Dissociation of antimicrobial and hemolytic activities of gramicidin S through N-methylation modification

Article abstract of DOI:10.1002/cmdc.201300232

β-Sheet antimicrobial peptides (AMPs) are well recognized as promising candidates for the treatment of multidrug-resistant bacterial infections. To dissociate antimicrobial activity and hemolytic effect of b-sheet AMPs we hypothesize that N-methylation of the intramolecular hydrogen bond(s)-forming amides could improve their specificities for microbial cells over human erythrocytes. We utilized a model b-sheet antimicrobial peptide gramicidin S (GS) to study the N-methylation effects on the antimicrobial and hemolytic activities. We synthesized twelve N-methylated GS analogues by replacement of residues at the β-strand and β-turn regions with N-methyl amino acids and tested their antimicrobial and hemolytic activities. Our experiments showed that the HC50 values increased fivefold compared with that of GS when the internal hydrogen-bonded leucine residue was methylated. Neither hemolytic effect nor antimicrobial activity changed when proline alone was replaced with N-methylalanine in the b-turn region. However analogues containing N-methylleucine at β-strand and N-methylalanine at β-turn regions exhibited a fourfold increase in selectivity index compared to GS. We also examined the conformation of these N-methylated GS analogues using ¹H NMR and circular dichroism (CD) spectroscopy in aqueous solution and visualized the backbone structures and residue orientations using molecular dynamics simulations. The results show that N-methylation of the internal hydrogen bond-forming amide affected the conformation backbone shape and side chain orientation of GS.

Full text of DOI:10.1002/cmdc.201300232

CHEMMEDCHEM
FULL PAPERS
DOI: 10.1002/cmdc.201300232
Dissociation of Antimicrobial and Hemolytic Activities of
Gramicidin S through N-Methylation Modification
Yangmei Li,* Nina Bionda, Austin Yongye, Phaedra Geer, Maciej Stawikowski, Predrag Cudic,
Karina Martinez, and Richard A. Houghten*[a]
b-Sheet antimicrobial peptides (AMPs) are well recognized as
promising candidates for the treatment of multidrug-resistant
bacterial infections. To dissociate antimicrobial activity and he-
molytic effect of b-sheet AMPs, we hypothesize that N-methyl-
ation of the intramolecular hydrogen bond(s)-forming amides
could improve their specificities for microbial cells over human
erythrocytes. We utilized a model b-sheet antimicrobial pep-
tide, gramicidin S (GS), to study the N-methylation effects on
the antimicrobial and hemolytic activities. We synthesized
twelve N-methylated GS analogues by replacement of residues
at the b-strand and b-turn regions with N-methyl amino acids,
and tested their antimicrobial and hemolytic activities. Our ex-
periments showed that the HC50 values increased fivefold com-
pared with that of GS, when the internal hydrogen-bonded
leucine residue was methylated. Neither hemolytic effect nor
antimicrobial activity changed when proline alone was re-
placed with N-methylalanine in the b-turn region. However, an-
alogues containing N-methylleucine at b-strand and N-methyl-
alanine at b-turn regions exhibited a fourfold increase in selec-
tivity index compared to GS. We also examined the conforma-
tion of these N-methylated GS analogues using 1H NMR and
circular dichroism (CD) spectroscopy in aqueous solution, and
visualized the backbone structures and residue orientations
using molecular dynamics simulations. The results show that
N-methylation of the internal hydrogen bond-forming amide
affected the conformation, backbone shape, and side chain ori-
entation of GS.
Introduction
The emergence of multidrug-resistant bacteria has led to an
urgent need for the development of new antibiotics. b-Sheet
antimicrobial peptides (AMPs), for example, tachyplesin,[1] pro-
tegrin-1,[2] arenicin,[3] lactoferricin B,[4] Ib-AMP1,[5] gomesin,[6]
and RTD-1,[7] are a class of host defense peptides that potently
inhibit the growth of Gram-positive and Gram-negative patho-
gens. These peptides present key structural elements for anti-
parallel b-sheet strand and amphiphilic character for biological
activity. The mechanism of action of b-sheet AMPs is generally
accepted as the disruption of the bacterial cell membrane
through a detergent-like behavior. Resistance is less likely to
develop when this mechanism is involved since it requires sig-
nificant changes of bacterial membrane composition. Beside
the disruption of membranes, many b-sheet AMPs were also
found that inhibited intracellular targets.[8] Thus, b-sheet AMPs
are well recognized as promising candidates for developing
novel antibiotics for the treatment of multidrug-resistant bac-
teria infections.
of peptide is caused by proteolysis in vivo and can be substan-
tially decreased by introduction of d-amino acids, cyclization,
and other chemical manipulation modifications. Toxicity of b-
sheet AMPs is usually associated with hemolysis, which is the
result of strong amphiphilic structure of these peptides. There-
fore, tuning amphiphilicity to improve their selectivity for bac-
terial cells over human erythrocytes plays a key role in the de-
velopment of b-sheet AMPs for therapeutic use, in particular
systemic use.
Intramolecular hydrogen bonds between an antiparallel b-
strand stabilize the conformation of b-sheet AMPs and amphi-
philicity thereof. Methylation of internal hydrogen-bonded
amides in the antiparallel b-strand disrupts the intramolecular
hydrogen bond(s) and reduces the structural order (Figure 1a).
We hypothesize that disruption of the intramolecular hydrogen
bond(s) may perturb amphiphilicity of b-sheet AMPs, therefore
improving selectivity between microbial and host cells. To illus-
trate how N-methylation impacts biological activity of b-sheet
AMPs, we utilized gramicidin S (GS) as a model b-sheet AMP to
test this hypothesis. GS is a typical cyclic b-sheet AMP, which
has the antiparallel b-sheet strand stabilized by two b-hairpin
motifs made up of d-Phe-Pro. Two cationic Orn residues are on
one side of the molecule and four hydrophobic Leu and Val
residues are on the opposite side (Figure 1b). Thus GS serves
as an excellent model for studying b-sheet antimicrobial pep-
tides.
Degradation and toxicity are two major hurdles in develop-
ing b-sheet AMPs for therapeutic use. Generally, degradation
[a] Dr. Y. Li, Dr. N. Bionda, Dr. A. Yongye, P. Geer, Dr. M. Stawikowski,
Dr. P. Cudic, Dr. K. Martinez, Dr. R. A. Houghten
Torrey Pines Institute for Molecular Studies
11350 SW Village Parkway, Port Saint Lucie, FL 34987 (USA)
E-mail: yli@tpims.org
Houghten@tpims.org
GS is currently limited to topical application due to lack of
selectivity required for a human antibiotic for systemic use. To
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cmdc.201300232.
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2013, 8, 1865 – 1872 1865

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
Figure 1. a) The b-Sheet conformation of antimicrobial peptides (I) and hydrogen-bond disruption by N-methylation (II). b) Structure of GS.
dissociate the antimicrobial activity from the hemolytic activity
Results and Discussion
of GS, structural modifications, in particular on the b-hairpin
Synthesis
and the ring size, have been carried out to improve the selec-
tivity.[9] Replacing d-Phe residues at the turn regions with sur-
rogates have been reported to strongly modulate the thera-
peutic index of GS.[10] A series of ring-expanded analogues
having 12-, 14-, and 16-peptidyl residues have been synthe-
sized. Their conformation, amphiphilicity, and self-association
propensity have been investigated in detail.[11] Although many
studies have been done to improve the selectivity of GS, the
effects of N-methylation of amides, particularly different types
of amides, on biological activities and backbone conformations
have not yet been known.
To investigate the effects of N-methylation at specific sites on
the biological activity of GS, we designed 12 N-methylated GS
(NMe-GS) analogues. The sequences of each analogue are
shown in Scheme 1. Six out of twelve analogues were mono-
methylated (NMe-1 to -6), including four compounds (NMe-
1 to -4) each consisting of a single amino acid substituted with
a N-methyl-l-alanine (MeAla)—a MeAla scan, one compound
(NMe-5) containing a Leu substituted with an N-methyl-l-leu-
cine (MeLeu), and one compound (NMe-6) having a d-Phe sub-
stituted with an N-methyl-d-phenylalanine (MedPhe). Four ana-
logues (NMe-7 to -10) were dimethylated, each compound
containing one Pro replaced with a MeAla and a second amino
acid replaced with an N-methylated amino acid. Two ana-
logues were trimethylated (NMe-11 to -12), each having one
Pro substituted with a MeAla and two other amino acids sub-
stituted with N-methylated amino acids.
In this study, we synthesized GS analogues in which N-me-
thylated amino acids were used at b-strand region to replace
residues whose amide protons are known to be either involved
in internal hydrogen bonds or are exposed externally. Since
Pro delivers a strong effect on conformation, dominates at b-
turn region in proteins and peptides, and shows structural sim-
ilarity to N-methyl-l-alanine
(MeAla),[12] we also replaced the
Pro at b-turn regions with
MeAla. We tested the antimicro-
bial activity and the hemolytic
effect for these N-methylated GS
(NMe-GS) analogues. We exam-
ined their conformation using
1H NMR and circular dichroism
(CD) spectroscopy in aqueous
solution, and visualized the
backbone structure and residue
orientation through molecular
dynamics simulations. Herein,
we present the results from the
biological investigation as well
as the conformational studies of
these NMe-GS analogues.
Scheme 1. Synthesis of GS and the N-methylated analogues. Reagents and conditions: a) Boc-Leu-OH/PyBOP/DIEA
(5:5:7.5 equiv), overnight; b) TFA/CH2Cl2 (55% v/v), 30 min, then DIEA/CH2Cl2 (5% v/v); c) Boc-Orn(Z)-OH/DIC/HOBt
(5:5:5 equiv), 2 h; d) TFA/CH2Cl2 (55% v/v), 30 min, then DIEA/CH2Cl2 (5% v/v); e) repeat step c) and d) to complete
resin-bound linear peptides; f) anhyd HF, 08C, 2 h; g) DIEA/DMF (10% v/v), overnight.
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2013, 8, 1865 – 1872 1866

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
Table 1. Antimicrobial and hemolytic activity of GS and N-methylated analogues.
Compd
MIC90 [mm][a]
HC50 [mm][b]
Type of amide
methylated
Residue
alternation
tR [min][c]
Gram (+) pathogens[d]
MDRSA VRSA MRSE
Gram (ꢀ) pathogens[e]
E. coli
K. pne.
Salm.
GS
1.8 3.6 1.8
28
56
>50.4
>37
28
34.6
45
116
28
3.6
56
3.6
>37
3.6
8.7
5.2
116
7
3.5
28
14.3ꢁ1.45
>56[f]
–
–
4.557
4.124
5.637
3.709
4.512
4.420
4.589
4.110
4.482
4.594
4.514
4.453
4.129
NMe-1
NMe-2
NMe-3
NMe-4
NMe-5
NMe-6
NMe-7
NMe-8
NMe-9
NMe-10
NMe-11
NMe-12
56 28
56
>56
>50.4
>37
29
34.6
22.5
116
28
24.5
30.4
internal
external
internal
Pro
internal
external
internal
internal
external
Pro, Pro
Internal, Pro
Internal, Pro
MeAla5
3.1
>37[g]
3.6
12.6
>37
3.1
18.6
1.8
17.1ꢁ0.5
MeAla4
39.0ꢁ3.0
14.2ꢁ0.75
75.4ꢁ0.63
13.0ꢁ1.0
MeAla3
3.6
8.7
MeAla2
8.7
1.7
MeLeu5
2.6
10.4
1.7
MePhe6
58
7
3.5
116
3.5
6.1
14.5
1.8
0.4
>116
MeAla2, MeAla5
MeLeu5, MeAla7
MeAla2, Mephe6
MeAla2, MeAla7
MeAla2, MeLeu5, MeAla7
MeAla2, MeAla5, Mephe6
69.2ꢁ1.3
14.4ꢁ0.65
16.3ꢁ2.1
94.1ꢁ9.9
48
1.8
3.6
0.9
30.4
56.6
57.4
3.6
14.2
114.8
7.1
7.1
3.5
56.6
>114.8
114.8
114.8
57.4
>114.8
[a] MIC90: minimum inhibitory concentration that inhibits 90% of bacterial isolates after 20 h, relative to a control culture. Data are the representative
values of two duplicate experiments performed in duplicate, and the experimental error is one MIC interval. [b] HC50: 50% hemolytic concentration is the
minimum concentration required for lysis 50% of human erythrocytes. Data represent the mean of representative values of two duplicate experiments per-
formed in triplicate ꢁSD. [c] Retention time (tR) is determined by reversed-phase HPLC; column: Luna 5m C18, 50ꢁ4.6 mm; method: linear gradient 5!
95% solvent B (0.1% formic acid in CH3CN) over 6 min; solvent A (0.1% formic acid in H2O). [d] Gram-positive pathogens included multidrug-resistant
S. aureus ATCC-BAA-44 (MDRSA), vancomycin-resistant S. aureus (VRSA), methicillin-resistant S. epidermidis (VRSE). [e] Gram-negative pathogens included
E. coli k-12 (E. col.), K. pneumonia (K. pne.), Salmonella (Salm.). [f] Sample concentrations ranging from 62 to 0.5 mgmLꢀ1, exhibited 4.3% hemolysis at 56 mm
(62 mgmLꢀ1). [g] Sample concentrations ranging from 42 to 0.3 mgmLꢀ1
.
We synthesized GS and NMe-GS analogues using a solid-
phase synthetic approach as shown in Scheme 1. Polystyrene-
based mercaptomethylphenyl resin was used as support for
the synthesis. The tert-butoxycarbonyl (Boc)-protected amino
acid was coupled stepwise on the resin to form the resin-
bound decapeptide thioester. Because thioesters are resistant
to hydrogen fluoride (HF), the protecting groups of the resin-
bound peptide thioesters were removed with anhydrous HF.[13]
The desired cyclic peptides were then released through cycliza-
tion/cleavage by the treatment of the resin bound peptide
thioesters with N,N-diisopropylethylamine (DIEA; 10% v/v in
N,N-dimethylformamide, DMF). GS synthesized using this
method exhibited identical molecular weight and HPLC reten-
tion time to a standard GS sample (Sigma–Aldrich), indicating
the synthetic GS is exactly same as the standard one.
remained almost unchanged when Leu was replaced with
MeLeu (NMe-5). This result suggests that Leu may be essential
for the antimicrobial activity. The antimicrobial activity was
also not affected when a single externally exposed residue,
Orn or d-Phe, was replaced with N-methylated amino acids
(NMe-2 and NMe-6). Since NMe-2 has a Orn4!MeAla4 replace-
ment, which is not a conservative replacement, this unaffected
antimicrobial activity could possibly be explained by two ef-
fects: methylation of the externally exposed amide did not
affect the antimicrobial activity, or only one positively charged
Orn was essential for keeping the antimicrobial activity. Re-
placement of Pro with MeAla (NMe-4 and NMe-10) did not
change the antimicrobial activity as compared to GS as well.
Hemolytic activity
The hemolytic activity for each compound was also examined.
The minimal concentration of each compound required for
50% lysis of human red blood cells (HC50) was determined
(Table 1). The HC50 value of the synthetic GS, 16 mgmLꢀ1, is
identical to the standard GS in our experiment, and is similar
to that reported.[15] As seen from the HC50 values, N-methylated
amino acid substitutions exhibited site-specific effects on the
hemolytic activity as well. Substitution of the internal hydro-
gen bond-forming residues with N-methylated amino acids,
either MeAla or the matching N-methylated amino acid, in-
creased HC50 values up to fivefold compared to that of GS, in-
dicating the hemolytic effects of those NMe-GS analogues de-
creased. Substitution of the residues having externally exposed
amide protons with N-methylated amino acids did not affect
the HC50. The HC50 value remained unchanged when Pro was
the only residue that was replaced by MeAla, no matter if one
Antimicrobial activity
The antimicrobial activities of the synthetic GS and the NMe-
GS analogues were tested against a variety of multidrug-resist-
ant Gram-positive and Gram-negative bacteria using a standard
microdilution broth assay.[14] Bacterial strains included multi-
drug-resistant Staphylococcus aureus (MDRSA) BAA-44, vanco-
mycin-resistant S. aureus (VRSA), methicillin-resistant S. epider-
midis (MRSE), Escherichia coli, Klebsiella pneumonia, and Salmo-
nella. Minimum inhibitory concentration (MIC90) of each com-
pound against each strain was determined (Table 1). Substitu-
tion with N-methylated amino acid at varying sites affected
the antimicrobial activity differently. Compared to GS, the aver-
age MIC90 value against Gram-positive bacteria increased 15–
20-fold when one internal hydrogen bond-forming residue,
Leu or Val, was replaced with MeAla (NMe-1 and NMe-3), but
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2013, 8, 1865 – 1872 1867

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
or both Pro were replaced (NMe-4 and NMe-10), but remarka-
bly increased when an internal hydrogen bond-forming resi-
due was also methylated (NMe-7, NMe-8, and NMe-11), sug-
gesting that N-methylation of internal hydrogen bond-forming
residues is responsible for the decrease of the hemolytic activi-
ty.
failed to see correlations between hydrophobicity and antimi-
crobial/hemolytic activity in these peptides. Thus, we believed
the conformational change may be more important in affect-
ing the antimicrobial activity of the NMe-GS analogues.
NMR study
N-Methylated amino acid substitutions at different positions
on GS exhibited varying effects on the selectivity and specifici-
ty. Selectivity indices (SI), designated as the ratio of HC50 to
MIC90 value, were calculated and are shown in Figure 2. NMe-8
To determine how methylation affected the backbone confor-
mation of GS, we studied the conformations of GS, NMe-5,
NMe-10, and NMe-11 using NMR analysis.[16] 1H NMR experi-
ments revealed that NMe-10
(MeAla2, MeAla7) kept the C2
symmetry of GS, but NMe-5
(MeLeu5) and NMe-11 (MeAla2,
MeLeu5,
MeAla7)
did
not
(Table 2). Comparing the values
3
for coupling constants JNHꢀCH of
NaHs of each residue in NMe-10
3
with those in GS, we found JNHꢀ
values remained almost iden-
CH
tical except for d-Phe, which is
3.2 Hz in GS and 2.8 Hz in NMe-
3
10. The increased JNHꢀCH value
Figure 2. Selectivity index (SI) and methylated gramicidin S. SI=HC50/MIC90. S. aureus ATCC-BAA-44 (MDRSA), van-
comycin-resistant S. aureus (VRSA), methicillin-resistant S. epidermidis (VRSE), E. coli k-12 (E. col.), K. pneumonia
(K. pne.), Salmonella (Salm.).
suggests NMe-10 has slightly
distorted b-turn regions. Varia-
ble-temperature 1H NMR experi-
ments showed the temperature
shift coefficients (Dd/DT) of the NaHs of Orn, Leu, and Val in
NMe-10 decreased compared with those in GS (Table 2). The
temperature shift coefficient of Leu NaH in NMe-10 decreased
most significantly to ꢀ4.3 ppbKꢀ1, which is close to the indica-
tor value of ꢀ4.5 ppbKꢀ1 that separates hydrogen-bonded and
nonhydrogen-bonded protons,[17] indicating Leu NaH formed
much weaker internal hydrogen bond in NMe-10 than in GS.
Taken together, these data suggest that NMe-10 has a distorted
b-turn and b-sheet conformation compared with GS. These re-
sults support our hypothesis that substitution of Pro with
MeAla decreases structural rigidity.
and NMe-11 exhibited significantly increased SI values against
the entire panel of Gram-positive and -negative bacteria tested
in this work. Other methylated analogues generally showed
similar SI values to the one for GS, except for NMe-5 and NMe-
9, whose SI values against MRSE remarkably increased. Because
NMe-8, NMe-11, and NMe-5 all have a Leu5!MeLeu5 replace-
ment, leading to a disruption of the internal hydrogen bond-
ing between Leu5 NaH···O=CVal3, methylation of internal hydro-
gen bond-forming Leu is most likely responsible for the im-
proved selectivity for bacteria cells over human erythrocytes.
Antimicrobial and hemolytic activities of peptides are report-
ed to link to the hydrophobicity—higher hydrophobicity of
AMP was correlated with stronger antimicrobial and hemolytic
activity.[15] To determine whether hydrophobicity affected anti-
microbial and hemolytic activity of the NMe-GS analogues, we
compared reversed-phase HPLC retention time (tR) with MIC90
and HC50 values separately for each analogue (Table 1), but
NMe-5 and NMe-11 were generated by replacing Leu5!
MeLeu5 in GS and in NMe-10, respectively. Both of them lost
the C2 symmetry and differentiated nine amide protons. NMe-
3
5 and NMe-11 had parallel JNH-CH values for the corresponding
residues, suggesting these two analogues had comparable
3
backbone conformation. The JNHꢀCH values of NaHs in NMe-5
3
Table 2. Amide proton chemical shifts d [ppm], coupling constants JNHꢀCH [Hz], and temperature shift coefficients Dd/DT [ppbKꢀ1].[a]
3
3
3
3
Compd
d, JNHꢀCH, (Dd/DT)
d, JNHꢀCH, (Dd/DT)
d, JNHꢀCH, (Dd/DT)
d, JNHꢀCH, (Dd/DT)
d-Phe1,6
Orn4,9
Leu5,10
Val3,8
GS
NMe-10
9.03, 3.2, (ꢀ7.8)
8.64, 9.2, (ꢀ5.3)
8.59, 9.2, (ꢀ6.1)
8.31, 8.8, (ꢀ2.9)
8.17, 9.2, (ꢀ4.3)
7.21, 9.6, (ꢀ1.9)
7.20, 9.6, (ꢀ2.6)
9.20, 2.8, (ꢀ8.0)
d-Phe1
d-Phe6
Orn4
Orn9
Leu10
Val3
Val8
NMe-5
NMe-11
NMe-8
8.92, –,[b] (ꢀ13.1)
9.21, –,[b] (ꢀ8.5)
8.92, –,[b] (ꢀ10.6)
7.70, 7.2, (ꢀ8.1)
8.08, 6.8, (ꢀ9.6)
8.26, 6.0, (ꢀ10.7)
8.75, 9.2, (ꢀ4.3)
8.70, 9.6, (ꢀ4.0)
8.32, 7.2, (ꢀ4.7)
8.23, 9.6, (ꢀ3.9)
8.29, 8.8, (ꢀ4.0)
8.73, 9.6, (ꢀ3.6)
8.04, 8.8, (ꢀ2.5)
8.04, 9.2, (ꢀ2.8)
8.16, 9.2, (ꢀ3.0)
6.92, 8.8, (ꢀ1.9)
6.57, 8.4, (ꢀ1.4)
6.59, 8.4, (ꢀ0.7)
7.22, 9.2, (ꢀ3.2)
7.20, 9.2, (ꢀ0.3)
7.21, 9.2, (ꢀ3.0)
[a] NMR experiments were performed in [D6]DMSO. Chemical shifts and 3JNHꢀCH values were measured at 305 K. [b] Broad singlet.
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2013, 8, 1865 – 1872 1868

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
and NMe-11 were comparable to the values in GS and NMe-10
for the corresponding residues except for d-Phe, which
changed tremendously from ~3 Hz in GS/NMe-10 to ~7 Hz for
d-Phe6, and singlet for d-Phe1 in NMe-5/NMe-10. These sub-
spectrum of NMe-2 exhibited a broad negative minimum at
207–213 nm, and NMe-3 shifted the negative minimum to
202 nm. The CD spectra of the rest of the analogues show
minimum ellipticity at 217 nm and a shoulder at 207 nm, with
the ellipticity ratio of q207:q217 ranging from 0.76 to 0.93. The
significant decrease of ellipticity in the 200–210 nm range sug-
gested the disruption of the b-turn part and reflected the con-
formational changed by N-methylation. These conformational
changes detected by CD spectra were correlated with the re-
sults from our NMR study. However, we have not observed the
difference or trend between methylation of internal hydrogen-
bonded and externally exposed amides by CD analysis. The
conformational change of these analogues presented in CD
spectra may also result from the formation of cis-peptide bond
that was facilitated by methylation.[19]
3
stantial changes of JNHꢀCH values indicated the significant back-
bone conformational changes at the b-turn regions. Since the
only difference between NMe-5 versus GS and NMe-11 versus
NMe-10 is the methylation at Leu5, we therefore concluded
methylation at Leu5 caused the conformational change in the
b-turn region in NMe-5 and NMe-11. The temperature shift co-
efficients of amide NaHs in NMe-5 and NMe-11 also exhibited
a comparable pattern in that Leu10, Val8 and Val3 are in the
range of ꢀ0.3–3.2 ppbKꢀ1, indicating these three NaHs are in-
ternal hydrogen-bonded; d-Phe1, d-Phe6, Orn4, and Orn9 are in
the range of ꢀ3.9 to ꢀ13.1 ppbKꢀ1, suggesting these protons
are solvent-exposed in both analogues. But the temperature
shift coefficients of Val8 in NMe-11 dramatically increased to
ꢀ0.3 ppbKꢀ1 compared with those of NMe-10 (ꢀ2.6), NMe-5
(ꢀ3.2), and GS (ꢀ1.9), suggesting intramolecular hydrogen
bonding between Val8 NaH···C=O MeLeu5 was significantly
strengthened.
Molecular dynamics (MD) simulations
To determine how N-methylations affected the conformational
properties of these analogues, we selected three with the best
SI values (NMe-5, NMe-8, and NMe-11) and one (NMe-4) whose
SI value was comparable to that of GS. The effects on the
shapes of the rings were investigated by computing backbone
radii of gyration (Rg; see figure S1B). It is evident that the Rg
value of NMe-8 and NMe-11 were larger compared with GS,
NMe-4 and NMe-5. Noteworthy, NMe-5, NMe-8 and NMe-11
show lower hemolytic properties. Therefore, a scalar property
such as Rg may not convey the whole picture. Thus, to identify
specific overall backbone properties, the trajectories were clus-
tered using a backbone root-mean-square deviation (RMSD)
cut-off of 0.5 ꢂ to define similar conformations using
8000 frames because of size limitations in generating the
square matrix for clustering (see figure S2 for the backbone
structures of the centroids from the top five clusters of each
trajectory and their fractional contribution to the total popula-
tion). Although the shapes of the rings and their populations
reflected the values of the radius of gyration, similar atoms of
the most populated clusters of GS and NMe-4 share the same
prolate and oblate, and are more slender (see figure S2A and
S2B). Although the Rg value of NMe-5 is comparable with that
of GS, its dimensions are different (see figure S2E), tending to
have a wider oblate. NMe-8 and NMe-11 share the same pro-
late as GS, but their oblates tend to be broader, similar to
NMe-5, presumably because they are more flexible due to the
absence of one (NMe-8) or two (NMe-11) Pro couple with the
disruption of backbone interactions by N-methylation of Leu5.
GS has been described as an amphiphilic cyclic peptide,
with a polar side comprising the Orn residues, and a lipophilic
surface formed by the Val and Leu residues (Figure 1). Given
that changes in the backbone conformation may affect the
presentation of side chains, we investigated the effects of
these analogues on side chain presentation (Figure 3). The side
chains of the polar Orn residues are shown in Figure 3a,
whereas those of the nonpolar residues are portrayed in Figur-
es 3b,c for the five most populated clusters of each peptide.
The Orn side chains were oriented predominantly below
a transverse plane of the view in Figure 3a for GS and all the
NMe-8 is a dimethylated (MeAla2, MeLeu5)-GS analogue,
which kept MeLeu5, but differed from NMe-5 in MeAla2!Pro2
and from NMe-11 in Pro7!MeAla7. The major sequence differ-
ence between NMe-8 and NMe-5/NMe-11 are Pro$MeAla al-
1
ternates at residue 2 or residue 7. H NMR analysis showed that
3
NMe-8 was similar to NMe-5 and NMe-11 in JNH-CH values, the
3JNH-CH value of d-Phe6 increased to 6 Hz, and d-Phe1 was a sin-
glet, suggesting significant conformational changes at b-turn
regions. The temperature shift coefficients of NaHs of NMe-8
were also similar to those of NMe-5 and NMe-11—showing
NaHs of Leu10, Val8 and Val3 were hydrogen-bonded, and NaHs
of d-Phe1, d-Phe6, Orn4, and Orn9 were solvent-exposed. The
temperature shift coefficient of Val3 significantly increased to
ꢀ0.7 ppbKꢀ1, suggesting NaH of Val3 was involved in strong in-
ternal hydrogen bonding.
Conformation of the NMe-GS analogues in aqueous solution
To examine the methylation effects on conformation of NMe-
GS analogues, circular dichroism (CD) spectroscopy was record-
ed for each analogue in an aqueous solution (see figure S5A
for monomethylated NMe-GS analogues and figure S5B for di-
and trimethylated NMe-GS analogues). The CD spectra for GS
and the 12 NMe-GS analogues differed in shape and intensity.
GS, NMe-4 (MeAla2), and NMe-10 (MeAla2, MeAla7) presented
similar CD spectra, with a negative minimum at 207 nm and
a shoulder at 217 nm for a combination of type II b-turn and
pleated b-sheet as reported elsewhere.[10a,18] These data
showed that the overall conformation of GS did not change
when MeAla was replaced with Pro. The ratio of ellipticity of
q207:q217 for GS, NMe-4, and NMe-10 is 1.04, 1.04, and 1.00, re-
spectively. The slightly decreased ratio of q207:q217 suggested
NMe-10 presented a disturbed b-turn and b-sheet conforma-
tion compared with GS in aqueous solution. All the other
NMe-GS analogues showed major conformational changes, be-
cause their ellipticities were substantially changed. The CD
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2013, 8, 1865 – 1872 1869

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
Conclusions
Gramicidin S (GS) has all the
structural elements required for
b-sheet antimicrobial peptides
(AMPs)—the b-turn and the anti-
parallel b-strand. Methylation of
its internal hydrogen-bonded
amide at b-strand region termi-
nated the specific hydrogen
bonding, thereby affecting the
subtle balance on backbone
conformation, side chain orienta-
tion, and amphiphilicity of GS.
This in turn decreased the he-
molytic activity of the N-methy-
lated products. Antimicrobial ac-
tivity and hemolytic effect fur-
ther dissociated when residues
were methylated at both b-
Figure 3. Side chain dispositions of GS and N-methylated analogues. a) Display of the Orn side chain. The Orn resi-
dues are shown in blue. The foremost residue in a) is P2. b) Display of hydrophobic side chains of l-amino acids.
c) Display of hydrophobic side chains of d-Phe. The backbone carbon atoms of GS and each N-methylated ana-
logue are overlapped and shown in green and yellow in a) and b), respectively. The images were generated using
PyMOL (version 1.5).
peptides. However, there was a relative increase in the pres-
ence of Orn side chains along the transverse plane in the me-
thylated analogues. These differences will invariably change
the polar surface of the analogues when compared with that
of GS. Interestingly, for NMe-8 and NMe-11, Orn4 was oriented
towards a sagittal plane, whereas its orientation in NMe-4 was
similar to GS. A close inspection of each individual cluster re-
vealed that in the second most populated cluster of NMe-5,
Orn9 was exclusively distributed close to the sagittal plane.
This may have some bearing with the observed hemolytic
properties of NMe-5, NMe-8, and NMe-11.
strand and b-turn regions. The results provide insight into
modification of other b-sheet antimicrobial peptides to im-
prove their selectivity between bacterial cell and host cell, and
impact the development of b-sheet antimicrobial peptides to
therapeutic agents for clinical use.
Experimental Section
Chemistry
The Val and Leu residues were oriented above the transverse
plane in GS, whereas this configuration was disrupted in the
analogues (Figure 3b). Leu10 was displaced towards the trans-
verse plane in all the analogues. We saw no association be-
tween the selectivity indices of these peptides with their hy-
drophobic group orientations.
Synthesis: All N-methylated gramicidin S (NMe-GS) analogues were
prepared by solid-phase peptide synthesis using mercaptomethyl-
phenyl polystyrene resin (100 mg each, Sigma–Aldrich, loading
~2.0 mmolSꢀ1 gꢀ1) as support. Parallel solid-phase synthesis of pep-
tide a-thioesters were carried out using the “teabag” approach.[20]
The first amino acid, Boc-Leu-OH (5 equiv), was coupled on the
resin by using benzotriazol-1-yl-oxytripyrrolidinophosphonium hex-
afluorophosphate (PyBOP; 5 equiv) and diisopropylethylamine
(DIEA; 7.5 equiv) in N,N-dimethylformamide (DMF; 10 mL) over-
night. Stepwise peptide synthesis was carried out using a standard
diisopropylcarbodiimide (DIC)/N-hydroxybenzotriazole (HOBt) cou-
pling protocol using Boc chemistry with the exception of the N-
methylated amino acids and the following amino acids, which
were coupled using PyBOP/DIEA activation for 2 h. After peptide
chain elongation, the resin-bound peptide was treated with anhyd
HF for 2 h at 08C. Following evaporation of anhyd HF with a gas-
eous nitrogen stream and washing of the resin-bound peptide
with CH2Cl2, the cyclic products was released by cyclization/cleav-
age in 10% v/v DIEA in DMF. All crude products were purified by
HPLC. The LC-MS profile for each NMe-GS analogue is shown in
the Supporting Information (figure S4).
While it may be expected that the MeAla2 analogue (NMe-4)
will affect the backbone conformations of local residues such
as Leu10 and d-Phe1, N-methylation of Leu5 affected the confor-
mational properties of the distal Leu10 (NMe-5; Figure 3b) and
d-Phe1 (NMe-8; Figure 3c). In Figure 3c, the effect of the
MeAla2 substitution on the spatial display of d-Phe1 is evident.
When the MeLeu5 substitution was performed on NMe-4 to
generate NMe-8, the side chain distribution of d-Phe1 reverted
to that observed in GS. It should be noted that NMe-8 exhibit-
ed similar space distributions of the d-Phe side chains with ref-
erence to those of GS, but possessed significantly reduced he-
molytic activity. Taken together, the differences between the
orientations of the Orn residues and the absence of a trend in
the display of the hydrophobic groups in the low and high he-
molytic peptides, suggested that the overall consequence of
presentation of the Orn groups may be important for optimiz-
ing hemolytic properties.
NMR: All NMR spectra were recorded on a Bruker Avance III 400
high performance digital NMR spectrometer. All spectra were ob-
tained at a concentration of 10 mm in [D6]DMSO.
Circular dichroism (CD): All CD spectra were recorded on
a JASCO 810 spectropolarimeter at 258C using a cell with 0.1 cm
path length. The spectra were acquired in a wavelength range of
190–250 nm at 1 nm bandwidth using four accumulations and
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2013, 8, 1865 – 1872 1870

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
200 nmminꢀ1 scanning speed. All spectra were obtained using
0.1–0.2 mm concentrations in H2O.
hemolysis is reported as HC50, where the absorbance from water
(10 mL) containing no compound was used as 0% hemolysis. Plates
were incubated for 1 h at 378C. PBS (100 mL) was added to each
well, and the plates were centrifuged (10 min, 1000 g). Superna-
tants (100 mL) were transferred into a new plate, and absorbance
at 405 nm was measured.
Molecular dynamics (MD) simulations: The starting conformation
of GS was extracted from its crystal structure complex with savi-
nase (PDBID: 1SVN) resolved at 1.5 ꢂ. The analogues were generat-
ed by substituting each residue of GS with the corresponding resi-
due of the methylated analogue. The cyclic peptides were im-
mersed in a box of TIP3P water molecules. A distance of 15 ꢂ sepa-
rated the peptides from the edges of the box. The equilibration
and production stages were performed as described previously;[21]
however, production lasted 85 ns. Of note, the CHARMM parameter
sets employed included CMAP terms derived for d-amino acids.[22]
NAMD v2.7b1 was employed to perform the simulations.[23] Snap-
shots were collected every picosecond for post-processing. Equili-
bration of the MD simulations was ascertained by monitoring back-
bone root mean-squared deviations (RMSDs) over the course of
the simulations (S1A). The RMSDs showed that the backbone
cyclic systems were no longer sampling new conformers. The first
5 ns of the trajectories were discarded as additional equilibration
because of fluctuations in the backbone of GS observed during
this time.
Abbreviations
AMP, antimicrobial peptide; Boc, tert-butoxycarbonyl; CD, cir-
cular dichroism; DIC, diisopropylcarbodiimide; DIEA, diisopro-
pylethylamine; DMF, N,N-dimethylformamide; GS, gramicidin S;
HOBt, N-hydroxybenzotriazole; hRBC, human red blood cells;
MDR, multidrug-resistant; MIC, minimum inhibitory concentra-
tion; NMe-GS, N-methylated GS; PyBOP, benzotriazol-1-yl-oxy-
tripyrrolidinophosphonium hexafluorophosphate; RMSD, root-
mean-square deviation; SI, selectivity index; TFA, trifluoroacetic
acid.
Acknowledgements
Biological analysis
We acknowledge support of this work from the State of Florida
(USA), Executive Office of the Governor’s Development of Eco-
nomic Opportunity.
Antimicrobial activity: A total of three Gram-positive and three
Gram-negative bacterial strains were used, including multidrug-re-
sistant (MDR) bacterial strains: Staphylococcus aureus ATCC BAA-44,
Staphylococcus aureus Mu50 (VRSA) ATCC 700699, Staphylococcus
epidermidis (MRSE) ATCC 27626, Escherichia coli K-12 ATCC 29181,
Klebsiella pneumonia K6 ATCC 700603, and Salmonella. Determina-
tion of the antibacterial activity of the synthetic GS and NMe-GS
analogues was performed in sterile 96-well flat-bottomed polystyr-
ene plates by the standard microdilution broth method based on
the Clinical and Laboratory Standards Institute (CLSI) protocol.[14]
Tests were performed using Mꢃller–Hinton broth (MHB) without di-
lution. Controls on each plate were media without bacteria, bacte-
rial inoculum without antimicrobials added, and bacterial inoculum
containing methicillin or vancomycin. Concentrations of synthetic
GS and NMe-GS analogues 1–12, as well as control antibiotics,
were in the range 1–128 mgmLꢀ1. All samples were loaded in du-
plicate, and the average OD value was taken for calculating mini-
mum inhibitory concentration (MIC) values. Each sample was
tested twice. Stock solutions of synthesized analogues 1–12 were
prepared in H2O. Plates were loaded with 90 mL bacterial suspen-
sion (with initial OD600 of 0.001) of the tested microorganism, and
10 mL aliquots of twofold serial dilutions of the synthetic GS and
the NMe-GS analogues 1–12 or control antibiotics. Plates were
then incubated at 378C overnight with gentle shaking. Inhibition
of bacterial growth was determined by measuring OD600; a decrease
in OD600 indicates inhibition of bacterial growth.
Keywords: antimicrobial
activity
and
selectivity
·
conformation · drug discovery · molecular modeling · peptide
methylation
[1] K. Kawano, T. Yoneya, T. Miyata, K. Yoshikawa, F. Tokunaga, Y. Terada, S.
Iwanaga, J. Biol. Chem. 1990, 265, 15365–15367.
[2] V. N. Kokryakov, S. S. Harwig, W. A. Panyutich, A. A. Shevchenko, G. M.
Aleshina, O. V. Shamova, H. A. Korneva, R. I. Lehrer, FEBS Lett. 1993, 327,
231–236.
[3] T. V. Ovchinnikova, G. M. Aleshina, S. V. Balandin, A. D. Krasnosdemb-
skaya, M. L. Markelov, E. I. Frolova, Y. F. Lenova, A. A. Tagaev, E. G. Kras-
nodembsky, V. N. Kokryakov, FEBS Lett. 2004, 577, 209–214.
[4] P. M. Hwang, N. Zhou, X. Shan, C. H. Arrowsmith, H. J. Vogel, Biochem-
istry 1998, 37, 4288–4298.
[5] S. U. Patel, R. Osborn, S. Rees, J. M. Thornton, Biochemistry 1998, 37,
983–990.
[6] P. I. Silva, Jr., S. Daffre, P. Bulet, J. Biol. Chem. 2000, 275, 33464–33470.
[7] M. Trabi, H. J. Schirra, D. J. Craik, Biochemistry 2001, 40, 4211–4221.
¯
[8] a) T. Mogi, H. Ui, K. Shiomi, S. Omura, K. Kita, FEBS Lett. 2008, 582,
¯
2299–2302; b) T. Mogi, Y. Murase, M. Mori, K. Shiomi, S. Omura, M. P.
Paranagama, K. Kita, J. Biochem. 2009, 146, 491–499.
[9] a) G. M. Grotenbreg, E. Spalburg, A. J. de Neeling, G. A. van der Marel,
H. S. Overkleeft, J. H. van Boom, M. Overhand, Bioorg. Med. Chem. 2003,
11, 2835–2841; b) A. W. Tuin, D. K. Palachanis, A. Buizert, G. M. Groten-
breg, E. Spalburg, A. J. de Neeling, R. H. Mars-Groenendijk, D. Noort,
G. A. van der Marel, H. S. Overkleeft, M. Overhand, Eur. J. Org. Chem.
2009, 4231–4241.
[10] a) C. Solanas, B. G. de La Torre, M. Fernꢄndez-Reyes, C. M. Santiveri, M. A.
Jimꢅnez, L. Rivas, A. I. Jimꢅnez, D. Andreu, C. Cativiela, J. Med. Chem.
2009, 52, 664–674; b) C. Solanas, B. G. de La Torre, M. Fernꢄndez-Reyes,
C. M. Santiveri, M. A. Jimꢅnez, L. Rivas, A. I. Jimꢅnez, D. Andreu, C. Cati-
viela, J. Med. Chem. 2010, 53, 4119–4129.
Hemolytic effect: Human red blood cells (hRBCs) were diluted with
phosphate-buffered saline (PBS) to 0.56%. The synthetic GS and
GS-NMe analogues 1–12 were dissolved in H2O. hRBCs (90 mL)
were placed into each well of the clear, flat-bottom 96-well plate,
followed by the addition of analogue solution (10 mL to final pep-
tide concentrations of 1–128 mgmLꢀ1). Assays were performed in
triplicate, and each experiment was repeated twice. Hemolysis was
monitored by measuring the absorbance of the released hemoglo-
bin at 405 nm. 100% hemolysis was obtained by adding Triton X-
100 aqueous solution (10 mL) to the above-prepared hRBC suspen-
sion at a final concentration of 0.5% (v/v). The minimum concen-
tration of GS and NMe-GS analogues that required to cause 50%
[11] E. J. Prenner, M. Kiricsi, M. Jelokhani-Niaraki, R. N. A. H. Lewis, R. S.
Hodges, R. N. McElhaney, J. Biol. Chem. 2005, 280, 2002–2011.
[12] B. Laufer, J. Chatterjee, A. O. Frank, H. Kessler, J. Pept. Sci. 2009, 15,
141–146.
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2013, 8, 1865 – 1872 1871

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
[13] a) Y. Li, Y. Yu, M. Giulianotti, R. A. Houghten, J. Comb. Chem. 2008, 10,
613–616; b) J. A. Camarero, A. Adeva, T. W. Muir, Lett. Pept. Sci. 2000, 7,
17–21.
[14] N. Bionda, M. Stawikowshi, R. Stawikowska, M. Cudic, F. Lꢆpez-Vallejo,
D. Treitl, J. Medina-Franco, P. Cudic, ChemMedChem 2012, 7, 871–882.
[15] a) T. Wieprecht, M. Dathe, M. Beyermann, E. Krause, W. L. Maloy, D. L.
MacDonald, M. Bienert, Biochemistry 1997, 36, 6124–6132; b) Y. Chen,
M. T. Guarnieri, A. I. Vasil, M. L. Vasil, C. T. Mant, R. S. Hodges, Antimicrob.
Agents Chemother. 2007, 51, 1398–1406.
Hodges, J. E. Meissner, U. E. Hassenstein, Biophys. J. 2008, 95, 3306–
3321.
[19] C. McInnes, L. H. Kondejewski, R. S. Hodges, B. D. Sykes, J. Biol. Chem.
2000, 275, 14287–14292.
[20] R. A. Houghten, Proc. Natl. Acad. Sci. USA 1985, 82, 5131–5135.
[21] A. B. Yongye, Y. Li, M. A. Giulianotti, Y. Yu, R. A. Houghten, K. Martꢇnez-
Mayorga, J. Comput.-Aided Mol. Des. 2009, 23, 677–689.
[22] E. R. Turpin, J. D. Hirst, Chem. Phys. Lett. 2012, 543, 142–147.
[23] J. C. Phillips, R. Braun, W. Wang, J. Gumbart, E. Tajkhorshid, E. Villa, C.
Chipot, R. D. Skeel, L. Kalꢅ, K. Schulten, J. Comput. Chem. 2005, 26,
1781–1802.
[16] a) E. M. Krauss, S. I. Chan, J. Am. Chem. Soc. 1982, 104, 6953–6961; b) K.
Yamada, S.-s. Shinoda, H. Oku, K. Komagoe, T. Katsu, R. Katakai, J. Med.
Chem. 2006, 49, 7592–7595.
[17] N. J. Baxter, M. P. Williamson, J. Biomol. NMR 1997, 9, 359–369.
[18] a) M. Kawai, H. Yamamura, R. Tanaka, H. Umemoto, C. Ohmizo, S. Higu-
chi, T. Katsu, J. Pept. Res. 2005, 65, 98–104; b) M. Jelokhani-Niaraki, R. S.
Received: May 21, 2013
Published online on September 10, 2013
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2013, 8, 1865 – 1872 1872