Our antimicrobial and antifungal activity studies of these peptides showed that BM-1 was much more active than IB-367 against Gram-positive bacteria and fungi, whereas BM-2 was inactive.
Trang 1and inactive analogues of the protegrin-1 peptide
Sylwia Rodziewicz-Motowidło1, Beata Mickiewicz1, Katarzyna Greber2, Emilia Sikorska1, Łukasz Szultka2, El_zbieta Kamysz1and Wojciech Kamysz2
1 Faculty of Chemistry, University of Gdan´sk, Poland
2 Faculty of Pharmacy, Medical University of Gdan´sk, Poland
Introduction
The search for new drugs and target sites has
gener-ated interest in a group of short polypeptides,
antimi-crobial peptides (AMPs), compounds that can combat
bacterial infections [1], and have a broad spectrum of
activity against bacteria, fungi and protozoa [2,3]
AMPs are positively charged molecules (there are also
a few negatively charged ones [4,5]) isolated from a
variety of animals and plants, where they participate
in natural defence mechanisms AMPs are usually
highly amphipathic molecules with hydrophobic and
hydrophilic moieties segregated into distinct patches
on the molecular surface Topologically, they can be
grouped into linear and cysteine-bridged peptides Further subdivided according to the number of disul-phide bridges in their structure, cysteine-bridged AMPs include the protegrins, first isolated in 1993 from porcine leucocytes [6] Protegrins are active against bacteria (Escherichia coli, Staphylococcus aureus [7], Pseudomonas aeruginosa, Chlamydia trachomatis, Nei-sseria gonorrhoeae [8]), yeasts (Candida albicans [6]) and viruses (HIV-1 [9]) The protegrin family contains the following peptides: protegrin-1, protegrin-2, prote-grin-3, protegrin-4 and protegrin-5 (PG-1–PG-5) [10] They are produced from a family of antimicrobial
Keywords
antimicrobial peptides; IB-367; NMR;
protegrin-1 analogues; three-dimensional
structure
Correspondence
S Rodziewicz-Motowidło, Faculty of
Chemistry, University of Gdan´sk,
Sobieskiego 18, 80-952 Gdan´sk, Poland
Fax: (+48 58) 523 54 72
Tel: (+48 58) 52 35 430
E-mail: sylwia@chem.univ.gda.pl
(Received 11 August 2009, revised 6
November 2009, accepted 9 December
2009)
doi:10.1111/j.1742-4658.2009.07544.x
The natural antimicrobial cationic peptide protegrin-1 displays a broad spectrum of antimicrobial activity and rapidly kills pathogens by interacting with their cell membrane We investigated the structure–activity relation-ships of three protegrin-1 analogues: IB-367
(RGGLCYCRGRFCVCVGR-NH2), BM-1 (RGLCYCRGRFCVCVG-NH2) and BM-2 (RGLCYRPRFV CVG-NH2) Our antimicrobial and antifungal activity studies of these peptides showed that BM-1 was much more active than IB-367 against Gram-positive bacteria and fungi, whereas BM-2 was inactive The BM-1 peptide showed fourfold reduced haemolysis relative to IB-367, an addi-tional advantage of this peptide In addition, BM-1 was about 15% cheaper than IB-367 to synthesize The absence of two cysteine residues in the BM-2 sequence could be the main reason for its unstable conformation and antimicrobial inactivity The solution structures of these peptides were determined in dimethyl sulphoxide using two-dimensional NMR and restrained molecular dynamics calculations IB-367 and BM-1 formed short, antiparallel, b-hairpin structures connected by a type II¢ b-turn The shorter, inactive BM-2 analogue exhibited major and minor conformations (predominantly unordered) in the NMR spectra and was much more flexible
Abbreviations
AMP, antimicrobial peptide; CSI, chemical shift index; MIC, minimal inhibitory concentration; NOESY, nuclear Overhauser effect
spectroscopy; PG-1, protegrin-1; ROESY, rotating frame Overhauser effect spectroscopy.
Trang 2peptide precursors known as cathelicidins [11], which
are synthesized as the C-terminal portion of a
cathelin-containing proregion The N-terminal cathelin domain
of the precursor is highly conserved at both the amino
acid and nucleotide sequence levels; this conservatism
is emphasized by both inter- and intraspecific
compari-sons Conversely, the sequence of the C-terminal
pep-tide carrying the antimicrobial activity is highly
variable It has been shown that activated porcine
neu-trophils release intact pro-protegrin, which is inactive
as an antimicrobial It is then processed extracellularly
by elastase to form antimicrobial protegrin [12]
PG-1 is an 18-amino-acid peptide with an amidated
C-terminus It is thought to form an antiparallel
b-sheet constrained by two disulphide bridges, Cys6–
Cys15 and Cys8–Cys13 (1PG1 [13] and 1ZY6 [14] in
the Protein Data Bank) [7,15] Containing six
posi-tively charged arginine residues, the native sequence of
PG-1 is highly cationic The distribution of
hydropho-bic and hydrophilic residues at the peptide surface is a
structural feature required for the cytolytic activity of
PG-1 Structure–activity relationship studies of several
hundred PG-1 analogues were analysed to determine
the role of individual hydrophobic and hydrophilic
residues in antimicrobial activity, i.e to gain an
under-standing of the relationship between the primary and
secondary structure of protegrins and their microbial
activities, and to identify a protegrin analogue for
clin-ical development [16] The presence of the b-hairpin
structure was found to be crucial to the antimicrobial
activity of the protegrin The analogues – linearized or
with amino acid substitutions eliminating hydrogen
bonding across the b-sheet – showed a reduced
biologi-cal activity, especially in the presence of physiologibiologi-cal
concentrations of NaCl [17,18] However, Tam et al
[19] reported that the activity of nondisulphide-bonded
analogues could be restored by the cyclization of the
peptide backbone In addition, Harwig et al [17]
found that, in peptides containing one disulphide
bond, the ‘bullet’ analogue (cysteines one and four
linked by a disulphide bond) had an activity
compara-ble with that of PG-1, whereas the ‘kite’ analogue
(cysteines two and three linked by a disulphide bond)
was less active In addition, the maintenance of the
amphiphilicity of the b-sheet is essential The cationic
and hydrophobic clusters in PG-1 have been shown to
be the structural features required for antibacterial
activity [16,20] Analogues with reduced positive
charge tend to be less active, which may imply that the
binding of a cationic surface to a lipopolysaccharide
is a key mechanistic step in the killing of bacteria
[16,20] The conformations of the structural features
determining the antimicrobial activity of protegrins
were calculated for 62 peptides and correlated with their experimental activity against six microbe species (E coli, N gonorrhoeae (Strain F-62), N gonorrhoeae (Strain FA-19), Listeria monocytogenes, C albicans,
P aeruginosa), as well as their haemolytic and cyto-toxic activities [20] Based on broad structure–activity relationship studies, only one analogue of PG-1 –
IB-367 – was selected for clinical development as a topical agent to prevent the oral mucositis associated with cancer therapy [16,21,22] It displays a broad spectrum
of activity, rapid microbicidal action and limited ability to induce resistance IB-367 kills a broad spec-trum of bacteria and fungi, including those resistant to conventional antimicrobial drugs, by attaching to and destroying the integrity of the lipid cell membrane [23]
In addition, IB-367 demonstrates enhanced bactericidal and fungicidal activity compared with that of native protegrins [16,24] It could therefore be an interesting compound for the inhibition of bacterial translocation and endotoxin release in obstructive jaundice [25] IB-367 is a 17-amino acid peptide with an amidated C-terminus (Fig 1): a peptide with such C-terminus displays greater biological activity than an analogue without one [16] Compared with other protegrin pep-tides of comparable activity, IB-367 has three advanta-ges in chemical synthesis: (a) most importantly, it contains an achiral amino acid residue (glycine) at position 9 in the b-turn – the problem of racemisation
is thus avoided [16]; (b) it has only four arginine resi-dues compared to six in PG-1 – this is significant, as arginine is expensive to purchase; (c) it contains 17 amino acids compared to 18 in PG-1
This paper elucidates the structure-activity relation-ship in IB-367 and two other analogues of PG-1
Fig 1 Amino acid sequences of PG-1 and its analogues IB-367, BM-1 and BM-2.
Trang 3(BM-1 and BM-2) (see the sequences in Fig 1) using
2D-NMR spectroscopy and molecular dynamics The
results of these investigations are compared with
pub-lished structural information about PG-1 and other
antimicrobial peptides in an attempt to understand
which structural features are responsible for the high
biological activity of AMPs
Results
Design of the new BM-1 and BM-2 analogues
Our aim was to design new analogues of PG-1 with
biological activity comparable with or better than that
of PG-1, but cheaper (about 15% less) to synthesize
chemically on a large scale On the basis of the
struc-ture–activity relationship studies of other PG-1
ana-logues (see introductory paragraphs), we designed two
PG-1 analogue sequences The first analogue, BM-1,
contains four cysteines occupying the
nonhydrogen-bonded sites of the natural b-hairpin core; Gly3, Arg4
and Arg18 were removed from the amino acid
sequence Although previous studies on protegrin
ana-logues have shown that a positive charge in the loop is
essential for activity [16], we also replaced Arg10 with
a glycine residue (see Fig 1) We reasoned that, by
removing these residues, we would retain the cationic
nature of the peptide (+4 under physiological
condi-tions), but only in the loop (Arg7, Arg9) and in the
N-terminal fragment (Arg1), not in the C-terminal
fragment In PG-1, the cationic arginine residues
occupy the loop, N- and C-termini In this way, we
endeavoured to reduce the cost of synthesis (arginine is
very expensive)
The second analogue, BM-2, was designed in such a
way that the relative importance of the rigidity and
charge at the turn of the b-hairpin in the context of a
single disulphide bridge could be assessed In BM-2,
we removed Gly3 and Arg18 as in BM-1 We also
removed the two cysteines forming the disulphide bond
proximal to the turn Although the turn structure of
single disulphide variants of PG-1 is less well defined
and their activity is intermediate relative to that of
PG-1 [16], we wanted to see how removal of the two
cysteines (shortening the amino acid sequence) would
affect their structure and activity We also wanted to
promote b-hairpin formation by inducing a b-turn,
and so we replaced Arg10 with Pro10 (see Fig 1)
Although the proline residue is known to have high
frequencies in b-turn formation [26,27], the
incorpora-tion of l-proline at posiincorpora-tion i + 1 of the reverse
turn prevents b-hairpin formation as a result of
incompatibilities with the intrinsic right-handed twist
of b-strands [28–30] Presumably, therefore, the BM-2 analogue would have an undefined structure and reduced antimicrobial activity
An analogue similar to the BM-2 amino acid sequence has been suggested previously by Lai et al [18] They designed a peptide in which two proximal cysteines were replaced with branched residues (threo-nine) with a high intrinsic preference for b-sheet conformation, and with a d-proline residue instead of
an arginine residue incorporated at position i + 1 of the reverse turn (see peptide 10 in [18]) The amino acid sequence of BM-2 also has a similar profile to the bullet variants studied by Harwig et al [17]
Antimicrobial and haemolytic activity PG-1, IB-367, BM-1 and BM-2 were characterized with regard to their antibacterial activity against Gram-positive bacteria, Gram-negative bacteria and fungi (see Table 1) All the test organisms were human pathogens, good selections for the initial screening of antimicrobial⁄ antifungal activity IB-367 and BM-1 exhibited antimicrobial activity against all the bacteria, but were less active against the fungi (see Table 1) In contrast, BM-2 showed a marked decrease in activity against all the bacteria Surprisingly, BM-1 was far more active than IB-367 against Gram-positive bacte-ria and fungi, showing better inhibition than IB-367 of
S aureus, S epidermidis and fungi The antimicrobial activity of IB-367 and BM-1 against Gram-negative bacteria was similar, however Interestingly, BM-2, inactive against bacteria, displayed a better antifungal activity than either IB-367 or BM-1 against Aspergil-lus niger Comparison of the PG-1 and BM-1 peptides showed that BM-1 exhibited slightly increased activity against Gram-positive bacteria and slightly decreased activity against Gram-negative bacteria and fungi than did PG-1
The comparison of the C albicans minimal inhibi-tory concentrations (MICs) obtained in this work with the values found by Barchiesi et al [31] showed much lower values found by Barchiesi et al [31] (2.0–
32 lgÆmL)1compared with 256 lgÆmL)1in our work) The differences between these MICs resulted from the fact that Barchiesi et al [31] used different experimen-tal conditions and another strain of C albicans Barchiesi et al [31] used RPMI 1640 medium, Mops buffer (pH 7.2) and 50% reduction of the initial inocu-lum They also used another strain of C albicans ATCC 90029 In our work, we used Sabouraud 5% glucose medium, pH 7.4 and 99% reduction of the initial inoculum In addition, we used the reference strain of C albicans (C albicans ATCC 10231)
Trang 4The haemolytic activity (see Table 2), using human
red blood cells as targets, was measured for PG-1,
IB-367, BM-1 and BM-2 peptides The IC50 results
showed fourfold reduced haemolysis of the BM-1
pep-tide (32 lgÆmL)1) relative to IB-367 (8 lgÆmL)1) BM-2
was not cytotoxic when tested against human red
blood cells (> 256 lgÆmL)1)
NMR results
The two-dimensional NMR spectra of the peptides,
obtained via the standard sequential assignment
meth-ods developed by Wu¨thrich [32], were assigned
(Figs S1 and S2, see Supporting information) The
proton and carbon aC chemical shifts, 3JNH–aH
cou-pling constants and amide-proton temperature
coeffi-cients are given in Tables S1–S4 (see Supporting
information) The1H and13C NMR chemical shifts of
IB-367 and BM-1 were well dispersed, a property
char-acteristic of b-sheet structures [32] Moreover, the good
dispersion of the1H chemical shifts permitted the
iden-tification of all the protons in the amino acid
side-chains, which was of further help in obtaining the
v torsion angles and in precise structural calculations
One distinct set of residual proton resonances in all
spectra was displayed for IB-367 and BM-1 The
chemical shifts of IB-367 and BM-1 were very similar
(Tables S1 and S2, see Supporting information), except for Phe10, which indicates that this phenylalanine resi-due in BM-1 points in the opposite direction to that in IB-367 The chemical shifts of IB-367 and BM-1 were also similar to the previously published data for PG-1 acquired at room temperature in water and in deuter-ated dimethyl sulphoxide [7,13] The dispersion of chemical shifts in BM-2 was not as good as in IB-367 and BM-1; there were also two sets of signals (major and minor) in the NMR spectra Analysis of rotating frame Overhauser effect spectroscopy (ROESY) and nuclear Overhauser effect spectroscopy (NOESY) revealed the trans geometry of all the peptide bonds in IB-367 and BM-1; the geometry of the Arg6–Pro7 pep-tide bond (major conformation) in BM-2 was cis The
dHa–Ha(i,i + 1) connectivity characteristic of the cis peptide bond was seen in the NOESY spectrum, and the relevant chemical exchange cross-peaks of this bond were present in the ROESY spectrum, indicating its cis–trans isomerization (trans geometry in the minor species) The cis–trans isomerization of the Xaa–Pro peptide bonds located in the turns is a common feature of peptides [33]; it is hardly surprising to find this conformational equilibrium in BM-2 As men-tioned above, at least two sets of proton resonances were present in the NMR spectrum of BM-2 Com-plete analysis, however, would require a separate conformational analysis for each set of resonances; too few proton resonances were obtained for the minor species to perform reliable three-dimensional structural calculations
The chemical shift index (CSI), defined as the devia-tion of the random-coil chemical shift from the experi-mental value, is a very sensitive indicator of the secondary structure of peptides and proteins [34] In
Table 2 Haemolytic activity (lgÆmL)1) (IC50 values –
concentra-tions that cause 50% haemolysis) of the PG-1, IB-367, BM-1 and
BM-2 peptides.
Table 1 Antimicrobial activity of the PG-1, IB-367, BM-1 and BM-2 peptides MBC⁄ MFC, minimal bactericidal ⁄ fungicidal concentration; MIC, minimal inhibitory concentration.
Organism
Gram-positive bacteria
Gram-negative bacteria
Fungi
Trang 5the case of IB-367, most of the amino acids had CSIs
of –1 (Leu4–Arg8 and Arg10–Arg17), a value
charac-teristic of b-sheet structure formation (see Fig 2) The
CSI pattern of IB-367 resembled that of PG-1, where
the Leu5–Arg10 and Cys13–Gly17 regions exhibited
typical b-strand chemical shifts, whereas the Arg9–
Arg11 region exhibited deviations from the b-strand
chemical shifts [7,13] The CSIs of the aC atoms in
BM-1 and BM-2 displayed no regularity, indicating a
predominantly random structure
Figure 2 summarizes the NOE pattern, vicinal
cou-plings 3JNH–aH and temperature coefficients of the
amide protons in the investigated peptides In IB-367
and BM-1, the presence of strong daN(i,i + 1) and
weak (or absence of) dMN(i,i + 1) and dbN(i,i + 1)
NOE connectivities in the Gly3–Arg8 and Cys14–
Gly16 regions of IB-367 and in the Gly2–Cys6 and
Val12–Val14 regions of BM-1 indicates a significant
population of conformers with dihedral angles in
the b-strand region of (/,w) space [35–37] In both
peptides, the weak or absent dNN(i,i + 1) NOE effects
indicate an unordered structure in the N-terminal
frag-ments and bend structures in the Arg10–Val13 region
of IB-367 and the Cys6–Cys11 region of BM-1 The
daa(5–14; 7–12) of IB-367 and daa(4–13; 6–11) of BM-1
NOEs strongly suggest a disulphide pattern in both
peptides; this results from subsequent calculations
Several long-range NOEs for IB-367 and BM-1
(Fig 2) were found, which involved residues from
N- and C-termini, in agreement with a two-stranded
antiparallel b-structure Moreover, the high values of
the vicinal coupling constants 3JNH–aH (> 9.0 Hz)
in whole peptide sequences and the temperature
coefficients of many amide protons higher than –
3.0 ppbÆK)1 (Fig S2A, B, Tables S1 and S2, see
Supporting information) strongly confirmed the pres-ence of a b-sheet structure in IB-367 and BM-1 Inspection of the NOE pattern of BM-2 (major con-formation, Fig 2C) showed a lack of diagnostic
dMN(i,i + 1) NOEs and provided evidence for the unstructured conformation of this peptide in the mid-dle part of the peptide (Tyr5–Val10) Some strong
daN(i,i + 1) NOEs in the Gly2–Cys4 and Val10–Gy13 regions suggested a b-structure In addition, the lack of hydrogen bonds in BM-2 indicated a flexible structure
Finally, no NOE cross-peaks, indicative of oligo-meric association in solution, could be detected, consis-tent with the high abundance of positively charged residues (four arginines in IB-367 and three arginines
in BM-1 and BM-2) in the primary structure of the peptides
Structural analysis Conformational analysis was performed for the three PG-1 analogues, in an attempt to correlate their struc-ture and activity in comparison with native PG-1, which has been shown previously to form a highly stable, rigid b-hairpin [7,13,18] IB-367 also adopts a well-defined b-hairpin structure, as expected from the sequence simi-larities with PG-1 (Fig 3A, B) The 300 structures of IB-367 were well defined, with an rmsd value of the
Ca atoms of all residues of 2.57 A˚, falling to 1.30 A˚ in the Cys5–Cys14 region (Table S5, see Supporting information) The solution structure of IB-367 consisted
of two antiparallel b-strands in the Tyr6–Arg8 and Phe11–Val13 regions linked by two residues – Gly9 and Arg10 The Arg8–Phe11 fragment formed a well-defined type II¢ b-turn, stabilized by a hydrogen bond between
Fig 2 CSIs relative to sodium 3-(trimethylsilyl)-(2,2,3,3- 2 H4)-propionate, summary of intra- and inter-residual NOEs among the backbone NH,
aH and bH, vicinal coupling constants3J HN–Ha measured in deuterated dimethyl sulphoxide at 22 C, and temperature coefficients of amide protons measured in deuterated dimethyl sulphoxide at 22, 25, 27, 30, 32, 35 and 37 C for IB-367 (A), BM-1 (B), BM-2 major (C) and BM-2 minor (D) CSIs were equal to zero or were not calculated for amino acid residues with open squares Bar height indicates the strength of the NOE correlation as strong, medium or weak Filled squares show3J NH–Ha coupling constants > 9.0 Hz, and filled circles the temperature coefficients of amide protons higher than –3.0 ppbÆdeg)1.
Trang 6the CO group of Arg8 and the NH group of Phe11,
found in all the calculated structures The b-sheet was
strongly stabilized by five regular backbone–backbone
hydrogen bonds – NH(Tyr6)–O(Val13), NH(Arg8)–
O(Phe11), NH(Arg10)–O(Arg8), NH(Phe11)–O(Arg8)
and NH(Val13)–O(Tyr6) – found between the
antiparal-lel strands in most of the calculated structures All
of these structures adopted a very similar b-hairpin structure in the middle region of the molecule with unordered N- and C-termini In addition, the N- and C-terminal fragments pointed in opposite directions as a result of the electrostatic repulsions of Arg1 and Arg17 (see Fig 3) The side-chains in all the structures were well defined, particularly in the middle region of the peptide (Fig 3A), owing to the presence of numerous interstrand NOEs In contrast, the side-chains at the N- and C-termini displayed large conformational variability Two interstrand disulphide bridges adopted
a well-defined, right-handed conformation IB-367 formed a characteristic amphipathic structure, display-ing a hydrophobic face formed by the bulked, hydrophobic residues (Leu4, Tyr6, Phe11, Val13, Val15) located on the concave surface of the peptide Two apolar disulphide bridges and charged Arg17 side-chains formed the second face of the peptide Two hydrophilic regions were located at the two spatial tips of the molecule, at both termini with Arg1 and Arg17, and in the turn in the presence of Arg8 and Arg10 (see Figs 3 and 4)
Our conformational studies showed that BM-1 formed a twisted b-sheet structure, similar to that of native PG-1 and IB-367 (Fig 3C, D) The structures of BM-1 were well defined in the backbone, with rmsd values of the Ca atoms of all residues of 2.60 A˚ and 1.82 A˚ in the Cys4–Cys13 region (Table S6, see Supporting information) The BM-1 structures con-sisted of two antiparallel b-strands in the Leu3–Cys6 and Cys11–Cys13 regions, linked by Arg7–Phe10 residues The turn region was formed by a type II¢ b-turn, as in IB-367 b-Hairpin stabilization was guar-anteed by three regular backbone–backbone hydrogen bonds – NH(Leu3)–O(Cys13), NH(Cys13)–O(Leu3) and NH(Cys13)–O(Cys4) – between the disulphide bridges in most calculated structures Although the backbone of BM-1 was well defined, the side-chains were much less clearly defined than in IB-367 (cf Fig 3A, C) Two interstrand disulphide bridges adopted a right-handed or extended conformation In most of the calculated structures, the disulphide bonds were located at the same peptide face, but, in some, the disulphide bonds were on the opposite side of the pep-tide face Thus, it was difficult to state unequivocally which residues were located on any given face of the peptide BM-1 also formed a characteristic amphipathic structure, which displayed a hydrophobic face formed
by the hydrophobic residues and the disulphide bridges, and a second polar face formed by charged Arg1, Arg7 and Arg9 side-chains (see Figs 3 and 4) In general, BM-1 was more flexible than PG-1 and IB-367, but, as
Fig 3 Superimposed conformations of the aC atoms of residues
Cys5–Cys13 of IB-367 (130 conformations) (A), Cys4–Cys13 of
BM-1 (93 conformations) (C) and Cys4–CysBM-1BM-1 of BM-2 (95
conforma-tions) (E) The most populated families of conformations are
shown Averaged structures of IB-367 (B), BM-1 (D) and BM-2 (F)
peptides The backbone is shown in ribbon representation, the
side-chains in stick representation Arginine residues are shown in
blue, disulphide bonds in yellow.
Trang 7in the last two peptides, two hydrophilic regions were
located at the two spatial tips of the molecule, at the
N-terminus with Arg1 and in the turn in the presence
of Arg7 and Arg9 (see Figs 3 and 4)
The shorter analogue, BM-2, was the most flexible of
all the peptides studied in this work BM-2 formed
major and minor conformations – this was easily read
from the NMR spectra The calculated structures of the
major BM-2 conformation were predominantly
unor-dered, especially in the turn region of the peptide with
the cis peptide bond between Arg6 and Pro7 (Fig 3E,
F) There were no hydrogen bonds in the calculated
structures and no regularities in the secondary structure
The rmsd values of all the aC atoms in the 300
calcu-lated structures were 2.71 A˚ and 1.54 A˚ in the Cys4–
Cys11 region (Table S7, see Supporting information)
The conformational ensemble of peptides,
deter-mined by molecular dynamics simulation restraints
from NMR experiments, was clustered into families
Ten families were found for IB-367 and BM-2, and six
for BM-1, at an rmsd cut-off of 5.0 A˚, one of which
was dominant (130 molecules) for IB-367, one (93
mol-ecules) for BM-1 and two (95 and 65 molmol-ecules) for
BM-2 The conformations in all the families for IB-367
and BM-1 had one feature in common: the central part
of the structure was better defined than the C- and
N-terminal parts The conformational differences
between the structural families of these peptides
applied mainly to the varied structures of the N- and
C-termini All the features of the structures calculated
for all three peptides were in very good agreement with
the experimental NMR data
Discussion
The presence of a cationic, amphiphilic b-sheet is key
to maintaining the biological activity and stability of
PG-1-like peptides (IB-367 and BM-1) The highly
flexible analogue without a b-sheet structure (BM-2)
has no antimicrobial activity Previous studies on
prote-grin variants have also shown that the antimicrobial
activity is highly dependent on b-hairpin stabilization
by disulphide bonds or backbone cyclization [16–
19,38,39] IB-367 and BM-1 share characteristic
physi-cochemical properties with most antimicrobial peptides,
adopting a b-hairpin-like structure with two disulphide
bridges [39,40] Sequence alignments revealed great
sim-ilarities between IB-367 or BM-1 and PG-1 from
por-cine leucocytes [6], gomesin from mygalomorph spider
haemocytes [41] and androctonin from scorpions [42]
All have a molecular mass of approximately 2 kDa,
including a rather high percentage (> 20%) of basic
residues Their three-dimensional structures are
stabi-lized by two internal disulphide bridges Most have
a broad spectrum of antimicrobial activity against various microorganisms
Comparison of the PG-1 structure from the Protein Data Bank (1PG1 [13] in water and 1ZY6 [14] in dodecylphosphocholine micelles) with the structures of IB-367 and BM-1 obtained here reveals several charac-teristic differences (see Fig 4) The b-sheet structures
of our peptides are shorter than that of PG-1 The positive charge in PG-1 in water turns almost the whole of one side of its structure into a cationic surface, whereas, in PG-1 in dodecylphosphocholine micelles, IB-367 and BM-1, the positive charge is distributed separately The b-sheet structures of IB-367 and BM-1, as in all the PG-1 peptide family, are char-acteristically amphipathic, with one surface hydrophilic and one hydrophobic Such a structure is essential to both Gram-positive and Gram-negative antimicrobial activity [16] PG-1, IB-367 and BM-1 form four-resi-due b-turns, but, in PG-1, there is an atypical b-turn, possibly caused by the presence of Arg10 in position
i+ 1 of the turn, whereas, in IB-367 and BM-1, a type II¢ b-turn is formed In PG-4, IB-367 and BM-1 analogues, the Arg10 residue is replaced by a glycine and, in PG-5, by a proline residue Glycine and proline residues are better suited than arginine to induce a canonical b-turn conformation [43] In addition, there
is only one cationic Arg1 residue on the N-terminus
in BM-1, rather than two arginines (Arg1 on the N-terminus and Arg18 in PG-1 or Arg17 in IB-367 on the C-terminus), one at the N- and the other at the C-terminus; this is sufficient to ensure the cationic nature of the peptide at its terminus The structure of BM-1 is very compact, like the PG-1 structure, whereas that of IB-367 is more expanded These struc-tural features could be responsible for the better antimicrobial activity of BM-1 than IB-367 against Gram-positive bacteria
Natural b-hairpin-like antimicrobial peptides other than PG-1 (gomesin [41], tachyplesin I [44], polyphemu-sin I [45] and androctonin [46]), with two disulphide bridges, are structurally similar to IB-367 and BM-1 Like PG-1, gomesin and polyphemusin contain 18 amino acids, but tachyplesin contains 17 residues and androctonin is significantly longer with 25 residues Although the spacing of the cysteine residues differs in these peptides from those studied here, all the molecules adopt a similar rigid plated b-sheet structure Androc-tonin has an unequal number of residues on each strand between the two bridges – five in the N-terminal strand and three in the C-terminal strand This causes
a greater twist in the b-sheet of androctonin com-pared with the other peptides Despite such differences,
Trang 8the rmsd value of the coordinates of the peptide
b-strands (IB-367, BM-1, PG-1, gomesin, tachyplesin I,
polyphemusin I and androctonin), when superimposed
on the backbone atoms, is approximately 6 A˚
Comparison of the hydrophilic⁄ hydrophobic
proper-ties on the molecule surfaces shows that the structures
of IB-367, BM-1, PG-1, gomesin, tachyplesin I and
polyphemusin I share two highly hydrophilic and
positively charged poles in the N- and C-terminal
regions and in the turn Androctonin also has a highly
hydrophilic and positively charged turn and
N-termi-nus, but, in contrast, the C-terminus containing
Pro24–Tyr25 is hydrophobic [46] There is a large difference in the distribution of hydrophobic⁄ hydro-philic potentials on the b-sheet surface between the tails and the turn The central portion of IB-367, BM-1 and PG-1 is particularly hydrophobic, as it con-tains only apolar residues distributed on either side of the b-sheet The b-sheet in gomesin is divided into two nonequivalent faces: the hydrophobic side-chains are clustered on the concave face, whereas the two polar side-chains flanked by the apolar disulphide bridges are located on the other face [41] In tachyplesin I and polyphemusin I, the cationic arginine and lysine
Fig 4 Structures of PG-1 (Protein Data
Bank code 1PG1 [13]), PG-1 (Protein Data
Bank code 1ZY6 [14]), IB-367, BM-1 and
BM-2 The backbone is shown in ribbon
representation, the side-chains of arginine
(blue) and cysteine (yellow) residues in stick
representation The electrostatic potential
was calculated on the van der Waals’
surface Positive potential is shown in blue,
neutral in grey.
Trang 9residues are also located in the central part of the
b-hairpin structures In androctonin, the highly twisted
character of the b-sheet does not suggest a clear
dichotomy in the distribution of polar and apolar
residues [46] Differences in the distribution of
hydro-philic and hydrophobic residues at the surface of the
peptides may indicate different modes of action on the
membrane This may also account for differences in
the haemolytic activity of the peptides A better
under-standing of the mode of action of these peptides is
crucial for the development of a new generation of
antibiotics
It is known that, in the absence of both disulphide
bonds, or even one disulphide bond, the b-sheet
struc-ture is less stable, and the antimicrobial activity is
much reduced [16–19,47,48] This was also found in
BM-2, which has no b-hairpin structure and is inactive
against bacteria The single disulphide bond and the
proline residue in position i + 1 of the reverse b-turn
prevent b-hairpin formation and are responsible for
the great flexibility of BM-2 in solution Lai et al [18]
obtained similar results with their analogue 12
Interestingly, BM-2 is more active than IB-367 and
BM-1 against A niger; the considerable plasticity of
the BM-2 structure may permit better activity against
this fungus
The current study shows that, with its broad spectrum
of antimicrobial activity, especially against
Gram-positive bacteria, the BM-1 analogue could be a good
molecule for clinical development Moreover, the BM-1
peptide shows fourfold reduced haemolysis relative
to IB-367, an additional advantage of this peptide
This peptide is easy to synthesize; it is 15% cheaper to
produce than IB-367
Materials and methods
Peptide synthesis
The peptides were solid-phase synthesized on Polystyrene
AM-RAM resin (0.76 mmolÆg)1, Rapp Polymere, Tu¨bingen,
Germany) using 9-fluorenylmethoxycarbonyl chemistry [49],
all the relevant reagents being obtained from Sigma-Aldrich
(Poznan´, Poland) The procedure was as follows: (a) 2 and
20 min deprotection steps using 20% piperidine in
dimethyl-formamide in the presence of 1% Triton X-100; (b) the
cou-pling reactions were carried out with the protected amino
acid diluted in dimethylformamide in the presence of 1%
Triton X using diisopropylcarbodiimide as coupling reagent
in 1-hydroxybenzotriazole for 2 h The completeness of each
coupling reaction was monitored by the chloranil test [50]
If positive, the coupling reaction was repeated using
O-(benzotriazol-1-yl)-N,N,N¢,N¢-tetramethyluronium
tetra-fluoroborate and 1-hydroxybenzotriazole in diisopropyleth-ylamine and mixed for 2 h The protected peptidyl resin was treated with a mixture of 90% trifluoroacetic acid, 2.5% ethanedithiol, 2.5% phenol, 2.5% water and 2.5% triiso-propylsilane for 2 h After cleavage, the solid support was removed by filtration, and the filtrate was concentrated under reduced pressure The cleaved peptide was precipi-tated with diethyl ether The linear product was oxidized with 0.1 m I2in CH3OH The peptide was purified by HPLC
on a Knauer K501 two-pump system with a Kromasil C8 column [10· 250 mm; particle diameter, 5 lm; pore size,
100 A˚; flow rate, 5 mLÆmin)1; gradient, 10–60% A⁄ 120 min (A, 0.1% trifluoroacetic acid in acetonitrile; B, 0.1% aque-ous trifluoroacetic acid), absorbance at 226 nm] The result-ing fractions with a purity of better than 96–98% were tested by HPLC (Hypersil C18 column, 4.6· 250 mm) The peptides were analysed by matrix-assisted laser desorption ionization-time of flight mass spectrometry
Organism and antimicrobial assay
The reference strains were supplied by the Polish Collection
of Microorganisms (Polish Academy of Sciences, Institute
of Immunology and Experimental Therapy, Wrocaw, Poland)
MICs were determined using a broth microdilution method with Mueller–Hinton broth (Becton Dickinson, Le Pont de Claix, France) at initial inocula of 5· 105 colony-forming units (cfu)ÆmL)1 for bacteria, and Sabouraud 5% dextrose broth pH 7.4 (Sigma-Aldrich) at initial inocula of
5· 103
cfuÆmL)1 for fungi, according to the procedures of the Clinical and Laboratory Standards Institute (formerly National Committee on Clinical Laboratory Standards) Polystyrene 96-well plates (Sigma-Aldrich) were incubated
in air at 37C for 18 h (bacteria) and at 25 C for 72 h (fungi) MIC was taken as the lowest drug concentration at which observable growth was inhibited The minimum bac-tericidal concentration was taken as the lowest concentra-tion of each drug resulting in > 99.9% reducconcentra-tion of the initial inoculum Experiments were performed in triplicate
on three different days
Haemolytic activity
Freshly collected human blood was washed with NaCl⁄ Pi
(pH 7.4) until the supernatant became colourless A suspen-sion was made of 0.5 mL packed cells in 10 mL of NaCl⁄ Pi Peptides were dissolved in NaCl⁄ Pi and serially diluted on a 96-well polystyrene microtiter plate The final concentration of peptides ranged from 256 to 0.5 lgÆmL)1 Twenty microlitres of cell suspension were added As a positive control (100% lysis), a 10% solution of Triton X was used After incubation for 4 h at 37C, haemolysis was observed and compared with the positive control
Trang 10(Triton X) A positive result (IC50) was defined when 50%
of haemolysed red blood cells were taken
NMR experiments
The NMR samples were prepared by dissolving 6 mg peptide
in 0.5 mL deuterated dimethyl sulphoxide (peptide
concentration, 3 mm) The same solvent was used in
conformational studies of native PG-1 [7] to enable a further
three-dimensional structural comparison of the studied
peptides with PG-1 Deuterated dimethyl sulphoxide was
also used in conformational studies of other bioactive
peptides A sample contained a small amount of
trifluoroace-tic acid in order to downfield shift the vestigial water signal
and to retard amide-proton exchange Chemical shifts
are given relative to sodium 3-(trimethylsilyl)-(2,2,3,3-2H4
)-propionate, the internal chemical shift standard
1H-NMR spectra (Varian Unity+ 500 NMR
spectro-meter, Varian Inc., Palo Alto, CA, USA) were obtained at
a proton frequency of 500 MHz and the following
temper-atures: 22, 25, 27, 30, 32, 35 and 37C Two-dimensional
spectra, including double-quantum filtered correlation
spec-troscopy, total correlation spectroscopy (60 ms), NOESY
(100 and 200 ms), ROESY (200 ms) and1H–13C
heteronucle-ar single quantum coherence spectroscopy, were obtained at
22C NMR data were processed with vnmr [51] and
analy-sed with xeasy software [52] Both ROESY spectra (200 ms)
were compared with the NOESY spectra (200 ms) and no
additional peaks demonstrating the appearance of a
spin-diffusion effect were found The NOESY spectra showed
better quality and, for this reason, were used for further
calculations Assignments were carried out according to
standard procedures, including spin-system identification
and sequential assignment [32] In the case of the
one-dimen-sional NMR spectra, 16 000 data points were collected and a
spectral width of 6 kHz was used The two-dimensional
homonuclear experiments were measured using a proton
spectral width of 4.5 kHz, collecting 2000 data points
3JNH-aH vicinal coupling constants were determined by
two-dimensional double-quantum filtered correlation
spec-troscopy experiments Distance constraints and coupling
constants were used in the habas program [53] of the
dyanapackage [53] to generate /, w and v1dihedral angle
constraints and stereospecific assignments Dihedral angle
constraints were calculated from the Karplus equation with
A= 6.4, B =)1.4 and C = 1.9 [54]
NOE intensities were determined from the NOESY
(200 ms) spectra of the PG-1 analogues NOE volumes
were integrated and calibrated with xeasy software [52]
After internal calibration, the cross-peaks from the NOESY
experiments were converted into upper distance limits with
the caliba program of the dyana package [53]
Based on the experimental chemical shifts of aC nuclei,
CSIs were calculated relative to the sodium
3-(trimethylsi-lyl)-(2,2,3,3-2H4)-propionate reference [55] For cysteines,
the reference random-coil chemical shift was not reported
in [55]; hence, CSIs of cysteine amino acid residues were not calculated
The temperature dependence of the amide proton chemi-cal shifts was measured in order to determine whether any
of the amide protons were involved in hydrogen-bonding interactions The temperature coefficients (Dd⁄ DT) of the amide proton chemical shifts were measured from one-dimensional NMR spectra for the following temperatures:
22, 25, 27, 30, 32, 35 and 37C About 80% of all hydro-gen-bonded amides in proteins occur in the range )5 to
0 ppbÆdeg)1, and their average value is )3.2 ± 2.0 ppbÆdeg)1 [56] In our studies, we used the criterion of hydrogen bond formation by amide protons as a value higher than )3.0 ppbÆdeg)1, and all values more negative than )3.0 ppbÆdeg)1 indicated a lack of hydrogen bond formation
Structures of the peptides studied
The structures of the peptides studied were determined with xplor software, Version 3.1 [57], each structure being produced using distance and torsion angle restraints For the xplor three-dimensional structure calculations, the NOESY experiments provided 558 distance restraints for IB-367, 427 for BM-1 and 213 for the major conformation
of BM-2 The habas program provided 36, 24 and 21 torsion angles for IB-367, BM-1 and BM-2 (major confor-mation), respectively The structures were calculated with the standard xplor program modules, as well as the charmm force field [58] in vacuo, starting from a random structure For all three molecules, 300 cycles of simulated annealing were carried out, each with 27 000 iterations of
80 ps with 3 fs steps The molecule was maintained at
1000 K for 50 ps and annealed at 100 K for 29 ps In the last 200 iterations (1 ps), energy was minimized with Powell’s algorithm [59] During simulated annealing refine-ment, the molecule was slowly cooled from 1000 to 100 K over 30 ps Finally, 300 energy-minimized conformations were obtained
The set of final conformations was clustered using the graphic molmol program [60]; this was also used to draw, analyse and display the electrostatic potential on the van der Waals’ surface All conformations for each peptide were divided into families at an rmsd cut-off of 5.0 A˚
Acknowledgements
The authors wish to thank the State Committee for Scientific Research for grants DS 8440-4-0172-10 and
DS 8452-4-0135-9 This research was conducted using the resources of the Linux cluster at the Informatics Centre of the Metropolitan Academic Network (IC MAN) in Gdan´sk, Poland Beata Mickiewicz expresses