(2023 - Strauss - PanK III Drug Target

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

Research Article

ChemMedChem doi.org/10.1002/cmdc.202200630

www.chemmedchem.org

Multifaceted Target Specificity Analysis as a Tool in


Antimicrobial Drug Development: Type III Pantothenate
Kinases as a Case Study
Siobhan Ernan Brigg,[a] Lizbé Koekemoer,[a] Leisl A. Brand,[a] and Erick Strauss*[a]

The research and development of a new antimicrobial drug be done. Here we developed Multifaceted Target Specificity
using a target-based approach raises the question of whether Analysis (MTSA) and applied it to type III pantothenate kinase
any resulting hits will also show activity against the homolo- (PanKIII) – an essential enzyme required for coenzyme A
gous target in other closely related organisms. While an biosynthesis in a wide range of pathogenic bacteria – as a case
assessment of the similarities of the predicted interactions study to establish if targeting a specific organism’s PanKIII would
between the identified inhibitor and the various targets is an lead to a narrow- or broad-spectrum agent. We propose that
obvious first step in answering this question, no clear and MTSA is a useful tool and aid for directing new target-based
consistent framework has been proposed for how this should antimicrobial drug development initiatives.

Introduction bacterial pathogens.[1] Pantothenate kinase (PanK) is the first


enzyme in the five-step CoA biosynthesis pathway, catalysing
The search for new antimicrobial drugs, especially ones that can the ATP-dependent phosphorylation of pantothenic acid
be used against the growing number of antimicrobial-resistant (Pan 1), also known as vitamin B5 (Scheme 1).[2] It is a target of
pathogens, is an ongoing and urgent endeavour – albeit one ongoing interest for antimicrobial drug development, partic-
that requires a significant investment of resources. However, ularly due to the differences between the PanKs of eukaryotic
the return on this investment and the potential clinical impact and prokaryotic organisms.[1a,b] Of the three types of PanK that
of a new antimicrobial drug may be dependent on whether it have been described thus far, two (types I and III) occur
shows narrow- or broad-spectrum activity. This raises several exclusively in bacteria, and are structurally and mechanistically
key questions: If the development process is following a target- divergent from the type II PanK found predominantly in
based approach focused on the discovery of a potent inhibitor eukaryotes, including humans (the only exception being the
of the target in one organism, what is the likelihood that an occurrence of type II PanKIIs in some Staphylococcae).[1d,2–3]
identified hit will also show activity against the homologous Importantly, type III PanKs (PanKIII) appear to predominate in
target in other organisms? Analysis of the structural similarities bacteria, with ~ 44 % of known species exclusively harbouring a
of the interactions between the drug and the various targets PanKIII.[2,3d,4] Among these are several pathogens included in the
should be an important component in providing an answer, but WHO priority list of antibiotic-resistant bacteria, such as
how can this be done in a consistent and reproducible manner? Acinetobacter baumannii and Pseudomonas aeruginosa (both
Are there other comparative analyses that would complement listed as Priority 1: critical) and Helicobacter pylori and Campylo-
such an approach, and that could be executed using readily bacter spp. (both Priority 2: high).[5] This suggests that PanKIII
available data? enzymes may be particularly good targets for the development
We, and others, have had a long-standing interest in the of novel antimicrobials for the treatment of problematic drug-
development of compounds inhibiting the biosynthesis of the resistant infections. However, would an inhibitor developed
essential metabolic cofactor coenzyme A (CoA) in a range of against one organism’s PanKIII enzyme also show potent activity
against another?
While bacterial type I PanK (PanKI) enzymes – especially the
[a] Dr. S. E. Brigg, Dr. L. Koekemoer, L. A. Brand, Prof. E. Strauss
Department of Biochemistry
Mycobacterium tuberculosis PanKI (MtPanKI) – have been shown
Stellenbosch University
Matieland 7602 (South Africa)
E-mail: estrauss@sun.ac.za
Supporting information for this article is available on the WWW under
https://doi.org/10.1002/cmdc.202200630
This article belongs to the Joint Special Collection “Biological and Medicinal
Chemistry in Africa”.
© 2023 The Authors. ChemMedChem published by Wiley-VCH GmbH. This is
an open access article under the terms of the Creative Commons Attribution Scheme 1. Pantothenate kinase (PanK) catalyses the first step in coenzyme A
Non-Commercial License, which permits use, distribution and reproduction (CoA) biosynthesis, the ADP-dependent phosphorylation of pantothenate
in any medium, provided the original work is properly cited and is not used (Pan, 1) to form 4’-phospho-pantothenic acid (P-Pan, 2). The β-alanine
for commercial purposes. moiety of Pan is highlighted in blue, its pantoyl moiety in red.

ChemMedChem 2023, 18, e202200630 (1 of 10) © 2023 The Authors. ChemMedChem published by Wiley-VCH GmbH
18607187, 2023, 7, Downloaded from https://chemistry-europe.onlinelibrary.wiley.com/doi/10.1002/cmdc.202200630 by Mcgill University Library, Wiley Online Library on [01/05/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Research Article
ChemMedChem doi.org/10.1002/cmdc.202200630

to be a highly tractable drug target based on the successful establish the likely target spectrum of newly developed
identification of inhibitors showing submicromolar potency,[1e] inhibitors.
this is not the case with PanKIIIs. These enzymes have not been
pursued for inhibitor discovery in any high-throughput screen-
ing initiatives, and the only rational inhibitor design study Results and Discussion
conducted to date focused on preparing ATP-analogues as
potential inhibitors.[6] However, a characteristic of PanKIII Bioinformatic analysis: Comparison of full sequences
enzymes is a poorly defined ATP binding site that has minimal
contacts with the adenine moiety, resulting in KM values for ATP We initiated the MTSA by considering the primary sequence
that are in the millimolar range in most cases.[3b,c,4] This limits diversity of PanKIII enzymes. The Pfam family “Type III Pan-
the opportunities for the design of tight-binding ATP mimics, tothenate Kinase” (Pan_kinase, PF03309) currently contains
and the best such competitive inhibitor reported to date had a 6260 full sequences with an average domain length of 200.1
Ki of 164 μM determined for the Bacillus anthracis PanKIII amino acids and 31 % average identity for the full alignment.
(BaPanKIII), which also has the lowest known KM for ATP The InterPro entry for Type III pantothenate kinase (Type_III_
(510 μM) of characterized PanKIIIs.[6] PanK; IPR004619 lists ~ 30,000 matching entries; we selected the
In contrast, the Pan 1 binding site of PanKIII enzymes is 367 entries listed as reviewed in UniProt as a representative
enclosed on the side that is distal to the site of catalysis, i. e. on sample of the family, and aligned their sequences using Clustal
the side where Pan’s carboxyl group is bound. This feature is Omega. The resulting alignment was subsequently used to
unique to the PanKIIIs, as most of the type I and II PanK enzymes construct an approximately-maximum-likelihood phylogenetic
that have been studied to date have a low level of substrate tree using FastTree. The unrooted tree (Figure 1A) shows that
specificity, allowing them to accept a range of Pan 1 analogues PanKIII enzymes separate into several distinct clades, with the
– especially amides of Pan 1 – as substrates.[7] This includes the sequences from representative organisms such as the Gram-
alternative CoA precursor pantetheine (PantSH) and the known negative P. aeruginosa (Pa) and the Gram-positive Bacillus
antimicrobial pantothenamides (PanAms). However, neither of anthracis (Ba) being found on opposite sides of the tree. We
these are accepted by PanKIII enzymes. also considered the location of seven other PanKIIIs from
While the highly defined Pan 1 binding site presents some organisms of interest, i. e. those that are relevant as antibiotic-
obvious challenges for inhibitor development, we also consid- resistant pathogens and/or that have been structurally charac-
ered that it offers the opportunity to discover a new terized. The locations of the sequences for the PanKIIIs of A.
antimicrobial compound that shows activity against a range of baumannii (Ab), Burkholderia cenocepacia (Bc), Burkholderia
pathogens that rely on a PanKIII for their de novo CoA biosyn- thailandensis (Bt), Campylobacter jejuni (Cj), H.pylori (Hp), Legion-
thesis. In this assessment, we were led by the assumption that ella pneumophila (Lp) and Thermotoga maritima (Tm), we
the binding site for such a relatively small molecule with limited observe obvious clustering of some sequences (e. g. Ba, Lp and
structural features is likely to be largely conserved regarding its Tm) that also show high level of sequence identity (Figure 1B).
size, shape, and ligand binding interactions in a set of In other cases the associations are less clear, with PaPanKIII
homologous enzymes. As such, a target-based PanKIII inhibitor showing the same level of sequence identity with AbPanKIII
discovery initiative offers an ideal opportunity to develop and (that is found in the same branch) as BcPanKIII.
test a structured analysis method – one based on several Taken together, the primary sequence similarity analysis
independent variables – to assess and predict the likelihood suggested that while broad-spectrum inhibition of PanKIII
that an inhibitor developed against one PanKIII would also enzymes by a single inhibitor is unlikely, it may be possible that
target others. closely related enzymes will be affected by the same com-
Here we present our proposal for such a method, called pound.
Multifaceted Target Specificity Analysis (MTSA). MTSA relies first
on bioinformatic and structural analyses that uses information
from sequence databases and the Protein Data Bank (PDB), and Bioinformatic analysis: Comparison of Pan binding site
combines it with systems-specific experimental information residues
(e.g. kinetic data). We applied MTSA to the PanKIII enzymes as a
case study, demonstrating that the in silico analyses provide Although the same enzyme may show significant sequence
sufficient information to allow for the identification of groups of diversity across many organisms, it is likely that the residues
similar enzymes. The small sample size and requirements for forming the active site and participating in catalysis would
assay optimization for individual enzymes meant that the remain conserved. We therefore performed a second bioinfor-
experimental data was only able to provide a small contribution matic analysis that focused on these residues, and specifically
to the overall analysis. those lining the Pan 1 binding site.
Taken together, our study provides specific insights that The experimentally determined structure of PaPanKIII with
may guide the development of antimicrobial compounds that Pan 1 bound (PDB 2F9 W) shows that the binding site is found
target PanKIII enzyme in pathogenic organisms of interest, while at the interface of the two protomers that constitute the
presenting MTSA as a tool that can more generally be used to homodimeric enzyme, with residues from both protomers lining
the site.[3b] Due to the system’s symmetry, two binding sites are

ChemMedChem 2023, 18, e202200630 (2 of 10) © 2023 The Authors. ChemMedChem published by Wiley-VCH GmbH
18607187, 2023, 7, Downloaded from https://chemistry-europe.onlinelibrary.wiley.com/doi/10.1002/cmdc.202200630 by Mcgill University Library, Wiley Online Library on [01/05/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Research Article
ChemMedChem doi.org/10.1002/cmdc.202200630

Asp101 is poised to deprotonate the 4’-OH group of Pan to


initiate catalysis.
These residues are some of the 29 found within 6.0 Å of the
Pan ligand in the PaPanKIII co-crystal structure. Comparing these
residues to the corresponding sequence location in the profile
hidden Markov model (HMM) for the PF03309 Pfam family
(Figure 2B) indicates that some of these residues are highly
conserved across all sequences, while for others at least the
polar/non-polar nature is maintained.
We next surveyed the Pan binding sites of the eight other
PanKIIIs of interest through a comparative analysis in the
Schrödinger software package using its Biologics Multiple
Sequence Viewer tool and by performing a ligand contact
annotation. Doing the latter identified the residues within 6.0 Å
of the Pan binding site of the six enzymes with experimentally
determined structures (PDB ids: Ba: 2H3G,[3c] Bc: 5B8H, Bt: 4O5F,
Cj: 2NRH, Lp: 3DJC and Tm: 3BEX[3d]); these were compared to
the 29 contact residues identified for PaPanKIII through a
multiple sequence alignment that also included the two PanKIII
sequences for which structures have not yet been determined
(Ab and Hp) (Supporting Information Figure 1). To simplify the
comparative analysis, a pseudo-sequence was generated for
each enzyme that only contained its Pan 1 binding site residues;
multiple sequence alignment of these pseudo-sequences
allowed for a direct visual comparison of the various PanKIII’s
Pan 1 binding site residues in the context of the profile HMM
(Figure 2B). This analysis shows that although the nine PanKIII
enzymes of interest largely reflect the relative occurrence of
Figure 1. A) Unrooted approximately-maximum-likelihood phylogenetic tree particular residues in a specific position as predicted by the
of 367 reviewed entries in UniProt belonging to the Pfam family “Type III profile HMM, some important deviations do occur. For example,
Pantothenate Kinase” (Pan_kinase, PF03309). The locations of the PanKIII in its position 60 PaPanKIII has a Glu, while all the other enzymes
enzymes of interest to this study are indicated by circles, with the
abbreviations referring to the host species: Acinetobacter baumannii (Ab), except one have a Val (BtPanKIII has an Ala). While this residue is
Bacillus anthracis (Ba), Burkholderia cenocepacia (Bc), Burkholderia thailanden- not in close contact with the Pan ligand, such differences may
sis (Bt), Campylobacter jejuni (Cj), Helicobacter pylori (Hp), Legionella pneumo- affect the overall active site architecture.
phila (Lp), Pseudomonas aeruginosa (Pa) and Thermotoga maritima (Tm).
B) Sequence identity matrix for the nine PanKIII enzymes of interest. Values To establish the relationship between the Pan binding sites
are the percentage residues that are identical in a pair-wise comparison of of the various PanKIIIs, a phylogram was generated based on
any two sequences. the alignment of the pseudo-sequences. The result (Figure 2C)
confirms the outcome of the full sequence analysis, showing
that the nine enzymes can be sorted into three groups: Pa, Bt
formed in this manner. A 2D ligand–protein interaction diagram and Bc in Group 1, Ba, Lp and Tm in Group 2, and Ab, Hp and Cj
highlights the main interactions between the Pan ligand and in Group 3. However, the branch lengths and node positions
the PaPanKIII protein in one of these sites (small, non-significant also indicate that while the associations within Group 2 are
differences are observed between the two active sites, mainly relatively close, Pa in Group 1 and Ab in Group 3 are further
regarding the position of Arg129’ and His156’) (Figure 2A). In removed from the other group members, and could be
the well-defined Pan binding pocket a Tyr residue (Tyr92 in considered members of a separate group. Overall, the outcome
PaPanKIII) acts as a cap that forms the closed binding site of the bioinformatic analysis of the PanKIII Pan binding site
characteristic of the majority of PanKIII enzymes. The cap is an largely reflects that of the full sequence (Figure 1) yet provides
important feature of the binding site, limiting the size of ligands further insights into specific, local protein–ligand interactions
that can bind and explaining why Pan analogues such as that could be useful in drug development.
pantetheine and PanAms are excluded from the site.
In addition to its function as a cap, the OH of the Tyr
residue, together with Thr180’ and Arg102, forms hydrogen Structural analysis: Comparison of Pan binding site
bonding and ionic interactions with the carboxylic acid moiety interactions and architectures
of Pan 1. Asn9, Thr157’ and the backbone carbonyl of Val55
interacts with the 2’-OH of Pan with hydrogen bonds bridged We next considered the actual interactions between the
by a water molecule, as does His156’ with its 4’-OH group. proteins and the Pan 1 ligand and the Pan binding site
architectures in the available crystal structures of the proteins

ChemMedChem 2023, 18, e202200630 (3 of 10) © 2023 The Authors. ChemMedChem published by Wiley-VCH GmbH
18607187, 2023, 7, Downloaded from https://chemistry-europe.onlinelibrary.wiley.com/doi/10.1002/cmdc.202200630 by Mcgill University Library, Wiley Online Library on [01/05/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Research Article
ChemMedChem doi.org/10.1002/cmdc.202200630

Figure 2. A) PaPanKIII‘s Pan binding site, depicted using a 2D ligand-protein interaction diagram based on the co-crystal structure of PaPanKIII bound to Pan 1
(PDB: 2F9W).[3b] The Pan ligand is shown with carbon atoms in green; residues making hydrophilic interactions with the ligand are shown as ball and stick
structures and are labelled in green; residues involved in hydrophobic interactions are shown as lashed semi-circles and are labelled in red. Water molecules
are shown as light blue spheres. Ionic and hydrogen bonding interactions are highlighted with dashed lines, with the interaction distance indicated in Å.
Figure created using LigPlot +.[8] B) Comparative analysis of the Pan binding site residues of PanKIII enzymes of interest (host organism abbreviations as in
Figure 1). The 29 residues found within 6 Å of the Pan ligand in the PaPanKIII structure is used as reference; their residue numbers are shown above the Pa
sequence. The corresponding profile HMM of the PF03309 Pfam family is shown above each residue; below is the multiple sequence alignment of the
pseudo-sequences generated from the corresponding residues in the other eight PanKIIIs of interest. Residues corresponding to one of the two most abundant
residues in the profile HMMs are coloured with a matching background. Homology is indicated with the symbols * (identical) and : (similar). C) Phylogram
based on the multiple sequence alignment shown in B), with the three groups of PanKIII enzymes indicated.

of interest; this was done to establish to what extent the location of some of the residues lining the binding site pocket
outcome of the bioinformatic analyses are reflected in the being visible. This suggests some level of flexibility in the Pan 1
protein structures. For four of the enzymes (Bc, Bt, Pa and Tm) binding sites of these enzymes. For the comparative analysis
structures with Pan bound has been determined, while for Ba, one of the two active sites were chosen for each of these
Cj and Lp only apo structures are available. To allow for these structures.
latter three structures to also be included in the analysis, the Visualisation of the enzymes‘ Pan binding sites by structural
structure of Pan was docked into their putative Pan 1 binding alignment shows a limited number of direct interactions
sites using Maestro of the Schrödinger software suite. This was between active site residues and the Pan ligand, in agreement
done by deleting all water molecules from the binding site prior with what was observed for PaPanKIII (Figure 3). Specifically, all
to docking, refining the residue side chain rotamers to match the structures show an interaction between an Asp (Pa: Asp101)
those seen in the complexes with ligand bound and by utilising and the 2’-OH and 4’-OH groups; this is to be expected as the
positional and hydrogen bond constraints to ensure that the deprotonation of the 4’-OH group initiates catalysis, and this
ligand binds in the expected configuration. For the BaPanKIII Asp residue is conserved in all PanKIII. The other direct
structure (PDB: 2H3G),[3c] a loop from residue Ser163 to Ile171 interactions are also with highly conserved residues: a Tyr (Pa:
that forms part of the Pan 1 binding site is missing; this was Tyr92), a backbone amide of a Gly (Pa: Gly99), an Arg (Pa:
filled in using the Biologics Crosslinks Proteins calculation in Arg102) and Thr (Pa: Thr180’) that all interact with the carboxyl
Maestro. group of Pan. The only major deviation is for TmPanKIII that has
The structures with Pan 1 bound were prepared using the a Val residue instead of the Tyr;[3d,4] because of this, and since
Protein Preparation wizard in Maestro and then compared with T. maritima is not a pathogen of interest, this structure was
a focus on their Pan binding sites. For the Bc, Bt and Pa excluded from further analyses.
structures the two Pan 1 binding sites of the respective proteins Comparison of the enzymes’ Pan 1 binding sites in regard
were found not to be identical, with small movements in the to the relative localization and conformation of these residues,

ChemMedChem 2023, 18, e202200630 (4 of 10) © 2023 The Authors. ChemMedChem published by Wiley-VCH GmbH
18607187, 2023, 7, Downloaded from https://chemistry-europe.onlinelibrary.wiley.com/doi/10.1002/cmdc.202200630 by Mcgill University Library, Wiley Online Library on [01/05/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Research Article
ChemMedChem doi.org/10.1002/cmdc.202200630

Figure 3. Comparison of the Pan binding sites of the PanKIII enzymes of interest (host organism abbreviations as in Figure 1) and the relative localization and
conformation of the residues interacting with the Pan ligand in these sites as viewed on Maestro (Schrödinger, Inc). Binding sites are visualized from the
experimentally determined structures with the Pan ligand bound (PDB ids: Pa: 2F9W, Bc: 5B8H and Bt: 4O5F) or of the available apo structures with the Pan
ligand docked into the putative binding sites using Maestro (PDB ids: Ba: 2H3G, Lp: 3DJC and Cj: 2NRH). Each box contains the enzymes belonging to one of
the three groups as identified by the bioinformatic analysis shown in Figure 2, and also shows the overlay of the Pan ligand binding poses as observed for
that group of enzymes. Enzyme residue structures are labelled and shown as stick structures with carbon atoms in grey, nitrogen atoms in blue, oxygen
atoms in red and hydrogen atoms in white. The Pan ligands in the binding sites are shown as ball-and-stick structures with the carbon atoms in green; for the
binding pose overlays the carbon atoms of the Pan seen in each enzyme’s binding site is coloured differently (for Group 2, Pan in Lp is green and purple in
Ba).

indicate that the enzymes belonging to the same groups also Bt (Group 1) structures, which all have the β-alanine moiety of
have the most similar binding site architectures (Figure 3). As the Pan 1 ligand adopting a cis configuration (i. e., the pantoyl
expected, the very closely related Bc and Bt have very similar amide relative to the carboxylate) (Figure 3). However, for Ba
binding sites that also resemble the site of the other Group 1 and Lp the Pan 1 ligand took on very different conformations.
enzyme, Pa. In the same way the Group 2 enzymes Ba and Lp For Ba, the ligand binding pose is very similar to those seen for
also have similar binding site, although for Ba the Tyr and Thr the Group 1 structures, likely because of the displacement of its
residues are displaced away from the ligand, creating more Tyr and Thr residues. For Lp, the energy minimised ligand
space around the carboxyl group. While we considered that this binding pose shows the β-alanine moiety in a configuration
difference could be due the ligand being modelled into the Ba that approaches a trans dihedral angle, as this allows for
active site (instead of the binding site being identified from an optimal interactions between the Pan carboxyl and its Tyr96
actual co-crystal structure), the activity analysis conducted with and Arg106, and between the pantoyl amide carbonyl and
various Pan 1 analogues (results discussed below) showed that Gly106. Interestingly, the ligand pose in the Cj structure is also
it is the only PanKIII that could accept as substrates analogues similar to that seen for Lp, in spite of the differences in binding
with extended modifications to the carboxyl group. This site architectures.
suggests that the Pan binding site of Ba is indeed larger than
those of other PanKIIIs. Finally, analysis of the binding site of the
Cj enzyme (only example of Group 3) is clearly different from Structural analysis: Comparison of solvent accessible cavities
the others. leading to Pan binding site
The differences in the binding site architectures also impact
on the binding pose of the Pan 1 ligand. Overlay of the ligand Next, we compared the structures in regards to the solvent
structures again confirm the similarity between the Pa, Bc and accessible cavity that can be observed above the β-alanine

ChemMedChem 2023, 18, e202200630 (5 of 10) © 2023 The Authors. ChemMedChem published by Wiley-VCH GmbH
18607187, 2023, 7, Downloaded from https://chemistry-europe.onlinelibrary.wiley.com/doi/10.1002/cmdc.202200630 by Mcgill University Library, Wiley Online Library on [01/05/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Research Article
ChemMedChem doi.org/10.1002/cmdc.202200630

moiety of the Pan 1 ligand. This cavity is a feature of all the opening towards the solvent. For Lp (Group 2), the opening is
PanKIII structures, and a comparative analysis of its shape and shifted towards the pantoyl moiety, with only the β-carbon of
electrostatic potential distribution could give a good indication the β-alanine moiety barely visible, and with the cavity opening
of whether the enzymes would respond similarly to a particular having more negatively charged residues. Finally, representing
inhibitor structure. Group 3, the Cj structure’s cavity opens showing the carboxyl
Visualization of this cavity by alignment of the various group and the α -carbon of the β-alanine moiety, with a wider
structures so as to give the same perspective on the Pan entrance lined by hydrophilic residues.
binding site, highlights the key similarities and differences In addition to the above-mentioned cavity, the Pan 1
(Figure 4). For this analysis, the Ba structure was not included, binding site also has an opening to the ATP binding site (not
as the loop from residue Ser163 to Ile171 that is missing in this shown). This opening varies in the size and shape between the
structure does not allow for a reasonable visualization to be different enzymes, with Bc and Bt from Group 1 and Cj from
prepared. For the remaining structures, the analysis shows that Group 3 having considerably smaller cavities compared to the
the Group 1 structures (Pa, Bt and Bc) all have a cavity with an other enzymes.
opening that exposes both carbon atoms of the β-alanine
moiety, suggesting that modifications at these positions would
be tolerated by these structures. In addition, the cavity is Kinetic analysis: Comparing the relative affinity and activity
relatively narrow, with positively charged residues lining its towards Pan

Finally, we set out to experimentally perform a comparative


kinetic analysis of representative enzymes from all three groups
to determine their apparent KM and Vmax values for Pan 1 using
a fixed amount of ATP. For these experiments a high
concentration of ATP (5 mM) was chosen based on previous
reports of the KMATP being in the range of 500 μM to 10 mM for
various PanKIII enzymes,[2] and because this value is close to a
physiologically relevant intercellular ATP concentration. For the
analysis, we chose the Group 1 enzymes PaPanKIII and BtPanKIII,
the Group 2 enzymes BaPanKIII and LpPanKIII and the Group 3
enzymes AbPanKIII and HpPanKIII. Unfortunately, the purified
LpPanKIII enzyme proved to be inactive using the standard assay
conditions, and it could not be included in the analysis.
The kinetic profiles for the remaining five enzymes were
obtained using a range of Pan 1 concentrations, with the
activity monitored using the standard pyruvate kinase/lactate
dehydrogenase kinase assay that depends on the oxidation of
NADH. From these profiles the respective apparent kinetic
parameters were determined by fitting the Michaelis-Menten
equation to the data obtained from several independent
experiments; the reported parameters are the average values
that were obtained (Table 1). We found that some enzymes
(e. g. BaPanKIII) had a large variation in the apparent kcat value,
mainly due to batch-to-batch differences in specific activity; the
reason for this remains unclear, although we did observe that
certain enzymes lost activity soon after purification. The

Table 1. Apparent kinetic parameters for Pan 1 for selected PanKIII enzymes.
Enzyme Group KMapp kcatapp kcat/KMapp N
[μM][a] [s 1][a] [mM 1 s 1]

Figure 4. Images of the Pan binding sites of the PanKIII enzymes of interest PaPanKIII 1 8.44 � 1.96 3.12 � 0.48 497 � 124 8
(host organism abbreviations as in Figure 1) showing the solvent accessible BtPanKIII 1 30.8 � 1.3 0.50 � 0.01 16.4 � 0.6 3
cavity leading to the opening that exposes the Pan ligand in the region of BaPanKIII 2 68.3 � 4.0 3.40 � 1.72 47.2 � 22.5 3
its β-alanine moiety (indicated on ball-and-stick structure of the ligand). AbPanKIII 3 10.7 � 0.2 2.05 � 0.04 185 � 14 2
HpPanKIII 3 44.3 � 12.3 3.58 � 0.48 108 � 25 5
Snapshots taken from Discovery Studio Visualiser generated from the crystal
structures as listed in Figure 3 using the same perspective from the [a] Apparent kinetic parameters determined with [ATP] = 5 mM by fitting the
overlayed structures. Enzyme surface coloured according to electrostatic Michaelis-Menten equation to the activity profiles. Values are the average of
potential (blue, positive; red, negative) with grey indicating hydrophobic the number of independent experiments (N) indicated in the last column with
surfaces. the error indicating SEM (N � 3) or range/2 (N = 2).

ChemMedChem 2023, 18, e202200630 (6 of 10) © 2023 The Authors. ChemMedChem published by Wiley-VCH GmbH
18607187, 2023, 7, Downloaded from https://chemistry-europe.onlinelibrary.wiley.com/doi/10.1002/cmdc.202200630 by Mcgill University Library, Wiley Online Library on [01/05/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Research Article
ChemMedChem doi.org/10.1002/cmdc.202200630

apparent KM values determined for Pan was in the range of compounds were evaluated at 1 mM and their activity
~ 10–70 μM. This gave apparent specificity constants (kcat/KMapp) compared to the activity observed towards 30 μM Pan (a
of between 16.4 mM 1 s 1 and ~ 500 mM 1 s 1 (for the two concentration roughly equal to the average KM value of the test
Group 1 enzymes BtPanKIII and PaPanKIII, respectively). The enzymes). These and the remaining compounds were subse-
Group 3 enzymes AbPanKIII and HpPanKIII had intermediate quently also tested to determine if they inhibit the activity of
specificity constants. While it is possible that the enzymes from the test enzymes acting on 30 μM Pan; for these tests, the
some groups have improved catalytic activity compared to compounds were present at 10-fold the substrate concentration
others, the relatively small data set, the inherent variability in (i. e., at 300 μM). The results are summarized as a heatmap
specific activity observed for different batches of purified (Figure 5), with green boxes indicating which compounds
enzyme and the apparent need to optimize assay conditions for showed at least 25 % activity as alternative substrates relative to
each enzyme make it difficult to use the kinetic data to draw the Pan test reaction, or at least 50 % inhibition of an enzyme’s
any firm conclusions in this respect.

Applying the MTSA: Pan analogues as test compounds

Taken together, the bioinformatic, structural and kinetic analysis


components of the MTSA confirms that PanKIII enzymes cannot
be considered as a monolithic target, where inhibitors designed
to be active against the PanKIII of one organism can be expected
to show activity against any other. Instead, the MTSA indicates
that the PanKIIIs can be sorted according to groups, with the
best predictor of these groupings being the alignment of the
pseudo-sequence created from the residues lining the binding
pocket of interest; the structural and kinetic analyses provide
more nuanced information regarding variations within these
groups. The MTSA predicts that PanKIII enzymes belonging to
the same group are more likely to show similar sensitivities to a
particular inhibitor than those from other groups.
To put this prediction to the test, we evaluated the activity
of the five purified enzymes using pantetheine (PantSH, 3), a
naturally occurring Pan analogue and alternative CoA precur-
sors, as well as a range of synthetic Pan analogues with various
structural modification. These included modifications in a) the
carboxyl group, with the -COOH exchanged for -OH (4), -CONH2
(5), -CONHNH2 (6) or -SO2 (7), in b) the β-alanine moiety, with
one methylene removed (8), one methylene added (9), the
configuration fixed to trans by addition of a double bond (10),
the addition of -CH3 groups to the α-carbon (11–13) or β-
carbon (14) or of a -CF3 group to the β-carbon (15), and in
c) the 4’-OH group, with preparation of the 4’-deoxy (16), 4’-
amino (17) and 4’-methoxy (18) variants, as well as 4’-phospho-
pantothenic acid (P-Pan, 2), the reaction product. The analogues
were specifically chosen in light of the protein-ligand inter-
actions observed in the Pan binding site (Figure 3), with
selected modifications predicted to either strengthen or disrupt
one or more of these interactions, and by considering the size
and shape of the cavity that opens above the β-alanine moiety
of Pan (Figure 4), as the Pan analogues with modifications in
this moiety could exploit new binding interactions made
available by the cavity. Consequently, these analogues would Figure 5. Heatmap showing the relative responses of the indicated PanKIII
enzymes (host organism abbreviations as in Figure 1) belonging to the three
allow for a direct evaluation of the ligand binding interactions groups identified by the bioinformatic analysis (Figure 2). The Pan analogues
and the extent to which these show similarities between 2–18 and the PanKIII product P-Pan 3 were tested as alternative substrates
enzymes belonging to the same group. (at 1 mM) or as inhibitors (at 300 μM). For activity as substrates, the box
colour indicates the activity relative to a reference reaction with 30 μM Pan
As compounds 3–15 all contain a free 4’-OH group, these (green > 25 %, orange 5–25 %, red < 5 %). For inhibitor activity, the box
were first evaluated to determine if they are accepted as colour indicates the relative reduction in activity of the same reference
alternative substrates by the enzymes; for this test, all reaction (green > 50 %, orange 20–50 %, red < 20 %).

ChemMedChem 2023, 18, e202200630 (7 of 10) © 2023 The Authors. ChemMedChem published by Wiley-VCH GmbH
18607187, 2023, 7, Downloaded from https://chemistry-europe.onlinelibrary.wiley.com/doi/10.1002/cmdc.202200630 by Mcgill University Library, Wiley Online Library on [01/05/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Research Article
ChemMedChem doi.org/10.1002/cmdc.202200630

activity towards Pan. Compounds that showed less than 5 % modifications on the α-carbon was predicted; while this was
activity as substrates or less than 20 % inhibition are indicated observed, with HpPanKIII showing activity towards all the α-
with red boxes, with those with intermediates responses methylated analogues, both enzymes also showed good activity
indicated with orange boxes. towards the rac-β-methylated 14 and HpPanKIII towards rac-β-
The heatmap shows that relatively few enzymes showed CF3 15. This suggests that the Cj structure on which the Group 3
substrate or inhibitor activity towards the analogues with analysis was based may not be the best representative of this
modifications in the carboxyl group, and then only intermediate group, and that having a structure of another Group 3 enzyme
activity. This is to be expected considering the several would be advantageous to refine the analysis and predictions.
interactions between the protein and the carboxyl group While the predictions of the structural analysis of the
observed in the Pan binding site (Figures 2 and 3). As solvent accessible cavities were therefore largely borne out by
substrates, the Group 2 enzyme BaPanKIII showed activity the outcomes, and the enzymes showing responses broadly
towards three analogues (3, 6 and 7), in agreement with its aligning with the predictions for their groups, this was not the
binding site being more open due to the displacement of its case for the analysis of β-alanine modified Pan analogues as
Tyr100 and Thr184’ residues. As inhibitors, only PaPanKIII inhibitors. The ability to act as an inhibitor implies that the
(Group 1) and AbPanKIII (Group 3) were affected, each by three compound is able to bind to the Pan binding site, yet fails to be
Pan analogues. phosphorylated (or is trapped following phosphorylation). This
The Pan analogues modified in the β-alanine moiety gave distinction between acting as a substrate compared to an
the most promising results when tested as substrates, in inhibitor is likely influenced by the exact binding pose of the
agreement with the potential highlighted by the structural ligand, with substitutions on the β-alanine moiety causing the
analyses of the ligand binding interactions (Figure 3) and the ligand to orient itself differently to take advantage of the cavity
solvent accessible cavities leading to the Pan binding site above this group. Such changes in binding pose are likely the
(Figure 4). The α-Pan (8) with one methylene group less than basis for the observed inhibition.
Pan was accepted as a substrate by all the enzymes except We found that with one exception, all the inhibitor hits
BtPanKIII. This is not surprising as the modification can easily be were again observed for PaPanKIII and AbPanKIII (as was the case
accommodated while likely still maintaining some of the many for the analogues with carboxyl group modifications) with the
interactions with the carboxyl group. However, only PaPanKIII latter enzyme showing inhibition by two compounds (8 and 11)
accepted HoPan (9) – which has one additional methylene – as that the other accepted as substrates instead. That PaPanKIII
a substrate, in agreement with the analysis that the interactions accepts the (R)-α-methylated 11 as a substrate while being
with the carboxyl are many and varied in most of the enzymes. inhibited by the (S)-α-methylated 12 can be explained by the
These interactions would likely be disrupted due to the structural analysis of the cavity above the β-alanine moiety: in
displacement of the carboxyl group in HoPan relative to Pan. the Pan-bound PaPanKIII structure it is the pro-R hydrogen on
The Pan analogue 10 with a double bond in the β-alanine the α-carbon that points to the cavity entrance, and introducing
moiety in the (E)-configuration is a natural product inhibitor of a methyl group in this configuration is therefore easily
the S. aureus type II PanK called CJ-15,801;[9] it was included accommodated without affecting the overall binding pose of
here to establish if the predicted ligand binding pose config- the ligand. However, having the methyl in the S-configuration
uration can be harnessed to ensure or disrupt activity. As the will require the ligand to adapt its conformation, and this likely
ligand binding pose analysis indicated that Group 3 enzymes impacts on catalysis.
prefer having Pan adopt a configuration of its β-alanine moiety Although PaPanKIII and AbPanKIII belong to different groups
that approaches trans, we predicted that 10 might be preferred according to their original assignment, the full sequence
as a substrate by the AbPanKIII and HpPanKIII enzymes. However, bioinformatic analysis did suggest these enzymes are closely
only the Group 1 enzyme BtPanKIII and Group 2’s BaPanKIII related (Figure 1), and the phylogram based on the pseudo-
showed average activity toward this analogue, while none of sequence indicated that both these enzymes are only distantly
the enzymes were inhibited. This suggested that ligand binding related to their other assigned group members (Figure 2C). The
pose is probably not a good basis for inhibitor design. current lack of an AbPanKIII structure makes it difficult to further
The cavity analysis specifically predicted that the Group 1 probe the basis for the association between these two enzymes
enzymes would accept modifications at both carbons of the β- and to rationalize the results of the inhibitor tests. However, it
alanine moiety; indeed, both PaPanKIII and BtPanKIII showed does highlight the power of the bioinformatic analyses of the
activity towards (R)-α-methylated 11 and rac-β-methylated 14, MTSA in suggesting alternative associations when the primary
with PaPanKIII showing activity towards the α,α -dimethylated ones fail to make useful predictions.
13 as well. Interestingly, PaPanKIII is inhibited by (S)-α-meth- Finally, the compounds 16–18 with modifications in the
ylated 12, instead of acting on it as a substrate. The cavity 4’-OH group (which therefore precludes catalysis) acted as
analysis also predicted that the Group 2 enzymes (based on the inhibitors of all the enzymes except BaPanKIII. This enzyme was
Lp structure) would be more likely to accept analogues likely not inhibited because of its less enclosed Pan binding
modified on the β-carbon. This is the case, with BaPanKIII site, which would allow the modified Pan analogues to be
showing > 25 % activity towards the rac-β-methylated 14, accommodated without much affect on catalysis. The reaction
although it also shows intermediate activity towards the α- product P-Pan (2) acted as an inhibitor of PaPanKIII and
methylated 11–13. For the Group 3 enzymes a preference for

ChemMedChem 2023, 18, e202200630 (8 of 10) © 2023 The Authors. ChemMedChem published by Wiley-VCH GmbH
18607187, 2023, 7, Downloaded from https://chemistry-europe.onlinelibrary.wiley.com/doi/10.1002/cmdc.202200630 by Mcgill University Library, Wiley Online Library on [01/05/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Research Article
ChemMedChem doi.org/10.1002/cmdc.202200630

AbPanKIII, again highlighting the potential link between these Interactive Tree Of Life (iTOL).[12] The sequence identity matrix for
two enzymes in inhibitor design. the eight PanKIII sequences of interest was obtained from the
Clustal Omega alignment of the sequences.

Conclusion Bioinformatic analysis of Pan binding site residues


The residues within 6.0 Å of the Pan ligand in the PaPanKIII co-
Antimicrobial drug development is fraught with difficulties and crystal structure (PDB: 2F9W)[3b] was identified using Maestro of the
obstacles, especially when the goal is to target more than one Schrödinger software suite (Release 2020-4; Schrödinger, Inc). The
type of pathogen. We developed a tool that can be used to residues were correlated with the corresponding residues in the
facilitate the process of drug design by systematically consider- profile HMM for the PF03309 Pfam family (the signature, as
ing the similarities and differences between the same drug visualised using Skylign).[13] The alignment of the eight PanKIII
sequences of interest was obtained in Meastro using its Biologics
target in different host organisms. MTSA depends on a
Multiple Sequence Viewer tool (Supporting Information Figure 1);
combination of complementary bioinformatic, structural and a this alignment was based on the overlay of the structures of the
kinetic analyses to establish the likelihood that a compound PanKIII of interest. The pseudo-sequence of the Pan binding site
developed to inhibit the enzyme of one organism would also residues was obtained from a ligand contact annotation. The
show activity against another. With the exception of the kinetic phylogram was obtained from the Clustal Omega alignment of the
analyses, all the facets of the MTSA can be performed in silico sequences.
using information from sequence and structural databases. In
practice we found that the in silico analyses proved to have Structural analyses
better predictive value than the experimentally determined
All the in silico structural analyses and calculations were performed
kinetic analysis, most likely due to the comparatively small
using Maestro of the Schrödinger software suite (Release 2020-4;
sample size of the latter, and the need to optimize assay Schrödinger, Inc). The description of the analyses and how they
conditions for individual enzymes. This suggests that MTSA can were performed is given in the main text.
be used to make predictions without a significant investment of
time and effort on experimental studies that may or may not
Kinetic parameter analyses
provide useful feedback.
We applied MTSA to the type III PanKs of several pathogenic The activity of five purified PanKIII enzymes (expression and
bacteria of interest, based on the proven potential of CoA purification conditions are given in Supporting Information) was
biosynthesis as a selective antimicrobial target. The outcome evaluated towards a range of Pan 1 concentrations (2-fold serial
dilution from 200 μM). All reactions were performed in triplicate at
shows the potential utility of MTSA, in that it clearly indicates
25 °C in clear 96-well plates (Greiner Bio-One). The decrease in
that PanKIII enzymes are diverse in regard to the targeted NADH was measured with a Thermo Varioskan multimode plate
binding site, and therefore that an inhibitor of one enzyme reader at 340 nm over a period of 5 min. Each well contained
would not necessarily also show activity against another. 100 mM HEPES (pH 7.5), 1 mM MgCl2 (10 mM for BtPanKIII assays),
However, the MTSA made it possible to organise the enzymes 60 mM NH4Cl, 2.0 mM PEP, 0.5 mM NADH, 5 mM ATP, 2.75 U lactate
dehydrogenase (LDH; Roche), 2.0 U pyruvate kinase (PK; Roche),
into groups that would likely show similar responses, and for
4.5 μg of PanKIII enzyme and the appropriate concentration of
which predictions can be made about the kind of compound substrate in a total volume of 300 μL. Reactions were initiated by
structures each of these groups would respond to. Our the addition of substrate to the other assay components (pre-mixed
experimental testing of these predictions demonstrated that as a master mix). A set of negative controls (milliQ water instead of
while they are not necessarily accurate in all respects, they Pan) was included; the average reading of these wells was used to
provide useful guidance that may reduce the burden of perform a blank subtraction on the data prior to analysis. The
apparent kinetic parameters were obtained by fitting the
experimental design and effort, and may facilitate the discovery
Michaelis Menten equation to the resulting activity profile data
of inhibitors with the desired target product profile. We using non-linear regression analysis in SigmaPlot version 14 (Systat
therefore propose that MTSA could be a useful addition to the Software, Inc). The analysis was repeated with independent batches
toolbox of the antimicrobial drug development medicinal of enzyme to obtain average parameter values.
chemist.
Chemistry

Experimental Section The test compounds used in this study were obtained as follows:
Pan (1), (PantSH, 3), d-panthenol (4) and dL-pantoyltaurine sodium
salt (7) were obtained from Sigma-Aldrich (Merck, KGaA).
MTSA
P-Pan (2)[14] and analogues 5,[15] 6,[1c] 8,[16] 9,[16] 10,[9,17] 11,[18] 12,[18]
and 14[1c] were prepared according to the procedures reported in
Full sequence bioinformatic analysis the indicated references. The synthetic procedures used to prepare
compounds 13,[19] 15, 16,[19] 17[20] and 18 are described in the
The set of UniProt reviewed sequences retrieved for the InterPro
Supporting Information.
entry for Type III pantothenate kinase (Type_III_PanK; IPR004619)
were aligned using Clustal Omega,[10] and the alignment used to
construct an approximately-maximum-likelihood phylogenetic tree
using FastTree 2.1.[11] The unrooted tree was visualized using

ChemMedChem 2023, 18, e202200630 (9 of 10) © 2023 The Authors. ChemMedChem published by Wiley-VCH GmbH
18607187, 2023, 7, Downloaded from https://chemistry-europe.onlinelibrary.wiley.com/doi/10.1002/cmdc.202200630 by Mcgill University Library, Wiley Online Library on [01/05/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Research Article
ChemMedChem doi.org/10.1002/cmdc.202200630

Substrate/inhibitor screening assays Keywords: pantothenate kinase · coenzyme A · antimicrobials ·


drug design · structure-activity relationships
General method
All reactions were performed at 25 °C in clear 384-well plates. The [1] a) C. Spry, K. Kirk, K. J. Saliba, FEMS Microbiol. Rev. 2008, 32 (1), 56–106;
decrease in NADH was measured with a Thermo Varioskan multi- b) W. J. Moolman, M. de Villiers, E. Strauss, Biochem. Soc. Trans. 2014, 42
mode plate reader at 340 nm over a period of 5 min. Each well (4), 1080–1086; c) L. Barnard, K. J. Mostert, W. A. L. van Otterlo, E.
contained 100 mM HEPES (pH 7.5), 10 mM MgCl2, 60 mM NH4Cl, Strauss, ACS Infect. Dis. 2018, 4 (5), 736–743; d) J. Guan, L. Barnard, J.
2.0 mM PEP, 0.3 mM NADH, 7.5 mM ATP, 2.75 U lactate dehydro- Cresson, A. Hoegl, J. H. Chang, E. Strauss, K. Auclair, Bioorg. Med. Chem.
genase, 2.0 U pyruvate kinase (PK), 0.35 μM PanKIII enzyme and the 2018, 26 (22), 5896–5902; e) H. S. Butman, T. J. Kotzé, C. S. Dowd, E.
appropriate concentration of substrate and/or inhibitor in a total Strauss, Front. Cell. Infect. Microbiol. 2020, 10, 605662.
[2] E. Strauss, in Comprehensive Natural Products II Chemistry and Biology,
volume of 50 μL.
Vol. 7 (Eds.: L. Mander, H.-W. Liu), Elsevier, Oxford, 2010, pp. 351–410.
[3] a) L. A. Brand, E. Strauss, J. Biol. Chem. 2005, 280 (21), 20185–20188;
b) B. S. Hong, M. K. Yun, Y. M. Zhang, S. Chohnan, C. O. Rock, S. W.
Substrate screening assays White, S. Jackowski, H. W. Park, R. Leonardi, Structure 2006, 14 (8), 1251–
1261; c) N. I. Nicely, D. Parsonage, C. Paige, G. L. Newton, R. C. Fahey, R.
Wells were loaded with 5 μL of compound at 10 mM (final Leonardi, S. Jackowski, T. C. Mallett, A. Claiborne, Biochemistry 2007, 46
concentration 1 mM) and the reaction was initiated by addition of (11), 3234–3245; d) K. Yang, E. Strauss, C. Huerta, H. Zhang, Biochemistry
the other components (45 μL) in a master mix. A negative control 2008, 47 (5), 1369–1380.
(5 μL of milliQ water) was included; the average reading of these [4] K. Yang, Y. Eyobo, L. A. Brand, D. Martynowski, D. Tomchick, E. Strauss,
wells was used to perform a blank subtraction on the data. All H. Zhang, J. Bacteriol. 2006, 188 (15), 5532–5540.
compounds were evaluated in triplicate, and the average rates [5] E. Tacconelli, E. Carrara, A. Savoldi, S. Harbarth, M. Mendelson, D. L.
Monnet, C. Pulcini, G. Kahlmeter, J. Kluytmans, Y. Carmeli, M. Ouellette,
compared to that of a reference assay containing 30 μM Pan.
K. Outterson, J. Patel, M. Cavaleri, E. M. Cox, C. R. Houchens, M. L.
Grayson, P. Hansen, N. Singh, U. Theuretzbacher, N. Magrini, A. O.
Aboderin, S. S. Al-Abri, N. Awang Jalil, N. Benzonana, S. Bhattacharya,
Inhibitor screening assays A. J. Brink, F. R. Burkert, O. Cars, G. Cornaglia, O. J. Dyar, A. W. Friedrich,
A. C. Gales, S. Gandra, C. G. Giske, D. A. Goff, H. Goossens, T. Gottlieb, M.
Wells were loaded with 5 μL of 300 μM Pan (30 μM final) and 5 μL Guzman Blanco, W. Hryniewicz, D. Kattula, T. Jinks, S. S. Kanj, L. Kerr, M.-
of 3 mM test compound (300 μM final) were loaded into a well and P. Kieny, Y. S. Kim, R. S. Kozlov, J. Labarca, R. Laxminarayan, K. Leder, L.
the reaction was initiated by addition of the other components Leibovici, G. Levy-Hara, J. Littman, S. Malhotra-Kumar, V. Manchanda, L.
(40 μL) in a master mix. A negative control (10 μL of milliQ water) Moja, B. Ndoye, A. Pan, D. L. Paterson, M. Paul, H. Qiu, P. Ramon-Pardo,
was included; the average reading of these wells was used to J. Rodríguez-Baño, M. Sanguinetti, S. Sengupta, M. Sharland, M. Si-
perform a blank subtraction on the data. All compounds were Mehand, L. L. Silver, W. Song, M. Steinbakk, J. Thomsen, G. E. Thwaites,
evaluated in triplicate, and the average rates compared to that of a J. W. M. van der Meer, N. Van Kinh, S. Vega, M. V. Villegas, A. Wechsler-
Fördös, H. F. L. Wertheim, E. Wesangula, N. Woodford, F. O. Yilmaz, A.
reference assay containing only 30 μM Pan.
Zorzet, Lancet Infect. Dis. 2018, 18 (3), 318–327.
[6] A. S. Rowan, N. I. Nicely, N. Cochrane, W. A. Wlassoff, A. Claiborne, C. J.
Hamilton, Org. Biomol. Chem. 2009, 7 (19), 4029–4036.
[7] M. de Villiers, L. Barnard, L. Koekemoer, J. L. Snoep, E. Strauss, FEBS J.
2014, 281, 4731–4753.
Acknowledgements [8] R. A. Laskowski, M. B. Swindells, J. Chem. Inf. Model. 2011, 51 (10), 2778–
2786.
The authors thank Tertius Ras for help with bioinformatic [9] R. van der Westhuyzen, J. C. Hammons, J. L. Meier, S. Dahesh, W. J.
Moolman, S. C. Pelly, V. Nizet, M. D. Burkart, E. Strauss, Chem. Biol. 2012,
analyses, and Dr. Leanne Barnard for assistance in the synthesis of 19 (5), 559–571.
compound 17. Access to the Schrödinger platform was provided [10] F. Sievers, A. Wilm, D. Dineen, T. J. Gibson, K. Karplus, W. Li, R. Lopez, H.
via a license to the Centre for High Performance Computing McWilliam, M. Remmert, J. Söding, J. D. Thompson, D. G. Higgins, Mol.
Syst. Biol. 2011, 7 (1), 539.
(CHPC) of the National Integrated Cyber-infrastructure System [11] M. N. Price, P. S. Dehal, A. P. Arkin, PLoS One 2010, 5 (3), e9490.
(NICIS) of South Africa. LK and SEB received bursary support from [12] I. Letunic, P. Bork, Nucleic Acids Res. 2021, 49 (W1), W293–W296.
the South African National Research Foundation (NRF) and the [13] T. J. Wheeler, J. Clements, R. D. Finn, BMC Bioinf. 2014, 15 (1), 7.
[14] E. Strauss, C. Kinsland, Y. Ge, F. W. McLafferty, T. P. Begley, J. Biol. Chem.
H.B. Thom trust. This work was supported by a grant from the NRF 2001, 276 (17), 13513–13516.
(Grant No. 93430). [15] G. C. Fischer, R. H. Turakhia, C. J. Morrow, J. Org. Chem. 1985, 50 (12),
2011–2019.
[16] M. van Wyk, E. Strauss, Org. Biomol. Chem. 2008, 6 (23), 4348–4355.
[17] R. Domingo, R. van der Westhuyzen, A. R. Hamann, K. J. Mostert, L.
Conflict of Interest Barnard, T. Paquet, E. T. Tjhin, K. J. Saliba, W. A. L. van Otterlo, E. Strauss,
MedChemComm 2019, 10 (12), 2118–2125.
[18] C. Spry, L. Barnard, M. Kok, A. K. Powell, D. Mahesh, E. T. Tjhin, K. J.
The authors declare no conflict of interest. Saliba, E. Strauss, M. de Villiers, ACS Infect. Dis. 2020, 6 (7), 1844–1854.
[19] S. R. Chirapu, C. J. Rotter, E. L. Miller, M. V. Varma, R. L. Dow, M. G. Finn,
Curr. Top. Med. Chem. 2013, 13 (7), 837–842.
[20] V. M. Kopelevich, L. N. Bulanova, V. I. Gunar, Tetrahedron Lett. 1979, 20
Data Availability Statement (40), 3893–3894.

The data that support the findings of this study are available
from the corresponding author upon reasonable request.
Manuscript received: November 20, 2022
Revised manuscript received: February 4, 2023
Accepted manuscript online: February 7, 2023
Version of record online: February 22, 2023

ChemMedChem 2023, 18, e202200630 (10 of 10) © 2023 The Authors. ChemMedChem published by Wiley-VCH GmbH

You might also like