European Biophysics Journal

https://doi.org/10.1007/s00249-019-01346-7

ORIGINAL ARTICLE

NMR model structure of the antimicrobial peptide maximin 3

Silvia Benetti1,2   · Patrick Brendan Timmons1   · Chandralal M. Hewage1 

Received: 5 November 2018 / Revised: 20 December 2018 / Accepted: 2 January 2019

© European Biophysical Societies’ Association 2019

Abstract
Maximin 3 is a 27-residue-long cationic antimicrobial peptide found in the skin secretion and brain of the Chinese red-belly
toad Bombina maxima. The peptide is of biological interest as it possesses anti-HIV activity, not found in the other maximin
peptides, in addition to antimicrobial, antitumor and spermicidal activities. The three-dimensional structure of maximin
3 was obtained in a 50/50% water/2,2,2-trifluoroethanol-d3 mixture using two-dimensional NMR spectroscopy. Maximin
3 was found to adopt an α-helical structure from residue G1 to A22, and a coil structure with a helical propensity in the
C-terminal tail. The peptide is amphipathic, showing a clear separation between polar and hydrophobic residues. Interac-
tions with sodium dodecyl sulfate micelles, a widely used bacterial membrane-mimicking environment, were modeled using
molecular dynamics simulations. The peptide maintained an α-helical conformation, occasionally displaying a flexibility
around residues G9 and G16, which is likely responsible for the peptide’s low haemolytic activity. It is found to preferentially
adopt a position parallel to the micellar surface, establishing a number of hydrophobic and electrostatic interactions with it.

Keywords  Antimicrobial peptide maximin · NMR · AMP modeling

Introduction parasites and viruses (Gordon et al. 2005). Some peptides

Antimicrobial peptides (AMPs), involved in the non-specific
response of the innate immune system (Xia et al. 2018),
are present in all kinds of organisms, from viruses, bacte-
ria and plants (Maróti Gergely et al. 2011; da Silva Pereira
et al. 2018) to vertebrates (Avila 2017) and invertebrates
(El Samak et al. 2018). The majority of these peptides are
found in amphibians, with more than a 1000 reported in the
antimicrobial peptide database, APD3 (https​://aps.unmc.
edu/AP/main.php). Bacterial resistance to antibiotics is the
motivation for the search of new antimicrobial compounds
that can be used without the development of resistance.
AMPs are interesting both as substitutes of or in combina-
tion with antibiotics because of their immediate response
and activity against many pathogens such as bacteria, fungi,
* Chandralal M. Hewage
chandralal.hewage@ucd.ie
1
UCD School of Biomolecular and Biomedical Science, UCD
Centre for Synthesis and Chemical Biology, UCD Conway
Institute, University College Dublin, Belfield, Dublin 4,
Ireland
2
Department of Chemical Sciences, Università Degli Studi Di
Padova, Via Marzolo 1, 35131 Padova, Italy
are also important for wound healing, immune activation and
inflammatory processes (Yang et al. 2002, 2004), or show
other biological activities such as anti-cancer (Deslouches
et al. 2017), anti-HIV (Pace et al. 2017) and spermicidal
(Tanphaichitr et al. 2016).
AMPs have a primary sequence typically ranging between
10 and 50 residues and usually possess no secondary struc-
ture in water. They adopt a secondary structure in the bacte-
rial membrane and in non-polar environments. Secondary
structure can be used to classify antimicrobial peptides.
Most, such as magainin and LL-37, are α-helical. Other
structures include β-sheet, which include human α-defensins,
α–β structures, such as β-defensins, and extended structures,
such as indolicidin. Many AMPs have two common features:
amphipathicity and a global positive charge, thanks to which
they interact with the bacterial cell membranes (Hsiao et al.
2018). Prokaryotic cell membranes are negatively charged
because of their composition; lipopolysaccharides (LPS) are
found in Gram-negative bacterial membrane, and phospho-
lipids and acidic polysaccharides (teichoic acids) in Gram-
positive bacterial one (Beveridge 1999; Weidenmaier and
Peschel 2008). In fact, the interaction between peptide and
prokaryotic cell membrane is electrostatically driven and
increasing the former positive charge results in a higher

13
Vol.:(0123456789)

antibacterial activity (Xie et al. 2017). Charges higher than
a certain value, dependent on the peptide considered, result
in a decrease in antimicrobial activity (Zelezetsky and Tossi
2006).
There are not many known cases of bacterial resistance
(Yeaman and Yount 2003; Perron et al. 2006) because of
the non-specific mechanisms for killing bacteria that usually
involve an interaction with the membrane to disrupt it. There
are three main mechanisms of membrane permeation: car-
pet-like (or detergent-like), barrel-stave and toroidal. They
all lead to leaking of cellular components and dissipation
of the electrical potential. The first step is always the bind-
ing with the membrane surface by electrostatic interaction.
After a threshold concentration is reached, the peptides,
respectively, disrupt the bilayer curvature and disintegrate
the membrane; or they form a pore, with the hydrophobic
residues facing the lipid core region of the bilayer and the
polar ones making up the interior region of the lumen; or
they bend the membrane from the outer towards the inner
leaflet, forming a toroidal pore made up of the lipid head
groups and the peptides (Shai 2002; Brogden 2005). The
threshold concentration required to disrupt the bilayer may
also be dependent on the bacterial target’s ability to repair
damaged cell membranes (Sani et al. 2015). Another impor-
tant feature to consider is their haemolytic activity; some
AMPs are also lytic to mammalian cells. Human erythro-
cytes are considered as representatives of eukaryotic cells
and their membrane is composed of neutral lipids such as
phosphatidylcholine/lecithin and sphingomyelin (Verkleij
et al. 1973; Nouri-Sorkhabi et al. 1996). Some peptides are
toxic to human cells and the interaction with human cells
was suggested to be hydrophobic instead of electrostatic.
Hydrophobicity is, therefore, a key feature for toxicity and
it can be changed to modulate the selectivity, alongside the
net charge (Yin et al. 2012; Son et al. 2013).
Structural studies of AMPs have been carried out in our
laboratory for various amphibian peptides, including XT-7
(Subasinghage et al. 2010), ranatuerin-2CSa (Subasinghage
et al. 2008) and alyteserin-1c (Subasinghage et al. 2011). In
search of a peptide with potential to target many diseases,
our attention was focused on maximin family peptides.
Maximin 3 (GIGGKILSGLKTALKGAAKELASTYLH) is a
27-residue AMP isolated from the skin secretions (Lai et al.
2002) and the brain of the Chinese red-belly toad Bombina
maxima (Liu et al. 2011). Three families of peptides were
isolated from this amphibian: maximins, maximins H (Lai
et al. 2002) and maximins S (Wang et al. 2005). Maximin 3
shows little haemolytic activity and possesses antimicrobial,
antitumor and spermicidal activities like the other maximins,
but a higher anti-HIV activity, making it an interesting anti-
microbial peptide for further study (Lai et al. 2002). In this
report, structural properties of maximin 3 have been studied
using NMR spectroscopy and molecular modeling.
13
European Biophysics Journal
Materials and methods
Materials and NMR sample preparation
3-Trimethylsilyl propionic acid (TSP) and 2,2,2-trifluoro-
ethanol (TFE-d3) of analytical grade were obtained from
Sigma-Aldrich (Ireland).
The peptide maximin 3 (MW =2698  g ­ m ol −1 ,
purity>98%) was purchased from ProteoGenix (Paris).
Peptide (5 mg) was dissolved in 0.6 mL of 50% (v/v) TFE-
d3–H2O solution, resulting in a peptide concentration of
3.09 mM.
NMR spectroscopy
The NMR measurements were performed at 298  K on
a Bruker Avance 600 NMR spectrometer with a 5-mm
inverse probe head at a 1 H resonance frequency of
600.13 MHz.
One-dimensional (1D) and two-dimensional (2D)
NMR spectra were acquired with a relaxation delay of
4.0 s and 1.5 s, respectively. The 2D phase-sensitive total
correlation spectroscopy (TOCSY) (Bax and Davis 1985),
nuclear Overhauser effect spectroscopy (NOESY) (Kumar
et al. 1980) and natural abundance 1H–13C heteronuclear
single-quantum coherence spectroscopy (1H–13C-HSQC)
(John et al. 1992) experiments were performed with an
acquisition time of 0.28 s, 0.57 s and 0.14 s, respectively,
and mixing time of 80 ms for TOCSY and 200  ms for
NOESY. The 1H spectral width was 7.2 kHz for all the 2D
data, and the 13C spectral width was 21.1 kHz. The spectra
were acquired with 8, 16 and 32 transients for each of the
2048, 1024 and 1024 ­t1 increments into 512 complex data
points for the TOCSY, NOESY and HSQC, respectively.
The sine-squared window function was applied for pro-
cessing in all the 2D spectra and the 1H signal of TSP was
used as chemical shift reference. Spectra were processed
with the Bruker TopSpin program, version 2.1 (Bruker
BioSpin, Germany).
Structure calculation
Spectral analysis and NOESY peak integration were
accomplished with NMRFAM-SPARKY software, ver-
sion 3.131 (Lee et al. 2015). Through the calibration of
nuclear Overhauser effect intensity (nOe) versus distance
bond routine of backbone, methyl proton and flexible side
chain classes, distance constraints were obtained from
European Biophysics Journal
NOESY peak volume integrals using CALIBA (Güntert,
Braun and Wüthrich 1991) labeling protons that could not
be stereospecifically assigned as pseudoatoms.
Using the distance constraints obtained that do not
account for fixed and limit-violating distances and weight-
ing them using 1 kJ m ­ ol−1 Å−2 as force constants (­ kNOE),
100 structures were randomly generated using the CYANA
program (Güntert and Buchner 2015). These model struc-
tures were further subjected to 20,000 steps of simulated
annealing and 20,000 of conjugated gradient minimization.
The final step involved 2000 cycles of conjugated gradi-
ent energy minimization on the 20 peptide structures with
lowest target function values, while restraining backbone
atoms with CHARMM22 force field (MacKerell et al. 2004)
in NAMD, version 2.12 (Phillips et al. 2005). The struc-
tural analysis of the energy-minimized structures was con-
ducted using VMD (visual molecular dynamics), version
1.9.3 (Humphrey et al. 1996). PROCHECK (Laskowski et al.
1993) and wwPDB (Berman et al. 2003) were used to deter-
mine the stereochemical quality and the structural statistics
of the 20 final structural models.
Molecular dynamics simulation
A peptide–SDS micelle system was constructed using VMD
(Humphrey et al. 1996) by aligning the centers of mass of
the medoid maximin 3 structure and the SDS micelle. Coor-
dinates of the SDS micelle were obtained from Jakobtorwei-
hen et al. (Jakobtorweihen et al. 2013). The system was then
solvated with TIP3P water (Jorgensen et al. 1983). Finally,
chloride ions were added to neutralize the system.
The system was minimized and simulated using the
CHARMM36 all-atom force field (MacKerell et al. 1998,
2004) within NAMD version 2.12 (Phillips et al. 2005). All
the calculations took place in the NPT ensemble, utilizing
the Langevin piston Nose–Hoover method (Martyna et al.
1994; Feller et al. 1995) with periodic boundary conditions.
Bonds to hydrogen atoms were constrained to a fixed value
with the SHAKE algorithm (Ryckaert et al. 1977). The inte-
gration time step was set to 2 fs. Long-range non-bonded
interactions were calculated up to a switching distance of 8.5
Å, beyond which a smooth switching function truncated the
energy to a cutoff of 11 Å. The computation of long-range
electrostatic interactions was performed for every step by the
PME method (Darden et al. 1993), while the non-bonded list
was updated every step.
The system was minimized for 2000 conjugate gradi-
ent steps with peptide backbone atoms fixed, and 2000
steps with the α-carbons restrained. The system was
heated to 310  K over 6000 steps and then the volume
equilibrated with the Langevin piston at 1 atm for 24,000
steps, followed by another 24,000 steps without restraints.
Finally, the system was simulated for 81 ns.
Results and discussion
Conformational analysis by NMR
The structure of maximin 3 was determined using 2D
NMR spectroscopy in TFE-d3–water, which is known to
induce formation of secondary structure by enhancing the
α-helical character (Shiraki et al. 1995). The collected
spectra showed well-dispersed peaks, indicating the pres-
ence of an ordered secondary structure. The TOCSY and
1
H–13C HSQC spectra were used to identify the individual
residue spin systems. Beginning with the peptide’s three
unique residues, E20, Y25 and H27, sequence-specific
resonance assignments were carried out using the ­HN–HN
and the ­H N–H α regions (Figs. 1 and 2, respectively) of
the NOESY spectrum, resulting in the characterization of
all the individual spin systems. The chemical shift of the
amide proton could not be found for G1 because of chemi-
cal exchange.
Table 1 lists all the 1H chemical shifts of maximin 3.
The intermolecular nOe connectivities are clearly identi-
fied in Fig. 3, with line thickness proportional to their
intensity. There are many d­ αN(i, i + 3), ­dαβ(i, i + 3), ­dαN(i,
i + 4) and ­d αβ(i, i + 4) connectivities throughout, which
are typical of an α-helical secondary structure, and some
­dαN(i, i + 2) connectivities in the N-terminal tail region,
which may be suggestive of a tighter helix. There is a
notable lack of d­ αN(i, i + 3) and d­ αN(i, i + 4) connectivities
spanning G16, which indicates the presence of a kink at
that position (Fig. 4).
Molecular modeling
NOE cross-peak intensities were converted into distance
restraints, from which 100 random structures were obtained,
of which the 20 with the lowest target function values were
further energy-minimized. The structural statistics of the cal-
culated models are summarized in Table 2. As illustrated in
Figs. 5 and 6, maximin 3 possesses two α-helical segments,
separated by a kink at G16. Helix I extends from G1–K15,
and helix II extends from G16–A22, with S23 incorporated
into the helix in some of the structures. Helix I displays
some curvature, centerd on G9, and is more ordered than
helix II (G16–A22), with a smaller RMSD for both the back-
bone and the heavy atoms. The C-terminal tail region shows
disordered character.
13
 European Biophysics Journal

Fig. 1  Amide region of the

200 ms NOESY spectrum
of maximin 3 in 50% TFE-
d3–H2O mixed solvent system
with ­dNN(i,i + 1) connectivities
labeled (TOPSPIN)

Fig. 2  Fingerprint region of the

200 ms NOESY spectrum of
maximin 3 in 50% TFE-d3–H2O
mixed solvent system with
backbone connectivities labeled
(TOPSPIN)

13
European Biophysics Journal

Table 1  1H chemical shifts Amino acid NH Hα Hβ Other protons

(ppm) identified for every
residue of maximin 3 in 50% Gly1 3.979, 4.006
TFE-d3/H2O
Ile2 8.505 4.187 1.920 γ1 1.285, 1.518; γ2 0.974; δ 0.917
Gly3 8.461 3.864, 3.919
Gly4 8.163 3.899
Lys5 7.967 4.159 1.977 γ 1.476, 1.554; δ 1.705
Ile6 7.955 3.859 2.000 γ1 1.215; γ2 0.934; δ 0.873
Leu7 8.170 4.169 1.739, 1.787 γ 1.644; δ 0.903
Ser8 8.046 4.198 3.998, 4.037
Gly9 8.110 3.927
Leu10 8.297 4.195 1.746, 1.817 γ 1.554; δ 0.877
Lys11 8.326 3.967 1.912, 1.971 γ 1.451; δ 1.705; ε 2.931
Thr12 7.856 4.396 3.942
Ala13 8.145 4.172 1.554
Leu14 8.470 4.164 1.891 γ 1.573; δ 0.871
Lys15 7.993 4.089 1.946; 1.999 γ 1.495; δ 1.675, 1.732; ε 2.976
Gly16 8.122 3.910
Ala17 8.138 4.211 1.525
Ala18 8.198 4.082 1.529
Lys19 7.894 4.069 1.976 γ 1.470, 1.577; δ 1.721; ε 2.991
Glu20 8.062 4.124 2.212, 2.249 γ 2.491
Leu21 8.462 4.166 1.863 γ 1.663, 1.770; δ 0.914
Ala22 8.308 4.149 1.524
Ser23 7.945 4.315 4.013, 4.052
Thr24 7.741 4.168 4.175 γ 1.108
Tyr25 7.788 4.562 2.975, 3.130 δ1 7.164; δ2 6.792; ε1 7.149; ε2 6.777
Leu26 7.817 4.324 1.537, 1.696 δ 1.619; γ 0.866, 0.904
His27 7.703 4.547 3.202, 3.316 δ2 7.225; ε1 8.499

Δδ values are found for shorter hydrogen bonds and vice

Fig. 3  Short and medium range connectivities for maximin 3 in 50%
TFE-d3–H2O (CYANA)
Chemical shift analysis
The chemical shift deviation of ­HN protons was calculated
(Fig. 4a) as the difference between the observed chemical
shift and the random coil chemical shift (Δδ=δ obs-δ rc).
Amide proton chemical shifts are connected with hydrogen
bond length (Wagner et al. 1983) shown in Fig. 4b: higher
versa. The amide proton chemical shifts have the expected
periodicity of 3–4 (Kuntz et al. 1991) and, as expected,
the most negative Δδ values are those of hydrophilic resi-
dues, K5, S8, T12, K15 and K19, and the most positive
Δδ values are the ones of hydrophobic residues, I6, L7,
L10, L14, A18, L21 and A22 (Zhou et al. 1992). K11,
which has a higher than expected chemical shift, is an
exception to this trend, meaning that the amide proton of
K11 has a shorter-than-expected hydrogen bond with the
carbonyl group of S7. This finding is consistent with there
being a region of curvature centerd on G9, thereby bring-
ing the S7 and K11 residues closer together. The hydrogen
bond lengths in the C-terminal tail region (T24-H27) are
outside those that would be typical of an α-helix. This is
confirmed by the investigation of the torsion angles found
in the calculated structural models; the (φ,ψ) angles do
not conform to those expected of residues in an α-helix.
The data presented in this study demonstrate that maxi-
min 3 possesses an α-helical structure between residues
G1–A22/S23, while the C-terminal tail residues are in
a coil conformation. Similar patterns were observed for

13
 European Biophysics Journal

Table 2  Average structural statistics of the 20 structural models of other natural peptides, including GO 1–1 (Suh et al. 1996),
maximin 3 maximin 4 (Toke et al. 2011) and the lasioglossin series of
NOEs peptides (Čeřovský et al. 2009) (Fig. 5).

Intraresidue 165
Interresidue, sequential 140
Molecular dynamics
Interresidue, non-sequential 111
Ensemble RMSD values, Å, SA ± SDa
The NMR-determined structural model of maximin 3 was
Backbone
used for a molecular dynamics simulation in SDS micelles
 All residues 0.465 ± 0.140
to model its orientation in and interactions with the bacterial
 Helix I (G1–K15) 0.356 ± 0.162
membrane-mimicking environment. During the simulation,
 Helix II (G16–A22) 0.474 ± 0.191
the peptide was found to adopt a completely α-helical con-
 Tail (T24–H27) 0.690 ± 0.205
formation, from G1 to H27. The adoption of a more ordered
Heavy atoms
secondary structure, especially in the C-terminal tail region,
 All residues 1.030 ± 0.192
can be accounted for by the electrostatic interactions present
 Helix I (G1–K15) 0.781 ± 0.208
in SDS complementing the hydrophobic forces that are also
 Helix II (G16–A22) 0.935 ± 0.341
present in TFE (Roland Montserret et al. 2000).
 Tail (T24–H27) 1.573 ± 0.266
Interestingly, in the micellar environment, the peptide
Ramachandran plot a­ nalysisb
exhibits marked flexibility around G9 and G16, at times dis-
 Residues in most favored regions 85.49%
playing a curvature there. The glycine-induced flexibility is
 Residues in additionally allowed regions 14.51%
likely responsible for the peptide’s low haemolytic activity;
 Residues in generously allowed regions 0%
Gly to Ala substitutions which reduce conformational flex-
 Residues in disallowed regions 0%
ibility and promote α-helical character have previously been
Average energies (kcal/mol)
found to increase peptides’ non-specific interactions with
 Ebond 15.811
zwitterionic cell membranes and potentially increase toxicity
 Eangle 64.693
against eukaryotic cells (Idiong et al. 2011).
 Eimprp 4.096
During the simulation, the peptide moved from its initial
 EVdW − 49.921
position at the center of the micelle to a pose parallel to
 Edihed 69.485
the micelle surface, as illustrated in Fig. 6. The amphip-
 Etotal 4.692
athic nature of the peptide allows the side chains of all the
a
 RMSD values from VMD non-polar residues to be involved in hydrophobic interac-
b
 Based on PROCHECK tions with the micelle interior, while the hydrophilic amino
acids face the solvent, and the positively charged lysine and

Fig. 4  a Amide chemical shift deviation plot of maximin 3 in 50% bond distance as calculated from the chemical shift deviations (Δδ =
TFE-d3–H2O. The observed H ­ N chemical shift was compared with 19.2dN−3–2.3) (Wagner et al. 1983) (ORIGIN) 
the random coil chemical shift standards (Δδ=δobs-δrc). b Hydrogen

13
European Biophysics Journal
Fig. 5  a Solution structure of maximin 3 in 50% TFE-d3–H2O show-
ing the best fit superposition of 20 backbone conformations. b Rib-
bon representation of the medoid solution conformation showing the
amphipathic helical motif, with the hydrophobic and polar faces on
the left and right, respectively (VMD)  the dynamic N-terminal helix of maximin 4 was proposed
Fig. 6  Final pose of maximin 3 in a simulated SDS micelle. Lysine
and histidine side residues are shown interacting with the negatively
charged sulfate headgroups. Positively charged residues are colored
in blue, negatively charged residues colored in red, polar residues
colored in purple, glycine residues colored in orange and hydrophobic
residues colored in gray (VMD)  in the C-terminal.
histidine residues interact with the negatively charged sulfate
headgroups.
Structural comparison between maximin 3
and maximin 4
Maximin 4 is the only other maximin whose structure has
been determined (Toke et al. 2011), and the only maximin
which has been investigated for potential clinical applica-
tions, namely by incorporation into antibacterial hydrogels
(Laverty et al. 2012). It has a helix–kink–helix configuration,
both in methanol and in negatively charged SDS micelles,
with an interhelical angle of 70–90°. While maximin 3 also
possesses a kink, the interhelical angle is comparatively
minimal at 6°. Toke and co-workers attribute part of the
stabilization of the maximin 4 kink to hydrophobic contacts
between the residues A12, A13 on helix I and L17 on helix
II. This stabilization is not present in maximin 3 as it pos-
sesses a polar residue, threonine, in position 12 and a residue
with a shorter side chain, alanine, in position 17.
It was also suggested that if maximin 4 was a linear, rigid
α-helix, there would not be a well-defined polar face. The
presence of the hinge allows the formation of two units:
one more hydrophobic and the other more polar. Maximin
3, on the other hand, possesses a well-defined distinction
between the polar and hydrophobic faces, which allows it
to lie parallel to the negatively charged SDS micelle surface
with a comparatively minimal interhelical angle. Meanwhile,
to insert into the hydrophobic core of the membrane, while
the more polar C-terminal helix was found to assume a tilted
orientation and be in contact with the lipid headgroups. The
mechanism of membrane disruption was, thus, proposed to
be funnel-like, with pores formed by peptides aligned this
way (Toke et al. 2011). The peptides’ different orientation in
prokaryotic cell membranes may be explained by the helix
I of maximin 3 having a larger polar angle, 116°, compared
to 45° in maximin 4—a difference caused by maximin 3
having the polar residues lysine and threonine at positions 5
and 12, respectively.
Biology
An array of other peptides in the maximin family has been
characterized, although none have had their three-dimen-
sional structures determined. There are two features that dif-
ferentiate maximin 3 from most other peptides of the maxi-
min family: it terminates with a histidine amino acid and
not with an asparagine or a glutamine, and is not amidated
13

The histidine termination of maximin 3 has been pro-
posed to be correlated to its anti-HIV activity, found only
for this maximin (Lai et al. 2002). Other maximin peptides
terminating with a histidine have been discovered since then
in the brain of the same amphibian (Liu et al. 2011), but
they have not been tested for anti-HIV activity. While no
direct correlation had been found between amidation and
efficiency/selectivity of AMPs, it was demonstrated that
peptides with a low hydrophobicity gradient, that lay paral-
lel to the membrane surface and disrupt it by the carpet-like
mechanism, increase their toxicity when their C termini are
amidated (Dennison et al. 2009).
Maximin family residue substitution analysis
Maximin 32 differs from maximin 3 by only one amino acid;
it has a glycine in position 8 instead of serine. This substitu-
tion results in a decrease in activity against Gram-positive
and Gram-negative bacteria, and yeast (Lai et al. 2002; Liu
et al. 2011). The change can be attributed to either glycine’s
property of being a helix breaker, or a reduction in the polar-
ity of the polar face. Two substitutions in maximin 32, K11R
and S23A, make maximin 45 as active against bacteria as
maximin 3. The restored activity of maximin 45 could come
either from R11 or A23. Arginine and lysine side chains are
similarly long and charged, while alanine and serine have
different hydrophobicity. There are two cases in the maximin
family in which there is a single substitution at position 23,
S23F; this leads from maximin 5 to maximin 42 and from
maximin 28 to maximin 39. In both cases, the substitution
of serine with phenylalanine results in reduced antibacterial
activity, which indicates the importance of having a polar
residue at this position.
Comparison of maximin 3 and magainin 2
One of the most studied antimicrobial peptides is magainin
2. It can be found in the skin secretion of the African clawed
frog Xenopus laevis, and like maximin 3, it has a net charge
of +3. It has antibacterial, antiviral, antifungal, antipara-
sitic, spermicidal, antimalarial, and antitumor activities and
a selective toxicity (Zasloff 1987; Wojcik et al. 2000; Albiol
Matanic and Castilla 2004; Lehmann et al. 2006; Mor 2009;
Sinha et al. 2016). Comparing the antimicrobial activity
against the same microorganisms as maximin 3, the latter
has lower MICs, while both show little haemolytic activity.
Comparing the peptides’ primary structures, the first
apparent difference is the length: magainin 2 has 23 amino
acids, making it 4 residues shorter than maximin 3. Align-
ing the two primary structures (Fig. 7), the similarity is low,
but it is pertinent to note that glutamate occurs at equivalent
13
European Biophysics Journal
Fig. 7  Multiple sequence alignment of selected maximin peptides and
magainin 2 (CLUSTAL)
positions in both peptides, E19 for magainin 2 and E23
for maximin 3. As for the secondary structure, they both
are α-helices apart from the termini (Gesell et al. 1997).
Magainin 2 is known for dimerizing in an antiparallel ori-
entation on binding to phospholipid bilayers (Wakamatsu
et al. 2002). The dimers are assumed to be stabilized by
dipole–dipole interactions, aromatic–aromatic interactions
between the side chains of F5 and F12/F16 of the other helix,
and electrostatic interactions between E19 and positively
charged residues found near the N terminus (Wakamatsu
et al. 2002). Considering the similarities found between
these peptides, maximin 3 could also dimerize, aided by
ionic or π-cation interactions between Glu/Tyr and positively
charged residues either at or near the N or C terminus.
In conclusion, this study is the first to determine the
three-dimensional structure of maximin 3 and underline the
importance of its α-helical structure and amphipathicity for
the bioactivity of this peptide. It was found that the peptide
adopts a helix–kink–helix conformation, followed by a dis-
ordered C-terminal tail. Elucidation of the solution structure
of maximin 3 will facilitate further studies to determine its
mechanism of action, while the residue substitution analysis
may lead to design of novel analogs containing additional
amino acid substitutions with improved biological activity
and therapeutic indices.
Acknowledgements  CH is grateful to John O’Brien and Manuel
Ruether at Trinity College Dublin for NMR facilities, University Col-
lege Dublin for a Research Scholarship to PBT, and ICHEC for access
to supercomputer facilities. SB and PBT are joint first authors of this
work. The solution structure of maximin 3 was deposited to the PDB
at the RSCB with deposition code 6HZ2.
References
Albiol Matanic VC, Castilla V (2004) Antiviral activity of antimi-
crobial cationic peptides against Junin virus and herpes sim-
plex virus. Int J Antimicrob Agents 23:382–389. https​://doi.
org/10.1016/j.ijant​imica​g.2003.07.022
Avila EE (2017) Functions of antimicrobial peptides in vertebrates.
Curr Protein Pept Sci. https​://doi.org/10.2174/13892​03717​66616​
08131​62629​
European Biophysics Journal
Bax A, Davis DG (1985) MLEV-17-based two-dimensional homonu-
clear magnetization transfer spectroscopy. J Magn Reson 65:355–
360. https​://doi.org/10.1016/0022-2364(85)90018​-6
Berman H, Henrick K, Nakamura H (2003) Announcing the worldwide
Protein Data Bank. Nat Struct Mol Biol 10:980–980. https​://doi.
org/10.1038/nsb12​03-980
Beveridge TJ (1999) Structures of gram-negative cell walls and their
derived membrane vesicles. J Bacteriol 181:4725–4733
Brogden KA (2005) Antimicrobial peptides: pore formers or metabolic
inhibitors in bacteria? Nat. Rev. Microbiol. 3:238–250
Čeřovský V, Buděšínsky M, Hovorka O et al (2009) Lasioglossins:
three novel antimicrobial peptides from the venom of the eusocial
bee Lasioglossum laticeps (Hymenoptera: Halictidae). ChemBio-
Chem 10:2089–2099. https​://doi.org/10.1002/cbic.20090​0133
da Silva Pereira L, do Nascimento VV, de Fátima Ferreira Ribeiro S,
et al (2018) Characterization of Capsicum annuum L. leaf and
root antimicrobial peptides: antimicrobial activity against phy-
topathogenic microorganisms. Acta Physiol Plant 40:107. https​
://doi.org/10.1007/s1173​8-018-2685-9
Darden T, York D, Pedersen L (1993) Particle mesh Ewald: An N
⋅log( N ) method for Ewald sums in large systems. J Chem Phys
98:10089–10092. https​://doi.org/10.1063/1.46439​7
Dennison SR, Harris F, Bhatt T et al (2009) The effect of C-terminal
amidation on the efficacy and selectivity of antimicrobial and
anticancer peptides. Mol Cell Biochem 332:43–50. https​://doi.
org/10.1007/s1101​0-009-0172-8
Deslouches B, Di YP, Deslouches B et al (2017) Antimicrobial pep-
tides with selective antitumor mechanisms: prospect for anticancer
applications. Oncotarget 8:46635–46651. https:​ //doi.org/10.18632​
/oncot​arget​.16743​
Feller SE, Zhang Y, Pastor RW, Brooks BR (1995) Constant pressure
molecular dynamics simulation: the Langevin piston method. J
Chem Phys 103:4613–4621. https​://doi.org/10.1063/1.47064​8
Gesell J, Zasloff M, Opella SJ (1997) Two-dimensional 1H NMR
experiments show that the 23-residue magainin antibiotic peptide
is an α-helix in dodecylphosphocholine micelles, sodium dode-
cylsulfate micelles, and trifluoroethanol/water solution. J Biomol
NMR 9:127–135. https​://doi.org/10.1023/A:10186​98002​314
Gordon YJ, Romanowski EG, McDermott AM (2005) A review of anti-
microbial peptides and their therapeutic potential as anti-infective
drugs. Curr Eye Res 30:505–515. https​://doi.org/10.1080/02713​
68059​09686​37
Güntert P, Buchner L (2015) Combined automated NOE assignment
and structure calculation with CYANA. J Biomol NMR 62:453–
471. https​://doi.org/10.1007/s1085​8-015-9924-9
Güntert P, Braun W, Wüthrich K (1991) Efficient computation of three-
dimensional protein structures in solution from nuclear magnetic
resonance data using the program DIANA and the supporting
programs CALIBA, HABAS and GLOMSA. J Mol Biol 217:517–
530. https​://doi.org/10.1016/0022-2836(91)90754​-T
Hsiao Y-W, Hedström M, Losasso V et al (2018) Cooperative modes of
action of antimicrobial peptides characterized with atomistic sim-
ulations: a study on cecropin B. J Phys Chem B 122:5908–5921.
https​://doi.org/10.1021/acs.jpcb.8b019​57
Humphrey W, Dalke A, Schulten K (1996) VMD: Visual molecular
dynamics. J Mol Graph 14:33–38. https​://doi.org/10.1016/0263-
7855(96)00018​-5
Idiong G, Won A, Ruscito A et al (2011) Investigating the effect of a
single glycine to alanine substitution on interactions of antimi-
crobial peptide latarcin 2a with a lipid membrane. Eur Biophys J
40:1087–1100. https​://doi.org/10.1007/s0024​9-011-0726-z
Jakobtorweihen S, Ingram T, Smirnova I (2013) Combination of COS-
MOmic and molecular dynamics simulations for the calculation of
membrane-water partition coefficients. J Comput Chem 34:1332–
1340. https​://doi.org/10.1002/jcc.23262​
John BK, Plant D, Webb P, Hurd RE (1992) Effective combina-
tion of gradients and crafted RF pulses for water suppression
in biological samples. J Magn Reson 98:200–206. https​://doi.
org/10.1016/0022-2364(92)90125​-Q
Jorgensen WL, Chandrasekhar J, Madura JD et al (1983) Comparison
of simple potential functions for simulating liquid water. J Chem
Phys 79:926–935. https​://doi.org/10.1063/1.44586​9
Kumar A, Ernst RR, Wüthrich K (1980) A two-dimensional nuclear
overhauser enhancement (2D NOE) experiment for the elucidation
of complete proton-proton cross-relaxation networks in biological
macromolecules. Biochem Biophys Res Commun 95:1–6. https:​ //
doi.org/10.1016/0006-291X(80)90695​-6
Kuntz ID, Kosen PA, Craig EC (1991) Amide chemical shifts in many
helices in peptides and proteins are periodic. J Am Chem Soc
113:1406–1408. https​://doi.org/10.1021/ja000​04a05​0
Lai R, Zheng YT, Shen JH et al (2002) Antimicrobial peptides from
skin secretions of Chinese red belly toad Bombina maxima. Pep-
tides 23:427–435. https​://doi.org/10.1016/S0196​-9781(01)00641​
-6
Laskowski RA, MacArthur MW, Moss DS et al (1993) PROCHECK:
a program to check the stereochemical quality of protein struc-
tures. J Appl Crystallogr 26:283–291. https​://doi.org/10.1107/
S0021​88989​20099​44
Laverty G, Gorman SP, Gilmore BF (2012) Antimicrobial peptide
incorporated poly(2-hydroxyethyl methacrylate) hydrogels for the
prevention of Staphylococcus epidermidis-associated biomaterial
infections. J Biomed Mater Res Part A 100A:1803–1814. https​://
doi.org/10.1002/jbm.a.34132​
Lee W, Tonelli M, Markley JL (2015) NMRFAM-SPARKY: Enhanced
software for biomolecular NMR spectroscopy. Bioinformatics
31:1325–1327. https​://doi.org/10.1093/bioin​forma​tics/btu83​0
Lehmann J, Retz M, Sidhu SS et al (2006) Antitumor activity of the
antimicrobial peptide magainin II against bladder cancer cell
lines. Eur Urol 50:141–147. https​://doi.org/10.1016/J.EURUR​
O.2005.12.043
Liu R, Liu H, Ma Y et al (2011) There are abundant antimicrobial
peptides in brains of two kinds of Bombina Toads. J Proteome Res
10:1806–1815. https​://doi.org/10.1021/pr101​285n
MacKerell AD, Bashford D, Bellott M et al (1998) All-atom empirical
potential for molecular modeling and dynamics studies of pro-
teins. J Phys Chem B 102:3586–3616. https​://doi.org/10.1021/
jp973​084f
MacKerell AD, Feig M, Brooks CL (2004) Extending the treatment
of backbone energetics in protein force fields: limitations of gas-
phase quantum mechanics in reproducing protein conformational
distributions in molecular dynamics simulation. J Comput Chem
25:1400–1415. https​://doi.org/10.1002/jcc.20065​
Maróti Gergely G, Kereszt A, Kondorosi É, Mergaert P (2011) Natural
roles of antimicrobial peptides in microbes, plants and animals.
Res Microbiol 162:363–374. https​://doi.org/10.1016/j.resmi​
c.2011.02.005
Martyna GJ, Tobias DJ, Klein ML (1994) Constant pressure molecular
dynamics algorithms. J Chem Phys 101:4177–4189. https​://doi.
org/10.1063/1.46746​8
Montserret R, McLeish MJ, Böckmann A, et al (2000) Involvement
of electrostatic interactions in the mechanism of peptide folding
induced by sodium dodecyl sulfate binding. Biochemistry. https​
://doi.org/10.1021/BI000​208X
Mor A (2009) Multifunctional host defense peptides: antipara-
sitic activities. FEBS J 276:6474–6482. https​://doi.org/10.111
1/j.1742-4658.2009.07358​.x
Nouri-Sorkhabi MH, Wright LC, Sullivan DR, Kuchel PW (1996)
Quantitative31P nuclear magnetic resonance analysis of the
phospholipids of erythrocyte membranes using detergent. Lipids
31:765–770. https​://doi.org/10.1007/BF025​22893​
13

Pace BT, Lackner AA, Porter E, Pahar B (2017) The role of defensins
in HIV pathogenesis. Mediat Inflamm 2017:1–12. https​://doi.
org/10.1155/2017/51869​04
Perron GG, Zasloff M, Bell G (2006) Experimental evolution of resist-
ance to an antimicrobial peptide. Proc Biol Sci 273:251–256. https​
://doi.org/10.1098/rspb.2005.3301
Phillips JC, Braun R, Wang W et al (2005) Scalable molecular dynam-
ics with NAMD. J Comput Chem 26:1781–1802. https​://doi.
org/10.1002/jcc.20289​
Ryckaert J-P, Ciccotti G, Berendsen HJ (1977) Numerical integration
of the cartesian equations of motion of a system with constraints:
molecular dynamics of n-alkanes. J Comput Phys 23:327–341.
https​://doi.org/10.1016/0021-9991(77)90098​-5
El Samak M, Solyman SM, Hanora A (2018) Antimicrobial activity of
bacteria isolated from Red Sea marine invertebrates. Biotechnol
Rep. https​://doi.org/10.1016/J.BTRE.2018.E0027​5
Sani M-A, Henriques ST, Weber D, Separovic F (2015) Bacteria may
cope differently from similar membrane damage caused by the
Australian tree frog antimicrobial peptide maculatin 1.1. J Biol
Chem 290:19853–19862. https:​ //doi.org/10.1074/jbc.M115.64326​
2 enhances spermicidal effects of magainin-2-amide. Contracep-
Shai Y (2002) Mode of action of membrane active antimicrobial pep-
tides. Biopolym Pept Sci Sect 66:236–248
Shiraki K, Nishikawa K, Goto Y (1995) Trifluoroethanol-induced sta-
bilization of the α-helical structure of β-lactoglobulin: implica-
tion for non-hierarchical protein folding. J Mol Biol 245:180–194.
https​://doi.org/10.1006/JMBI.1994.0015
Sinha S, Singh A, Medhi B, Sehgal R (2016) Systematic review: insight
into antimalarial peptide. Int J Pept Res Ther 22:325–340. https​://
doi.org/10.1007/s1098​9-016-9512-1
Son M, Lee Y, Hwang H et al (2013) Disruption of interactions between
hydrophobic residues on nonpolar faces is a key determinant in
decreasing hemolysis and increasing antimicrobial activities of
α-helical amphipathic peptides. ChemMedChem 8:1638–1642.
https​://doi.org/10.1002/cmdc.20130​0264
Subasinghage AP, Conlon JM, Hewage CM (2008) Conformational
analysis of the broad-spectrum antibacterial peptide, ranatuerin-
2CSa: identification of a full length helix-turn-helix motif. Bio-
chim Biophys Acta Proteins Proteomics 1784:924–929. https​://
doi.org/10.1016/j.bbapa​p.2008.02.019
Subasinghage AP, Conlon JM, Hewage CM (2010) Development of
potent anti-infective agents from Silurana tropicalis: conforma-
tional analysis of the amphipathic, alpha-helical antimicrobial
peptide XT-7 and its non-haemolytic analogue [G4K]XT-7. Bio-
chim Biophys Acta Proteins Proteom 1804:1020–1028. https​://
doi.org/10.1016/J.BBAPA​P.2010.01.015
Subasinghage AP, O’Flynn D, Conlon JM, Hewage CM (2011) Con-
formational and membrane interaction studies of the antimicro-
bial peptide alyteserin-1c and its analogue [E4K]alyteserin-1c.
Biochim Biophys Acta Biomembr 1808:1975–1984. https​://doi.
org/10.1016/j.bbame​m.2011.04.012
Suh J-Y, Lee K-H, Chi S-W et al (1996) Unusually stable helical kink
in the antimicrobial peptide—a derivative of gaegurin. FEBS Lett
392:309–312. https​://doi.org/10.1016/0014-5793(96)00840​-X
Tanphaichitr N, Srakaew N, Alonzi R et al (2016) Potential use of
antimicrobial peptides as vaginal spermicides/microbicides. Phar-
maceuticals 9:13. https​://doi.org/10.3390/ph901​0013
Toke O, Bánóczi Z, Király P et al (2011) A kinked antimicrobial pep-
tide from Bombina maxima. I. Three-dimensional structure deter-
mined by NMR in membrane-mimicking environments. Eur Bio-
phys J 40:447–462. https​://doi.org/10.1007/s0024​9-010-0657-0
13
European Biophysics Journal
Verkleij A, Zwaal RF, Roelofsen B et al (1973) The asymmetric dis-
tribution of phospholipids in the human red cell membrane. A
combined study using phospholipases and freeze-etch electron
microscopy. Biochim Biophys Acta Biomembr 323:178–193.
https​://doi.org/10.1016/0005-2736(73)90143​-0
Wagner G, Pardi A, Wuethrich K (1983) Hydrogen bond length
and proton NMR chemical shifts in proteins. J Am Chem Soc
105:5948–5949. https​://doi.org/10.1021/ja003​56a05​6
Wakamatsu K, Takeda A, Tachi T, Matsuzaki K (2002) Dimer struc-
ture of magainin 2 bound to phospholipid vesicles. Biopolymers
64:314–327. https​://doi.org/10.1002/bip.10198​
Wang T, Zhang J, Shen JH et al (2005) Maximins S, a novel group
of antimicrobial peptides from toad Bombina maxima. Biochem
Biophys Res Commun 327:945–951. https​://doi.org/10.1016/j.
bbrc.2004.12.094
Weidenmaier C, Peschel A (2008) Teichoic acids and related cell-wall
glycopolymers in Gram-positive physiology and host interac-
tions. Nat Rev Microbiol 6:276–287. https​://doi.org/10.1038/
nrmic​ro186​1
Wojcik C, Sawicki W, Marianowski P et  al (2000) Cyclodextrin
tion 62:99–103. https​://doi.org/10.1016/S0010​-7824(00)00143​-8
Xia X, Cheng L, Zhang S et al (2018) The role of natural antimicro-
bial peptides during infection and chronic inflammation. Antonie
Van Leeuwenhoek 111:5–26. https​: //doi.org/10.1007/s1048​
2-017-0929-0
Xie J, Zhao Q, Li S et al (2017) Novel antimicrobial peptide CPF-C1
analogs with superior stabilities and activities against multidrug-
resistant bacteria. Chem Biol Drug Des 90:690–702. https​://doi.
org/10.1111/cbdd.12988​
Yang D, Biragyn A, Kwak LW, Oppenheim JJ (2002) Mammalian
defensins in immunity: more than just microbicidal. Trends Immu-
nol 23:291–296. https:​ //doi.org/10.1016/S1471-​ 4906(02)02246-​ 9
Yang D, Biragyn A, Hoover DM et al (2004) Multiple roles of anti-
microbial defensins, cathelicidins, and eosinophil-derived neuro-
toxin in host defense. Annu Rev Immunol 22:181–215. https:​ //doi.
org/10.1146/annur​ev.immun​ol.22.01270​3.10460​3
Yeaman MR, Yount NY (2003) Mechanisms of antimicrobial pep-
tide action and resistance. Pharmacol Rev 55:27–55. https​://doi.
org/10.1124/pr.55.1.2
Yin LM, Edwards MA, Li J et al (2012) Roles of hydrophobicity and
charge distribution of cationic antimicrobial peptides in peptide–
membrane interactions. J Biol Chem 287:7738–7745. https​://doi.
org/10.1074/jbc.M111.30360​2
Zasloff M (1987) Magainins, a class of antimicrobial peptides from
Xenopus skin: isolation, characterization of two active forms, and
partial cDNA sequence of a precursor. Proc Natl Acad Sci USA
84:5449–5453. https​://doi.org/10.1073/PNAS.84.15.5449
Zelezetsky I, Tossi A (2006) Alpha-helical antimicrobial peptides-
Using a sequence template to guide structure-activity relationship
studies. Biochim. Biophys. Acta - Biomembr. 1758:1436–1449
Zhou NE, Zhu BY, Sykes BD, Hodges RS (1992) Relationship between
amide proton chemical shifts and hydrogen bonding in amphip-
athic.alpha.-helical peptides. J Am Chem Soc 114:4320–4326.
https​://doi.org/10.1021/ja000​37a04​2
Publisher’s Note Springer Nature remains neutral with regard to
jurisdictional claims in published maps and institutional affiliations.