0.04 achieved with
acyl
the guanidine-based
theacyl
acylguanidine-based
guanidine-based compound 18 [69].
compound
compound Further
18[69].
18 development
[69].Further
Further of 23of
development
development by Wyeth
of23
23 led toled
byWyeth
by Wyeth compound
led 24 in
totocompound
compound
Figure 21.
24 in Figure
24 in Figure 21. 21.
Figure21.
Figure 21.Structure
Structureofofthe
theimproved
improvedaminopyridine-base
aminopyridine-basecompound
compound24
24developed
developedby
byWyeth.
Wyeth.
Wyeth.
The
Thepyrimidine
The pyrimidinelinked
pyrimidine linkedby
linked byan
by anoxygen
an oxygen atatpara
oxygenat paraposition
para positionof
position ofofthe
theP
the PP phenyl allow
allowthe
phenylallow
11 1phenyl theestablishment
the establishmentof
establishment ofof
hydrogen-bonding
hydrogen-bonding interactions
interactions with
with Ser229
Ser229 at
atS3
S3region.
region. The
The oxo-ethyl-hydroxyl
oxo-ethyl-hydroxyl at
at
hydrogen-bonding interactions with Ser229 at S3 region. The oxo-ethyl-hydroxyl at position-3 of theposition-3
position-3of
ofthe
the
aminopyridine moiety allow an 0 region and selectivity against other
aminopyridinemoiety
aminopyridine moietyallow
allowananimproved
improved ligand
improvedligand affinity
affinityin
ligandaffinity ininS1
S1’
S1’ regionand
region andselectivity
selectivityagainst
againstother
other
proteases
proteases(Figure
proteases (Figure22)
(Figure 22)[69].
22) [69].
[69].
6
6 6¶
6¶
3RWHQF\
3RWHQF\
6HOHFWLYLW\
6HOHFWLYLW\
3RWHQF\
3RWHQF\
Figure22.
Figure 22.Interactions
Interactionsofofthe
theimproved
improvedaminopyridine-base
aminopyridine-basecompound
compound24
24developed
developedby
byWyeth
Wyethwith
with
thecatalytic
the catalyticsite
siteofofBACE-1.
BACE-1.Adapted
Adaptedfrom
from[16].
[16].
Thearomatic
The aromaticrings
ringslinked
linkedtotothe
thepyrrole
pyrrolering
ringestablished
establishedvan
vander
derWaals
Waalsinteractions
interactionsatatthe
theS1
S1
andS2’
and S2’pockets,
pockets,while
whilethe
thepyrrole
pyrrolering
ringofofthe
theligand
ligandpoints
pointstoward
towardthe
theFLAP
FLAPregion,
region,presumably
presumably
makingaaΔ-edge
making Δ-edgestacking
stackinginteraction
interactionwith
withTyr71
Tyr71[69].
[69].Compound
Compound24 24showed
showedananICIC5050(BACE-1)
(BACE-1)==40
40
nMand
nM and>100-fold
>100-foldand
and>500
>500fold
foldselectivity
selectivityagainst
againstBACE-2
BACE-2andandCatD,
CatD,respectively.
respectively.
6 6¶
3RWHQF\
6HOHFWLYLW\at the S1 and
The aromatic rings linked to the pyrrole ring established van der Waals interactions
S20 pockets, while the pyrrole ring of the ligand points toward the FLAP region, presumably making a
π-edge stacking interaction with Tyr71 [69]. Compound 24 showed an IC50 (BACE-1) = 40 nM and
>100-fold and >5003RWHQF\
fold selectivity against BACE-2 and CatD, respectively.
Aminothiazine- and Aminooxazoline-Based Inhibitors

Figure 22. Interactions of the improved aminopyridine-base compound 24 developed by Wyeth with
F. Hoffmann-La Roche (Roche) developed a series of aminothiazine-based inhibitors starting
the catalytic site of BACE-1. Adapted from [16].
with the aminothiazine fragment 25 as a hit (Figure 23). X-ray analysis of hit 25 co-crystallized
with The
BACE-1 showed
aromatic ringsthat bothtonitrogens
linked of the
the pyrrole ringprotonated
establishedamidine
van dermoiety form tight hydrogen
Waals interactions at the S1
bonds to the catalytic aspartates. Interactive three-dimensional (3D) modeling revealed
and S2’ pockets, while the pyrrole ring of the ligand points toward the FLAP region, presumably that the S3
pocket
making a Δ-edge stacking interaction with Tyr71 [69]. Compound 24 showed an IC50 (BACE-1) ring.
of the catalytic site of BACE-1 could be best reached by meta-substitution on the phenyl = 40
This meta-substitution
nM and >100-fold and >500 was fold
made with an amide
selectivity againstlinker to enable
BACE-2 the formation
and CatD, of a hydrogen bond
respectively.
between the NH of the amide bond with the backbone carbonyl oxygen of Gly291 (Figure 24).
Aminothiazine- and Aminooxazoline-Based Inhibitors

F. Hoffmann-La Roche (Roche) developed a series of aminothiazine-based inhibitors starting
with the aminothiazine fragment 25 as a hit (Figure 23). X-ray analysis of hit 25 co-crystallized with
BACE-1 showed that both nitrogens of the protonated 25 amidine moiety form tight hydrogen bonds to
the catalytic aspartates. Interactive three-dimensional (3D)μM
IC50 (BACE-1) = 41.2 modeling revealed that the S3 pocket of
the catalytic site of BACE-1 could be best reached by meta-substitution on the phenyl ring. This meta-
Figure 23.
Figure 23. HTS
HTS aminothiazine fragment hit
aminothiazine fragment hit 25.
25.
substitution was made with an amide linker to enable the formation of a hydrogen bond between the
NH of the amide bond with the backbone carbonyl oxygen of Gly291 (Figure 24).
Additionally, it leads to a conformationally favorable fixation of the two aromatic rings in an
Additionally, it leads to a conformationally favorable fixation of the two aromatic rings in an
almost planar arrangement. Extensive SAR studies led to the discovery of inhibitor 26, a highly active
almost planar arrangement. Extensive SAR studies led to the discovery of inhibitor 26, a highly active
in vitro compound (BACE-1 enzyme IC50 = 27 nM, cellular assay IC50 = 2 nM). However, it turned out
in
vitro compound (BACE-1 enzyme IC50 = 27 nM, cellular assay IC50 = 2 nM). However, it turned out
to be a good P- gp substrate (P-gp efflux ratio (P-gp ER) = 5.5), leading to unsatisfactory results in the
to be a good P- gp substrate (P-gp efflux ratio (P-gp ER) = 5.5), leading to unsatisfactory results in the
in vivo model (<10% reduction of Aβ40 levels with an oral dose of 30 mg/kg) [26].
in vivo model (<10% reduction of AΆ40 levels with an oral dose of 30 mg/kg) [26].
+\GURJHQ ERQGV ZLWK
FDWDO\WLFDVSDUWDWHV
3RWHQF\
,QWHUDFWLRQZLWKWKH6
UHJLRQ
3RWHQF\
+\GURJHQERQGZLWK*O\
$OORZVDIDYRUDEOHFRQIRUPDWLRQ
26
Figure 24. SAR of optimized inhibitor 26 developed by Roche.
In order to improve brain penetration, Roche continued developing compound 26. A set of
In order to improve brain penetration, Roche continued developing compound 26. A set of
modifications and SAR studies led to the discovery of an aminooxazine-based inhibitor containing a
modifications and SAR studies led to the discovery of an aminooxazine-based inhibitor containing a
CF group. Compound 26 has a high basicity (pK = 9.8), giving rise to potential phospholipidosis and
CF33 group. Compound 26 has a high basicity (pKaa = 9.8), giving rise to potential phospholipidosis and
leading to the high efflux rate by P-gp. The introduction of fluorines into the headgroup showed to
be efficacious in reducing pKa about 3.5 log units and improving brain penetration.
3RWHQF\
,QWHUDFWLRQZLWKWKH6
UHJLRQ
3RWHQF\
leading to the high efflux rate by P-gp. The introduction+\GURJHQERQGZLWK*O\
of fluorines into the headgroup showed to be
efficacious in reducing pKa about 3.5 log units and improving $OORZVDIDYRUDEOHFRQIRUPDWLRQ
brain penetration.
Compound 27 (Figure 25) was the most potent compound, combining favorable in vitro properties
(cellular IC50 (BACE-1) = 2 nM, P-gp ER = 1.9) with 26a reduction of Aβ40 of 78% at a dose of 1 mg/kg in
mice. With regards to its toxicological profile, compound 27 did not inhibit cytochromes P450 (CYP450)
3A4, 2D6, and 2C9 (IC50 > 25 µM) and showed to be selective against other aspartyl proteases such as
human CatD to
In order /Eimprove
and the peptidases renin andRoche
brain penetration, pepsin (IC50 > 200
continued µM) [70]. compound
developing Amgen reported
26. Aaset
series
of
of small-molecule
modifications and SARBACE-1 inhibitors
studies including
led to the discovery a xanthene core with a spirocyclic
of an aminooxazine-based aminooxazoline
inhibitor containing a
head
CF group.
3 group. The company
Compound 26 hasstarted with lead
a high basicity (pKcompound
a = 9.8), giving28 (Figure 26) which
rise to potential showed a goodand
phospholipidosis Aβ
lowering activity in a rat PD model, however it also has a low therapeutic window
leading to the high efflux rate by P-gp. The introduction of fluorines into the headgroup showed to to QTc prolongation,
consistent
be with
efficacious in in vitro activity
reducing on the
pKa about 3.5human
log unitsether-a-go-go
and improving genebrain
(hERG) ion channel [68,71].
penetration.
%UDLQSHQHWUDWLRQ
Pharmaceuticals 2018, 11, 5HGXFHVS.DDQG3JS(5 19 of 31
1 mg/kg in mice. With regards to its toxicological profile, 27 compound 27 did not inhibit cytochromes
P450 (CYP450) 3A4, 2D6, and 2C9 (IC50 > 25 μM) and showed to be selective against other aspartyl
proteases such as human CatD /E and the peptidases renin and pepsin (IC50 > 200 μM) [70]. Amgen
reported a series of small-molecule BACE-1IC 50 (BACE-1) = 12 nM
inhibitors including a xanthene core with a spirocyclic
aminooxazoline head group. The company started Cell IC50 (AΆ40)
with lead= 2compound
nM 28 (Figure 26) which showed
P-gp ER = 1.9
a good AΆ lowering activity in a rat PD model, however it also has a low therapeutic window to QTc
prolongation, consistent
Figure
Figure25. with in vitro
25.Oxazine-based
Oxazine-based activity27
compound
compound on
27 theaahuman
with
with ether-a-go-go
trifluoromethyl
trifluoromethyl group gene (hERG)
groupdeveloped
developed by ion channel
byRoche.
Roche.
[68], [71].
Compound 27 (Figure 25) was the most potent compound, combining favorable in vitro
properties (cellular IC50 (BACE-1) = 2 nM, P-gp ER = 1.9) with a reduction of AΆ40 of 78% at a dose of
28
IC50 (BACE-1) = 2nM
Cell IC50 (BACE-1) = 25 nM
hERG Ki= 660 nM
Figure 26. Spirocyclic aminooxazoline lead compound 28.

Figure 26. Spirocyclic aminooxazoline lead compound 28.
Introduction of polar groups is used as a common strategy to reduce hERG binding affinity.
Introduction of polar groups is used as a common strategy to reduce hERG binding affinity.
However, it increases TPSA, which is strongly correlated with increased P-gp recognition. A series
However, it increases TPSA, which is strongly correlated with increased P-gp recognition. A series
of SAR studies were conducted in order to determinate how to balance hERG, P-gp ER and BACE-1
of SAR studies were conducted in order to determinate how to balance hERG, P-gp ER and BACE-1
potency by identifying regions of the molecule where polarity could be incorporated to minimize hERG
potency by identifying regions of the molecule where polarity could be incorporated to minimize
activity without leading to a significant efflux or a potency decrease. This balance was accomplished by
hERG activity without leading0 to a significant efflux or a potency decrease. This balance was
introducing polarity at the P2 site and at the same time reducing the TPSA of the P3 group (Figure 27).
accomplished by introducing polarity at the P2’ site and at the same time reducing the TPSA of the
The introduction of a fluorine in position 4 of the xanthene ring also improved BACE-1 enzymatic and
P3 group (Figure 27). The introduction of a fluorine in position 4 of the xanthene ring also improved
cellular potencies, attributed to a favorable hydrogen-bond interaction between the 4-fluor atom and
BACE-1 enzymatic and cellular potencies, attributed to a favorable hydrogen-bond interaction
NH of Trp76 of BACE-1 [71].
between the 4-fluor atom and NH of Trp76 of BACE-1 [71].
+ 1
2
1 3
3
<
2 ;
(28) (30)
IC50 (BACE-1) = 2nM IC50 (BACE-1) = 0.3nM
Cell IC50 (BACE-1) = 25 nM Cell IC50 (BACE-1) = 4.0 nM
potency by identifying regions of the molecule where polarity could be incorporated to minimize
hERG activity without leading to a significant efflux or a potency decrease. This balance was
accomplished by introducing polarity at the P2’ site and at the same time reducing the TPSA of the
P3 group (Figure 27). The introduction of a fluorine in position 4 of the xanthene ring also improved
BACE-1 enzymatic
Pharmaceuticals 2019,and
12, 41cellular potencies, attributed to a favorable hydrogen-bond interaction 20 of 31
between the 4-fluor atom and NH of Trp76 of BACE-1 [71].
+ 1
2
1 3
3
<
2 ;
(28) (30)
IC50 (BACE-1) = 2nM IC50 (BACE-1) = 0.3nM
Cell IC50 (BACE-1) = 25 nM Cell IC50 (BACE-1) = 4.0 nM
hERG Ki= 660 nM hERG Ki= >15 000 nM
Figure 27. Spirocyclic aminooxazoline developed by Amgen [68,71].

Figure 27. Spirocyclic aminooxazoline developed by Amgen [68,71].
Inhibitor 30 was orally administered to a number of species and showed a robust reduction of
Inhibitor 30 was orally administered to a number of species and showed a robust reduction of
CNS Aβ40 levels (74 % and 75 % for CSF and brain, respectively) when orally administered in Dawley
CNS AΆ40 levels (74 % and 75 % for CSF and brain, respectively) when orally administered in Dawley
rats. It showed a bioavailability of 50% and 43% in rat and Cynomolgus monkey, respectively, and no
rats. It showed a bioavailability of 50% and 43% in rat and Cynomolgus monkey, respectively, and
significant effect on the QTc interval at a maximum dose of 12 mg/kg [72].
no significant effect on the QTc interval at a maximum dose of 12 mg/kg [72].
Among the several aminothiazine/aminooxazoline based inhibitors developed by different
Among the several aminothiazine/aminooxazoline based inhibitors developed by different
pharmaceutical
pharmaceutical research research teams,
teams, just ajust a limited
limited numbernumber of compounds
of compounds were ablewereto able
reachtoclinical
reach clinical
Pharmaceuticals
evaluation. 2018, 11, 28 depicted the structure of two aminothiazines (31–32) and one aminooxazine
Figure 20 of 31(33)
evaluation. Figure 28 depicted the structure of two aminothiazines (31–32) and one aminooxazine
(33)based
basedcompounds
compounds with BACE-1 inhibitory
with BACE-1 inhibitoryactivity
activitythat
thatentered
enteredinin clinical
clinical trials.
trials.
31 32 33
LY2811376 (Eli Lilly) LY2886721 (Eli Lilly) RG7129 (Roche)

IC50 (BACE-1) = 240 nM IC50 (BACE-1) = 20.3 nM IC50 (BACE-1) = 30 nM
Figure 28. Aminothiazines 31–32 and aminooxazine 33 based compounds evaluated in clinical trials.
Figure 28. Aminothiazines 31-32 and aminooxazine 33 based compounds evaluated in clinical trials.
LY2811376
LY2811376 (31)(31) is an
is an aminothiazine
aminothiazine basedbased compound
compound andand waswasthethe
firstfirst clinical
clinical BACE-1
BACE-1 inhibitor
inhibitor
developed by Eli Lilly, which began phase I clinical trials in 2009. The inhibitor was given given
developed by Eli Lilly, which began phase I clinical trials in 2009. The inhibitor was to 61 to
61 healthy
healthy individuals
individuals to investigate
to investigate the safetytheandsafety and tolerability
tolerability at singleat single
doses fromdoses
5 tofrom
500 mg 5 to
as500
oralmg
as oral capsules [73]. LY2811376 (31) showed to be well tolerated with
capsules [73]. LY2811376 (31) showed to be well tolerated with no serious adverse effects reported. no serious adverse effects
reported. The maximum concentration was reached 2 h post dose in
The maximum concentration was reached 2 h post dose in plasma and 5 hours post dose in CSF, and plasma and 5 h post dose in CSF,
and a dose-dependent
a dose-dependent reduction reduction
of AΆ40 of Aβ40
and AΆ42 and Aβ42 was observed.
was observed. With With
a singlea single
dosedoseof 90ofmg
90 mg it was
it was
observed a Aβ40 reduction of 80% after 7 h and 54 % after 12–14
observed a AΆ40 reduction of 80% after 7 hours and 54 % after 12–14 h, in plasma and CSF, h, in plasma and CSF, respectively.
However, in
respectively. a rat toxicologic
However, in a rat study performed
toxicologic studyinperformed
parallel with the phase
in parallel I clinical
with studies,
the phase a retinal
I clinical
pathology at doses ≥ 30 mg/kg was observed, characterized by
studies, a retinal pathology at doses ǃ 30 mg/kg was observed, characterized by cytoplasmic cytoplasmic accumulations of finely
accumulations of finely granular autofluorescent material dispersed within the retinal epithelium. Asthe
granular autofluorescent material dispersed within the retinal epithelium. As a consequence,
undergoing clinical
a consequence, trials of LY2811376
the undergoing clinical trials(31)
ofwere terminated
LY2811376 and the
(31) were compound
terminated anddidthenot proceed to
compound
did not proceed to phase II. This toxic effect was attributed to off-target effects of LY2811376aspartic
phase II. This toxic effect was attributed to off-target effects of LY2811376 (31) against other (31)
acid proteases, namely CatD, as demonstrated by a subsequent study
against other aspartic acid proteases, namely CatD, as demonstrated by a subsequent study using using LY2811376 (31) in BACE1
_/_ mice [74]. As a safety procedure, all the study participants were contacted for follow-up exams
LY2811376 (31) in BACE1 _/_ mice [74]. As a safety procedure, all the study participants were contacted
6–100 months
for follow-up exams after6-10 conclusion
months after of the trial. Fortunately,
conclusion all the volunteers
of the trial. Fortunately, revealed no
all the volunteers clinically
revealed
no significant observations
clinically significant [74].
observations [74].
LY2886721 (32), also an an
LY2886721 (32), also aminothiazine
aminothiazine based
based inhibitor,
inhibitor, was wasthethe second
second clinical
clinical candidate
candidate tested
tested
in human subjects and the first BACE-1 inhibitor to reach phase II clinical trials. LY2886721 (32) waswas
in human subjects and the first BACE-1 inhibitor to reach phase II clinical trials. LY2886721 (32)
tested
tested in six
in six phase
phase I clinical
I clinical studies
studies for for
the the evaluation
evaluation of safety,
of its its safety, tolerability,
tolerability, andand pharmacology,
pharmacology,
evolving a total of 150 healthy volunteers. Fourteen days of daily dosing of a 70 mg strength reduced
CSF AΆ40 by 74 % and CSF AΆ42 by 71 %. No safety concerns were apparent in dosing up to six
weeks [75]. Based on the satisfactory results found in the phase I clinical trials, in March 2012 Lilly
started a phase II study to examine the safety, tolerability, and PD effects of LY2886721 (32) in patients
with mild AD [76]. During the study, routine safety monitoring detected abnormal liver enzyme
evolving a total of 150 healthy volunteers. Fourteen days of daily dosing of a 70 mg strength reduced
CSF Aβ40 by 74 % and CSF Aβ42 by 71 %. No safety concerns were apparent in dosing up to six
weeks [75]. Based on the satisfactory results found in the phase I clinical trials, in March 2012 Lilly
started a phase II study to examine the safety, tolerability, and PD effects of LY2886721 (32) in patients
with mild AD [76]. During the study, routine safety monitoring detected abnormal liver enzyme
elevations in2018,
Pharmaceuticals 4 patients
11, of 70. Consequently, Lilly voluntarily terminate the phase II study and the
21 of 31
clinical development of LY2886721 (32) was halted. Again, the toxicity of this BACE-1 inhibitor was
Aminoimidazole-Based
considered Inhibitors
to be an off-target effect of the compound unrelated to BACE-1 inhibition [75,77].
RG7129 (RO5598887, 33) is an aminooxazine-based compound developed by Roche which entered
Starting with the aminoimidazole HTS hit 34 (Figure 29) Merck developed a series of
in clinical trials phase I in September 2011. Preclinical evaluation showed high potency against BACE-1
aminoimidazole-based inhibitors represented in Figure 30 [16].
(IC50 = 30 nM). It also showed selectivity against CatD and CatE, pepsin and renin (>1000 fold) but
not against BACE-2 (IC50 = 40 nM). Three phase I clinical trials have been completed between 2012
and 2013, however no results were reported. In October 2013, Roche terminated the development of
RG7129 (33) but did not provide an explanation for its cessation [78]. In 2014, in a peer reviewed paper
Roche reported a mouse study supporting the use of a combination treatment of BACE-1 inhibitor
RG7129 (33) and the anti-Aβ antibody gantenerumab, 34suggesting a future clinical evaluation of these
two compounds combined [79].
Pharmaceuticals 2018, 11, Figure 29. Merck’s aminoimidazole HTS hit. 21 of 31
Aminoimidazole-Based Inhibitors
Molecular modeling
Aminoimidazole-Based studies showed that the amino group of the imidazole heterocycle was
Inhibitors
Starting with the aminoimidazole HTSaspartic
responsible for the binding with the catalytic hit 34acid
(Figure 29) of
residues Merck developed
BACE-1. a series
It was also of
observed
Starting with theinhibitors
aminoimidazole-based aminoimidazole HTSin hit
represented 34 30
Figure (Figure
[16]. 29) Merck developed a series of
that 2-methoxy-5-nitro substituents on the benzyl subunit led to potent inhibitors, such as compound
aminoimidazole-based inhibitors represented in Figure 30 [16].
35 (Figure 30), however, their presence was also responsible for a high P-gp ER. The following
synthesis of compound 36 allowed a significant reduction in P-gp ER and Merck moved forward to
improve the potency of this inhibitor. The introduction of a conformational constraint in compound
37 allowed an increase in potency of 5 fold due to additional hydrophobic interactions with the flap
region of BACE-1 [80]. Additionally, the introduction of a fluor group in compound 38 allowed
additional hydrophobic interactions and increased 34 the potency in about 6-fold to an IC50 (BACE-1) =
63 nM. This inhibitor has a reduced P-gp ER of 3.6, suggesting viable brain penetration [80].
Figure 29. Merck’s aminoimidazole HTS hit.
Figure 29. Merck’s aminoimidazole HTS hit.
Molecular modeling studies showed that the3RWHQ 3RWHQ

amino group of the imidazole heterocycle was
%UDLQ Conformational Additional
responsible for the binding with the catalytic aspartic acid residues of BACE-1. It was
hydrophobic
also observed
constraint
that 2-methoxy-5-nitro substituents on the benzyl subunit led to potent inhibitors, such as compound
35 (Figure 30), however, their presence was also responsible for a high P-gp ER. The following
synthesis of compound 36 allowed a significant reduction in P-gp ER and Merck moved forward to
improve the potency of this inhibitor. The introduction of a conformational constraint in compound
37 allowed an increase in potency of 5 fold due to additional hydrophobic interactions with the flap
region of BACE-1 [80]. Additionally, the introduction of a fluor group in compound 38 allowed
35 36 37 38
additional hydrophobic interactions and increased the potency in about 6-fold to an IC50 (BACE-1) =
IC50 (BACE-1) = 470 nM IC50 (BACE-1) = 1800 nM IC50 (BACE-1) = 350 IC50 (BACE-1) = 63 nM
63 nM. This inhibitor has a reduced P-gp ER of 3.6, suggesting viable brain penetration [80].
P-gp ER = 20 P-gp ER = 4.0 nM P-gp ER = 3.6
P-gp ER = 4.0
3RWHQ
3RWHQ
%UDLQ Figure 30. Merck’s
Figure30. Merck’s aminoimidazole-based
aminoimidazole-based inhibitors.
Conformationalinhibitors. Additional
constraint hydrophobic
Molecular modeling
Wyeth reported thestudies
designshowed that the amino
and synthesis group
of potent of the imidazole
BACE-1 inhibitors heterocycle was
with a bicyclic
responsible for the binding with the catalytic aspartic acid residues of BACE-1. It
aminoimidazole scaffold, based on the HTS hit compound 39 (Figure 31). It represents a bicyclic was also observed
that 2-methoxy-5-nitro
aminoimidazole core substituents
and a biphenylon the moiety,
benzyl subunit
and a led to potent inhibitors,
suboptimal potency (IC such
50 as compound
(BACE-1)) in
35 (Figure 30),range.
micromolar however,
X-raytheir presence
studies showwas
thatalso
theresponsible forthe
hit occupies a high P-gp
center ER. The following
of BACE-1 synthesis
binding pocket (S1,
of
S2’compound
regions) in36anallowed a significant
orientation where thereduction in P-gp ERportion
aminoimidazole and Merck moved
of the ligandforward
directlytointeracts
improvewith
the
potency of this inhibitor. The introduction of a conformational constraint
the catalytic-site aspartic acids (Asp32 and Asp228) via hydrogen interactions. in compound 37 allowed
35 36 37 38
an increase in potency of 5 fold due to additional hydrophobic interactions with the flap region of
IC50 (BACE-1) = 470 nM IC50 (BACE-1) = 1800 nM IC50 (BACE-1) = 350 IC50 (BACE-1) = 63 nM
P-gp ER = 20 P-gp ER = 4.0 nM P-gp ER = 3.6
P-gp ER = 4.0

Figure 30. Merck’s aminoimidazole-based inhibitors.
BACE-1 [80]. Additionally, the introduction of a fluor group in compound 38 allowed additional
hydrophobic interactions and increased the potency in about 6-fold to an IC50 (BACE-1) = 63 nM.
This inhibitor has a reduced P-gp ER of 3.6, suggesting viable brain penetration [80].
Wyeth reported the design and synthesis of potent BACE-1 inhibitors with a bicyclic
aminoimidazole scaffold, based on the HTS hit compound 39 (Figure 31). It represents a bicyclic
aminoimidazole core and a biphenyl moiety, and a suboptimal potency (IC50 (BACE-1)) in micromolar
range. X-ray studies show that the hit occupies the center of BACE-1 binding pocket (S1, S20 regions) in
an orientation where the aminoimidazole portion of the ligand directly interacts with the catalytic-site
aspartic acids
Pharmaceuticals (Asp32
2018, 11, and Asp228) via hydrogen interactions. 22 of 31
39
39
IC50 (BACE-1) = 38 μm
IC50 (BACE-1) = 38 μm
)LJXUH%LF\FOLFDPLQRLPLGD]ROHKLWFRPSRXQG
Figure 31. Bicyclic aminoimidazole hit compound 39.
)LJXUH%LF\FOLFDPLQRLPLGD]ROHKLWFRPSRXQG
Thestructure
The structurealso
alsoshows
showsthatthatthethe S3region
regioncould
could beexplored
exploredthrough
throughthethesubstitution
substitution ofthe the
The structure also shows that the S3S3region could bebeexplored through the substitution ofofthe
moiety occupying
moietyoccupying
occupying the S1 pocket, indicating an opportunity to build off the phenyl moiety into this S3
moiety thethe
S1 S1 pocket,
pocket, indicating
indicating an opportunity
an opportunity to buildto off
build
theoff the phenyl
phenyl moiety intomoietythis into
S3
region
this S3and and
region improve binding
and improve affinity [81]. In order to achieve the unoccupied S3 region the meta-
region improve binding binding affinity
affinity [81]. In [81].
orderIntoorder to achieve
achieve the unoccupied
the unoccupied S3 region S3 the
region the
meta-
substitution of the
meta-substitution of phenyl
the ring
phenyl was
ring explored
was explored with
with benzyl
benzyl analogues
analogues allowing
allowing an
an improvement of
improvement of
substitution of the phenyl ring was explored with benzyl analogues allowing an improvement of
ligand potency
ligandpotency
potencyaboutabout 5-fold.
about5-fold. Greater
5-fold.Greater results
Greaterresults were
resultswere obtained
wereobtained
obtainedwith with a pyridine
witha apyridine moiety
pyridinemoiety where
moietywhere
wherethe the pyridine
thepyridine
pyridine
ligand
nitrogeninteracts
nitrogen interactswith
withSer229
Ser229through
throughthe theconserved
conservedwater wateratatS3S3pocket
pocket(Figure
(Figure32).
32).TheTheintroduction
introduction
nitrogen interacts with Ser229 through the conserved water at S3 pocket (Figure 32). The introduction
of a 2-fluorine
2-fluorine waswas centered on subtle metabolic properties (additional information not disclosed). In
ofofa a2-fluorine centered
was centered onon subtle
subtle metabolic
metabolic properties
properties (additional
(additional information
information not disclosed).
not disclosed). In
additional
In additional SAR studies exploring the phenyl ring that projects into the S2´pocket,
0 it was discovered
additional SARSAR studies
studies exploring
exploring thethe phenyl
phenyl ringring
thatthat projects
projects into
into thethe S2 pocket,
S2´pocket, it itwas
wasdiscovered
discovered
thatthe
that thetrifluoromethoxy
trifluoromethoxysubstituted
substitutedcompound
compound40 40shows
showsan anapproximate
approximate15-fold
15-foldimprovement
improvementof of
that the trifluoromethoxy substituted compound 40 shows an approximate 15-fold improvement of
BACE-1
BACE-1potency potency (comparing
potency (comparing with compound 39). The improved compound 40 (R-enantiomer)
BACE-1 (comparingwith withcompound
compound TheThe
39).39). improved
improved compound
compound40 (R-enantiomer)
40 (R-enantiomer) showed
ashowed
high a high for
potency potency
BACE-1 for inhibition
BACE-1 inhibition
and and >100-fold
>100-fold selectivity selectivity
over the over
otherthe other structurally
structurally related
showed a high potency for BACE-1 inhibition and >100-fold selectivity over the other structurally
related aspartyl
aspartyl proteases proteases
BACE-2, BACE-2,
CatD, CatD,
renin, andrenin, and[81].
pepsin pepsin [81].
related aspartyl proteases BACE-2, CatD, renin, and pepsin [81].
- Improve metabolic
- Improve metabolic
properties
properties
3RWHQF\
3RWHQF\ 3RWHQF\
3RWHQF\
- Additional interactions - Improve interactions in S2’
- Additional interactions - Improve interactions in S2’
in S3 pocket pocket
in S3 pocket pocket
40
40
IC50 (BACE-1) = 20 nM
IC50 (BACE-1) = 20 nM
IC50 (BACE-2) = 3600 nM
IC50 (BACE-2) = 3600 nM
IC50 (cathepsin D) = 6900 nM
IC50 (cathepsin D) = 6900 nM
IC50 (renin)= 2080 nM
IC50 (renin)= 2080 nM
Figure 32. Optimized aminoimidazole-based inhibitor 40 from Wyeth.
Independently Figure 32. Optimized

from32.
Wyeth, aminoimidazole-based
AstraZeneca also developed inhibitor 40inhibitors
a series40of from Wyeth.
Figure Optimized aminoimidazole-based inhibitor from Wyeth.based on compound
39, herein represented by compounds 41 and 42 in Figure 33.
Independently from Wyeth, AstraZeneca also developed a series of inhibitors based on
Independently from Wyeth, AstraZeneca also developed a series of inhibitors based on
compound 39, herein represented by compounds 41 and 42 in Figure 33.
compound 39, herein represented by compounds 41 and 42 in Figure 33.
42
41 (R -enantiomer)
42
pIC50 (BACE-1)=
41 7.11 pIC(R50 (BACE-1)=
-enantiomer) 7.50
Cell
pICpIC (sAPPΆ) =7.11
(BACE-1)=
50 50 7.46 Cell
pICpIC (sAPPΆ) =7.50
(BACE-1)=
50 50 8.10
Efflux ratio = 3.5
Cell pIC50 (sAPPΆ) = 7.46 Efflux ratio = 0.8
Cell pIC50 (sAPPΆ) = 8.10
Figure 33.Efflux ratio = 3.5
Aminoimidazole-based Efflux
inhibitors 41–42 developed by ratio = 0.8
AstraZeneca.
Figure
Figure 33.33. Aminoimidazole-based
Aminoimidazole-based inhibitors
inhibitors 41–42
41–42 developed
developed byby AstraZeneca.
AstraZeneca.
Compound 41, with a p-difluoromethyl ether substitution on one of the phenyl rings and an m-
alkynyl Compound
substituent
Compound 41,41,
on with
the
with a p-difluoromethyl
aother, showed good
p-difluoromethyl ether
potency
ether substitution
in bothon
substitution on one
enzymatic
one of
of theandthecellular
phenylphenyl
rings rings
assays
and an and
withm-an
pICm-alkynyl
50 values
alkynyl substituent
of 7.11 and
substituent onother,
7.46,
on the the other,
respectively.
showed showed good
Cell membrane
good potency potency inenzymatic
both enzymatic
inpermeability,
both as determined and cellular
and cellular by a Caco-2
assays assays
with
with
assay, pIC
was 8.4 values
x 10 of
-6 cm/s, 7.11
and and
the 7.46,
efflux respectively.
ratio was 3.5, Cell membrane
indicating potential
pIC50 values of 7.11 and 7.46, respectively. Cell membrane permeability, as determined by a Caco-2
50 permeability,
for BBB as
penetration.determined
The
by a Caco-2 assay, was 8.4 x 10 -6 cm/s, and the efflux ratio was 3.5, indicating potential for
crystal structure
assay, was 8.4 x 10of cm/s,
-6 compound
and the41efflux in complex
ratio waswith BACE-1 displayed
3.5, indicating potential for theBBBinteraction
penetration. of theThe
BBB penetration.
aminoimidazole
crystal structuremoiety of The crystal
with
compound thestructure
41catalytic ofAsp32
in complex compound 41 in complex
and BACE-1
with Asp228 residues,
displayed with BACE-1
binding
the in adisplayed
interaction flap-open
of the the
interaction
conformation,
aminoimidazole of the aminoimidazole
allowing
moietyTrp76 with theto be moiety with
in position
catalytic Asp32the catalytic
forand
hydrogen Asp32
Asp228bonding and Asp228
residues,tobinding residues,
the oxygen in a of binding
the p- a
flap-open in
flap-open
difluoromethyl conformation,
conformation, ether. allowing allowing
TheTrp76
alkynyl Trp76
tosubstituent to
be in position be in position
of thefor
second for hydrogen
ring extends
hydrogen bonding bonding
intotothe to the
theS3oxygen oxygen
of theofp-
pocket [16,82]. the
p-difluoromethyl
Replacement
difluoromethyl ofether.
ether. the The
Thealkyne alkynyl
alkynyl chain substituent
of inhibitor
substituent of41
of the the second
with
second ring extends
a fluorinated
ring extends into
the the
propyl
into S3 pocket
ether
S3 pocket resulted [16,82].
[16,82]. in
inhibitor Replacement
42. This inhibitor
Replacement of the alkyne
showedchain
of the alkyne chain
similar of inhibitor
of potency 41 with
inhibitoras41compound a fluorinated
41 but an propyl
with a fluorinated propyl
enhanced ether
etherefflux resulted
ratio of
resulted inin
inhibitor
0.8, which42.
inhibitor 42. 4This
isThis inhibitor
times lower.
inhibitor showed
showedHowever,similaran
similar potency
in vivo
potency as compound
as compound
assessment41 41inbut
buta an enhanced
mouse
an model
enhanced efflux
using
efflux ratio
ratio ofof
oral0.8,
0.8, which is 4 times lower. However, an in vivo assessment in a mouse model using oralof
which is
administration 4 timesof lower.
compound However,
42 an
showed in vivo
a low assessment
brain/plasmain a mouse
ratio ofmodel
0.18, using
showing oral administration
that the results
compound
obtained in the
administration 42 ofshowed
Caco-2
compounda lowefflux
cells brain/plasma
42 showedtest were ratio
a low of 0.18, showing
underestimated.
brain/plasma that
This
ratio the
oflack results
0.18,ofshowingobtained
brain that in
penetrationthethe Caco-2
was
results
cells efflux
attributed
obtained in test
to the
P-gp were
Caco-2 underestimated.
efflux, once
cells the test
efflux This lackunderestimated.
coadministration
were of brainofpenetration
42 with Thisawas P-gp
lack attributed
brain to
ofinhibitor P-gp
resulted
penetration efflux, once
inwas
a
the coadministration
brain/plasma
attributed toratio P-gpofefflux, of 42
2.34 [82].with a P-gp inhibitor resulted in a brain/plasma
once the coadministration of 42 with a P-gp inhibitor resulted in a ratio of 2.34 [82].
Compound
Compound
brain/plasma ratio AZD3293
AZD3293
of (LY3314814,
(LY3314814,
2.34 [82]. Lanabecestat,
Lanabecestat, 43,43, Figure
Figure 34)34)is is
anan aminoimidazole
aminoimidazole basedbased
compound
compoundCompound developed
developed AZD3293byby AstraZeneca
AstraZeneca
(LY3314814, which
which reached
reached
Lanabecestat, phase
phase
43, I clinical
I clinical
Figure 34) trials
trials
is in in
an 2012
2012 [83].
[83].
aminoimidazole based
compound developed by AstraZeneca which reached phase I clinical trials in 2012 [83].
Figure 34. 34.

Figure AZD3293 (LY3314814,
AZD3293 lanabecestat,
(LY3314814, 43)43)
lanabecestat, from AstraZeneca.
from AstraZeneca.
Figure 34. AZD3293 (LY3314814, lanabecestat, 43) from AstraZeneca.
TheThe resultsfrom
results from thethe first
first two
twophase
phase I clinical studies
I clinical are currently
studies are currentlyavailable [84]. It [84].
available was composed
It was
of (1)
composed a single
of 1) aascending
single dose
ascending study
dose evaluating
study doses
evaluating of 1–750
doses mg
of
The results from the first two phase I clinical studies are currently available [84]. It was with
1–750 a
mgfood-effect
with a component
food-effect
(n = 72), and
component
composed (n (2)
of =1) aa single
72), 2-week amultiple
2-week ascending
and 2) ascending multiple
dose study dose study
ascending evaluating
dose
evaluating ofdoses
study evaluating
doses 1–750of 15 orwith
doses
mg 50 of
mg a15once daily
or 50 mg or
food-effect
once70 daily
mg once
component weekly
or(n70= mg in elderly
72),once
and weekly
2) subjects
a 2-weekin elderly
multiple n = 31),
(Partsubjects
1,ascending and
(Part 1, 15, 50,
31),orand
n =study
dose 150 mg50,
15,
evaluating once daily
ordoses
150 mg in15
of patients
once with
daily
or 50 mg
in mild
once to
patients moderate
daily orwith70 mg AD
mildonce (Part 2,
to moderate n =
weekly in AD 16). Results
elderly(Part showed
2, n =(Part
subjects that
16). 1, AZD3293
Results
n = 31),showed (43) was
and 15, that generally
50, orAZD3293 well
150 mg once tolerated
(43) daily
was
inup to the
generally
patients highest
well
with doses
tolerated
mild to given.
up to the
moderateNo notable
highest
AD (Part food
doses effects
2,given. were
No
n = 16). observed.
notable
Results foodFor
showed single
effects
thatweredoses ≥5 (43)
observed.
AZD3293 ≥70%
mg, For
was
reduction
single doses was
ǃ5 observed
mg, ǃ70% in mean
reduction plasma
was Aβ40
observed andin Aβ42
mean concentrations,
plasma AΆ40
generally well tolerated up to the highest doses given. No notable food effects were observed. For with
and prolonged
AΆ42 suppression
concentrations,
withforprolonged
singleup to 3 ǃ5
doses weeks
mg, at the highest
suppression
ǃ70% for updose
reduction towas level
3 weeks studied.
observed Following
at theinhighest
mean dosemultiple
plasma level
AΆ40 doses,
studied.
and robust
Following
AΆ42 reductions
multiple in
concentrations,
plasma
doses,
with (≥64%
robust
prolonged at 15 mg
reductions
suppression in and for≥up
plasma 78% toat3 ≥
(ǃ64% at5015mg)
weeks at and
mg ǃ78%(≥
CSF
andhighest
the 51%
ǃ50 at
atdose mg)15and
level mg CSF
studied. ≥Following
and (ǃ51%76% at 15 ≥50mgmg)andAβ
multiple
ǃ76%peptides
at ǃ50 were
mg) seen,
AΆ including
peptides wereprolonged
seen, suppression
including even
prolonged with a once
suppression
doses, robust reductions in plasma (ǃ64% at 15 mg and ǃ78% at ǃ50 mg) and CSF (ǃ51% at 15 mg and weekly
even dosing
with a regimen
once [84].
weekly
dosing
ǃ76% atregimen
ǃ50 mg)[84]. AΆ peptides were seen, including prolonged suppression even with a once weekly
dosing regimen [84].

conducted. They evaluated a new tablet formulation and possible interactions of this inhibitor with
several drugs commonly prescribed in the elderly, namely warfarin and dabigatran, midazolam, as
well as simvastatin and donepezil [85]. Currently, AZD3293 (43) is being evaluated in two phase III
clinical trials (NCT02245737 and NCT02783573) sponsored by an alliance between AstraZeneca and
Pharmaceuticals
Eli Lilly [83].2019, 12, 41multi-center, randomized, double-blind, placebo-controlled studies are testing
These 24 of 31
the disease-modifying
Pharmaceuticals 2018, 11, potential of AZD3293 (43) at daily doses 20 or 50 mg for 18 to 24 months, in
24 of 31
overIn4,000
2015patients
and 2016, withfour
mildadditional
cognitive impairment
phase I trials due to AD
with and of
a total mild
175AD [84]. volunteers were
healthy
In 2015 and 2016, four additional phase I trials with a total of 175 healthy volunteers were
conducted. They evaluated a new tablet formulation and possible interactions of this inhibitor with
Iminothiadiazinane
conducted. Dioxide-Based
They evaluated Inhibitors
a new tablet formulation and possible interactions of this inhibitor with
well asMerck developed
simvastatin a series of
and donepezil iminothiadiazinane
[85]. Currently, AZD3293 dioxide
(43) isbased inhibitors in
being evaluated represented
two phase III by
well as simvastatin and donepezil [85]. Currently, AZD3293 (43) is being evaluated in two phase III
compound
clinical trials45, verubecestat.and
(NCT02245737 TheNCT02783573)
starting point was an iminopyrimidinone
sponsored scaffold
by an alliance between (44) which and
AstraZeneca was
clinical trials (NCT02245737 and NCT02783573) sponsored by an alliance between AstraZeneca and
modified
Eli in These
Lilly [83]. order multi-center,
to improve binding affinity
randomized, and explore
double-blind, a new intellectual
placebo-controlled property
studies space [86]
are testing the
Eli Lilly [83]. These multi-center, randomized, double-blind, placebo-controlled studies are testing
(Figure 35).
disease-modifying potential of AZD3293 (43) at daily doses 20 or 50 mg for 18 to 24 months, in over
the disease-modifying potential of AZD3293 (43) at daily doses 20 or 50 mg for 18 to 24 months, in
4,000 Although
patients withthe mild
PD activities of compound
cognitive impairment due44towere
AD andconsidered
mild AD favorable
[84]. for further clinical
over 4,000 patients with mild cognitive impairment due to AD and mild AD [84].
development, this compound did not advance. Metabolite profiling following oral administration to
Iminothiadiazinane
rats revealed biliary Dioxide-Based
excretion of Inhibitors
metabolites corresponding to direct glutathione addition and
Iminothiadiazinane Dioxide-Based Inhibitors
glutathione adducts derived
Merck developed from
a series ofoxidative metabolism.dioxide
iminothiadiazinane Additionally,
based the modest represented
inhibitors CatD selectivityby
(onlyMerck
21 fold)developed
was a series
considered to of
be iminothiadiazinane
inadequate and a dioxide
major based
limitation to inhibitors
its
compound 45, verubecestat. The starting point was an iminopyrimidinone scaffold (44) which was represented
progression [87]. In by
this
compound
modified in45,
way, an effort verubecestat.
was
order made
to improve The
in order starting
to remove
binding point
thewas
affinity an iminopyrimidinone
metabolically
and explore scaffold
labileintellectual
a new propynylpyridine (44)and
property which was
improve
space [86]
modified in order
CatD selectivity.
(Figure 35). to improve binding affinity and explore a new intellectual property space [86]
(Figure 35).
Although the PD activities of compound 44 were considered favorable for further clinical
development, this compound did not advance. Metabolite profiling following oral administration to
rats revealed biliary excretion of metabolites corresponding to direct glutathione addition and
glutathione adducts derived from oxidative metabolism. Additionally, the modest CatD selectivity
(only 21 fold) was considered to be inadequate and a major limitation to its progression [87]. In this
way, an effort was made in order to remove the metabolically labile propynylpyridine and improve
CatD selectivity.
44 45
Iminopyrimidinone based compound Verubecestat
IC50 (AΆ40) = 11 nM IC50 (AΆ40) = 2.1 nM
BACE-1/CatD = 21 BACE-1/CatD = >10.000
Figure 35.
Figure 35. Structures
Structures of
of iminopyrimidinone
iminopyrimidinonebased
basedcompound
compound44
44and
andverubecestat
verubecestat(45).
(45).
Although
Merck hadthe PD activities
previously of compound
disclosed 44 were 46
the iminohydantoin considered
(Figure 36) favorable for further
which although clinicala
represents
development, this compound
suboptimal activity, it displayeddidannot advance.
enhanced Metabolite
selectivity for profiling
BACE-1 overfollowing
CatD. oral administration
to rats revealed biliary excretion 44 of metabolites corresponding to 45 direct glutathione addition and
glutathione adducts derived frombased
Iminopyrimidinone compound
oxidative Verubecestat
metabolism. Additionally, the modest CatD selectivity
IC50 (AΆ40)
(only 21 fold) was considered to be= inadequate
11 nM and a majorIC 50 (AΆ40) = 2.1 nM
limitation to its progression [87]. In this
way, an effort was madeBACE-1/CatD = 21
in order to remove BACE-1/CatD = >10.000
the metabolically labile propynylpyridine and improve
CatD selectivity.
Figure 35. Structures of iminopyrimidinone based compound 44 and verubecestat (45).
Merck had previously disclosed the iminohydantoin 46 (Figure 36) which although represents a
Merck had
suboptimal previously
activity, disclosed
it displayed the iminohydantoin
an enhanced selectivity
46 for46BACE-1
(Figure 36)overwhich
CatD.although represents a
suboptimal activity, it displayed an enhanced selectivity for
BACE-1/vCatD = 83 BACE-1 over CatD.
Figure 36. Iminohydantoin 46 previously developed by Merck.
In an X-ray cocrystal structure of compound 46 with BACE-1, the interaction of the phenyl and
furanyl groups with the S1 and S3 pockets, respectively, and the amide N-H engaged in a hydrogen-
bonding interaction with the carbonyl of Gly230 were observed. Additionally, comparing the
46
BACE-1/vCatD = 83

In an X-ray cocrystal structure of compound 46 with BACE-1, the interaction of the phenyl and
furanyl groups with the S1 and S3 pockets, respectively, and the amide N-H engaged in a hydrogen-
bonding interaction with the carbonyl of Gly230 were observed. Additionally, comparing the

interactions of the iminopyrimidinone based compound 44 and the iminohydantoin 46 with BACE-
1, it was observed that the furanyl ring of the iminohydantoin is projected deeper into the S3
In an X-ray
subpocket (S3spcocrystal
) of BACE-1 structure of compound
in comparison with 46thewith BACE-1,
pyridyl ring ofthethe
interaction of the phenyl
iminopyrimidinone based
and furanyl groups
compound 44 [87]. with the S1 and S3 pockets, respectively, and the amide N-H engaged in a
hydrogen-bonding
Comparing interaction with the carbonyl
the crystal structures of BACE-1 of Gly230 wereit observed.
and CatD was observed Additionally,
that theircomparing
S3SP differ in
thetopology,
interactions
whereof the iminopyrimidinone based compound 44 and the iminohydantoin
CatD S3 in slightly smaller than the one of BACE-1. Hence, it was postulated
SP 46 with
BACE-1,
the differences in CatD selectivity of these two compounds is related with their interaction the
it was observed that the furanyl ring of the iminohydantoin is projected deeper into withS3the
subpocket sp
(S3 of) the
of BACE-1 in comparison with the pyridyl ring of the iminopyrimidinone based
S3 domains two proteases [87].
compound 44 [87].
In this way, Merck conducted a series of SAR studies in order to improve selectivity over CatD
Comparing the crystal structures of BACE-1 SP differ in
while maintaining high BACE-1 affinity, leadingand CatD
to the it was observed
discovery that their
of verubecestat (45,S3Figure 37). The
topology, SP
pyridylwhere
motif the CatD S3 in(45)
in verubecestat slightly smaller
allowed than the one
an improved of BACE-1.
selectivity over Hence,
CatD >10 it was
000.postulated
The 6-fluoro
thesubstituent
differences in in the
CatD selectivity
phenyl group of allowed
these twoancompounds
improvement is related withintheir
of 5-fold interaction
potency comparingwith the
withS3the
domains
defluoroof the two proteases
analogues and the [87].
iminothiadiazinane dioxide core enhanced permeability and reduced
In this way, Merck conducted
efflux ratio, improving in vivo potency a series[87].
of SAR studies in order to improve selectivity over CatD
while maintaining high BACE-1 affinity,
Verubecestat (45) showed good PK/PD properties leading to the discovery of verubecestat
in preclinical (45, Figure
species, allowing 37).
a profound
Theand
pyridyl motif reduction
sustained in verubecestat
of CSF (45) allowed
AΆ40 levelsanin
improved selectivity
cynomolgus over (72%
monkeys CatD and
>10 000.
81% The 6-fluoroat 3
reduction
substituent in the phenyl group allowed an improvement of 5-fold in potency
and 10 mg/kg, respectively). In a single-dose cardiovascular study telemetered monkeys it showed comparing with the
defluoro
no effectanalogues
on the QT and the iminothiadiazinane
interval and no induction of dioxide coreorenhanced
CYP 3A4 permeability
1A2 expression in human andhepatocytes
reduced
efflux
[87].ratio, improving in vivo potency [87].
1+ Selectivity over CatD
+1 1
1 6
2 Permeability and efflux ratio
&O
Potency
44
45
Iminopyrimidinone based
Verubecestat
compound
Figure 37. Drug development and SAR of verubecestat (45).

Figure 37. Drug development and SAR of verubecestat (45).
Verubecestat (45) showed good PK/PD properties in preclinical species, allowing a profound and
Despite the fact verubecestat (45) has selectivity over the proteases CatD, CatE, pepsin and renin
sustained reduction of CSF Aβ40 levels in cynomolgus monkeys (72% and 81% reduction at 3 and 10
(>45,000, >45,000, >45,000, and 15,000, respectively) it showed to be a potent inhibitor of BACE-2 (Ki
mg/kg, respectively). In a single-dose cardiovascular study in telemetered monkeys it showed no
(BACE-2) = 0.38 nM against Ki (BACE-1) = 2.2 nM). Although the specific functions of BACE-2 are
effect on the QT interval and no induction of CYP 3A4 or 1A2 expression in human hepatocytes [87].
currently not well known, recent reports associate BACE-2 with the process of pigmentation,
Despite the fact verubecestat (45) has selectivity over the proteases CatD, CatE, pepsin and renin
consistent with the lighter coat color observed in BACE-2 knockout mice [88]. This phenomenon was
(>45,000, >45,000, >45,000, and 15,000, respectively) it showed to be a potent inhibitor of BACE-2
observed in mice and rats treated with verubecestat (45), but not in the chronic toxicology studies
(Ki (BACE-2) = 0.38 nM against Ki (BACE-1) = 2.2 nM). Although the specific functions of BACE-2
performed in monkeys. Overall, the relatively benign phenotypes of BACE-2 knockout mice, current
are currently not well known, recent reports associate BACE-2 with the process of pigmentation,
understanding of the role of BACE-2 processing of its endogenous substrates, and the outcome of
consistent with the lighter coat color observed in BACE-2 knockout mice [88]. This phenomenon was
preclinical toxicology studies has mitigated concerns related to the lack of selectivity of verubecestat
observed in mice and rats treated with verubecestat (45), but not in the chronic toxicology studies
(45) over BACE-2 [87].
performed in monkeys. Overall, the relatively benign phenotypes of BACE-2 knockout mice, current
Verubecestat (45) advanced to phase I clinical trials in 2011 to assess safety, PK and PD in healthy
understanding of the role of BACE-2 processing of its endogenous substrates, and the outcome of
volunteers and in mild to moderate AD patients. Single and multiple doses were generally well
preclinical toxicology studies has mitigated concerns related to the lack of selectivity of verubecestat
tolerated and produced reductions in AΆ levels in the CSF >80% of both healthy human subjects and
(45) over BACE-2 [87].
AD patients. Notably, there were no reports of changes in skin or hair pigmentation as a potential
Verubecestat (45) advanced to phase I clinical trials in 2011 to assess safety, PK and PD in healthy
consequence of BACE-2 inhibition in any of the phase I studies; however, longer treatment time
volunteers and in mild to moderate AD patients. Single and multiple doses were generally well
would likely be required for pigmentation changes to manifest [87]. In 2012, verubecestat (45) entered
tolerated and produced reductions in Aβ levels in the CSF > 80% of both healthy human subjects and
into phase II/III clinical trials in people with mild to moderate AD (EPOCH trial), and in 2013 a phase
AD patients. Notably, there were no reports of changes in skin or hair pigmentation as a potential
III trial for prodromal AD (APECS trial) was also started [68].
consequence of BACE-2 inhibition in any of the phase I studies; however, longer treatment time would

likely be required for pigmentation changes to manifest [87]. In 2012, verubecestat (45) entered into
phase II/III clinical trials in people with mild to moderate AD (EPOCH trial), and in 2013 a phase III
trial for prodromal AD (APECS trial) was also started [68].
EPOCH trial was discontinued on February 2017 due to lack of efficacy, with researchers defending
that it is very unlikely to observe a clinical benefit in using a BACE-1 inhibitor in cases that a substantial
synaptic and neuron loss is already installed [89].
In February 2018, APECS clinical trial was also discontinued and Merck no longer listed
verubecestat in its research pipeline. Participants taking verubecestat scored worse than the placebo
group on the cognitive test Alzheimer’s Disease Cooperative Study-Activities of Daily Living
(ADCS-ADL), in a small but significant way. Further research in needed to understand the origin of
this negative effect, if it is reversible and if can be associated with certain patient populations or stages
of disease [90,91].
4. Conclusions
The discovery of disease modifying agents for AD has shown to be very challenging for medicinal
chemists. Compounds should exhibit improved CNS penetration by enhanced BBB permeability and
reduced P-gp ER. The physical chemistry characteristics needed for this PK properties need to be
balanced with target requirements, in which the improvements of these properties sometimes lead to a
reduction on binding affinity and potency. Additionally, off-target effects also need to be addressed,
being these features one of the principal causes for the discontinuation of clinical programs.
Medicinal chemistry teams have utilized high throughput and virtual screening followed by
chemical optimization trying to accommodate several physico-chemical nuances and discover new
compounds able to be safety and efficacy in modifying AD progression.
Taking into consideration the actual AD pipeline, a special focus has been given to BACE-1 as
a target for a disease modifying therapy for AD. The information currently available pointed the
inhibition of BACE-1 as a safety and efficacy target for Aβ reduction. All the adverse effects identified
up to date in the use of BACE-1 inhibitors were pointed as off-target effects, confirming BACE-1 as a
suitable target to explore. Nevertheless, medicinal chemists should perform extended SAR studies,
not only for potency and PK properties exploration, but also in terms of selectivity, providing the
discovery of compounds highly selective for BACE-1 against other related proteases such as CatD.
Another question stands on the adequacy of the pharmacological therapy with the stage of
Alzheimer disease (AD). The use of disease modifying therapies as BACE-1 inhibitors should be a
suitable option for early stages of AD where minimum neuronal loss and synaptic dysfunction are
observed. On the other hand, in a mild-to-moderate AD scenario it would possibly be too late in
the disease process for BACE-1 inhibition to be effective and a symptomatically therapeutic should
be preferred.
In sum, the future of AD will rely on the development of potent, selective and safety compounds
able to delay AD progression and neuronal impairment. At the same time, it is essential the
identification of specific biomarkers to allow an early and efficacious pharmacological intervention.
Funding: This work was partially supported through national funds provided by FCT/MCTES - Foundation
for Science and Technology from the Minister of Science, Technology and Higher Education (PIDDAC) and
European Regional Development Fund (ERDF) through the COMPETE – Programa Operacional Factores de
Competitividade (POFC) programme, under the Strategic Funding UID/Multi/04423/2013.
Acknowledgments: The authors thanks to Tino Rossi, László E. Kiss and Patrício Soares-da-Silva (Laboratory of
Chemistry, Research Department, BIAL-Portela & Cª, S.A., À Avenida da Siderurgia Nacional, 4745-457 Coronado
(S. Romão and S. Mamede, Portugal)) for their support on the development of this work.
References
1. Duthey, B. Background Paper 6.11 Alzheimer Disease and Other Dementias, Update on 2004; World Health
Organization: Geneva, Switzerland, February 2013; pp. 1–77.
2. Stelzmann, R.A.; Schnitzlein, H.N.; Murtagh, F.R. An english translation of alzheimer’s 1907 paper, “Uber
eine eigenartige erkankung der Hirnrinde”. Clin. Anat. 1995, 8, 429–431. [CrossRef] [PubMed]
3. Lane, C.A.; Hardy, J.; Schott, J.M. Alzheimer’s disease. Eur. J. Neurol. 2018, 25, 59–70. [CrossRef] [PubMed]
4. Eurostat. Population Structure and Ageing. Available online: http://ec.europa.eu/eurostat/statistics-
explained/index.php/Population_structure_and_ageing (accessed on 5 January 2019).
5. Alzheimer’s Association. 2017 Alzheimer’s disease facts and figures. Alzheimer’s Dement. 2017, 13, 325–373.
6. Schneider, J.; Arvanitakis, Z.; Leurgans, S.; Bennett, D. The Neuropathology of Probable Alzheimer’s Disease
and Mild Cognitive Impairment. Ann. Neurol. 2010, 66, 200–208. [CrossRef]
7. Villemagne, V.L.; Burnham, S.; Bourgeat, P.; Brown, B.; Ellis, K.A.; Salvado, O.; Szoeke, C.; Macaulay, S.L.;
Martins, R.; Maruff, P.; et al. Amyloid β deposition, neurodegeneration, and cognitive decline in sporadic
Alzheimer’s disease: A prospective cohort study. Lancet Neurol. 2013, 12, 357–367. [CrossRef]
8. Bateman, R.J.; Xiong, C.; Benzinger, T.L.S.; Fagan, A.M.; Goate, A.; Fox, N.C.; Marcus, D.S.; Cairns, N.J.;
Xie, X.; Blazey, T.M.; et al. Clinical and Biomarker Changes in Dominantly Inherited Alzheimer’s Disease.
N. Engl. J. Med. 2012, 367, 795–804. [CrossRef] [PubMed]
9. Cummings, J.L.; Morstorf, T.; Zhong, K. Alzheimer’s disease drug-development pipeline: Few candidates,
frequent failures. Alzheimer’s Res. Ther. 2014, 6, 1–7. [CrossRef]
10. Balson, R.; Gibson, P.R.; Ames, D.; Bhathal, P.S. Tacrine-Induced Hepatotoxicity. CNS drugs 1995, 4, 168–169.
[CrossRef]
11. Scarpini, E.; Schelterns, P.; Feldman, H. Treatment of Alzheimer’s disease: Current status and new
perspectives. Lancet Neurol 2003, 2, 539–547. [CrossRef]
12. FDA US Food and Drug Administration. Drug Approvals and Databases. Available online: https://www.
accessdata.fda.gov/drugsatfda_docs/nda/2003/21-487_Namenda.cfm (accessed on 5 January 2019).
13. Alzheimer-Association. Changing the Trajectory of Alzheimer’s Disease: How a Treatment by 2025 Saves
Lives and Dollars. Alzheimer’s Dement. 2013, 9, 208–245.
14. Cummings, J.; Lee, G.; Ritter, A.; Zhong, K. Alzheimer ’s disease drug development pipeline: 2018.
Alzheimer’s Dement. Transl. Res. Clin. Interv. 2018, 4, 195–214. [CrossRef]
15. Suzuki, K.; Wata, A.I.; Watsubo, T.I. The past, present, and future of disease-modifying therapies for
Alzheimer ’s disease. Proc. Jpn. Acad. Ser. B 2017, 93, 757–771. [CrossRef]
16. Ghosh, A.K.; Cárdenas, E.L.; Osswald, H.L. The Design, Development, and Evaluation of BACE1 Inhibitors
for the Treatment of Alzheimer’s Disease. Springer Int. Publ. Ag 2017, 24, 27–86.
17. Bursavich, M.G.; Harrison, B.A.; Blain, J.F. Gamma secretase modulators: New Alzheimer’s drugs on the
horizon? J. Med. Chem. 2016, 59, 7389–7409. [CrossRef]
18. Brouwers, N.; Sleegers, K.; Van Broeckhoven, C. Molecular genetics of Alzheimer’s disease: An update.
Ann. Med. 2008, 40, 562–583. [CrossRef]
19. Jonsson, T.; Atwal, J.K.; Steinberg, S.; Snaedal, J.; Jonsson, P.V.; Bjornsson, S.; Stefansson, H.; Sulem, P.;
Gudbjartsson, D.; Maloney, J.; et al. A mutation in APP protects against Alzheimer‘s disease and age-related
cognitive decline. Nature 2012, 488, 96–99. [CrossRef]
20. Hamada, Y.; Kiso, Y. New directions for protease inhibitors directed drug discovery. Biopolymers 2016, 106,
21. Hardy, J.; Selkoe, D. The Amyloid Hypothesis of Alzheimer’s Disease: Progress and Problems on the Road
to Therapeutics. Sci. Compass 2002, 297, 353–356. [CrossRef]
22. Karran, E.; Mercken, M.; De Strooper, B. The amyloid cascade hypothesis for Alzheimer’s disease:
An appraisal for the development of therapeutics. Nat. Rev. Drug Discov. 2011, 10, 698–712. [CrossRef]
23. Chen, G.F.; Xu, T.H.; Yan, Y.; Zhou, Y.R.; Jiang, Y.; Melcher, K.; Xu, H.E. Amyloid beta: Structure, biology and
structure-based therapeutic development. Acta Pharmacol. Sin. 2017, 38, 1205–1235. [CrossRef]
24. Wolfe, M.S. γ-Secretase inhibitors and modulators for Alzheimer’s disease. J. Neurochem. 2012, 120, 89–98.
[CrossRef] [PubMed]
25. Golde, T.E.; Koo, E.H.; Felsenstein, K.M.; Osborne, B.A.; Miele, L. γ-Secretase inhibitors and modulators.
Biochim. Biophys. Acta Biomembr. 2013, 1828, 2898–2907. [CrossRef] [PubMed]
26. Woltering, T.J.; Wostl, W.; Hilpert, H.; Rogers-Evans, M.; Pinard, E.; Mayweg, A.; Göbel, M.; Banner, D.W.;
Benz, J.; Travagli, M.; et al. BACE1 inhibitors: A head group scan on a series of amides. Bioorganic Med.
Chem. Lett. 2013, 23, 4239–4243. [CrossRef] [PubMed]
27. Sato, T.; Diehl, T.S.; Narayanan, S.; Funamoto, S.; Ihara, Y.; De Strooper, B.; Steiner, H.; Haass, C.; Wolfe, M.S.
Active γ-secretase complexes contain only one of each component. J. Biol. Chem. 2007, 282, 33985–33993.
[CrossRef]
28. Weidemann, A.; Eggert, S.; Reinhard, F.B.M.; Vogel, M.; Paliga, K.; Baier, G.; Masters, C.L.; Beyreuther, K.;
Evin, G. A novel ε-cleavage within the transmembrane domain of the Alzheimer amyloid precursor protein
demonstrates homology with notch processing. Biochemistry 2002, 41, 2825–2835. [CrossRef] [PubMed]
29. Jurisch-Yaksi, N.; Sannerud, R.; Annaert, W. A fast growing spectrum of biological functions of γ-secretase
in development and disease. Biochim. Biophys. Acta Biomembr. 2013, 1828, 2815–2827. [CrossRef] [PubMed]
30. Bai, X.; Yan, C.; Yang, G.; Lu, P.; Ma, D.; Sun, L.; Zhou, R.; Scheres, S.H.W.; Shi, Y. An atomic structure of
human γ-secretase. Nature 2015, 525, 212–217. [CrossRef]
31. Wolfe, M.S. Processive proteolysis by γ-secretase and the mechanism of Alzheimer’s disease. Biol. Chem.
2012, 393, 899–905. [CrossRef]
32. Zhao, G.; Cui, M.Z.; Mao, G.; Dong, Y.; Tan, J.; Sun, L.; Xu, X. Gamma-Cleavage Is Dependent on
zeta-Cleavage During the Proteolytic Processing of Amyloid Precursor Protein Within Its Transmembrane
Domain. J. Biol. Chem. 2005, 280, 37689–37697. [CrossRef]
33. Oehlrich, D.; Berthelot, D.J.C.; Gijsen, H.J.M. γ-Secretase modulators as potential disease modifying
anti-Alzheimer’s drugs. J. Med. Chem. 2011, 54, 669–698. [CrossRef]
34. Shimojo, M.; Sahara, N.; Mizoroki, T.; Funamoto, S.; Morishima-Kawashima, M.; Kudo, T.; Takeda, M.;
Ihara, Y.; Ichinose, H.; Takashima, A. Enzymatic characteristics of I213T mutant presenilin-1/γ-secretase in
cell models and knock-in mouse brains: Familial Alzheimer disease-linked mutation impairs γ-site cleavage
of amyloid precursor protein C-terminal fragment beta. J. Biol. Chem. 2008, 283, 16488–16496. [CrossRef]
[PubMed]
35. Okochi, M.; Tagami, S.; Yanagida, K.; Takami, M.; Kodama, T.S.; Mori, K.; Nakayama, T.; Ihara, Y.; Takeda, M.
γ-Secretase Modulators and Presenilin 1 Mutants Act Differently on Presenilin/γ-Secretase Function to
Cleave Aβ42 and Aβ43. Cell Rep. 2013, 3, 42–51. [CrossRef] [PubMed]
36. Chávez-Gutiérrez, L.; Bammens, L.; Benilova, I.; Vandersteen, A.; Benurwar, M.; Borgers, M.; Lismont, S.;
Zhou, L.; Van Cleynenbreugel, S.; Esselmann, H.; et al. The mechanism of γ-Secretase dysfunction in familial
Alzheimer disease. Embo J. 2012, 31, 2261–2274. [CrossRef] [PubMed]
37. Szaruga, M.; Veugelen, S.; Benurwar, M.; Lismont, S.; Sepulveda-Falla, D.; Lleo, A.; Ryan, N.S.; Lashley, T.;
Fox, N.C.; Murayama, S.; et al. Qualitative changes in human γ-secretase underlie familial Alzheimer’s
disease. J. Exp. Med. 2015, 212, 2003–2013. [CrossRef] [PubMed]
38. Kretner, B.; Fukumori, A.; Gutsmiedl, A.; Page, R.M.; Luebbers, T.; Galley, G.; Baumann, K.; Haass, C.;
Steiner, H. Attenuated Aβ42 responses to low potency γ-secretase modulators can be overcome for many
pathogenic presenilin mutants by second-generation compounds. J. Biol. Chem. 2011, 286, 15240–15251.
[CrossRef] [PubMed]
39. Coric, V.; Van Dyck, C.H.; Salloway, S.; Andreasen, N.; Brody, M.; Richter, R.W.; Soininen, H.; Thein, S.;
Shiovitz, T.; Pilcher, G.; et al. Safety and tolerability of the γ-secretase inhibitor avagacestat in a phase 2
study of mild to moderate Alzheimer disease. Arch. Neurol. 2012, 69, 1430–1440. [CrossRef] [PubMed]
40. Doody, R.S.; Raman, R.; Farlow, M.; Iwatsubo, T.; Vellas, B.; Joffe, S.; Kieburtz, K.; He, F.; Sun, X.;
Thomas, R.G.; et al. A Phase 3 Trial of Semagacestat for Treatment of Alzheimer’s Disease. N. Engl. J. Med.
2013, 369, 341–350. [CrossRef]
41. Woo, H.N.; Park, J.S.; Gwon, A.R.; Arumugam, T.V.; Jo, D.G. Alzheimer’s disease and Notch signaling.
Biochem. Biophys. Res. Commun. 2009, 390, 1093–1097. [CrossRef]
42. Gillman, K.W.; Starrett, J.E.; Parker, M.F.; Xie, K.; Bronson, J.J.; Marcin, L.R.; McElhone, K.E.; Bergstrom, C.P.;
Mate, R.A.; Williams, R.; et al. Discovery and evaluation of BMS-708163, a potent, selective and orally
bioavailable γ-secretase inhibitor. Acs Med. Chem. Lett. 2010, 1, 120–124. [CrossRef]
43. Crump, C.J.; Castro, S.V.; Wang, F.; Pozdnyakov, N.; Ballard, T.E.; Sisodia, S.S.; Bales, K.R.; Johnson, D.S.;
Li, Y.M. BMS-708,163 targets presenilin and lacks notch-sparing activity. Biochemistry 2012, 51, 7209–7211.
[CrossRef]
44. Reynaud, E. A subset of NSAIDs lower amyloidogenic Aβ42 independently of cyclooxygenase activity.
Nat. Educ. 2010, 3, 212–216.
45. Eriksen, J.L.; Sagi, S.A.; Smith, T.E.; Weggen, S.; Das, P.; McLendon, D.C.; Ozols, V.V.; Jessing, K.W.;
Zavitz, K.H.; Koo, E.H.; et al. NSAIDs and enantiomers of flurbiprofen target gamma-secretase and lower
Aβ42 in vivo. J. Clin. Investig. 2003, 112, 440–449. [CrossRef]
46. Imbimbo, B.P. Why did tarenflurbil fail in Alzheimer’s Disease? J. Alzheimer’s Dis. 2009, 17, 757–760.
[CrossRef]
47. Peretto, I.; Radaelli, S.; Parini, C.; Zandi, M.; Raveglia, L.F.; Dondio, G.; Fontanella, L.; Misiano, P.;
Bigogno, C.; Rizzi, A.; et al. Synthesis and biological activity of flurbiprofen analogues as selective inhibitors
β-Amyloid1-42 secretion. J. Med. Chem. 2005, 48, 5705–5720. [CrossRef]
48. Baumann, S.; Höttecke, N.; Schubenel, R.; Baumann, K.; Schmidt, B. NSAID-derived γ-secretase modulators.
Part III: Membrane anchoring. Bioorganic Med. Chem. Lett. 2009, 19, 6986–6990. [CrossRef] [PubMed]
49. Imbimbo, B.P.; Frigerio, E.; Breda, M.; Fiorentini, F.; Fernandez, M.; Sivilia, S.; Giardino, L.; Calzà, L.;
Norris, D.; Casula, D.; et al. Pharmacokinetics and pharmacodynamics of CHF5074 after short-term
administration in healthy subjects. Alzheimer Dis Assoc Disord 2013, 27, 278–286. [CrossRef]
50. Van Vleet, T.R.; Liu, H.; Lee, A.; Blomme, E.A.G. Acyl glucuronide metabolites: Implications for drug safety
assessment. Toxicol. Lett. 2017, 272, 1–7. [CrossRef]
51. Kounnas, M.Z.; Lane-Donovan, C.; Nowakowski, D.W.; Herz, J.; Comer, W.T. NGP 555, a γ-secretase
modulator, lowers the amyloid biomarker, Aβ42 in cerebrospinal fluid while preventing Alzheimer’s disease
cognitive decline in rodents. Alzheimer’s Dement. Transl. Res. Clin. Interv. 2017, 3, 65–73.
52. Mullan, M.; Crawford, F.; Axelman, K.; Houlden, H.; Lilius, L.; Winblad, B.; Lannfelt, L. A pathogenic
mutation for probable Alzheimer’s disease in the APP gene at the N–terminus of β–amyloid. Nat. Genet.
1992, 1, 345–347. [CrossRef]
53. Kimura, T.; Kawano, K.; Doi, E.; Kitazawa, N.; Shin, K.; Miyagawa, T.; Kaneco, T.; Ito, K.; Takaishi, M.;
Sasaki, T.; et al. Cinnamide Compound. Patent Application Publication No. WO 2005115990 A1,
8 December 2015.
54. Hashimoto, T.; Ishibashi, A.; Hagiwara, H.; Murata, Y.; Takenaka, O.; Miyagawa, T. E2012: A novel
gamma-secretase modulator-pharmacology part. Alzheimer’s Dement. 2010, 6, S242. [CrossRef]
55. Boggs, L.N.; Lindstrom, T.; Watson, B.; Sheehan, S.; Audia, J.E.; May, P.C.; Lilly, E. E2012: A novel
gamma-secretase modulator-mechanism of action. Alzheimer’s Dement. 2010, 6, S541–S542. [CrossRef]
56. Nagy, C.; Schuck, E.; Ishibashi, A.; Nakatani, Y.; Rege, B.; Logovinsky, V. E2012, a novel gamma-secretase
modulator, decreases plasma amyloid-beta (Aβ) levels in humans. Alzheimer’s Dement. 2010, 6, S574.
[CrossRef]
57. Nakano-Ito, K.; Fujikawa, Y.; Hihara, T.; Shinjo, H.; Kotani, S.; Suganuma, A.; Aoki, T.; Tsukidate, K.
E2012-induced cataract and its predictive biomarkers. Toxicol. Sci. 2014, 137, 249–258. [CrossRef] [PubMed]
58. Yu, Y.; Logovinsky, V.; Schuck, E.; Kaplow, J.; Chang, M.K.; Miyagawa, T.; Wong, N.; Ferry, J. Safety,
tolerability, pharmacokinetics, and pharmacodynamics of the novel γ-secretase modulator, E2212, in healthy
human subjects. J. Clin. Pharmacol. 2014, 54, 528–536. [CrossRef] [PubMed]
59. Pettersson, M.; Stepan, A.F.; Kauffman, G.W.; Johnson, D.S. Novel γ-secretase modulators for the treatment
of Alzheimer’s disease: A review focusing on patents from 2010 to 2012. Expert Opin. Ther. Pat. 2013, 23,
1349–1366. [CrossRef] [PubMed]
60. Voytyuk, I.; de Strooper, B.; Chávez-Gutiérrez, L. Modulation of γ- and β-Secretases as Early Prevention
Against Alzheimer ’ s Disease. Biol. Psychiatry 2018, 83, 320–327. [CrossRef] [PubMed]
61. Vassar, R.; Bennett, B.D.; Babu-Khan, S.; Kahn, S.; Mendiaz, E.A.; Rogers, G.; Citron, M. β-Secretase Cleavage
of Alzheimer’s Amyloid Precursor Protein by the Transmembrane Aspartic Protease BACE. Science 1999,
62. Park, H.; Min, K.; Kwak, H.S.; Koo, K.D.; Lim, D.; Seo, S.W.; Choi, J.U.; Platt, B.; Choi, D.Y. Synthesis, SAR,
and X-ray structure of human BACE-1 inhibitors with cyclic urea derivatives. Bioorganic Med. Chem. Lett.
2008, 18, 2900–2904. [CrossRef] [PubMed]
63. Cole, D.C.; Manas, E.S.; Stock, J.R.; Condon, J.S.; Jennings, L.D.; Aulabaugh, A.; Chopra, R.; Cowling, R.;
Ellingboe, J.W.; Fan, K.Y.; et al. Acylguanidines as small-molecule β-secretase inhibitors. J. Med. Chem. 2006,
64. Hernández-Rodríguez, M.; Correa-Basurto, J.; Gutiérrez, A.; Vitorica, J.; Rosales-Hernández, M.C. Asp32
and Asp228 determine the selective inhibition of BACE1 as shown by docking and molecular dynamics
simulations. Eur. J. Med. Chem. 2016, 124, 1142–1154. [CrossRef]
65. Hong, L.; Koelsch, G.; Lin, X.; Wu, S.; Terzyan, S.; Ghosh, A.K.; Zhang, X.C.; Tang, J. Structure of the protease
domain of memapsin 2 (β-secretase) complexed with inhibitor. Science 2000, 290, 150–153. [CrossRef]
66. Gerritz, S.W.; Zhai, W.; Shi, S.; Zhu, S.; Toyn, J.H.; Meredith, J.E.; Iben, L.G.; Burton, C.R.; Albright, C.F.;
Good, A.C.; et al. Acyl guanidine inhibitors of β-secretase (BACE-1): Optimization of a micromolar hit to a
nanomolar lead via iterative solid- and solution-phase library synthesis. J. Med. Chem. 2012, 55, 9208–9223.
[CrossRef]
67. Thomas, A.A.; Hunt, K.W.; Volgraf, M.; Watts, R.J.; Liu, X.; Vigers, G.; Smith, D.; Sammond, D.; Tang, T.P.;
Rhodes, S.P.; et al. Discovery of 7-tetrahydropyran-2-yl chromans: β-site amyloid precursor protein cleaving
enzyme 1 (BACE1) inhibitors that reduce amyloid β-protein (Aβ) in the central nervous system. J. Med. Chem.
2014, 57, 878–902. [CrossRef]
68. Prati, F.; Bottegoni, G.; Bolognesi, M.L.; Cavalli, A. BACE-1 Inhibitors: From Recent Single-Target Molecules
to Multitarget Compounds for Alzheimer’s Disease. J. Med. Chem. 2018, 61, 619–637. [CrossRef]
69. Malamas, M.S.; Barnes, K.; Hui, Y.; Johnson, M.; Lovering, F.; Condon, J.; Fobare, W.; Solvibile, W.; Turner, J.;
Hu, Y.; et al. Novel pyrrolyl 2-aminopyridines as potent and selective human β-secretase (BACE1) inhibitors.
Bioorganic Med. Chem. Lett. 2010, 20, 2068–2073. [CrossRef]
70. Hilpert, H.; Guba, W.; Woltering, T.J.; Wostl, W.; Pinard, E.; Mauser, H.; Mayweg, A.V.; Rogers-Evans, M.;
Humm, R.; Krummenacher, D.; et al. β-secretase (BACE1) inhibitors with high in vivo efficacy suitable for
clinical evaluation in Alzheimer’s disease. J. Med. Chem. 2013, 56, 3980–3995. [CrossRef]
71. Epstein, O.; Bryan, M.C.; Cheng, A.C.; Derakhchan, K.; Dineen, T.A.; Hickman, D.; Hua, Z.; Human, J.B.;
Kreiman, C.; Marx, I.E.; et al. Lead optimization and modulation of hERG activity in a series
of aminooxazoline xanthene β-site amyloid precursor protein cleaving enzyme (BACE1) inhibitors.
J. Med. Chem. 2014, 57, 9796–9810. [CrossRef]
72. Cheng, Y.; Brown, J.; Judd, T.C.; Lopez, P.; Qian, W.; Powers, T.S.; Chen, J.J.; Bartberger, M.D.; Chen, K.;
Dunn, R.T.; et al. An orally available BACE1 inhibitor that affords robust CNS Aβ reduction without
cardiovascular liabilities. Acs Med. Chem. Lett. 2015, 6, 210–215. [CrossRef]
73. U.S. National Library of Medicine. A Safety Study of LY2811376 Single Doses in Healthy Subjects.
Available online: https://clinicaltrials.gov/ct2/show/NCT00838084 (accessed on 18 March 2019).
74. May, P.C.; Dean, R.A.; Lowe, S.L.; Martenyi, F.; Sheehan, S.M.; Boggs, L.N.; Monk, S.A.; Mathes, B.M.;
Mergott, D.J.; Watson, B.M.; et al. Robust Central Reduction of Amyloid-β in Humans with an Orally
Available, Non-Peptidic -Secretase Inhibitor. J. Neurosci. 2011, 31, 16507–16516. [CrossRef]
75. May, P.C.; Willis, B.A.; Lowe, S.L.; Dean, R.A.; Monk, S.A.; Cocke, P.J.; Audia, J.E.; Boggs, L.N.; Borders, A.R.;
Brier, R.A.; et al. The potent BACE1 inhibitor LY2886721 elicits robust central Aβ pharmacodynamic
responses in mice, dogs, and humans. J. Neurosci. 2015, 35, 1199–1210. [CrossRef]
76. U.S. National Library of Medicine. Study of LY2886721 in Mild Cognitive Impairment Due to Alzheimer’s
Disease or Mild Alzheimer’s Disease. Available online: https://clinicaltrials.gov/ct2/show/NCT01561430
(accessed on 18 March 2019).
77. PRNewswire. Lilly Voluntarily Terminates Phase II Study for LY2886721, a Beta Secretase
Inhibitor, Being Investigated as a Treatment for Alzheimer’s Disease. 2013. Available online:
https://www.prnewswire.com/news-releases/lilly-voluntarily-terminates-phase-ii-study-for-ly2886721-
a-beta-secretase-inhibitor-being-investigated-as-a-treatment-for-alzheimers-disease-211452171.html
(accessed on 5 January 2019).
78. ALZFORUM. RG7129. Available online: https://www.alzforum.org/therapeutics/rg7129 (accessed on
5 January 2019).
79. Jacobsen, H.; Ozmen, L.; Caruso, A.; Narquizian, R.; Hilpert, H.; Jacobsen, B.; Terwel, D.; Tanghe, A.;
Bohrmann, B. Combined Treatment with a BACE Inhibitor and Anti-Aβ Antibody Gantenerumab Enhances
Amyloid Reduction in APP London Mice. J. Neurosci. 2014, 34, 11621–11630. [CrossRef]
80. Hills, I.D.; Katharine Holloway, M.; de León, P.; Nomland, A.; Zhu, H.; Rajapakse, H.; Allison, T.J.;
Munshi, S.K.; Colussi, D.; Pietrak, B.L.; et al. A conformational constraint improves a β-secretase inhibitor
but for an unexpected reason. Bioorganic Med. Chem. Lett. 2009, 19, 4993–4995. [CrossRef]
81. Malamas, M.S.; Erdei, J.; Gunawan, I.; Barnes, K.; Johnson, M.; Yu, H.; Turner, J.; Yun, H.; Wagner, E.;
Fan, K.; et al. Aminoimidazoles as potent and selective human β-secretase (BACE1) inhibitors. J. Med. Chem.
2009, 52, 6314–6323. [CrossRef]
82. Swahn, B.M.; Holenz, J.; Kihlström, J.; Kolmodin, K.; Lindström, J.; Plobeck, N.; Rotticci, D.; Sehgelmeble, F.;
Sundström, M.; Fälting, J.; et al. Aminoimidazoles as BACE-1 inhibitors: The challenge to achieve in vivo
brain efficacy. Bioorganic Med. Chem. Lett. 2012, 22, 1854–1859. [CrossRef]
83. AstraZeneca. AstraZeneca and Lilly Announce Alliance to Develop and Commercialise BACE Inhibitor
AZD3293 for Alzheimer’s Disease. 2014. Available online: https://www.astrazeneca.com/media-
centre/press-releases/2014/astrazeneca-lilly-bace-inhibitor-azd3293-alzheimers-disease-16092014.html#!
(accessed on 15 February 2019).
84. Cebers, G.; Alexander, R.C.; Haeberlein, S.B.; Han, D.; Goldwater, R.; Ereshefsky, L.; Olsson, T.; Ye, N.;
Rosen, L.; Russell, M.; et al. AZD3293: Pharmacokinetic and pharmacodynamic effects in healthy subjects
and patients with Alzheimer’s disease. J. Alzheimer’s Dis. 2017, 55, 1039–1053. [CrossRef]
85. ALZFORUM. Lanabecestat. Available online: https://www.alzforum.org/therapeutics/azd3293
86. Menting, K.W.; Claassen, J.A.H.R. β-secretase inhibitor; a promising novel therapeutic drug in Alzheimer’s
Disease. Front. Aging Neurosci. 2014, 6, 165. [CrossRef]
87. Scott, J.D.; Li, S.W.; Brunskill, A.P.J.; Chen, X.; Cox, K.; Cumming, J.N.; Forman, M.; Gilbert, E.J.;
Hodgson, R.A.; Hyde, L.A.; et al. Discovery of the 3-Imino-1,2,4-thiadiazinane 1,1-Dioxide Derivative
Verubecestat (MK-8931)-A β-Site Amyloid Precursor Protein Cleaving Enzyme 1 Inhibitor for the Treatment
of Alzheimer’s Disease. J. Med. Chem. 2016, 59, 10435–10450. [CrossRef]
88. Rochin, L.; Hurbain, I.; Serneels, L.; Fort, C.; Watt, B.; Leblanc, P.; Marks, M.S.; De Strooper, B.; Raposo, G.;
van Niel, G. BACE2 processes PMEL to form the melanosome amyloid matrix in pigment cells. Proc. Natl.
Acad. Sci. USA 2013, 110, 10658–10663. [CrossRef]
89. ALZFORUM. Verubecestat. Available online: https://www.alzforum.org/therapeutics/verubecestat
90. Merck. Merck Announces Discontinuation of APECS Study Evaluating Verubecestat (MK-8931) for the
Treatment of People with Prodromal Alzheimer’s Disease. Available online: https://investors.merck.
com/news/press-release-details/2018/Merck-Announces-Discontinuation-of-APECS-Study-Evaluating-
Verubecestat-MK-8931-for-the-Treatment-of-People-with-Prodromal-Alzheimers-Disease/default.aspx
(accessed on 13 March 2019).
91. ALZFORUM. Bump in the Road or Disaster? BACE Inhibitors Worsen Cognition. Available online: https://www.
alzforum.org/news/conference-coverage/bump-road-or-disaster-bace-inhibitors-worsen-cognition (accessed
on 13 March 2019).
pharmaceuticals
Article
Localization of 99mTc-GRP Analogs in
GRPR-Expressing Tumors: Effects of Peptide Length
and Neprilysin Inhibition on Biological Responses
Aikaterini Kaloudi 1 , Emmanouil Lymperis 1 , Panagiotis Kanellopoulos 1 , Beatrice Waser 2 ,
Marion de Jong 3 , Eric P. Krenning 4 , Jean Claude Reubi 2 , Berthold A. Nock 1 and
Theodosia Maina 1, *
1 Molecular Radiopharmacy, INRASTES, NCSR “Demokritos”, 15310 Athens, Greece;
katerinakaloudi@yahoo.gr (A.K.); mlymperis@hotmail.com (E.L.); kanelospan@gmail.com (P.K.);
nock_berthold.a@hotmail.com (B.A.N.)
2 Cell Biology and Experimental Cancer Research, Institute of Pathology, University of Berne,
CH-3010 Berne, Switzerland; waserpatho@rubigen.ch (B.W.); jean.reubi@pathology.unibe.ch (J.C.R.)
3 Department of Radiology & Nuclear Medicine Erasmus MC, 3015 CN Rotterdam, The Netherlands;
m.hendriks-dejong@erasmusmc.nl
4 Cytrotron Rotterdam BV, Erasmus MC, 3015 CN Rotterdam, The Netherlands; erickrenning@gmail.com
* Correspondence: maina_thea@hotmail.com; Tel.: +30-210-6503908

Abstract: The overexpression of gastrin-releasing peptide receptors (GRPRs) in frequently occurring

human tumors has provided the opportunity to use bombesin (BBN) analogs as radionuclide
carriers to cancer sites for diagnostic and therapeutic purposes. We have been alternatively exploring
human GRP motifs of higher GRPR selectivity compared to frog BBN sequences aiming to improve
pharmacokinetic profiles. In the present study, we compared two differently truncated human
endogenous GRP motifs: GRP(14–27) and GRP(18–27). An acyclic tetraamine was coupled at the
N-terminus to allow for stable binding of the SPECT radionuclide 99m Tc. Their biological profiles
were compared in PC-3 cells and in mice without or with coinjection of phosphoramidon (PA) to
induce transient neprilysin (NEP) inhibition in vivo. The two 99m Tc-N4 -GRP(14/18–27) radioligands
displayed similar biological behavior in mice. Coinjection of PA exerted a profound effect on in vivo
stability and translated into notably improved radiolabel localization in PC-3 experimental tumors.
Hence, this study has shown that promising 99m Tc-radiotracers for SPECT imaging may indeed
derive from human GRP sequences. Radiotracer bioavailability was found to be of major significance.
It could be improved during in situ NEP inhibition resulting in drastically enhanced uptake in
GRPR-expressing lesions.
Keywords: bombesin; gastrin-releasing peptide; gastrin-releasing peptide receptor; tumor targeting;

99m Tc-radioligand; metabolic stability; neprilysin-inhibition; phosphoramidon
1. Introduction
Gastrin-releasing peptide receptors (GRPRs) are overexpressed in several human malignancies
such as prostate cancer, mammary carcinoma, and lung cancer [1–10]. Consequently, they have
attracted considerable attention as potential biomolecular targets for diagnosis and therapy
with radionuclide carriers directed to GRPR-positive cancer lesions [11,12]. Originally, the frog
tetradecapeptide bombesin (BBN, Pyr-Gln-Arg-Tyr-Gly-Asn-Gln-Trp-Ala-Val-Gly-His-Leu-Met-NH2 )
and its truncated C-terminal octapeptide fragment BBN(7–14) have served as motifs for the
development of GRPR-targeting radioligands. However, BBN-like analogs bind with comparable

affinity not only to GRPR (BB2 R), but also to the neuromedin B (NMBR, BB1 R), another
member of the three mammalian bombesin receptor subtypes [1,2]. The above two subtypes
are pharmacologically distinguished by their selectivity for different endogenous human
homologs of amphibian BBN. Thus, the 27-mer GRP (H-Val-Pro-Leu-Pro-Ala-Gly-Gly-Gly-Thr-
Val-Leu-Thr-Lys-Met-Tyr-Pro-Arg-Gly-Asn-His-Trp-Ala-Val-Gly-His-Leu-Met-NH2 ), the 14-mer
GRP(14–27), and the C-terminal decapeptide GRP(18–27) fragments strongly bind to the GRPR,
whereas neuromedin B (NMB, H-Gly-Asn-Leu-Trp-Ala-Thr-Gly-His-Phe-Met-NH2 ) exhibits high
affinity for the NMBR [13]. The two GRPR and NMBR subtypes are physiologically expressed in the
human brain and the gut, especially in stomach, pancreas, and gastrointestinal tract, and they are
also implicated in cancer [14–16]. It is reasonable to assume, that radiolabeled BBN agonists of poor
GRPR selectivity will show increased levels of background radioactivity by virtue of their binding to
both GRPR and NMBR populations distributed in the body, especially in the abdomen. Furthermore,
additive GRPR- and NMBR-mediated effects in the gastrointestinal tract, such as abdominal smooth
muscle contraction and stimulation of gastrointestinal hormone secretion, are to be expected after
intravenous injection of BBN-like agonist radioligands [17–22].
In contrast to amphibian BBN-like motifs, the respective human homologs have surprisingly
remained unexploited as radionuclide carriers for targeting GRPR-positive cancer [13]. Motivated by
this gap in the inventory of GRPR-directed radioligands we have expanded our research efforts to
native human GRP sequences in order to explore their applicability in GRPR-targeted tumor diagnosis
and therapy. First, we introduced a small library of tetraamine derivatized GRP(18–27) analogs labeled
with the SPECT radionuclide 99m Tc [23,24]. Compared to previously reported 99m Tc-radiopeptides,
which are based on the full-length BBN or its truncated BBN(7–14) octapeptide fragment [25],
the 99m Tc-N4 -GRP(18–27) showed high GRPR selectivity and superior in vivo characteristics in
tumor-bearing mice, such as faster renal clearance and improved tumor to background ratios.
On the other hand, single or double amino acid substitutions in the decapeptide backbone exerted
pronounced effects on several biological properties, eventually affecting tumor targeting capabilities
and pharmacokinetics. In a following study, a series of differently truncated GRP sequences were
coupled to the universal chelator DOTA (1,4,7,10- tetraazacyclododecane-1,4,7,10-tetraacetic acid) and
labeled with 111 In. Receptor affinity, internalization efficiency and tumor uptake of these analogs were
favored both by longer peptide chain and by the presence of basic amino acids Lys13 and Arg17 in the
native GRP sequence [26].
Following this line of research, we herein introduced 99m Tc-N4 -GRP(14–27) and compared its
biological profile in PC-3 cells and mice models with 99m Tc-N4 -GRP(18–27) (Figure 1). It should
be noted that basic positions Lys13 and Arg17 in the native GRP sequence are now occupied by
the positively charged N4 +x /[99m Tc(O)2 (N4 )]+1 -moiety and not by the negatively charged DOTA.
This arrangement allows for comparisons with the DOTA-derivatized analogs and further studying
the influence of positive/negative charges in 13 and 17 positions of the GRP chain [26]. Next,
the selectivity of N4 -GRP(14–27) for each of the three mammalian bombesin receptor subtypes was
investigated applying receptor autoradiography in human excised biopsy samples, expressing one of
the GRPR, NMBR, and bombesin subtype 3 (BB3 R) receptors. Finally, the impact of in vivo stability
of 99m Tc-N4 -GRP(14–27) and 99m Tc-N4 -GRP(18–27) on tumor targeting and pharmacokinetics was
compared in mice. The role of neprilysin (NEP) [27] on the in vivo degradation of the two human
GRP-based sequences was monitored by HPLC analysis of blood samples collected without or with
coinjection of the NEP-inhibitor phosphoramidon (PA) [28,29], as previously described for BBN-like
radioligands [30–34]. The enhancement of radiotracer localization in experimental GRPR-positive
PC-3 tumors in mice during transient NEP inhibition induced by PA was assessed.
Figure 1. Chemical structure of (a) 99m Tc-N4 -GRP(14–27) (red) and (b) 99m Tc-N4 -GRP(18–27) (blue).
2. Results
2.1. Peptides and Radioligands

The bifunctional acyclic tetraamine chelator (6-(carboxy)-1,4,8,11-tetraazaundecane) was
covalently coupled by its carboxy functionality at the N-terminal Met14 of GRP(14–27) or the Gly18
of GRP(18–27) via an amide bond [24], generating two different length GRP analogs amenable
for labeling with the preeminent SPECT radionuclide 99m Tc. Labeling was typically proceeded by
a brief incubation with 99m TcO4 − generator eluate, SnCl2 as reducing agent, and citrate anions as
transfer ligand in alkaline pH at ambient temperature at molar activities of 20 to 40 MBq 99m Tc/nmol
peptide. Quality control of the radiolabeled products combined HPLC and ITLC analysis. The total
radiochemical impurities, comprising 99m TcO4 − , [99m Tc] citrate, and 99m TcO2 × H2 O, did not exceed
2%, while a single radiopeptide species was detected by RP-HPLC. In view of labeling yields >98%
and >99% radiochemical purity of the resultant 99m Tc-N4 -GRP(14–27) and 99m Tc-N4 -GRP(18–27),
the radioligands were used without further purification in all subsequent experiments. Representative
radiochromatograms of HPLC analysis of 99m Tc-N4 -GRP(14–27) and 99m Tc-N4 -GRP(18–27) are included
in Figure 2a,b, respectively.
Figure 2. Representative radiochromatograms of the radiolabeling reaction mixture of

(a) 99m Tc-N4 -GRP(14–27) (red) and (b) 99m Tc-N4 -GRP(18–27) (blue), confirming the quantitative
formation of high purity radioligands at tR = 14.9 min and tR = 13.3 min, respectively (HPLC system 1).
2.2. In Vitro Assays
2.2.1. Receptor Autoradiography in Human Tumor Samples

The selective affinities of N4 -GRP(14–27) for each of the three bombesin receptor subtypes
found in mammals were studied during in vitro competition binding assays against the universal
radioligand 125 I-[DTyr6 ,βAla11 ,Phe13 ,Nle14 ]BBN(6–14) [6]. Receptor autoradiography was applied
in cryostat sections of well characterized human cancers, preferentially expressing one of the
subtypes. As summarized in Table 1, N4 -GRP(14–27) showed high affinity for the GRPR expressed
in resected prostate carcinoma specimens (IC50 = 4.2 ± 1.0 nM, n = 3, vs. IC50 = 2.4 ± 1.0 nM,
n = 3 for N4 -GRP(18–27) [23]), very low affinity for NMBR present in ileal carcinoid biopsy samples
(IC50 = 72 ± 7.6 nM, n = 3, vs. IC50 = 106 ± 13 nM; n = 2 for N4 -GRP(18–27) [23]), and no affinity for the
BB3 R expressed in bronchial carcinoid samples (IC50 > 1000 nM, n = 3, identical to N4 -GRP(18–27) [23]).
Thus, N4 -GRP(14–27) similarly to N4 -GRP(18–27), displayed good selectivity for the GRPR. Hence,
the GRP-based analogs turned out to be more GRPR-preferring compared to BBN-based radioligands,
like Demobesin 3 (N4 -[Pro1 ,Tyr4 ]BBN) [25] or [DTyr6 ,βAla11 ,Phe13 ,Nle14 ]BBN(6–14) (Table 1).
Table 1. Affinities for the three human bombesin receptor subtypes.
IC50 s in nM
Peptide Conjugate
GRPR 1 NMBR 2 BB3 R 3
Universal ligand 4 1.5 ± 0.1 (3) 1.5 ± 0.2 (3) 3.5 ± 0.7 (3)
N4 -GRP(14–27) 4.2 ± 1.0 (3) 72 ± 7.6 (3) >1000 (3)
N4 -GRP(18–27) 2.4 ± 1.0 (3) 106 ± 13 (2) >1000 (3)
Demobesin 3 0.5 (2) 1.6 (2) >100 (3)
The data represents the mean (±SEM; n = 3) for the cold universal ligand and for the two N4 -GRP(14/18–27) analogs
and the mean (n = 2) for Demobesin 3. 125 I[DTyr6 ,βAla11 , Phe13 ,Nle14 ]BBN(6–14) was used as radioligand in all
experiments; 1 expressed in human prostate cancer, 2 in human ileal carcinoids and 3 in human lung carcinoids,
4 [DTyr6 ,βAla11 ,Phe13 ,Nle14 ]BBN(6–14) was used as cold universal ligand.
2.2.2. Binding Affinity for the Human GRPR

As shown in Figure 3a, N4 -GRP(14–27) and N4 -GRP(18–27) as well as the respective GRP(14–27)
and GRP(18–27) parent peptide references displaced [125 I-Tyr4 ]BBN from GRPR-sites on PC-3 cell
membranes in a monophasic and dose-dependent manner. The respective half-maximal inhibitory
concentration (IC50 ) values differed, yielding the following rank of decreasing receptor affinity:
N4 -GRP(14–27) (IC50 0.32 ± 0.03 nM) > GRP(14–27) (IC50 0.45 ± 0.02 nM) > N4 -GRP(18–27)
(IC50 0.63 ± 0.06 nM) > GRP(18–27) (IC50 1.66 ± 0.20 nM). We observe that the longer-chain peptides
consistently showed higher binding affinity to GRPR than their shorter chain counterparts. Moreover,
coupling of the positively charged acyclic tetraamine unit in the N-terminus of parent GRP(14/18–27)
references improved the affinity of resulting analogs to the GRPR, as previously reported for similarly
modified peptide analogs [35].
2.2.3. Internalization of 99m Tc-N4 -GRP(14–27) and 99m Tc-N4 -GRP(18–27) in PC-3 Cells
During incubation at 37 ◦ C in PC-3 cells, both 99m Tc-N4 -GRP(14–27) and 99m Tc-N4 -GRP(18–27)
were taken up by the cells via a GRPR-mediated process, as demonstrated by the lack of internalization
observed in the presence of excess [Tyr4 ]BBN. In both cases the bulk of cell-associated radioactivity
was found in the cells with 99m Tc-N4 -GRP(14–27) internalizing much faster in PC-3 cells compared to
99m Tc-N -GRP(18–27) at all time intervals (Figure 3b). For example, at 1 h, 12.7 ± 0.7% of total added
4
99m Tc-N -GRP(14–27) specifically internalized in the cells vs. 5.0 ± 0.3% of 99m Tc-N -GRP(18–27),
4 4
whereas at 2 h these values increased to 19.5 ± 1.4% and 6.9 ± 1.5%, respectively.
Figure 3. (a) [125 I-Tyr4 ]BBN displacement curves from gastrin-releasing peptide receptor (GRPR)-sites
on PC-3 cells after 1-h incubation at 22 ◦ C by N4 -GRP(14–27) (red solid line—IC50 0.32±0.03 nM),
GRP(14–27) (red dashed line—IC50 0.45 ± 0.02 nM), N4 -GRP(18–27) (blue solid line—IC50 0.63±0.06
nM) and GRP(18–27) (blue dashed line—IC50 1.66 ± 0.20 nM). (b) GRPR-specific internalization of
99m Tc-N -GRP(14–27) (red solid line) and 99m Tc-N -GRP(18–27) (blue solid line) in PC-3 cells during
4 4
incubation at 37 ◦ C at 15, 30, 60, and 120 min. Results represent average of cell internalized activity ± sd,
n = 3; data is corrected for nonspecific internalization in the presence of 1 µM [Tyr4 ]BBN.
2.3. In Vivo Comparison of 99m Tc-N4 -GRP(14–27) and 99m Tc-N4 -GRP(18–27)
2.3.1. Stability of 99m Tc-N4 -GRP(14–27) and 99m Tc-N4 -GRP(18–27) in Mice
The two 99m Tc-N4 -GRP(14–27) and 99m Tc-N4 -GRP(18–27) radiotracers exhibited distinct resistance
to degrading proteases after injection in mice. As revealed by HPLC analysis of blood samples collected
at 5 min postinjection (pi), 99m Tc-N4 -GRP(14–27) was found less stable (20.1 ± 4.5% intact, n = 3) than
the shorter chain 99m Tc-N4 -GRP(18–27) (31.0 ± 0.9% intact, n = 3). Representative radiochromatograms
are shown in Figure 4a,b, respectively.
Figure 4. Representative radiochromatograms of HPLC analysis of mouse blood samples collected 5 min
pi of (a) 99m Tc-N4 -GRP(14–27) (25.2% intact radiotracer; red dashed line) or (b) 99m Tc-N4 -GRP(18–27)
(31.8% intact radiotracer; blue dashed line) without PA coinjection; the respective radiochromatograms of
(c) 99m Tc-N4 -GRP(14–27) (63.1% intact radiotracer; tR = 29.6 min; red solid line) or (d) 99m Tc-N4 -GRP(18–27)
(68.1% intact radiotracer; tR = 25.1 min; blue solid line) with PA coinjection are also included; the tR of
parent radiopeptide was determined by coinjection with the respective radioligand sample in the column
(HPLC system 2) and is indicated here by the arrow.
It should be noted that coinjection of the NEP-inhibitor PA remarkably enhanced the in vivo
stability of 99m Tc-N4 -GRP(14–27) (66.5 ± 4.8% intact, n = 3) and 99m Tc-N4 -GRP(18–27) (70.8 ± 5.4%
intact, n = 3) in the circulation, revealing NEP as a major degrading protease for both radiotracers in
mice. Representative radiochromatograms are included in Figure 4c,d, respectively.
2.3.2. Biodistribution in PC-3 Xenograft-Bearing Mice

The biodistribution of 99m Tc-N4 -GRP(14–27) and 99m Tc-N4 -GRP(18–27) was studied in severe
combined immune deficiency (SCID) mice bearing human PC-3 xenografts expressing the human
GRPR. Subcutaneous tumors of suitable size developed in the flanks of mice about four weeks after
inoculation of a suspension of prostate adenocarcinoma PC-3 cells and biodistribution was conducted.
Cumulative biodistribution results for 99m Tc-N4 -GRP(14–27) at the 1-, 4-, and 24-h pi intervals are
summarized in Table 2, and are expressed as mean % injected dose per gram (%ID/g) values ± sd,
n = 4. The radiotracer washed rapidly from the blood and the background tissues predominantly
via the kidneys and the urinary system. High uptake was observed in the PC-3 tumor at 1-h pi
(10.20 ± 0.72%ID/g) that remained at comparably high levels at 4-h pi (8.41 ± 4.16%ID/g; p > 0.05),
declining by ~50% at 24-h pi (4.50 ± 0.69%ID/g). Tumor uptake at 4-h pi was significantly lower in the
animals treated with excess [Tyr4 ]BBN (0.62 ± 0.24%ID/g; p < 0.001), suggestive of a GRPR-mediated
process. Likewise, 99m Tc-N4 -GRP(14–27) highly localized in the GRPR-rich mouse pancreas via
a GRPR-specific process, as demonstrated by the lack of pancreatic uptake during GRPR-blockade by
coinjection of excess [Tyr4 ]BBN (35.24 ± 4.70%ID/g vs. 0.83 ± 0.24%ID/g in block; p < 0.001).
Table 2. Biodistribution data for 99m Tc-N4 -GRP(14–27), expressed as %ID/g mean ± sd, n = 4, in PC-3
xenograft-bearing SCID mice at 1-h, 4-h block, 4-h, and 24-h pi.
Tissue 1h1 4h1 24 h 1 4 h block 2

Blood 1.54 ± 0.17 0.10 ± 0.03 0.07 ± 0.01 0.08 ± 0.02
Liver 5.02 ± 0.46 3.75 ± 1.11 3.23 ± 0.78 6.12 ± 2.89
Heart 0.70 ± 0.07 0.11 ± 0.04 0.07 ± 0.01 0.28 ± 0.13
Kidneys 14.82 ± 2.22 4.83 ± 2.12 3.10 ± 0.73 4.85 ± 2.66
Stomach 1.02 ± 0.29 1.42 ± 0.93 0.78 ± 0.33 0.43 ± 0.18
Intestines 7.21 ± 0.52 7.61 ± 2.23 1.73 ± 0.23 1.45 ± 0.44
Muscle 0.29 ± 0.02 0.03 ± 0.01 0.06 ± 0.01 0.05 ± 0.02
Lungs 1.96 ± 0.33 0.49 ± 0.30 0.18 ± 0.04 0.65 ± 0.22
Pancreas 37.85 ± 1.95 35.24 ± 4.70 13.41 ± 0.77 0.83 ± 0.24
Tumor 10.20 ± 0.72 8.41 ± 4.16 4.50 ± 0.69 0.62 ± 0.24
1Animal groups injected with 180–370 kBq/10 pmol peptide; 2 block mice group coinjected with 40 nmol [Tyr4 ]
BBN for in vivo GRPR blockade.
Comparative biodistribution results for 99m Tc-N4 -GRP(14–27) and 99m Tc-N4 -GRP(18–27) at the
4-h pi interval are included in Table 3. Data from additional 4-h pi animal groups coinjected with
the NEP-inhibitor PA (300 µg) is also included in the Table. The radiotracers displayed similar tissue
distribution patterns. The observed higher tumor and pancreatic uptake of 99m Tc-N4 -GRP(14–27)
did not however differ significantly to that of 99m Tc-N4 -GRP(18–27) (p > 0.05) [24]. Treatment with
PA induced a drastic increase in tumor values for both radiotracers with 99m Tc-N4 -GRP(14–27)
showing superior tumor values than the shorter chain 99m Tc-N4 -GRP(18–27) (38.19 ± 4.79%ID/g vs.
28.37 ± 8.05%ID/g, respectively; p < 0.01). Pancreatic values increased by >3-fold for both radiotracers
as well. Thus, in agreement with previous studies on BBN-based analogs [31–34], NEP inhibition
likewise resulted in significant stabilization of GRP(14/18–27)-based radioligands in peripheral mouse

blood and notable improvement of localization in GRPR-expressing lesions in mice.
Table 3. Comparative biodistribution data for 99m Tc-N4 -GRP(14–27) and 99m Tc-N4 -GRP(18–27),
expressed as %ID/g mean ± sd, n = 4, in PC-3 xenograft-bearing SCID mice at 4-h pi (controls)
and 4-h pi after coinjection of PA.
99m Tc-N -GRP(14–27) 99m Tc-N -GRP(18–27)

4 4
Tissue
4h1 4 h PA 1,2 4 h1 4 h PA 1,2
Blood 0.10 ± 0.03 0.23 ± 0.08 0.13 ± 0.04 0.21 ± 0.04
Liver 3.75 ± 1.11 4.99 ± 1.53 1.02 ± 0.17 2.13 ± 0.37
Heart 0.11 ± 0.04 0.89 ± 0.06 0.14 ± 0.10 0.47 ± 0.27
Kidneys 4.83 ± 2.12 11.68 ± 1.61 6.01 ± 1.38 7.66 ± 1.39
Stomach 1.42 ± 0.93 3.82 ± 1.10 1.12 ± 0.75 4.02 ± 0.66
Intestines 7.61 ± 2.23 21.35 ± 1.68 7.28 ± 0.60 16.23 ± 3.90
Muscle 0.03 ± 0.01 0.06 ± 0.02 0.03 ± 0.01 0.06 ± 0.01
Lungs 0.49 ± 0.30 1.06 ± 0.19 0.28 ± 0.09 0.84 ± 0.26
Pancreas 35.24 ± 4.70 110.32 ± 8.76 32.18 ± 5.91 95.39 ± 20.34
Tumor 8.41 ± 4.16 38.19 ± 4.79 7.08 ± 1.29 28.37 ± 8.05
1Animal groups injected with 180–370 kBq/10 pmol peptide; 2 PA mice groups with animals coinjected with 300 µg
PA to in situ inhibit NEP.
3. Discussion
A considerable number of radiolabeled analogs of frog BBN have been developed for potential
application in the diagnosis and therapy of GRPR-expressing tumors in Homo sapiens [11,12].
This pursuit is based on the overexpression of GRPRs on the surface of malignant cells serving as
easily accessible biomolecular targets on cancer lesions [3–10]. Joining this effort, we have previously
introduced a series of BBN-like analogs, generated by covalently coupling of an acyclic tetraamine at the
N-terminus, Demobesin 3–6 [25]. Like native BBN, these peptide ligands displayed indistinguishable
binding affinities for the human bombesin receptor subtypes GRPR and NMBR, but no affinity for the
BB3 R. The respective radiotracers [99m Tc] Demobesin 3–6 specifically localized in GRPR-expressing
human PC-3 xenografts in mice. Moreover, one of the analogs, [99m Tc] Demobesin 4, was able to
visualize malignant lesions in a small number of prostate cancer patients with SPECT/CT [36].
We have recently expanded our search toward radioligands based on human endogenous GRP
sequences [23,24,26]. The latter are reported for higher GRPR selectivity compared to their frog
homolog BBN [13]. This approach has surprisingly remained unexplored up to now, although it
offers two major advantages. First, radioactivity levels in abdominal tissues physiologically
expressing both GRPR and NMBR subtypes would favorably decrease when GRPR-selective
radioligands are used [14–16]. Second, injection of GRPR-selective agonists will not activate the
NMBR subtype populations in the gut and is hence associated with less adverse effects [17–22].
Promising GRPR-selective radioligands based on receptor antagonists have been developed and
evaluated in animals and in human, showing excellent tumor targeting and pharmacokinetic
profiles [37,38]. Yet, the suitability and efficacy of noninternalizing GRPR antagonists for therapy
with low-range beta, alpha, or Auger emitters of high LET has not been established thus far [39,40].
Therefore, the study of radiolabeled GRPR-selective agonists is warranted and may provide
an alternative platform for the development of radioligands exhibiting new combinations of biological
features particularly suited for cancer theranostics.
As a part of this effort we herein introduced two analogs of endogenous truncated fragments
of human GRP carrying an acyclic tetraamine at their terminus, one based on the tetradecapeptide
GRP(14–27), and the other on a shorter decapeptide GRP(18–27) chain identical to human neuromedin
C (NMC) (Figure 1). As revealed during receptor autoradiography studies in excised biopsy samples
of well characterized human tumors expressing each of the GRPR, NMBR, and BB3 R subtypes
using the universal 125 I-[DTyr6 ,βAla11 ,Phe13 ,Nle14 ]BBN(6–14) radioligand, both N4 -GRP(14–27) and
N4 -GRP(18–27) analogs showed good selectivity for the GRPR subtype (Table 1). This finding is
in agreement with previous reports documenting the preference of human GRP and its C-terminal
native fragments for GRPR [13]. In contrast, the tetraamine derivatized BBN-like analogs Demobesin
3–6 do not distinguish between GRPR and NMBR, instead behaving like the native frog BBN [25].
The His20 of GRP corresponding to Gln7 of BBN seems to have significant impact on subtype
selectivity, followed by the Met14 -Tyr13 -Pro12 -Arg11 -tetrapeptide in GRP(14–27) corresponding to
the Pyr1 -Gln2 -Arg3 -Leu4 -counterpart in native BBN.
Both N4 -GRP(14–27) and N4 -GRP(18–27) analogs displayed sub-nM affinities for the GRPR during
competition assays against [125 I-Tyr4 ]BBN in PC-3 cell membranes, which were found to be increased
compared to unmodified GRP(14/18–27) lead structures (Figure 3a). It is interesting to note that the
presence of the positively charged N4 +x -moiety at positions 13 and 17 of the native GRP chain occupy
the two basic amino acids Lys13 and Arg17 . When these positions were taken up by negatively charged
DOTA instead, a drastic drop in binding affinity was observed (IC50 DOTA-GRP(14–27) = 6.6 ± 0.9 nM
and IC50 DOTA-GRP(18–27) = 112 ± 16 nM) [26]. In all cases, longer chain GRP(14–27) analogs
consistently displayed higher GRPR-affinity compared to their C-terminal decapeptide counterparts.
In agreement with this observation, the internalization rate of 99m Tc-N4 -GRP(14–27) in PC-3 cells was
clearly superior to that of 99m Tc-N4 -GRP(18–27) (Figure 3b).
A crucial feature for efficient tumor targeting is metabolic stability that will ensure sufficient
radioligand supply to tumor sites after injection in the living organism [31]. Recent studies have revealed
the significance of in vivo stability assessment of BBN-like radiotracers vs. in vitro determinations in
serum or tissue homogenate incubates to more accurately predict the actual protease(s) encountered
by the radioligand after entering the circulation. These studies have demonstrated NEP as a major
degrading protease of BBN and its radiolabeled analogs [41,42]. In a recently proposed approach,
NEP-inhibitors, like PA [28,29], coinjected with the BBN-based radioligand drastically increased
metabolic stability leading to notable enhancement of tumor uptake in animal models [31–34].
Following this rationale, we have tested the in vivo stability of the two 99m Tc-N4 -GRP(14–27) and
99m Tc-N -GRP(18–27) radiotracers in peripheral mouse blood collected 5 min pi by radioanalytical
4
HPLC (Figure 4). The longer chain 99m Tc-N4 -GRP(14–27) displayed somewhat poorer stability
(20.1 ± 4.5% intact, n = 3) compared to shorter chain 99m Tc-N4 -GRP(18–27) (31.0 ± 0.9% intact, n = 3).
Thus, the presence of additional amide bonds in the longer peptide chain offered further degradation
sites for attacking proteases. It is interesting to note, that coinjection of PA drastically increased the
stability of both analogs in the circulation, indicating NEP as a major protease in the catabolism of
GRP sequences as well.
Aiming to explore the effects of peptide chain as well as NEP inhibition on the pharmacokinetics
and tumor targeting capabilities of the two tracers, we have conducted biodistribution experiments in
immunosuppressed mice bearing human GRPR-expressing xenografts. Interestingly, the biodistribution
patterns of the two radiotracers did not clearly differ (Table 3). Compared to the previously reported
111 In-DOTA analogs [26], 99m Tc-N -GRP(14–27) showed higher tumor uptake. Comparable in vivo
4
tumor targeting was obtained only by 111 In-DOTA-GRP(13–27), carrying basic Lys at position 13.
However, kidney clearance was notably faster for the 99m Tc-radiotracer (4.83 ± 2.12%/g at 4 h pi)
than for the 111 In-DOTA-Lys13 -analog (46.02 ± 3.73%/g at 4 h pi) [26]. Treatment of mice with PA
exerted a profound effect on the in vivo profile of both 99m Tc-N4 -GRP(14–27) and 99m Tc-N4 -GRP(18–27)
(Table 2), is in agreement with previous findings for other BBN-like radioligands. The longer chain
99m Tc-N -GRP(14–27) exhibited the highest uptake in the experimental tumor, as a combined result of
4
higher internalization rate and pronounced in vivo stabilization induced by PA.
The present study has shown that GRP motifs of different chain length may provide new
opportunities for the development of promising GRPR-selective radioligands. The latter may prove to
be a useful asset in the arsenal of anti-GRPR radioactive drugs, by internalizing in cancer cells while
evading binding to and activation of bombesin receptor subtypes other than the GRPR in the gut.
Furthermore, it has shown the impact of in vivo metabolic stability in maximizing tumor localization.
Structural modifications in peptide lead-structures to improve the stability of radiopeptides, such as
amino acid replacements, cyclization, or changes of cleavable peptide bonds, may deteriorate other
important biological features, as for example receptor affinity, cell uptake, tumor targeting, and overall
pharmacokinetics. On the other hand, transient in situ inhibition of NEP represents a smart method
to accomplish this goal and warrants further efforts for translation in the clinic. Despite challenges
related to biosafety, regulatory and financial hurdles, once established in a proof-of-principal study,
this strategy is expected to boost the development and clinical application of other NEP-catabolized
radioligands, considerably saving costs and resources [31–34].
4.1. Peptides and Reagents

The N4 -GRP(14–27) and N4 -GRP(18–27) peptide conjugates were synthesized on the solid
support following a published method [24] and were provided by PiChem (Graz, Austria).
The [Tyr4 ]BBN (Tyr4 -bombesin, Pyr-Gln-Arg-Tyr-Gly-Asn-Gln-Trp-Ala-Val-Gly-His-Leu-Met-NH2 )
and GRP(14–27) (H-Met-Tyr-Pro-Arg-Gly-Asn-His-Trp-Ala-Val-Gly-His-Leu-Met-NH2 ) references
were purchased from PSL GmbH (Heidelberg, Germany), whereas GRP(18–27) (H-Gly-Asn-His-Trp-Ala-
Val-Gly-His-Leu-Met-NH2) was provided by PeptaNova GmbH (Sandhousen, Germany). Phosphoramidon
disodium dehydrate (N-(α-rhamnopyranosyloxyhydro xyphosphinyl) -L-leucyl-L-tryptophan × 2Na ×
2H2O; PA) was purchased from PeptaNova GmbH (Sandhousen, Germany).
Preparation and Quality Control of 99m Tc-N4 -GRP(14–27) and 99m Tc-N4 -GRP(18–27)
Lyophilized N4 -GRP(14–27) and N4 -GRP(18–27) were dissolved in bidistilled water to a final
1 mM concentration and bulk solutions were distributed in 50 µL aliquots in Eppendorf vials (Protein
LoBind Tube 1.5 mL; Eppendorf AG, Hamburg, Germany) and stored at −20 ◦ C. For 99m Tc labeling,
the following solutions were added into an Eppendorf tube containing 0.5 M phosphate buffer pH 11.5
(25 µL): 0.1 M sodium citrate (3 µL), [99m Tc]NaTcO4 (210 µL, 150–300 MBq) eluted from a commercial
99 Mo/99m Tc generator (Ultratechnekow, Tyco Healthcare, Petten, The Netherlands), N -GRP(14–27) or
4
N4 -GRP(18–27) stock solution (7.5 µL, 7.5 nmol), and finally, fresh SnCl2 solution in EtOH (5 µg, 5 µL).
After 30 min incubation at ambient temperature the reaction mixture was neutralized by addition of
1 M HCl (4 µL) and EtOH was added (25 µL).
Quality control of the radiolabeled products comprised radioanalytical HPLC and instant
thin-layer chromatography (ITLC). HPLC analyses were performed on a Waters Chromatograph
coupled to a 2998 photodiode array UV detector (Waters, Vienna, Austria) and a Gabi gamma
detector (Raytest RSM Analytische Instrumente GmbH, Germany). For analysis, a Waters Symmetry
Shield RP-18 cartridge column (5 µm, 3.9 mm × 20 mm) was eluted at a 1.0 mL/min flow rate with
the following gradient, 0% B to 40% B in 20 min, where A = 0.1% aq. trifluoroacetic acid (TFA),
B = MeCN (System 1). Under these conditions 99m TcO4 − eluted at 1.8 min and 99m Tc-N4 -GRP(14–27)
and 99m Tc-N4 -GRP(18–27) with a tR > 12 min. For the detection of reduced hydrolyzed technetium
(99m TcO2 ·× H2 O) ITLC was conducted on ITLC-SG strips (Pall Corporation, New York, NY, USA),
as previously described. The resultant 99m Tc-N4 -GRP(14–27) and 99m Tc-N4 -GRP(18–27) radioligands
were used without further purification in all subsequent experiments.
Radioiodination of [Tyr4 ]BBN was performed using 125 I ([125 I]NaI in 0.1 N NaOH (pH 12–14)
provided by MDS Nordion, Ottawa, ON, Canada) according to the chloramine-T methodology,
as previously described [34]. Methionine was added to the purified radioligand solution to prevent
reoxidation of Met14 to the corresponding sulfoxide and the resulting stock solution in 0.1% BSA-PBS
was kept at −20 ◦ C; aliquots thereof were used in competition binding assays (molar activity of
81.4 GBq/µmol).
4.2. In Vitro Assays
4.2.1. Cell Lines and Culture

Human androgen-independent prostate adenocarcinoma PC-3 cells endogenously expressing
the human GRPR (LGC Promochem, Teddington, UK) were used in the present study [43]. Cells
were cultured in Roswell Park Memorial Institute (RPMI)-1640 medium, supplemented with 10%
heat-inactivated fetal bovine serum (FBS), 100 U/mL penicillin, and 100 µg/mL streptomycin, and kept
in a controlled humidified atmosphere containing 5% CO2 at 37 ◦ C. Passages were performed weekly
using a trypsin/EDTA (0.05%/0.02% w/v) solution. All culture media were purchased from Gibco
BRL, Life Technologies and supplements were provided by Biochrom KG Seromed.
4.2.2. Receptor Autoradiography

Binding affinities of N4 -GRP(14–27) were determined by in vitro receptor autoradiography
performed on cryostat sections of well characterized human tumor tissues, prostate carcinomas
for GRPR, ileal carcinoids for NMBR and bronchial carcinoids for BB3 R, as previously described [23].
The universal radioligand 125 I-[DTyr6 ,βAla11 ,Phe13 ,Nle14 ]BBN(6–14) (2 Ci/mol; ANAWA, Wangen
Switzerland) was used as tracer known to identify all three bombesin receptor subtypes [6]. IC50 values
are given in nM ± SEM.
4.2.3. Competition Binding in PC-3 Cell-Membranes

Competition binding experiments against [125 I-Tyr4 ] BBN were performed with N4 -GRP(14–27)
and N4 -GRP(18–27), or with the unmodified parent peptide references GRP(14–27) and GRP(18–27)
in PC-3 cell membranes. For the assay, triplicates per concentration point (concentration range:
10−13 –10−6 M) of each test peptide were incubated together with the radioligand (~40,000 cpm per
assay tube at a 50 pM concentration) in PC-3 cell-membrane homogenates in a total volume of 300 µL
binding buffer (BB, 50 mM HEPES pH 7.4, 1% BSA, 5.5 mM MgCl2 , 35 µM bacitracin) for 1 h at
22 ◦ C in an Incubator-Orbital Shaker (MPM Instr. SrI, Bernareggio, Italy). Binding was interrupted by
ice-cold washing buffer (WB, 10 mM HEPES pH 7.4, 150 mM NaCl) and rapid filtration (Whatman
GF/B filters presoaked in BB) on a Brandel Cell Harvester (Adi Hassel Ing. Büro, Munich, Germany).
Filters were washed with ice-cold WB and counted in an automatic well-type gamma counter (NaI(Tl)
30 -crystal, Cobra Packard Auto-Gamma 5000 series instrument). The IC50 values were calculated
using nonlinear regression according to a one-site model applying the PRISM 2 program (Graph Pad
Software, San Diego, CA, USA).
4.2.4. Internalization Assay in PC-3 Cells

The internalization rates of 99m Tc-N4 -GRP(14–27) and 99m Tc-N4 -GRP(18–27) were compared in
PC-3 cells. Briefly, PC-3 cells were seeded in six-well plates (~1 × 106 cells per well) 24 h before
the experiment. Approximately 50,000 cpm of either 99m Tc-N4 -GRP(14–27) or 99m Tc-N4 -GRP(18–27)
(corresponding to 250 fmol total peptide in 150 µL of 0.5% BSA/PBS) was added alone (total) or in the
presence of 1 µM [Tyr4 ]BBN (nonspecific). Cells were incubated at 37 ◦ C for 15, 30, 60, and 120 min
and incubation was interrupted each time by placing the plates on ice, removing the supernatants and
rapid rinsing with ice-cold 0.5% BSA/PBS. Cells were then treated 2 × 5 min with acid wash buffer
(2 × 0.6 mL, 50 mM glycine buffer pH 2.8, 0.1 M NaCl) at room temperature and supernatants were
collected (membrane-bound fraction). After rinsing with 1 mL chilled 0.5% BSA/PBS, cells were lysed
by treatment with 1 N NaOH (2 × 0.6 mL) and lysates were collected (internalized fraction). Sample
radioactivity was measured in the γ-counter and percent internalized radioactivity was determined
vs. total added activity. Results represent the average values ± sd of three experiments performed
in triplicate.
4.3. Animal Studies
4.3.1. In Vivo Stability Tests

For stability experiments, healthy male Swiss albino mice (30 ± 5 g, NCSR “Demokritos” Animal
House Facility) were used. Test radioligand—99m Tc-N4 -GRP(14–27) or 99m Tc-N4 -GRP(18–27)—was
injected as a 100 µL bolus (37–74 MBq, 3 nmol total peptide) in the tail vein together with injection
solution (100 µL; control) or with a PA-solution (100 µL injection solution containing 300 µg PA).
Animals were euthanized and blood (0.5–1 mL) was directly withdrawn from the heart in an ice-cold
syringe and transferred in a prechilled EDTA and methionine-containing Eppendorf tube on ice.
Blood samples were centrifuged for 10 min at 2000 g/4 ◦ C and plasma was collected. After addition
of an equal volume of ice-cold MeCN the mixture was centrifuged for 10 min at 15,000 g/4 ◦ C.
The supernatant was concentrated under a N2 -flux at 60 ◦ C to 0.05–0.1 mL, diluted with saline (0.4 mL),
filtered through a 0.22 µm Millex GV filter (Millipore, Milford, CT, USA), and analyzed by RP-HPLC.
The Symmetry Shield RP18 (5 µm, 3.9 mm × 20 mm) column was eluted at a flow rate of 1.0 mL/min
with the following linear gradient (system 2): 0% B at 0 min to 10% B in 10 min and then in 40 min
to 30% B; A = 0.1% aq. TFA and B = MeCN. The tR of the intact radiopeptide was determined by
coinjection with the 99m Tc-N4 -GRP(14–27) and 99m Tc-N4 -GRP(18–27) reference in the HPLC.
4.3.2. Induction of PC-3 Xenografts in SCID Mice

A suspension containing freshly harvested human PC-3 cells (≈150 µL of a ≈1.2 × 107 cells) was
subcutaneously injected in the flanks of female SCID mice (15 ± 3 g, six weeks of age at the day of
arrival, NCSR “Demokritos” Animal House Facility). The animals were kept under aseptic conditions
and 4 weeks later developed well-palpable tumors (80–200 mg) at the inoculation sites.
4.3.3. Biodistribution in PC-3 Xenograft-Bearing SCID Mice

For the biodistribution study, animals in groups of 4 received via the tail vein a 100 µL bolus
of 99m Tc-N4 -GRP(14–27) (180–370 kBq, corresponding to 10 pmol total peptide) coinjected either
with injection solution (100 µL; control) or PA-solution (300 µg PA dissolved in 100 µL injection
solution; 4 h + PA), or with excess [Tyr4 ]BBN (100 µL injection solution containing 50 µg [Tyr4 ]BBN
for in vivo GRPR-blockade; 4 h block). Animals were euthanized at 1-, 4-, and 24-h pi; in the case of
99m Tc-N -GRP(18–27) two animal groups were included at the 4-h pi interval, namely the control and
4
PA groups described above. Mice were dissected; samples of blood, tumors, and organs of interest were
collected, weighed, and measured for radioactivity in the gamma counter. Intestines and stomach were
not emptied of their contents. Data was calculated as percent injected dose per gram tissue (%ID/g)
with the aid of standard solutions and represent mean values ± sd, n = 4. All animal experiments were
performed in compliance with national and European guidelines and approved by national authorities
(Prefecture of Athens, EL 25 BIO 021; #1609 and #1610).
Author Contributions: Conceptualization, B.A.N.; Methodology, B.A.N., T.M.N. and J.C.R.; Investigation, A.K.,
E.L., P.K., B.W., B.A.N., T.M.N. and J.C.R.; Writing—Original Draft Preparation, T.M.N.; Writing—Review and
Editing, B.A.N., M.d.J., E.P.K. and J.C.R.
Conflicts of Interest: The authors declare no conflicts of interest.
References
1. Kroog, G.S.; Jensen, R.T.; Battey, J.F. Mammalian bombesin receptors. Med. Res. Rev. 1995, 15, 389–417.
[CrossRef]
2. Jensen, R.T.; Battey, J.F.; Spindel, E.R.; Benya, R.V. International union of pharmacology. LXVIII. Mammalian
bombesin receptors: Nomenclature, distribution, pharmacology, signaling, and functions in normal and
disease states. Pharmacol. Rev. 2008, 60, 1–42. [CrossRef]
3. Markwalder, R.; Reubi, J.C. Gastrin-releasing peptide receptors in the human prostate: Relation to neoplastic
transformation. Cancer Res. 1999, 59, 1152–1159. [PubMed]
4. Körner, M.; Waser, B.; Rehmann, R.; Reubi, J.C. Early over-expression of GRP receptors in prostatic
carcinogenesis. Prostate 2014, 74, 217–224. [CrossRef] [PubMed]
5. Beer, M.; Montani, M.; Gerhardt, J.; Wild, P.J.; Hany, T.F.; Hermanns, T.; Muntener, M.; Kristiansen, G.
Profiling gastrin-releasing peptide receptor in prostate tissues: Clinical implications and molecular correlates.
Prostate 2012, 72, 318–325. [CrossRef] [PubMed]
6. Reubi, J.C.; Wenger, S.; Schmuckli-Maurer, J.; Schaer, J.C.; Gugger, M. Bombesin receptor subtypes in
human cancers: Detection with the universal radioligand 125 I-[d-Tyr6 ,beta-Ala11 ,Phe13 ,Nle14 ]bombesin(6-14).
Clin. Cancer Res. 2002, 8, 1139–1146.
7. Halmos, G.; Wittliff, J.L.; Schally, A.V. Characterization of bombesin/gastrin-releasing peptide receptors in
human breast cancer and their relationship to steroid receptor expression. Cancer Res. 1995, 55, 280–287.
[PubMed]
8. Gugger, M.; Reubi, J.C. Gastrin-releasing peptide receptors in non-neoplastic and neoplastic human breast.
Am. J. Pathol. 1999, 155, 2067–2076. [CrossRef]
9. Mattei, J.; Achcar, R.D.; Cano, C.H.; Macedo, B.R.; Meurer, L.; Batlle, B.S.; Groshong, S.D.; Kulczynski, J.M.;
Roesler, R.; Dal Lago, L.; et al. Gastrin-releasing peptide receptor expression in lung cancer. Arch. Pathol.
Lab. Med. 2014, 138, 98–104. [CrossRef]
10. Guinee, D.G., Jr.; Fishback, N.F.; Koss, M.N.; Abbondanzo, S.L.; Travis, W.D. The spectrum of
immunohistochemical staining of small-cell lung carcinoma in specimens from transbronchial and open-lung
biopsies. Am. J. Clin. Pathol. 1994, 102, 406–414. [CrossRef] [PubMed]
11. Moreno, P.; Ramos-Alvarez, I.; Moody, T.W.; Jensen, R.T. Bombesin related peptides/receptors and their
promising therapeutic roles in cancer imaging, targeting and treatment. Expert. Opin. Ther. Targets 2016,
12. Maina, T.; Nock, B.A. From bench to bed: New gastrin-releasing peptide receptor-directed radioligands and
their use in prostate cancer. PET Clin. 2017, 12, 205–217. [CrossRef] [PubMed]
13. Uehara, H.; Gonzalez, N.; Sancho, V.; Mantey, S.A.; Nuche-Berenguer, B.; Pradhan, T.; Coy, D.H.; Jensen, R.T.
Pharmacology and selectivity of various natural and synthetic bombesin related peptide agonists for human
and rat bombesin receptors differs. Peptides 2011, 32, 1685–1699. [CrossRef] [PubMed]
14. Vigna, S.R.; Mantyh, C.R.; Giraud, A.S.; Soll, A.H.; Walsh, J.H.; Mantyh, P.W. Localization of specific binding
sites for bombesin in the canine gastrointestinal tract. Gastroenterology 1987, 93, 1287–1295. [CrossRef]
15. Chave, H.S.; Gough, A.C.; Palmer, K.; Preston, S.R.; Primrose, J.N. Bombesin family receptor and ligand gene
expression in human colorectal cancer and normal mucosa. Br. J. Cancer 2000, 82, 124–130. [CrossRef]
16. Fleischmann, A.; Laderach, U.; Friess, H.; Buechler, M.W.; Reubi, J.C. Bombesin receptors in distinct tissue
compartments of human pancreatic diseases. Lab. Investig. 2000, 80, 1807–1817. [CrossRef] [PubMed]
17. Delle Fave, G.; Annibale, B.; de Magistris, L.; Severi, C.; Bruzzone, R.; Puoti, M.; Melchiorri, P.; Torsoli, A.;
Erspamer, V. Bombesin effects on human GI functions. Peptides 1985, 6 (Suppl. 3), 113–116. [CrossRef]
18. Bruzzone, R.; Tamburrano, G.; Lala, A.; Mauceri, M.; Annibale, B.; Severi, C.; de Magistris, L.; Leonetti, F.;
Delle Fave, G. Effect of bombesin on plasma insulin, pancreatic glucagon, and gut glucagon in man. J. Clin.
Endocrinol. Metab. 1983, 56, 643–647. [CrossRef]
19. Delle Fave, G.; Kohn, A.; De Magistris, L.; Annibale, B.; Bruzzone, R.; Sparvoli, C.; Severi, C.; Torsoli, A.
Effects of bombesin on gastrin and gastric acid secretion in patients with duodenal ulcer. Gut 1983, 24, 231–235.
[CrossRef]
20. Severi, C.; Jensen, R.T.; Erspamer, V.; D’Arpino, L.; Coy, D.H.; Torsoli, A.; Delle Fave, G. Different receptors
mediate the action of bombesin-related peptides on gastric smooth muscle cells. Am. J. Physiol. 1991,
260, G683–G690. [CrossRef]
21. Bitar, K.N.; Zhu, X.X. Expression of bombesin-receptor subtypes and their differential regulation of colonic
smooth muscle contraction. Gastroenterology 1993, 105, 1672–1680. [CrossRef]
22. Falconieri Erspamer, G.; Severini, C.; Erspamer, V.; Melchiorri, P.; Delle Fave, G.; Nakajima, T. Parallel
bioassay of 27 bombesin-like peptides on 9 smooth muscle preparations. Structure-activity relationships and
bombesin receptor subtypes. Regul. Pept. 1988, 21, 1–11. [CrossRef]
23. Nock, B.A.; Cescato, R.; Ketani, E.; Waser, B.; Reubi, J.C.; Maina, T. [99m Tc]Demomedin C, a radioligand
based on human gastrin releasing peptide(18–27): Synthesis and preclinical evaluation in gastrin releasing
peptide receptor-expressing models. J. Med. Chem. 2012, 55, 8364–8374. [CrossRef] [PubMed]
24. Marsouvanidis, P.J.; Maina, T.; Sallegger, W.; Krenning, E.P.; de Jong, M.; Nock, B.A. 99m Tc Radiotracers
based on human GRP(18–27): Synthesis and comparative evaluation. J. Nucl. Med. 2013, 54, 1797–803.
[CrossRef] [PubMed]
25. Nock, B.A.; Nikolopoulou, A.; Galanis, A.; Cordopatis, P.; Waser, B.; Reubi, J.C.; Maina, T. Potent bombesin-like
peptides for GRP-receptor targeting of tumors with 99m Tc: A preclinical study. J. Med. Chem. 2005, 48, 100–110.
[CrossRef] [PubMed]
26. Marsouvanidis, P.J.; Maina, T.; Sallegger, W.; Krenning, E.P.; de Jong, M.; Nock, B.A. Tumor diagnosis
with new 111 In-radioligands based on truncated human gastrin releasing peptide sequences: Synthesis and
preclinical comparison. J. Med. Chem. 2013, 56, 8579–8587. [CrossRef]
27. Roques, B.P.; Noble, F.; Dauge, V.; Fournie-Zaluski, M.C.; Beaumont, A. Neutral endopeptidase 24.11:
Structure, inhibition, and experimental and clinical pharmacology. Pharmacol. Rev. 1993, 45, 87–146. [PubMed]
28. Suda, H.; Aoyagi, T.; Takeuchi, T.; Umezawa, H. Letter: A thermolysin inhibitor produced by actinomycetes:
Phosphoramidon. J. Antibiot. (Tokyo) 1973, 26, 621–623. [CrossRef]
29. Oefner, C.; D’Arcy, A.; Hennig, M.; Winkler, F.K.; Dale, G.E. Structure of human neutral endopeptidase
(neprilysin) complexed with phosphoramidon. J. Mol. Biol. 2000, 296, 341–349. [CrossRef] [PubMed]
30. Marsouvanidis, P.J.; Melis, M.; de Blois, E.; Breeman, W.A.; Krenning, E.P.; Maina, T.; Nock, B.A.; de Jong, M.
In vivo enzyme inhibition improves the targeting of [177 Lu]DOTA-GRP(13-27) in GRPR-positive tumors in
mice. Cancer Biother. Radiopharm. 2014, 29, 359–367. [CrossRef] [PubMed]
31. Nock, B.A.; Maina, T.; Krenning, E.P.; de Jong, M. “To serve and protect”: Enzyme inhibitors as radiopeptide
escorts promote tumor targeting. J. Nucl. Med. 2014, 55, 121–127. [CrossRef]
32. Chatalic, K.L.; Konijnenberg, M.; Nonnekens, J.; de Blois, E.; Hoeben, S.; de Ridder, C.; Brunel, L.;
Fehrentz, J.A.; Martinez, J.; van Gent, D.C.; et al. In vivo stabilization of a gastrin-releasing peptide
receptor antagonist enhances pet imaging and radionuclide therapy of prostate cancer in preclinical studies.
Theranostics 2016, 6, 104–117. [CrossRef] [PubMed]
33. Maina, T.; Kaloudi, A.; Valverde, I.E.; Mindt, T.L.; Nock, B.A. Amide-to-triazole switch vs. In vivo
NEP-inhibition approaches to promote radiopeptide targeting of GRPR-positive tumors. Nucl. Med. Biol.
2017, 52, 57–62. [CrossRef]
34. Lymperis, E.; Kaloudi, A.; Sallegger, W.; Bakker, I.L.; Krenning, E.P.; de Jong, M.; Maina, T.; Nock, B.A.
Radiometal-dependent biological profile of the radiolabeled gastrin-releasing peptide receptor antagonist
SB3 in cancer theranostics: Metabolic and biodistribution patterns defined by neprilysin. Bioconjug. Chem.
2018, 29, 1774–1784. [CrossRef]
35. Nock, B.; Maina, T. Tetraamine-coupled peptides and resulting 99m Tc-radioligands: An effective route
for receptor-targeted diagnostic imaging of human tumors. Curr. Top. Med. Chem. 2012, 12, 2655–2667.
[CrossRef] [PubMed]
36. Mather, S.J.; Nock, B.A.; Maina, T.; Gibson, V.; Ellison, D.; Murray, I.; Sobnack, R.; Colebrook, S.; Wan, S.;
Halberrt, G.; et al. GRP receptor imaging of prostate cancer using [99m Tc]Demobesin 4: A first-in-man study.
Mol. Imaging Biol. 2014, 16, 888–895. [CrossRef] [PubMed]
37. De Castiglione, R.; Gozzini, L. Bombesin receptor antagonists. Crit. Rev. Oncol. Hematol. 1996, 24, 117–151.
[CrossRef]
38. Maina, T.; Nock, B.A.; Kulkarni, H.; Singh, A.; Baum, R.P. Theranostic prospects of gastrin-releasing peptide
receptor-radioantagonists in oncology. PET Clin. 2017, 12, 297–309. [CrossRef] [PubMed]
39. Kassis, A.I. Therapeutic radionuclides: Biophysical and radiobiologic principles. Semin. Nucl. Med. 2008,
38, 358–366. [CrossRef]
40. Wild, D.; Frischknecht, M.; Zhang, H.; Morgenstern, A.; Bruchertseifer, F.; Boisclair, J.; Provencher-Bolliger, A.;
Reubi, J.C.; Maecke, H.R. Alpha- versus beta-particle radiopeptide therapy in a human prostate cancer model
(213 Bi-DOTA-Pesin and 213 Bi-AMBA versus 177 Lu-DOTA-Pesin). Cancer Res. 2011, 71, 1009–1018. [CrossRef]
41. Linder, K.E.; Metcalfe, E.; Arunachalam, T.; Chen, J.; Eaton, S.M.; Feng, W.; Fan, H.; Raju, N.; Cagnolini, A.;
Lantry, L.E.; et al. In vitro and in vivo metabolism of Lu-AMBA, a GRP-receptor binding compound, and the
synthesis and characterization of its metabolites. Bioconjug. Chem. 2009, 20, 1171–1178. [CrossRef] [PubMed]
42. Shipp, M.A.; Tarr, G.E.; Chen, C.Y.; Switzer, S.N.; Hersh, L.B.; Stein, H.; Sunday, M.E.; Reinherz, E.L.
CD10/neutral endopeptidase 24.11 hydrolyzes bombesin-like peptides and regulates the growth of small
cell carcinomas of the lung. Proc. Natl. Acad. Sci. USA 1991, 88, 10662–10666. [CrossRef] [PubMed]
43. Reile, H.; Armatis, P.E.; Schally, A.V. Characterization of high-affinity receptors for bombesin/gastrin
releasing peptide on the human prostate cancer cell lines PC-3 and DU-145: Internalization of receptor
bound 125 I-(Tyr4 )bombesin by tumor cells. Prostate 1994, 25, 29–38. [CrossRef] [PubMed]
pharmaceuticals
Article
Microencapsulation of Salmonella-Specific
Bacteriophage Felix O1 Using Spray-Drying in a
pH-Responsive Formulation and Direct Compression
Tableting of Powders into a Solid Oral Dosage Form
Gurinder K. Vinner 1 , Zahra Rezaie-Yazdi 1 , Miika Leppanen 2 , Andrew G.F. Stapley 1 ,
Mark C. Leaper 1 and Danish J. Malik 1, *
1 Department of Chemical Engineering, Loughborough University, Loughborough LE11 3TU, UK;
G.Vinner@lboro.ac.uk (G.K.V.); Z.Rezaie-Yazdi@lboro.ac.uk (Z.R.-Y.); a.g.f.stapley@lboro.ac.uk (A.G.F.S.);
M.C.Leaper@lboro.ac.uk (M.C.L.)
2 Department of Physics, Department of Biological and Environmental Science, Nanoscience Center,
University of Jyväskylä, 40014 Jyväskylä, Finland; miika.j.leppanen@jyu.fi
* Correspondence: d.j.malik@lboro.ac.uk; Tel.: +44-(0)1509-222507

Abstract: The treatment of enteric bacterial infections using oral bacteriophage therapy can be
challenging since the harsh acidic stomach environment renders phages inactive during transit
through the gastrointestinal tract. Solid oral dosage forms allowing site-specific gastrointestinal
delivery of high doses of phages, e.g., using a pH or enzymatic trigger, would be a game changer for
the nascent industry trying to demonstrate the efficacy of phages, including engineered phages for gut
microbiome modulation in expensive clinical trials. Spray-drying is a scalable, low-cost process for
producing pharmaceutical agents in dry powder form. Encapsulation of a model Salmonella-specific
phage (Myoviridae phage Felix O1) was carried out using the process of spray-drying, employing
a commercially available Eudragit S100® pH-responsive anionic copolymer composed of methyl
methacrylate-co-methacrylic acid formulated with trehalose. Formulation and processing conditions
were optimised to improve the survival of phages during spray-drying, and their subsequent
protection upon exposure to simulated gastric acidity was demonstrated. Addition of trehalose to the
formulation was shown to protect phages from elevated temperatures and desiccation encountered
during spray-drying. Direct compression of spray-dried encapsulated phages into tablets was shown
to significantly improve phage protection upon exposure to simulated gastric fluid. The results
reported here demonstrate the significant potential of spray-dried pH-responsive formulations for
oral delivery of bacteriophages targeting gastrointestinal applications.
Keywords: antibiotic resistance; bacteriophages; direct compression; microparticles; pH-responsive;

spray drying; salmonella; tablets
1. Introduction
The emergence of antibiotic resistance in pathogenic bacteria is a serious global health threat.
Common enteric bacterial pathogens are becoming progressively resistant to frontline antibiotics.
The pipeline for the development of new classes of broad-spectrum antibiotics is not looking
promising [1]. In addition to treating gastrointestinal infections in humans, a safe and low-cost
strategy to reduce pathogen carriage in livestock and poultry is also needed. National health agencies
are increasingly banning general antibiotic use in animals grown for human consumption, e.g., see
the European Union (EU) directive on additives for use in animal nutrition [2]. There is an increasing

awareness of the need to move away from broad-spectrum antibiotics and use more specific treatments
which do not cause dysbiosis of the microbiome [3].
Lytic bacteriophages (phages) are viruses that infect and kill bacteria including antibiotic-resistant
ones in a highly species-specific manner. Commonly occurring gastrointestinal infections are caused
by several types of bacteria including Clostridium difficile, Escherichia coli, Salmonella spp., and Vibrio
cholera [4]. It is estimated that Salmonella alone accounts for 1.2 million foodborne illnesses in the United
States, with 23,000 hospitalisations and 450 deaths costing an estimated 365 million dollars in medical
costs each year [5]. Increasing centralisation and industrialisation of food supply increases the risk of
distribution of these hardy organisms. Antimicrobial resistance to “first-line” drugs is increasingly
common among Salmonella worldwide [6–9]. In animals, decolonisation of the gastrointestinal tract
from Salmonella may be beneficial for biocontrol to reduce dissemination of harmful bacteria through
the food chain, e.g., lairage-associated Salmonella transmission in pigs [10]. Phages incorporated in
solid oral dosage forms may be mixed in with animal feed for prophylactic or therapeutic applications.
The use of phage therapy is a particularly promising alternative to using broad-spectrum
antibiotics for acute enteric infections, because typically in such cases intestinal concentration
of the infecting bacteria is high and the causative agent and strain may be suitably diagnosed
using rapid diagnostic tools. Solid oral dosage forms that are capable of reliably delivering a
therapeutically high phage dose to the site of infection is a major barrier for the treatment of
gastrointestinal infections [11–14]. Phages are biological entities requiring protection from stresses
typically encountered during manufacturing and storage [15]. The harsh conditions encountered in the
human stomach (pH ~1), as well as exposure to bile and digestive enzymes in the gastrointestinal tract,
render phages inactive [12,13,16–20]. Previous studies on phage encapsulation focused on extrusion
methods, e.g., to produce hydrogel microparticles with variable levels of acid protection [16,17,21].
Recently, more sophisticated microfluidic methods were used, giving precise control over the
fabrication of microcapsules. Such methods, however, are difficult to scale-up, costly, and better
suited for high-value products [12,13]. There is a need for scalable low-cost methods for producing
stable oral dosage forms for delivering bacteriophages to the gastrointestinal tract.
Spray-drying is an industrially acceptable process used to manufacture dry powder forms carrying
bioactive agents such as proteins, peptides, attenuated antibodies, and phages [22]. Spray-drying
was previously used for producing phage-containing powders in sugar formulations suitable for
pulmonary delivery [23–25]. Published studies employing spray-drying to produce phage-containing
powders with pH-responsive characteristics are relatively rare; however, a previous study employing
spray-dried E. coli phages using pure Eudragit S100® showed a ~1 log loss in phage titre in the final
spray-dried powder [26]. The E. coli phage titre was shown to fall significantly (between 2–3 log
reduction) during storage over a period of one year at 20 ◦ C for different phages evaluated in the
study [26]. However, the focus of previous research was not on optimising the spray-drying conditions,
and evaluation of the effects of excipients on phage viability and storage stability were not adequately
addressed. Research is, therefore, needed to evaluate suitable formulations and optimum spray-drying
conditions to produce acid stable solid dosage forms for enteric delivery, and these are addressed
in this study. Dry powder forms are favoured due to their ease of handling and long-term storage
stability, e.g., at ambient temperatures, thereby avoiding the need for a cold supply chain for storage.
Encapsulating phages in a stable dry powder form opens possibilities for their use in oral solid dosage
forms, e.g., using direct compression to produce tablets for enteric delivery [27].
Commercial spray-dryers typically operate at high temperatures with powders exposed to
temperatures around or exceeding 100 ◦ C [28]. This ensures low residual moisture content in the
dried product which impacts positively on powder handling and storage stability. Exposing phages
to elevated temperatures can be detrimental to phage viability, resulting in the loss of phage titre in
the final dried powder. Phages were previously spray-dried at low outlet temperatures using small
laboratory-scale spray-dryers with outlet temperatures typically between 40 and 60 ◦ C and using
excipients such as trehalose to provide protection from thermal stresses [23,25]. Good excipients
such as trehalose prevent or minimise the effect of thermal stress on the active agents and act as
water-replacing agents. Phage proteins may undergo irreversible damage due to the removal of water,
which is essential for the maintenance of the hydrogen bonds necessary to stabilise their secondary
structure [24]. Trehalose replaces these hydrogen bonds and vitrifies the phages, yielding dry powders
with high glass transition temperatures [29].
The aim of the present study was to investigate the effect of spray-drying temperatures and
formulation parameters, varying the amount of trehalose and the pH-responsive polymer Eudragit
S100® to produce stable dry powders having a high amount of encapsulated phage and good storage
stability. The spray-dried powders were tableted using a direct compression process. Powders
and tablets containing phages were exposed to simulated gastric fluid to evaluate acid protection.
The work reported here allows evaluation of the suitability of these low-cost scalable methods for
phage encapsulation in oral solid dosage forms. This study addresses the urgent unmet need for the
development of scalable delivery methods to facilitate translation of phages from bench-to-bedside in
order to demonstrate the effectiveness of phage therapy in both humans and animals.
2.1. Model Bacterium and Phage

Salmonella and phage Felix O1 were used as the model bacterium and phage for this study.
Salmonella enterica ATCC19585 was purchased from LGC standards, EU. Phage Felix O1 was kindly
donated by Dr Cath Rees, University of Nottingham, UK [30]. An S. enterica strain was used to
propagate and enumerate Felix O1.
Bacterial growth and phage propagation were carried out using previously published
methods [13]. Briefly, a log-phase culture of Salmonella at an optical density (OD) of 0.2 (this typically
equates to a viable cell count of ~108 colony-forming units (CFU)/mL) was inoculated with Felix
O1 at a multiplicity of infection (MOI) of 0.01. The lysate was centrifuged at 2000× g and filtered
using a 0.2-µm pore size in-line syringe filter (Millipore, Watford, UK) and stored at 4 ◦ C until further
use. Plaque assays were used to enumerate phage concentration employing the double overlay agar
method; serial dilutions of phages were spotted on a bacterial lawn overlay. All measurements were
performed in triplicate. The plaque-forming units (PFU) were counted after incubation for 24 h at
37◦ C.
2.2. Spray-Drying Conditions and Formulations

Eudragit S100 was kindly supplied by Evonik Germany. D-(+)-Trehalose dihydrate was purchased
from Fisher Scientific (Loughborough, UK). Solutions containing different excipient (Eudragit S100
or trehalose) amounts were dissolved in 500 mL of deionised distilled water (dH2 O). For ease of
presentation, the following nomenclature is used in the manuscript: the amount of polymer (Eudragit,
denoted as “P”) to sugar (trehalose, denoted as “S”) corresponds to the dissolved percentage (w/v) of
polymer and sugar in the solutions used for spray-drying, e.g., PS21 refers to 2% (w/v) polymer and
1% (w/v) sugar in the solution. The following formulations with different proportions of trehalose and
polymer were evaluated in the study: PS04 (4% w/v trehalose), PS30 (3% w/v polymer), PS32 (3% w/v
polymer, 2% w/v trehalose), PS21 (2% w/v polymer, 1% w/v trehalose), and PS24 (2% w/v polymer,
4% w/v trehalose).
In order to dissolve Eudragit, the pH of the water was changed to alkaline (pH 12) via addition
of 4 M NaOH (Fisher Scientific, UK) to allow polymer dissolution, followed by pH adjustment to
pH 7 using 0.1 M HCl prior to addition of trehalose powder, its dissolution, and then addition
of bacteriophages to the solution. For each formulation, typically 1% (v/v) high-titre phage Felix
O1 (~1011 PFU/mL) was added to the solution, yielding phage titres of ~109 PFU/mL in the final
formulations. The phage-containing solutions were spray-dried using a commercially available
Labplant spray-dryer SD-06 (Labplant, UK Limited), which is a co-current dryer with a pneumatic
atomiser and a cylindrical drying chamber of dimensions 215 mm outer diameter and 420 mm height.
The air exit stream was passed through a high-efficiency particulate air (HEPA) filter prior to discharge.
The diameter of the atomization nozzle used throughout the work was 0.5 mm with the measured feed
liquid flow rate at 280 mL·h−1 and a drying gas air flow rate of ~20 L·s−1 . The air inlet temperatures
were set at 100 ◦ C, 120 ◦ C, 150 ◦ C, and 180 ◦ C resulting in corresponding air outlet temperatures of
56 ± 2 ◦ C, 66 ± 2 ◦ C, 82 ± 2 ◦ C, and 96 ± 2 ◦ C, respectively. The outlet temperature is only indicative
of the highest temperature the phages could be exposed to as dry powders in the collection bottle;
temperature in the collection bottle varied between 40 and 60 ◦ C.
2.3. Powder Storage

Following collection, the phage-containing spray-dried powders were stored in sealed screw top
bottles. These were stored at either 4 ◦ C or 23 ◦ C for a period of up to three months. At specific time
intervals, 0.1 g of powder was removed and dissolved in simulated intestinal fluid (SIF) (10 mg/mL
of pancreatin in 0.5 mM KH2 PO4 , pH 7) and the phage titre was enumerated using a plaque assay
(described above). Similarly, experiments involving powder or tablet exposure to simulated gastric
fluid (SGF) (composition 3.2 mg/mL of pepsin in 0.2 M NaCl, pH adjusted to pH 2 using 5 M HCl)
involved weighing a known amount of powder or tablet (typically 0.3 g) and these were subsequently
exposed to 10 mL of SGF for 2 h at 37 ◦ C. Following SGF exposure, the powder or tablet samples
were centrifuged at 2000× g to pellet the suspended solids. Supernatant SGF was removed using a
pipette. Then, 10 mL of SIF was added to the pelleted solid material, which was gently vortexed to
resuspend the solids, before being left in an orbital incubator shaker (Certomat, Sartorius, UK) at 37 ◦ C
and 120 rpm. Complete dissolution of the powders (SIF exposure for 3 h at 37 ◦ C) and tablets (SIF
exposure for 5 h at 37 ◦ C) was achieved. Thereafter, 10 µL of the supernatant was removed using a
micropipette; then, the sample was serially diluted and plaque assays were performed to measure
phage release.
2.4. Direct Compression Tableting of Spray-Dried Powders

Spray-dried powders at an inlet temperature of 150 ◦ C were used for tableting using a Riva
Minipress MII (UK) tabletting machine. Approximately 0.3 g of powder was loaded into the punch
hole and compressed at a force of 5 kN. The produced tablets were weighed before being stored at
4 ◦ C in sealed tubes for further analysis.
2.5. Ion Microscopy

To analyse the morphology of spray-dried powders, a representative sample from formulation
PS21 spray-dried at an inlet temperature 150 ◦ C was examined with ion microscopy. Dried powder
was applied on carbon tape attached to a sample stub, and any excess was blown away. A Zeiss Orion
NanoFab (University of Jyväskylä) with Ne+ beam and acceleration voltage of 10 kV was used to cut
the microparticle in half. Milling was done using a 45◦ tilted angle by setting the reduced raster scan
rectangle over the area to be removed and scanned until the material disappeared. Ion current values
used were typically ~20 pA, resulting in a total processing time of about 1 h. Following sample cutting
by milling, the sample stage was rotated through 180◦ and the cross-section was imaged with He+ .
An acceleration voltage of 33 kV, a current of 0.20 pA, 32 line averages, and 1 µs of dwell time were
used for He+ imaging. Flood-gun charge compensation was used during both milling and imaging.
2.6. Moisture Content

The average moisture content (wet basis) of the spray-dried powder was measured gravimetrically.
A known mass of sample (approximately 0.5 g) was placed in a dry ceramic crucible and dried in a
vacuum oven at 120 ◦ C for a period of 24 h. Measurements were done in triplicate. The sample was
then removed, cooled over silica gel in a sealed desiccator, and immediately weighed to limit water
absorption from the atmosphere. The initial and final masses were then used to calculate the wet basis
moisture content.
2.7. Particle Size Distribution

A Coulter LS 130 Particle Size Analyzer (Beckman Coulter Inc., High Wycombe, UK) was used to
determine the particle size distribution of the spray-dried powders. The device is laser-modulated
and uses the optical model Fraunhofer to detect laser diffraction caused by particles. A representative
sample from formulation PS21 was taken, loaded into a glass sample containing n-hexane (Sigma
Aldrich, Gillingham, U.K.) as a suspending medium. The cell was equipped with a stirrer to keep the
particles suspended in the solvent. The volume size and cumulative distribution were measured.
2.8. Differential Scanning Calorimetry (DSC) Analysis of Spray-Dried Trehalose Powders

Typically, ~15 mg of the spray-dried powder PS04 (stored upon collection in a dry atmosphere
over silica gel in a sealed desiccator) was placed in a pre-weighed DSC aluminium pan. The pan
was then hermetically sealed and weighed to 0.1-mg accuracy. The sample was placed in a Q10 DSC
(TA Instruments, Crawley, Sussex, UK) and scanned from 25 to 120 ◦ C at a programmed heating rate
of 10 ◦ C/min. An empty pan was used as a reference. All DSC measurements were carried out in
duplicate. The DSC instrument was calibrated for enthalpy and temperature using an indium standard
at the same scan rate. The glass transition temperature was computed from each thermal curve.
2.9. Statistical Analysis

Statistical analysis was carried out using Minitab version 18 (USA). Two-sample t-tests were
performed (n = 3) with reporting of p < 0.05 as statistically significant. Where multiple tests were done,
the value of alpha was adjusted using the Bonferroni correction. Error bars represent a single standard
deviation for the mean values of the replicates.
3. Results
3.1. Optimisation of Spray-Drying Formulation of Felix O1 Phage

Felix O1 at a phage titre of 5 × 109 PFU/mL was spray-dried individually in trehalose (PS04) and
Eudragit alone (PS30). The moisture content of PS04 spray-dried powders was typically in the range
of ~2–10% (w/w) and less than 5% (w/w) for inlet drying temperatures 150 ◦ C and above (Figure S1,
Supplementary Materials). The moisture content of PS30 spray-dried powders was considerably
higher in the range of ~15–25% (w/w). The effect of inlet air temperature on phage survival following
the spray-drying process was evaluated. The spray-dried phage-containing powders were exposed to
SIF for a period of 3 h or until complete dissolution of powder, and the resulting phage concentration
was measured (Figure 1). Trehalose was found to be an excellent excipient for the protection of Felix O1
phages exposed to outlet temperatures which were varied between 56 and 96 ◦ C. Phage titre of the feed
solution used for spray-drying was 5 × 109 PFU/mL, equating to a theoretical phage concentration
of ~1.7 × 109 PFU/g using the assumption that all the phage virions present in the original solution
remained viable in the final collected powder, accounting for the amount of dissolved solids in the
feed solution. There was no measurable loss in phage titre in the PS04 powders for the entire range of
drying temperatures evaluated (p > 0.05), and the mean phage titre in the powders was the same as
the theoretical yield of ~1 × 109 PFU/g. Felix O1 phages spray-dried in pure Eudragit S100 (PS30)
resulted in a significant loss in phage titre; typically, a ~4 log reduction in phage titres was observed at
all temperatures post spray-drying in PS30. A two-sample t-test of means indicated that the phage titre
in the powders at 180 ◦ C was lower in comparison with the phage yield in PS30 powders spray-dried
at lower temperatures. For example, at 100 ◦ C, the 95% confidence interval for the mean was between
1.8 × 105 and 3.6 × 105 PFU/g, compared with between 4.6 × 104 and 2.6 × 105 PFU/g at 180 ◦ C
(p < 0.05). There was no difference in phage titres for PS30 powders spray-dried at 120 ◦ C and 150 ◦ C
Pharmaceuticals 2019, 12,◦x FOR PEER REVIEW 6 of 14
compared with 100 C (p > 0.05).
Figure 1. Phage Felix O1 spray-dried at varying inlet air drying temperatures in formulations PS04 and
PS30
Figure(see1.text for description)
Phage followed by
Felix O1 spray-dried completeinlet
at varying release
air of phages
drying in simulatedinintestinal
temperatures fluid (SIF)
formulations PS04
(titre measured after 3 h of exposure to SIF). The phage titre of feed solution used for spray-drying
and PS30 (see text for description) followed by complete release of phages in simulated intestinal fluid was
9
× 10(titre 9
5(SIF) PFU/mL,
measuredequating
after to
3h a theoretical
of exposure final phageThe
to SIF). concentration
phage titre of feed×solution
of~1.7 10 PFU/g. used* for
Indicates
spray-
significant
drying wasdifference in means
5 × 109 PFU/mL, for samples
equating at a givenfinal
to a theoretical temperature compared with
phage concentration the×spray-dried
of ~1.7 109 PFU/g. *
sample
Indicates 100 ◦ C for the
at significant same formulation
difference in means for(psamples
< 0.05) at
using a two-sample
a given temperaturet-test. Error bars
compared with represent
the spray-
one standard deviation; all measurements were done in triplicate (n = 3).
dried sample at 100 °C for the same formulation (p < 0.05) using a two-sample t-test. Error bars
represent one standard deviation; all measurements were done in triplicate (n = 3).
The effect of combining trehalose (thermal protection and desiccation resistance) and Eudragit
S100®The
(pHeffect
resistance and pH trigger
of combining trehalosefor(thermal
release) on phage survival
protection followingresistance)
and desiccation spray-drying 150 ◦ C
andatEudragit
was
S100examined using formulations
® (pH resistance and pH triggerwithforvarying proportions
release) on phage of trehalose
survival (Figure 2).
following In comparison
spray-drying with
at 150 °C
phage survival in formulation PS04 (100% trehalose, 2.4 × 10 9 PFU/g), phage titres remained high at
was examined using formulations with varying proportions of trehalose (Figure 2). In comparison
2.5 × 10 9 PFU/g (PS24, 67% trehalose), 1.8 × 109 PFU/g (PS32, 40% trehalose), and 1.1 × 109 PFU/g
with phage survival in formulation PS04 (100% trehalose, 2.4 × 109 PFU/g), phage titres remained
(PS21,
high at 2.5 × 109 PFU/gA(PS24,
33% trehalose). two-sample t-test comparing
67% trehalose), 1.8 × 10the means
9 PFU/g for samples
(PS32, PS24 and
40% trehalose), andPS32
1.1 with
× 109
sample PS04 showed no statistical difference in means. However, there was
PFU/g (PS21, 33% trehalose). A two-sample t-test comparing the means for samples PS24 and PS32a difference for PS21 (95%
confidence interval for difference in means was 5 × 8 –2.2 × 109 PFU/g higher for PS04 compared
with sample PS04 showed no statistical difference in10
means. However, there was a difference for PS21
with PS21), i.e., similar in magnitude to the titre of PS21.
(95% confidence interval for difference in means was 5 × 10 –2.2 This
8 suggests that the
× 109 PFU/g proportion
higher of compared
for PS04 trehalose
in the polymer formulations does affect phage survival during spray-drying with higher
with PS21), i.e., similar in magnitude to the titre of PS21. This suggests that the proportion of trehalose trehalose
proportions
in the polymerin the powders PS24
formulations doesand PS32phage
affect yielding higherduring
survival phage spray-drying
titres. with higher trehalose
proportions in the powders PS24 and PS32 yielding higher phage titres. to SGF for 2 h followed by
Spray-dried formulations PS24, PS32, and PS21 were all exposed
quantification of the remaining viable phage after dissolution of the polymer in SIF. Felix O1 phages
encapsulated in formulations PS24 and PS32 were not sufficiently protected from acid exposure
at pH 2. Phage titre in the powders fell by ~2 log from ~109 PFU/g to ~107 PFU/g for PS24 and
PS32. The sample with a higher proportion of polymer content (PS21) showed a considerably smaller
reduction in phage titre upon acid exposure from 1.1 × 109 PFU/g to 2.2 × 108 PFU/g, i.e., around a
1 log reduction. The higher loss of phage viability in sample PS24 and PS32 compared with PS21 may
be attributed to the higher proportion of trehalose (67% trehalose in PS24 and 40% trehalose in PS32
compared with 33% in PS21) dispersed in the microparticle shell material; sugar may readily dissolve
Pharmaceuticals
upon exposure 2019,
to12, x FOR
SGF, PEER REVIEW
exposing phages
to the acid environment. 7 of 14
Figure 2. Concentration of encapsulated phage Felix O1 released from spray-dried powders after
Figure
complete 2. Concentration
dissolution in ofSIFencapsulated
(~3 h). Samples phage Felix O1
labelled SIFreleased from spray-dried
were exposed to SIF onlypowders
without after
acid
complete dissolution in SIF (~3 h). Samples labelled SIF were exposed to SIF only
exposure, whereas samples labelled simulated gastric fluid (SGF)/SIF were first exposed to simulated without acid
exposure,
gastric fluidwhereas
(pH 2)samples
for 2 h, labelled simulated
subsequently gastric fluid
centrifuged (SGF)/SIF
to remove the were first exposed and
acid supernatant, to simulated
then SIF
gastric fluidadded
(pH 7) was (pH 2)tofor
the2sample
h, subsequently
to dissolvecentrifuged
the polymer. to All
remove the acid were
formulations supernatant, and at
spray-dried then ◦C
150SIF
(pH
inlet 7) was addedcorresponding
temperature to the sample to 82 ◦
dissolve the temperature.
C outlet polymer. All *formulations were spray-dried
Indicates significant difference at
in 150 °C
means
inlet temperature
(p < 0.05) corresponding
using a two-sample to 82
t-test; °Cmeans
n.d. outlet notemperature. * Indicates
difference (p significant
> 0.05). Error differenceone
bars represent in
means
standard (p deviation;
< 0.05) using a two-sample t-test;
all measurements n.d. means
were done no difference
in triplicate (n = 3). (p > 0.05). Error bars represent
one standard deviation; all measurements were done in triplicate (n = 3).
3.2. Powder Characterisation
Spray-dried
Helium ion formulations
microscopy (HIM)PS24, PS32, and PS21
imaging of thewere all exposed
spray-dried to SGF (PS21)
powders for 2 hrevealed
followed the
by
quantification
morphology ofofthe the microparticles
remaining viable phage3).
(Figure after dissolutionmicroparticles
Spray-dried of the polymerwerein SIF.
alsoFelix O1 phages
examined for
encapsulated in formulations
their size distribution PS24
(see Figure S2,and PS32 were not
Supplementary sufficiently
Materials). Theprotected from
sizes of the PS21acid exposure at
microparticles
pH 2. Phage at
spray-dried titre
anin thetemperature
inlet powders fellof by150
~2 log from in
◦ C were ~10the
9 PFU/g to ~107 PFU/g for PS24 and PS32. The
range of 1–10 µm with a d50 value of 6 µm
sample
and a dwith
90
a
value higher
of 11 proportion
µm from of
the polymer
cumulative content (PS21)
particle size showed a considerably
distribution (Figure S2,smaller reduction
Supplementary
in phage titre upon acid exposure from 1.1 × 10 9 PFU/g to 2.2 × 108 PFU/g, i.e., around a 1 log
Materials). The results were consistent with sizes of particles observed using HIM. The microparticles
reduction.
were spheres The orhigher
flattenedlossspheres
of phage
withviability
a smooth,in sample PS24non-porous
defect-free, and PS32 compared with 3b).
surface (Figure PS21After
maythe
be
attributed to thethe
milling process, higher
core ofproportion of trehalose
the microparticles was(67%
foundtrehalose in PS24
to be hollow, and 40%
indicating thattrehalose in PS32
the phages were
compared with 33% in PS21) dispersed in the microparticle shell material; sugar may
encapsulated in the thin shell structure (Figure 3c). The red circle highlights the capsid heads of two readily dissolve
upon
phageexposure to SGF, exposing
virions protruding phages
out of the shellto the acid
matrix. environment.
Small “bumps” (indicated by the red arrow) on the
surface of the shell may potentially be phage virions entrapped inside the shell structure (Figure 3d).
3.2.
ThesePowder
bumpsCharacterisation
were not visible in control particles not containing phages (Figure 3e). The thickness of
the microparticle
Helium ion shell was estimated
microscopy (HIM)toimaging
be betweenof 200
the and 300 nm, which
spray-dried is similar
powders (PS21)in magnitude
revealed theto
phage dimensions
morphology of the (length ~200 nm).
microparticles The images
(Figure showed microparticles
3). Spray-dried no merging of microparticles or anyfor
were also examined surface
their
perforations (absence of blow holes).
size distribution (see Figure S2, Supplementary Materials). The sizes of the PS21 microparticles spray-
dried at an inlet temperature of 150 °C were in the range of 1–10 µm with a d50 value of 6 µm and a
d90 value of 11 µm from the cumulative particle size distribution (Figure S2, Supplementary
Materials). The results were consistent with sizes of particles observed using HIM. The microparticles
were spheres or flattened spheres with a smooth, defect-free, non-porous surface (Figure 3b). After
the milling process, the core of the microparticles was found to be hollow, indicating that the phages
were encapsulated in the thin shell structure (Figure 3c). The red circle highlights the capsid heads of
two phage virions protruding out of the shell matrix. Small “bumps” (indicated by the red arrow) on
the surface of the shell may potentially be phage virions entrapped inside the shell structure (Figure
3d). These bumps were not visible in control particles not containing phages (Figure 3e). The
thickness of the microparticle shell was estimated to be between 200 and 300 nm, which is similar in
magnitude to2019,
Pharmaceuticals phage
12, 43dimensions (length ~200 nm). The images showed no merging of microparticles
8 of 14
or any surface perforations (absence of blow holes).
Figure 3.3. Helium

Helium ion
ion microscopy
microscopy (HIM) images of spray-dried PS21 microparticles (inlet drying
temperature 150 ◦°C).
C). (a) Spray-dried microparticles were typically
typically <10 µm in size and did not display
surface defects such as blow holes. (b) Some particles were spherical, whereas others were flattened
appearance. (c) A spherical particle about 10 µm in size was cut
spheres, and some had a lens-shaped appearance.
half using
in half usingaaneon
neonionion beam
beam and,
and, after
after ◦ rotation,
180180° rotation,
waswas imaged
imaged with with a helium
a helium ion beam.
ion beam. Two
Two phage
phage particles
virion virion particles can be
can be seen in seen in left-hand
the top the top left-hand corner
corner (red (red(d)
circle). circle). (d) Expanded
Expanded view
view of red ofarea
box red
box area
shown in shown in frame
frame (c). Bumps(c).
in Bumps in the
the inside wallinside
of thewall of the
sphere weresphere
foundwere found in microparticles
in microparticles containing
containing
phages phages
imaged using imaged using
a higher a higher magnification
magnification (red arrow). (e)(red arrow).
Inside wall of(e)a control
Inside wall of a control
microparticle not
microparticle
containing not containing phages.
phages.
3.3. Direct Compression

3.3. Direct Compression Tableting
Tableting of
of Spray-Dried
Spray-Dried Phage
Phage
Spray-dried
Spray-dried microparticles
microparticleswereweretableted
tabletedusing
usingthethe
process of direct
process compression.
of direct compression.The spray-dried
The spray-
powder
dried powder had suitable processing characteristics, i.e., flowability and hardness toundergo
had suitable processing characteristics, i.e., flowability and hardness to undergo direct
direct
compression to produce tablets. The tablets had the following dimensions: diameter
compression to produce tablets. The tablets had the following dimensions: diameter of 1 cm, of 1 cm, thickness
of 0.4 cm, of
thickness and0.4average
cm, and weight of 0.3
average g (Figure
weight of 0.3S3,
g Supplementary Materials). The
(Figure S3, Supplementary tablets used
Materials). in the
The tablets
study
used inwere all visually
the study were identical
all visually(any deformed
identical (anyordeformed
chipped tablets weretablets
or chipped discarded).
were discarded).
Powders containing different proportions of trehalose were
Powders containing different proportions of trehalose were evaluated in evaluated in terms
terms ofof acid
acid stability
stability
following formation of tablets. The results showed there to be a significant
following formation of tablets. The results showed there to be a significant difference between difference between
the
the three formulations (Figure 4). Post-spray-drying release in SIF showed
three formulations (Figure 4). Post-spray-drying release in SIF showed titres of more than 1 × 109 titres of more than
× 109in
1PFU/g PFU/g in the spray-dried
the spray-dried powderspowders for all
for all three three formulations,
formulations, which
which was was similar
similar in magnitude
in magnitude to the
to the maximum theoretical phage titre yield (~1.7 × 10 9 PFU/g) based on the phages present in the
maximum theoretical phage titre yield (~1.7 × 109 PFU/g) based on the phages present in the original
original spray-drying
spray-drying solution surviving
solution surviving the spray-drying
the spray-drying process. The process.
phageThedosephage
loaded dose loaded
in each in each
tablet was
tablet was approximately 6 × 10 8 PFU per tablet, and no adverse effect of the compression force on
approximately 6 × 108 PFU per tablet, and no adverse effect of the compression force on subsequent
subsequent phage
phage viability wasviability was observed
observed (Figure 4).(Figure 4). A significant
A significant loss of loss of phage
phage titre was
titre was observed
observed for for
all
all formulations in powder form after exposure to SGF (pH 2) for 2 h. For PS21, phage titre fell from
1.7 × 109 PFU/g to 1.6 × 108 PFU/g, i.e., ~1 log, whereas, for PS24 and PS32, the phage titre fell from
~109 PFU/g to ~107 PFU/g, i.e., ~2 log. The effect of tableting on acid protection resulted in marked
improvement in acid stability for all three formulations. PS21 showed no observable loss in phage titre
following 2 h of exposure to SGF (Figure 4). PS24 and PS32 showed only a ~1 log reduction compared
formulations in powder form after exposure to SGF (pH 2) for 2 h. For PS21, phage titre fell from 1.7
× 109 PFU/g to 1.6 × 108 PFU/g, i.e., ~1 log, whereas, for PS24 and PS32, the phage titre fell from ~109
PFU/g to ~107 PFU/g, i.e., ~2 log. The effect of tableting on acid protection resulted in marked
improvement in acid stability for all three formulations. PS21 showed no observable loss in phage
titre following 2 h of exposure to SGF (Figure 4). PS24 and PS32 showed only a ~1 log reduction
compared with samples not exposed to SGF, along with a viable phage concentration in the tablets
of 1 × 108 PFU/g; this was ~1 log greater than that compared with PS24 and PS32 powders following
with samples
exposure to SGFnot (~10
exposed to SGF, along with a viable phage concentration in the tablets of 1 × 108
7 PFU/g).
PFU/g; this was ~1 log greater than that compared with PS24 and PS32 powders following exposure
to SGF (~107 PFU/g).
Figure 4. Encapsulated Felix O1 phages released in SIF from spray-dried powders and corresponding
tablets
Figure using three different
4. Encapsulated Felix formulations.
O1 phages releasedSamples labelled
in SIF SIF were exposed
from spray-dried powderstoandSIFcorresponding
only without
acid exposure,
tablets whereas
using three samples
different labelled SGF/SIF
formulations. Sampleswere exposed
labelled first exposed
SIF were to simulated gastric
to SIF fluid (pH
only without 2)
acid
for 2 h andwhereas
exposure, were then centrifuged,
samples labelledbefore the supernatant
SGF/SIF were exposed was withdrawn
first to simulatedandgastric
SIF was added
fluid (pHto2)the
for
sample.
2 h andPhage titrescentrifuged,
were then were measured after
before 3 hsupernatant
the of exposurewas to SIF for powders
withdrawn andandSIF 5was
h ofadded
exposure to
to the
SIF for tablets to ensure complete dissolution of tablets. * Indicates significant difference
sample. Phage titres were measured after 3 h of exposure to SIF for powders and 5 h of exposure to in means
(p < for
SIF 0.05) usingtoaensure
tablets two-sample t-test;
complete n.d. means
dissolution no difference
of tablets. (p >significant
* Indicates 0.05). Error bars represent
difference in meansone(p
standard deviation;
< 0.05) using all measurements
a two-sample were done
t-test; n.d. means in triplicate
no difference (p >(n = 3).Error bars represent one standard
0.05).
deviation; all measurements were done in triplicate (n = 3).
3.4. Storage Stability of Spray-Dried Phages in Powders
The most
3.4. Storage promising
Stability formulation
of Spray-Dried in in
Phages terms of acid stability (PS21) was compared with PS04 in
Powders
terms of storage stability. Spray-dried powders had different initial moisture contents with PS04 3%
The most promising formulation in terms of acid stability (PS21) was compared with PS04 in
(w/w) and PS21 9% (w/w) (Figure S1, Supplementary Materials). The powders were stored at two
terms of storage stability. ◦Spray-dried◦ powders had different initial moisture contents with PS04 3%
different temperatures (4 C and 23 C) for a period of three months. A significant loss in phage
(w/w) and PS21 9% (w/w) (Figure S1, Supplementary Materials). The powders were stored at two
titre was observed after three months of storage for both PS04 (titre falling from 2.4 × 109 PFU/g
different temperatures (4 °C and 23 °C) for a period of three months. A significant loss in phage titre
to 1.8 × 108 PFU/g) and PS21 (titre falling from 2.7 × 109 PFU/g to 2.4 × 108 PFU/g) at 23 ◦ C in
was observed after three months of storage for both PS04 (titre falling from 2.4 × 109 PFU/g to 1.8 ×
comparison with the phage titre immediately after spray-drying (Figure 5). There was no statistical
108 PFU/g) and PS21 (titre falling from 2.7 × 109 PFU/g to 2.4 × 108 PFU/g) at 23 °C in comparison with
difference in phage titres for samples stored for one month at 23 ◦ C for both PS04 and PS21 compared
the phage titre immediately after spray-drying (Figure 5). There was no statistical difference in phage
with titres immediately after production (0 months). There was no statistical difference in sample
titres for samples stored for one month at 23 °C for both PS04 and PS21 compared with titres
means for Felix O1 formulated in PS04 and stored at 4 ◦ C for a period of three months (starting
immediately after production (0 months). There was no statistical difference in sample means for
titre 2.4 × 109 PFU/g, and 1.9 × 109 PFU/g after three months). A slight decrease in phage titre
was recorded after three months of storage at 4 ◦ C for PS21 (titre falling from 2.7 × 109 PFU/g to
1.1 × 109 PFU/g).
Pharmaceuticals 2019, 12, x FOR PEER REVIEW 10 of 14
Felix O1 formulated in PS04 and stored at 4 °C for a period of three months (starting titre 2.4 × 109
PFU/g, and 1.9 × 109 PFU/g after three months). A slight decrease in phage titre was recorded after
three months
Pharmaceuticals of storage
2019, 12, 43 at 4 °C for PS21 (titre falling from 2.7 × 109 PFU/g to 1.1 × 109 PFU/g). 10 of 14
Figure 5. Storage results for formulations PS04 and PS21 used to spray-dry phage Felix O1. Phage titre
Figure 5. Storage results for formulations PS04 and PS21 used to spray-dry phage Felix O1. Phage
was measured after one month and three months of storage at 4 ◦ C and 23 ◦ C. * Indicates a significant
titre was measured after one month and three months of storage at 4 °C and 23 °C. * Indicates a
difference in means in comparison with the mean value immediately after spray-drying (0 months)
significant difference in means in comparison with the mean value immediately after spray-drying (0
using a two-sample t-test, (p < 0.05). Error bars represent one standard deviation; all measurements
months) using a two-sample t-test, (p < 0.05). Error bars represent one standard deviation; all
were done in triplicate (n = 3).
measurements were done in triplicate (n = 3).
4. Discussion
4. Discussion
Trehalose was previously shown to protect phages during the spray-drying process [24]. The role
Trehalose
of trehalose is was
bothpreviously shown
in stabilising to protect
phage proteinphages during the
conformation spray-drying
through hydrogen process
bonding[24]. and
The
role of trehalose
vitrification due is to both
a highin stabilising phage temperature
glass transition protein conformation through hydrogen
of the spray-dried powder bonding
[29]. High and
vitrification due to a high glass transition temperature of the spray-dried
residual moisture in samples spray-dried using relatively low drying temperatures results in lower powder [29]. High residual
moisture
glass in samples
transition spray-dried
temperatures using relatively of
and recrystallisation lowthedrying temperatures
trehalose, results in
which negatively lowerphage
affects glass
transition
stability temperatures
[24]. Spray-drying and Felix
recrystallisation of the trehalose,
O1 phage formulated which
in pure negatively
trehalose affects
resulted in phage stability
no significant
[24]. Spray-drying Felix O1 phage formulated in pure trehalose resulted in no
loss in phage titre at all spray-drying temperatures tested in the present study. The glass transition significant loss in phage
titre at all spray-drying temperatures tested in the present study. The glass
temperature of trehalose powders was previously correlated with residual moisture in the sample transition temperature of
trehalose
and powders
amorphous was previously
trehalose correlated
samples, with moisture with residual
content moisture
typically belowin the samplehaving
5% (w/w) and amorphous
Tg values
trehalose samples,
◦ with moisture content typically below 5% (w/w) having
greater than 50 C [31]. The glass transition temperatures for PS04 samples were typically above 60 C T g values greater than◦50
°C [31]. The glass

immediately transition
following temperatures
drying, due to lowfor PS04 samples
residual moisturewere typically
content and the above 60 °C immediately
amorphous nature of
following drying, due to low residual moisture content and the amorphous
spray-dried trehalose (Figure S4, Supplementary Materials). Phages encapsulated in a glassy matrix in nature of spray-dried
trehalose
samples (Figure
having S4, moisture
a low Supplementary
contentMaterials).
may result in Phages
betterencapsulated
storage stability in aatglassy
low and matrix
ambient in samples
storage
having a low moisture content may result in better storage stability at
temperatures (Figure 5). It was not possible to reliably measure the glass transition temperatures low and ambient storage
of
temperatures (Figure 5). It was not possible to reliably measure the glass
the composite trehalose–polymer PS21 microparticles. However, the moisture content of the samples transition temperatures of
the composite trehalose–polymer PS21 microparticles. However, the moisture
was found to be dependent on the spray drying temperatures (Figure S1, Supplementary Materials). content of the samples
was found to be
Spray-drying at adependent on the spray
drying temperature ofdrying
150 ◦ Ctemperatures
resulted in PS21 (Figure S1, Supplementary
powder having moisture Materials).
content
Spray-drying at a drying temperature of 150 °C resulted in PS21 powder
less than 10% (w/w) and high titres of viable phages in the dry powders. During the early stages having moisture content
of
less than
drying 10%the
where (w/w) and high
droplet titres
surface of viable
remains phageswith
saturated in the dry powders.
moisture Duringhumidity,
(100% relative the early RH),
stagestheof
drying where the droplet surface remains saturated with moisture (100%
droplet surface temperature is maintained at the wet bulb temperature, which is significantly lowerrelative humidity, RH), the
than the hot air temperature. As drying progresses, the droplet temperature begins to increase as water
diffusion to the droplet surface is not able to maintain 100% RH. As the air flow was co-current, the air
temperature dropped due to evaporative cooling. Once the moisture content of the particles drops, the
temperature of the particles begins to increase. However, the exposure period to high temperatures is
fairly short due to the short residence times of the particles in the dryer. Felix O1 phages spray-dried in
formulations containing ES100 and trehalose, e.g., PS21, remained viable at outlet drying temperatures
as high as 82 ◦ C. These results are particularly encouraging since industrial spray-dryers operate
at similar temperatures (80–100 ◦ C), which are markedly higher than those previously employed in
published studies using small laboratory-scale dryers [25,32]. However, individual phages may show
considerable differences in thermal stability and, therefore, spray-drying conditions may need to be
optimised appropriately [32].
Spray-drying Felix O1 in a formulation containing only Eudragit S100® (PS30) resulted in a
significant loss in phage activity in the final dried powders (Figure 1) compared with spray-drying
under identical conditions using pure trehalose (PS04). High residual moisture content in the polymer
only powders coupled with the thermal stress may play an important role with respect to phage
viability during spray-drying. In the present study, encapsulation of Felix O1 phage in a composite
matrix containing different proportions of trehalose and Eudragit S100® resulted in pH-responsive
microparticles with good retention of high titres of viable phages (~109 PFU/g) in the spray-dried
powders. Addition of trehalose in the polymer formulation afforded phages significant protection
from the thermal and desiccation stresses encountered during spray-drying compared with phages
formulated in the polymer without any trehalose present. PS21 powders had lower moisture content
compared with PS30 (Figure S1, Supplementary Materials). A higher proportion of trehalose in the
formulation resulted in higher phage titres, e.g., PS32 (40% (w/w) trehalose) was higher than PS21 (33%
(w/w) trehalose). The particle morphology of PS21 powders was spherical and defect-free (Figure 3).
The particles were hollow internally and had an outer skin with phages presumably encapsulated
within the shell (Figure 3). The absence of blow holes suggested that the drying temperature and rate of
drying were not excessive and, consequently, did not result in an increase in the internal water pressure
to burst the microparticles. Formulations resulting in a high proportion of trehalose in the structure
of the composite microparticles (PS32 and PS24) showed poor acid stability (Figure 2). However,
increasing the proportion of polymer in the microparticle shell (PS21) enhanced acid protection for
the phages exposed to SGF at pH 2 for 3 h (Figures 2 and 4). Acid stability of the spray-dried phage
powders was improved considerably by forming tablets using a direct compression tabletting process
routinely used in the pharmaceutical industry (Figure 4). The physical properties such as bulk density,
brittle fracture, and plastic behaviour of the spray-dried microparticles were suitable for the formation
of robust tablets (Figure S3, Supplementary Materials). The process of direct compression did not
adversely affect the phage titre (Figure 4). The proportion of trehalose in the formulation affected phage
acid stability in the tablets with the high polymer-containing formulation (PS21) showing complete
acid protection (Figure 4). Felix O1 phages were stable in formulation PS21 stored at 4 ◦ C over a period
of three months. The moisture content of the PS21 powders was higher ~9% (w/w) compared with
trehalose-only powders (PS04) under similar drying conditions, which may have adversely impacted
on storage stability (Figure S1, Supplementary Materials). Future work will evaluate phage stability in
PS21 powders dried at 180 ◦ C which had low moisture content similar in magnitude to PS04 (Figure S1,
Supplementary Materials). Optimisation of the proportion of trehalose and Eudragit S100® was needed
to ensure good thermal protection for the phages, attributed to trehalose during the spray-drying
process, whilst ensuring acid protection due to the presence of high amounts of polymer in the
microparticle shell encapsulating the phage. Formulation PS21 was found to be superior to PS24 and
PS32 in terms of protecting phages from acid exposure after forming tablets using direct compression.
Formulation PS21 afforded phage protection during thermal spray-drying, resulting in dry powders
with high viable phage titres showing good storage stability. PS21, therefore, fulfils the criteria of a
suitable formulation for production of acid stable oral solid dosage forms using spray-drying.
Spray-drying is a highly scalable industrial process which is suitable for manufacturing
encapsulated phages in aqueous polymer formulations, such as the one evaluated in the present
study. The spray-dried powders were spherical with good mechanical and flowability properties and
could be reliably tabletted into oral solid dosage forms suitable for enteric delivery. One limitation
of the present study is the short residence times and small particle sizes achieved using small
laboratory-scale spray-dryers. Bench-top spray-dryers are only capable of producing small particles,
typically <10 µm, similar to those produced in this study, which require drying times of only a few
seconds. Industrial-scale dryers produce larger particles ~50 µm and, therefore, have considerably
longer residence times [28]. Future work needs to evaluate the stability of the phages produced using
a pilot-scale spray-dyer resulting in larger particles.
The tableted phages would allow ease of use (good for patient compliance) and reliable delivery
of high titres of viable phages at the site of infection in the gastrointestinal tract. These are important
advantages and can be achieved using the relatively simple, highly scalable, and low-cost process
evaluated in the present study. Phage-containing tablets in standard blister packs would need to be
stored under refrigerated conditions, which is not ideal; therefore, further work is needed to improve
the formulation such that the tablets can be stored without the need for a cold supply chain. Future
work is also needed to evaluate the in vivo release characteristics of the phage-containing tablets and
the targeted delivery of phages at specific locations in the gastrointestinal tract.
Supplementary Materials: The following are available at http://www.mdpi.com/1424-8247/12/1/43/s1.

Author Contributions: G.K.V. and D.J.M. conceived and designed the experiments; G.K.V., Z.R.-Y., M.L. and
D.J.M. performed the experiments; D.J.M. and G.K.V. analyzed the data; D.J.M., M.C.L., M.L. and A.G.F.S.
contributed reagents/materials/analysis tools; G.K.V. and D.J.M. wrote the paper.
Funding: This research was funded by the UK Engineering and Physical Sciences Research Council (EPSRC),
grant number EP/M027341/1 and the APC was funded by EPSRC.
References
1. Freire-Moran, L.; Aronsson, B.; Manz, C.; Gyssens, I.C.; So, A.D.; Monnet, D.L.; Cars, O.; ECDC-EMA
Working Group. Critical shortage of new antibiotics in development against multidrug-resistant
bacteria—Time to react is now. Drug Resist. Updat. 2011, 14, 118–124. [CrossRef] [PubMed]
2. The European Parliament and the Council of The European Union. Regulation (EC) No 1831/2003 of the
European Parliament and of the Council of 22 September 2003 on Additives for Use in Animal Nutrition; European
Union: Brussels, Belgium, 2003.
3. Sheth, R.U.; Cabral, V.; Chen, S.P.; Wang, H.H. Manipulating Bacterial Communities by in situ Microbiome
Engineering. Trends Genet. 2016, 32, 189–200. [CrossRef] [PubMed]
4. Wikswo, M.E.; Kambhampati, A.; Shioda, K.; Walsh, K.A.; Bowen, A.; Hall, A.J. Outbreaks of Acute
Gastroenteritis Transmitted by Person-to-Person Contact, Environmental Contamination, and Unknown
Modes of Transmission—United States, 2009–2013. MMWR Surveill. Summ. 2015, 64, 1–16. [CrossRef]
[PubMed]
5. Scallan, E.; Hoekstra, R.M.; Angulo, F.J.; Tauxe, R.V.; Widdowson, M.A.; Roy, S.L.; Jones, J.L.; Griffin, P.M.
Foodborne Illness Acquired in the United States—Major Pathogens. Emerg. Infect. Dis. 2011, 17, 7–15.
[CrossRef] [PubMed]
6. Hohmann, E.L. Nontyphoidal Salmonellosis. Clin. Infect. Dis. 2001, 32, 263–269. [PubMed]
7. Hyeon, J.Y.; Chon, J.W.; Hwang, I.G.; Kwak, H.S.; Kim, M.S.; Kim, S.K.; Choi, I.S.; Song, C.S.; Park, C.;
Seo, K.H. Prevalence, Antibiotic Resistance, and Molecular Characterization of Salmonella Serovars in Retail
Meat Products. J. Food Prot. 2011, 74, 161–166. [CrossRef] [PubMed]
8. Liu, W.-B.; Chen, J.; Huang, Y.-Y.; Liu, B.; Shi, X.-M. Serotype, Genotype, and Antimicrobial Susceptibility
Profiles of Salmonella from Chicken Farms in Shanghai. J. Food Prot. 2010, 73, 562–567. [CrossRef]
9. Klemm, E.J.; Wong, V.K.; Dougan, G. Emergence of dominant multidrug-resistant bacterial clades: Lessons
from history and whole-genome sequencing. Proc. Natl. Acad. Sci. USA 2018, 115, 12872–12877. [CrossRef]
10. Wall, S.K.; Zhang, J.; Rostagno, M.H.; Ebner, P.D. Phage therapy to reduce preprocessing Salmonella infections
in market-weight swine. Appl. Environ. Microbiol. 2010, 76, 48–53. [CrossRef]
11. Malik, D.J.; Sokolov, I.J.; Vinner, G.K.; Mancuso, F.; Cinquerrui, S.; Vladisavljevic, G.T.; Clokie, M.R.J.;
Garton, N.J.; Stapley, A.G.F.; Kirpichnikova, A. Formulation, stabilisation and encapsulation of bacteriophage
for phage therapy. Adv. Colloid Interface Sci. 2017, 249, 100–133. [CrossRef]
12. Vinner, G.K.; Vladisavljevic, G.T.; Clokie, M.R.J.; Malik, D.J. Microencapsulation of Clostridium difficile specific
bacteriophages using microfluidic glass capillary devices for colon delivery using pH triggered release.
PLoS ONE 2017, 12, e0186239. [CrossRef] [PubMed]
13. Vinner, G.K.; Malik, D.J. High precision microfluidic microencapsulation of bacteriophages for enteric
delivery. Res. Microbiol. 2018, 169, 522–530. [CrossRef] [PubMed]
14. Andreatti Filho, R.L.; Higgins, J.P.; Higgins, S.E.; Gaona, G.; Wolfenden, A.D.; Tellez, G.; Hargis, B.M. Ability
of bacteriophages isolated from different sources to reduce Salmonella enterica serovar enteritidis in vitro and
in vivo. Poult. Sci. 2007, 86, 1904–1909. [CrossRef] [PubMed]
15. Jończyk, E.; Kłak, M.; Mi˛edzybrodzki, R.; Górski, A. The influence of external factors on
bacteriophages—Review. Folia Microbiol. (Praha) 2011, 56, 191–200. [CrossRef] [PubMed]
16. Ma, Y.; Pacan, J.C.; Wang, Q.; Xu, Y.; Huang, X.; Korenevsky, A.; Sabour, P.M. Microencapsulation of
Bacteriophage Felix O1 into Chitosan-Alginate Microspheres for Oral Delivery. Appl. Environ. Microbiol.
2008, 74, 4799–4805. [CrossRef] [PubMed]
17. Ma, Y.; Pacan, J.C.; Wang, Q.; Sabour, P.M.; Huang, X.; Xu, Y. Enhanced alginate microspheres as means of
oral delivery of bacteriophage for reducing Staphylococcus aureus intestinal carriage. Food Hydrocoll. 2012, 26,
18. Dini, C.; Islan, G.A.; de Urraza, P.J.; Castro, G.R. Novel Biopolymer Matrices for Microencapsulation of
Phages: Enhanced Protection Against Acidity and Protease Activity. Macromol. Biosci. 2012, 12, 1200–1208.
[CrossRef] [PubMed]
19. Tang, Z.; Huang, X.; Baxi, S.; Chambers, J.R.; Sabour, P.M.; Wang, Q. Whey protein improves survival and
release characteristics of bacteriophage Felix O1 encapsulated in alginate microspheres. Food Res. Int. 2013,
52, 460–466. [CrossRef]
20. Colom, J.; Cano-Sarabia, M.; Otero, J.; Aríñez-Soriano, J.; Cortés, P.; Maspoch, D.; Llagostera, M.
Microencapsulation with alginate/CaCO3 : A strategy for improved phage therapy. Sci. Rep. 2017, 7.
[CrossRef]
21. Tang, Z.; Huang, X.; Sabour, P.M.; Chambers, J.R.; Wang, Q. Preparation and characterization of dry powder
bacteriophage K for intestinal delivery through oral administration. LWT Food Sci. Technol. 2015, 60, 263–270.
[CrossRef]
22. Wanning, S.; Suverkrup, R.; Lamprecht, A. Pharmaceutical spray freeze drying. Int. J. Pharm. 2015, 488,
136–153. [CrossRef] [PubMed]
23. Matinkhoo, S.; Lynch, K.H.; Dennis, J.J.; Finlay, W.H.; Vehring, R. Spray-Dried Respirable Powders
Containing Bacteriophages for the Treatment of Pulmonary Infections. J. Pharm. Sci. 2011, 100, 5197–5205.
[CrossRef]
24. Vandenheuvel, D.; Meeus, J.; Lavigne, R.; Van Den Mooter, G. Instability of bacteriophages in spray-dried
trehalose powders is caused by crystallization of the matrix. Int. J. Pharm. 2014, 472, 202–205. [CrossRef]
25. Leung, S.S.Y.; Parumasivam, T.; Nguyen, A.; Gengenbach, T.; Carter, E.A.; Carrigy, N.B.; Wang, H.;
Vehring, R.; Finlay, W.H.; Morales, S.; et al. Effects of storage conditions on the stability of spray dried,
inhalable bacteriophage powders. Int. J. Pharm. 2017. [CrossRef]
26. Stanford, K.; McAllister, T.A.; Mazzocco, A.; Waddell, T.E.; Johnson, R.P. Oral delivery systems for
encapsulated bacteriophages targeted at Escherichia coli O157: H7 in feedlot cattle. J. Food Prot. 2010,
27. Jivraj, M.; Martini, L.G.; Thomson, C.M. An overview of the different excipients useful for the direct
compression of tablets. Pharm. Sci. Technol. Today 2000, 3, 58–63. [CrossRef]
28. Ameri, M.; Maa, Y.F. Spray drying of biopharmaceuticals: Stability and process considerations. Dry Technol.
2006, 24, 763–768. [CrossRef]
29. Grasmeijer, N.; Stankovic, M.; De Waard, H.; Frijlink, H.W.; Hinrichs, W.L.J. Unraveling protein stabilization
mechanisms: Vitrification and water replacement in a glass transition temperature controlled system.
Biochim. Biophys. Acta Proteins Proteom. 2013, 1834, 763–769. [CrossRef] [PubMed]
30. De Siqueira, R.S.; Dodd, C.E.R.; Rees, C.E.D. Evaluation of the natural virucidal activity of teas for use in the
phage amplification assay. Int. J. Food Microbiol. 2006, 111, 259–262. [CrossRef]
31. Chen, T.; Fowler, A.; Toner, M. Literature review: Supplemented phase diagram of the trehalose-water binary
mixture. Cryobiology 2000, 40, 277–282. [CrossRef] [PubMed]
32. Vandenheuvel, D.; Singh, A.; Vandersteegen, K.; Klumpp, J.; Lavigne, R.; Van Den Mooter, G. Feasibility
of spray drying bacteriophages into respirable powders to combat pulmonary bacterial infections. Eur. J.
Pharm. Biopharm. 2013, 84, 578–582. [CrossRef] [PubMed]

Volume 12 Edisi 1

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Volume 12 Edisi 1

Uploaded by

Copyright:

Available Formats

pharmaceuticals

Abstract: Tuberculosis (TB) is an infection caused by Mycobacterium tuberculosis, responsible for

Keywords: virtual screening; pharmacophore model; molecular docking; β-ketoacyl-ACP

Pharmaceuticals 2019, 12, 36; doi:10.3390/ph12010036 www.mdpi.com/journal/pharmaceuticals

2. Results and Discussion

Models Strain Energy (kcal/mol) Hbond Mol-qry

Figure 6. Distribution of compounds according to their affinity energy

Figure 8. Evaluationof of RMSDplots

Table 4. Hydrogen bonds formed in complexes.

Acceptor Hydrogen Donor Donor Occupancy (%) a Average Distance (Å)

2.4.2. Bind Free Energy KasA-Ligands

Table 5. Energy contributions to the free energy binding KasA-compounds.

Compound ∆EvdW ∆Eele ∆GGB ∆GNP ∆Gbind

3. Materials and Methods

3.2. Pharmacophore Models Construction

Pharmacophore-Based Virtual Screening

PdC = ±2σ (1)

3.3. Docking-Based Virtual Screening

KISS score = Ir /In (2)

3.4. Molecular Dynamics (MD) Simulations

Binding Free Energy Calculations

∆Gbind = ∆EMM + ∆Gsolv − T∆S (3)

∆EMM = ∆Einternal + ∆Eelectrostatic + ∆Evdw (4)

∆Gsolv can be obtained by solving Equation (5):

∆Gsolv = ∆GGB + ∆GSASA (5)

Pharmaceuticals 2019, 12, x; doi: www.mdpi.com/journal/pharmaceuticals

IC50 Values, Enzyme Assay (nM)

2.4. In Vitro Studies, Rodent Models, and Clinical Trials

Pharmaceuticals 2019, 12, 38; doi:10.3390/ph12010038 www.mdpi.com/journal/pharmaceuticals

2. HCV Genomic Functional RNA Domains

Pharmaceuticals 2019, 12, x 3 of 14

4. Functional Genomic RNA Domains as Therapeutic Targets

4.1.1. Aptamers Targeting HCV IRES

4.1.2. Anti-HCV CRE Aptamers

Funding: This research received no external funding.

Abstract: A rapid, high-yielding microwave-mediated synthetic procedure was developed and

Keywords: Microwave reactions; chemoselective; oxime; aminooxy; glycoclusters; multivalent

Pharmaceuticals 2019, 12, 39; doi:10.3390/ph12010039 www.mdpi.com/journal/pharmaceuticals

For the trivalent

2.1. Medium Scale Microwave-mediated Synthesis of Monovalent Sugar-Linkers

Sugar Additive % Yield % Yield

For further comparison, two traditional, non-microwave-mediated reactions to yield the

2.2. Large Scale Microwave-mediated Synthesis of Monovalent Sugar-Linkers

2.3. Multivalent Glycocluster Microwave-mediated Synthesis

4. Materials and Methods

4.1. General Methods

4.2. Synthesis of Monovalent Sugar-Linkers

Without Aniline-microwave or Traditional Heating

4.3. Synthesis of Trivalent Glycoclusters

Supplementary Materials: The following are available online at http://www.mdpi.com/1424-8247/12/1/39/s1,

Keywords: ezetimibe; simvastatin; co-amorphous; melt viscosity; solubility enhancement

Pharmaceuticals 2019, 12, 40; doi:10.3390/ph12010040 www.mdpi.com/journal/pharmaceuticals

2. Results and Discussion

2.1. Thermal Properties of Ezetimibe, Simvastatin And Its Binary System

System Te [K] Tl [K] Tg exp [K] Tg pred [K]

To determineTable how the polymeric

System Log10 η∞ B = DT0 T0

3. Materials and Methods

3. Materials and Methods

3.2. Preparation of Binary and Ternary Systems

3.3. Differential Scanning Calorimetry (DSC)

3.4. X-ray Diffraction (XRD)

3.5. Rheological Studies

5. Al-Husein, B.A.A. Molecular Mechanisms Regulating Simvastatin-Mediated Inhibition of Prostate Cancer