This article was originally published online as an accepted preprint. The ‘Published Online’ date corresponds to the preprint version. You can request a copy of the preprint by emailing the Biopolymers editorial office at firstname.lastname@example.org
Antimicrobial peptides are remarkably diverse, but the fundamental structural principle underlying all classes is their ability to adopt a shape in which clusters of hydrophobic and cationic amino acids are spatially organized in an amphipathic manner.1 On the basis of amino acid composition and three-dimensional structure, antimicrobial peptides can be divided into three classes. Peptides in class I, such as cecropin, magainin, and melittin, have linear α-helical structures with no cysteines. On the other hand, class II peptides, such as defensin, tachyplesin, and protegrin contain an even number of cysteine residues linked by disulfide bridges. The class III peptides, proline–arginine rich peptide (PR-39), tryptophan-rich peptide (indolicidin), and histidine-rich peptide (histatin), contain an unusually high proportion of one or two amino acids.1–5
Arenicin-1 (AR-1:RWCVYAYVRVRGVLVRYRRCW), a novel 21-residue antimicrobial peptide, was purified from coelomocytes of the marine polychaeta Arenicola marina (lugworm).6 This class II antimicrobial peptide contains a single disulfide bridge between Cys3 and Cys20, forming a large 18-residue ring.6 To determine the role of the disulfide bond and structure–activity relationships of arenicin, we synthesized AR-1 and its linear derivative, arenicin-1-S (AR-1-S: RWSVYAYVRVRGVLVRYRRSW), generated by substituting cysteines at positions 3 and 20 with serines (Table I). We evaluated their toxicities to bacteria and human erythrocytes, as well as leakage ability from liposomes composed of a mixture of phosphatidylethanolamine (PE) and phosphatidylglycerol (PG), which mimics bacterial cytoplasmic membranes. The three-dimensional structures of AR-1 and AR-1-S were determined by 1H NMR spectroscopy. Elucidation of the structure-activity relationships of these antibiotic peptides should facilitate the rational design of potent bacterial-selective antimicrobial peptides.
Table I. Amino Acid Sequences of AR-1 and its Linear Derivative, AR-1-S
AR-1 (arenicin-1) and its linear derivative, AR-1-S, were prepared using the standard Fmoc-based solid-phase method.7 Fmoc(9-fluorenylmethoxycarbonyl)-Trp-Wang-resin, Fmoc-amino acids and other reagents for peptide synthesis were purchased from Calbiochem-Novabiochem (La Jolla, CA). Dicyclohexylcarbodiimide and N-hydroxybenzotriazole were used as coupling reagents, and 10-fold excess Fmoc-amino acids were added during every coupling cycle. After cleavage and deprotection with a mixture of trifluoroacetic acid (TFA)/water/thioanisole/ethanedithiol/triisopropylsilane (81.5:5:5:5:2.5:1, v/v) for 2 h at room temperature, crude peptides were repeatedly extracted with diethyl ether. Reduced crude AR-1 was oxidized in ammonium acetate solution (0.1M ammonium acetate, 0.1M NaCl, 20 mM Na2HPO4, 1 mM EDTA, pH 8.0) in the presence of reduced and oxidized glutathiones (1:100:10 molar ratio of peptide:GSH:GSSG) for 24 h at room temperature.8 Oxidized crude AR-1 was purified by high-performance liquid chromatography (HPLC) on a preparative (15 μm, 20 × 250 mm2) C18 Vydac column using a water-acetonitrile gradient (0–80%) containing 0.05% TFA. The final purity of peptides (>98%) was assessed by HPLC on a Vydac C18 reversed-phase analytical column (5 μm, 4.6 × 250 mm2), and their identities confirmed by matrix-assisted laser-desorption ionization-time-of-flight mass spectrometry (Shimadzu, Japan) (Table I).
The antimicrobial activities of peptides were examined in sterile 96-well plates in a final volume of 100 μl, using the following procedure: briefly, aliquots (100 μl) of bacterial suspension at 2 × 106 colony-forming units/ml in 1% peptone were added to 100 μl of peptide solution (serial twofold dilutions in 1% peptone). After incubation for 18–20 h at 37°C, inhibition of bacterial growth was determined by measuring absorbance at 620 nm with a Micro plate auto reader EL 800 (Bio-Tek Instruments). The minimal inhibitory concentration (MIC) was defined as the lowest concentration of peptide required to inhibit bacterial growth. Two types of gram-negative bacteria (Escherichia coli [KCTC 1682] and Pseudomonas aeruginosa [KCTC 1637]) and two types of gram-positive bacteria (Staphylococcus aureus [KCTC 1621] and Staphylococcus epidermidis [KCTC 1917]) were purchased from the Korean Collection for Type Cultures (KCTC) at the Korea Research Institute of Bioscience and Biotechnology.
The hemolytic activities of peptides were examined against human red blood cells (h-RBC). Fresh h-RBCs were washed three times with phosphate buffered saline (PBS; 35 mM phosphate buffer containing 150 mM NaCl, pH 7.4) by centrifugation for 10 min at 1000g, and resuspended in PBS. Peptide solutions were added to 50 μl of h-RBC in PBS to obtain a final volume of 100 μl and a final erythrocyte concentration of 4% (v/v). The resulting suspension was incubated with agitation for 1 h at 37°C. Samples were centrifuged at 1000g for 5 min. Hemoglobin release was monitored by measuring absorbance at 405 nm of the supernatant. h-RBCs suspended in PBS and 0.1% Triton-X 100 were employed as controls for no hemolysis (blank) and 100% hemolysis, respectively. The percent hemolysis was calculated using the following equation:
Dye Leakage From Lipid Vesicles
Calcein-entrapped LUVs composed of EYPC/EYPG (7:3, w/w) were prepared by vortexing dried lipid in dye buffer solution (70 mM calcein, 10 mM Tris, 150 mM NaCl, 0.1 mM EDTA, pH 7.4). The suspension was freeze-thawed in liquid nitrogen for 10 cycles, and extruded through two-stacked polycarbonate filters of 100 nm pore size with a LiposoFast extruder (Avestin. Canada). Untrapped calcein was removed by gel filtration on a Sephadex G-50 column. Lipid vesicles were diluted ∼10-fold after passing through a Sephadex G-50 column. Eluted calcein-entrapped vesicles were further diluted to the desired final lipid concentration. Leakage of calcein from LUVs was monitored by measuring fluorescence intensity at an excitation wavelength of 490 nm and emission wavelength of 520 nm, using a model RF-5301PC spectrophotometer (Shimadzu, Kyoto, Japan). For determination of 100% dye release, 10% Triton-X100 in Tris-buffer (20 μl) was added to dissolve the vesicles. The percentage of dye leakage caused by peptides was calculated as follows:
where F is the fluorescence intensity achieved by peptides, and F0 and Ft are fluorescence intensities without peptides and with Triton X-100, respectively.
Perdeuterated DMSO (DMSO-d6) was purchased from Cambridge Isotope Laboratories (Andover, MA). Peptides were dissolved in 0.45 ml of DMSO to a concentration of 1.0 mM. Double quantum-filtered correlation spectroscopy (DQF-COSY), total correlation spectroscopy (TOCSY), and nuclear Overhauser effect spectroscopy (NOESY) were performed using the time-proportional phase incrementation method.9–14 TOCSY experiments were performed with 60 ms MLEV-17 spin-lock mixing pulses. Mixing times of 150 and 250 ms were used for NOESY experiments. For DQF-COSY experiments, 400 transients with 4 K complex data points were collected for each increment, and data along the t1 dimension were zero-filled to 4 K before two-dimensional Fourier transformation.3JHNα coupling constants were measured from DQF-COSY spectra with a spectral width of 5202.33 Hz and digital resolution of 1.27 Hz/point. Chemical shifts were expressed relative to the tetramethylsilane (TMS) signal at 0 ppm. To investigate intramolecular hydrogen bonding in the peptides, temperature coefficients were calculated from the TOCSY experiments at 298, 303, and 308 K. All NMR spectra were recorded on a Bruker Avance-400 spectrometer at Konkuk University and Bruker Avance-600 MHz spectrometer at Korea Basic Science Institute. NMR spectra were processed off-line using the FELIX software package (Accelrys, San Diego, CA) on an SGI workstation.
Distance constraints were extracted from NOESY spectra with mixing times of 150 and 250 ms. The volumes of the nuclear Overhauser effects (NOEs) between the two beta protons of the Tyr residue were used as the reference. All other volumes were converted into distances by assuming a 1/r6 distance dependence. All NOE intensities were divided into three classes, depending on their distance ranges (strong, 1.8–2.7 Å; medium, 1.8–3.3 Å; and weak, 1.8–5.0 Å).15, 16 Structure calculations were performed using CNSSOLVE GENERAL-RELEASE 1.1 (Yale University, New Haven, CT) with the topology and parameter sets ‘topallhdg’ and ‘parallhdg’, respectively. Standard pseudoatom corrections were applied to the nonstereospecifically assigned restraints,17 and an additional 0.5 Å was added to the upper bounds for NOEs involving methyl protons.18 A hybrid distance geometry-dynamical simulated annealing protocol19, 20 was applied to generate the structures. A total of 400 structures were generated. Among these, 20 structures with the lowest energies were selected for further analyses.
Antimicrobial and Hemolytic Activities
The MIC values of the peptides against two strains of gram-negative bacteria (Escherichia coli and Pseudomonas aeruginosa) and two stains of gram-positive bacteria (Staphylococcus aureus and Staphylococcus epidermidis) are listed in Table II. AR-1 displayed potent antimicrobial activity, with MIC values ranging from 2.0 to 8.0 μM. Interestingly, the linear derivative, AR-1-S, was fourfold less active against Pseudomonas aeruginosa than AR-1 (Table II). Moreover, AR-1-S was ∼ twofold less active against the other bacterial strains tested, compared to AR-1. Next, we assessed the cytotoxicity of the peptides against mammalian cells by measuring their ability to induce lyses of human erythrocytes. Dose-response curves for hemolytic activities of the peptides are shown in Figure 1. AR-1-S displayed lower hemolytic activity than AR-1. These results suggest that the disulfide bond in AR-1 plays an important role in antibacterial and hemolytic activities.
Table II. Antibacterial Activities of AR-1 and its Linear Derivative
Minimal Inhibitory Concentration (μM)
Escherichia coli [KCTC 1682]
Pseudomonas aeruginosa [KCTC 1637]
Staphylococcus aureus [KCTC 1621]
Staphylococcus epidermidis [KCTC 1917]
Peptide-Induced Dye Leakage From Liposomes
To determine whether the bacterial cell membrane is the major target of peptides, we measured their abilities to induce leakage of a fluorescent dye from PE/PG (7:3, w/w) liposomes, which mimic negatively charged bacterial membranes. Both peptides were particularly effective in inducing calcein leakage from PE/PG (7:3, w/w) liposomes (Figure 2). Our results suggest that the peptides bind to and disrupt bacterial cell membranes, leading to cell death.
Resonance Assignments of AR-1 and its Linear Derivative
Sequence-specific resonance assignments were deduced from DQF-COSY, TOCSY, and NOESY data.21 Figure 3 depicts NOESY spectra with sequential assignments for AR-1 and its derivative in the NHCαH region. The chemical shifts of AR-1 and AR-1-S in DMSO at 298 K are listed in Tables III and IV, respectively. It is difficult to discern similarities in the chemical shift patterns of AR-1 and AR-1-S.
Table III. Chemical Shift of AR-1 in DMSO at 298 K
Sequential NOE connectivities and other NMR data are illustrated in Figure 4. In case of AR-1, there are a number of long-range NOEs between the protons in the two-stranded antiparallel β-sheet, strong dαN (i,i + 1) NOEs, and large 3JHNα coupling constants, which are characteristics of the β-sheet structure linked by a turn and disulfide bond between Cys3 and Cys18. However, the linear derivative, AR-1-S, displayed few interresidual NOEs, owing to a more flexible structure than AR-1.
Structures of AR-1 and AR-1-S
To calculate the tertiary structures of AR-1 and AR-1-S, we employed experimental restraints, such as sequential (|i − j | = 1), medium-range (1 < |i − j | ⩽ 5), long-range (|i − j | > 5), intraresidual distance, and torsion angle restraints. Among the structures accepted with small deviations from idealized covalent geometry and experimental restraints (⩽0.05 Å for bonds, ⩽5° for angles, ⩽5° for chirality, ⩽0.3 Å from NOE restraints, and ⩽3° from torsion angle restraints), we analyzed 20 output structures with the lowest energy for each peptide. None of the structures had violations over 0.5 Å from the NOE distance restraints or 3° from dihedral angle restraints and all exhibited good covalent geometry. The statistics of the 20 final simulated annealing structures of AR-1 are listed in Table V. Superimposition of the 20 lowest energy structures of AR-1 over the backbone atoms (from Ala6 to Ile10 and Val13 to Tyr17) revealed root mean squared deviations from mean structures of 0.437 ± 0.155 Å for the backbone atoms (N, Cα, C′, O) and 1.182 ± 0.200 Å for all heavy atoms.
Table V. Structural Statistics and Mean Pairwise Root Mean Squared (rmsd) Deviations for the 20 Lowest Energy Structures of AR-1 in DMSO at 298 K
Restraints for Structure Calculation
Experiment Distance Restraints
Dihedral Angle Restraints
Rmsd From Experimental Geometry
0.040 ± 0.001
0.354 ± 0.055
Rmsd From Covalent Geometry
0.003 ± 9.665E-05
0.445 ± 0.011
0.319 ± 0.023
Average Energies (kcal/mol)
51.254 ± 2.012
14.548 ± 1.190
0.231 ± 0.075
6.533 ± 1.257
Rmsd From the Mean Structure
Backbone Atoms of All Residues (1–21)
1.159 ± 0.345
All Heavy Atoms of All Residues (1–21)
2.454 ± 0.309
Backbone Atoms of Residues (6–10, 13–17)
0.437 ± 0.155
All Heavy Atoms of Residues (6–10, 13–17)
1.182 ± 0.200
Since AR-1-S does not have a disulfide bond, it is more flexible than AR-1, resulting in the lack of interresidual NOEs and the 20 lowest energy structures of AR-1-S were not converged. Therefore, Figure 5 depicts the superposition of the 20 lowest energy structures of AR-1 only, over the backbone atoms in DMSO. Structural analyses with PROCHECK disclosed that AR-1 has a two-strand antiparallel β-sheet structure from Ala6 to Ile10 and Val13 to Tyr17, whereas AR-1-S displays a random coil structure.
The orientation of the hydrophobic and hydrophilic side-chains of AR-1 is depicted in Figure 6. When an amphipathic α-helical peptide forms an ion channel, the hydrophilic residues face inwards to contact the solvent and hydrophobic side-chains face the acyl chains of the hydrophobic lipid.22–25 The hydrophobic side-chains (colored red) protrude to one side, and the hydrophilic side chains (blue) protrude to the other side, confirming the amphipathic character of AR-1.
DISCUSSION AND CONCLUSIONS
To determine the structure-function relationship of AR-1, we synthesized a linear derivative, AR-1-S, by substituting both cysteines with serines. AR-1-S was less active against bacterial strains and displayed lower hemolytic activity than AR-1. Moreover, AR-1 was more effective in inducing calcein leakage from PE/PG (7:3, w/w) liposomes, which mimic bacterial cytoplasmic membranes, compared to AR-1-S. Our results suggest that the disulfide bond plays an important role in antibacterial and hemolytic activities.
The tertiary structures of AR-1 and AR-1-S in DMSO were determined by NMR spectroscopy. The overall fold of AR-1 consists of a two-stranded antiparallel β-sheet (Ala6 to Ile10 and Val13 to Tyr17) connected by a four-residue β-turn (Val10-Val13), while the linear derivative, AR-1-S, exhibits a random coil structure. We performed NMR experiments on arenicin in SDS and DPC micelles. However, data were obscured by severe spectral overlap and isotropic mixing in TOCSY was extremely inefficient. Therefore, all subsequent structural studies were performed in DMSO, which mimics the hydrophobic environment of the biological membrane to some extent. As shown in Figure 4, the chemical shift index (CSI) of the Cα proton of AR-1 shows dense grouping of “1” uninterrupted by “−1”, strong dαN (i,i + 1) NOEs, and large 3JHNα coupling constants, characteristic of β-sheet structure. Additionally, long-range NOEs for AR-1 are the salient traits of antiparallel β-sheet structure.26
The mechanisms of action of α-helical antimicrobial peptides have been extensively studied. However, relatively little is known about the mechanism by which peptides with β-sheet structures and disulfide bonds cause membrane damage. Structure-function relationships of the cationic disulfide-linked antimicrobial peptides are rather complex. For example, in protegrin with two antiparallel β-sheet structures, both the disulfide bridge and a well-defined β-sheet structure are necessary for antimicrobial activity.27 However, in the case of tachyplesin, the rigid disulfide-bonded β-sheet structures is not absolutely essential for antimicrobial activity, and maintenance of the peptide hydrophobic–hydrophilic balance may be the critical parameter.28 Analogs designed by the deletion of all cysteine residues of tachyplesin I retained antimicrobial activity, which led to the proposal that the two disulfide bonds do not contribute significantly to antimicrobial activity.28 Similarly, the linear AR-1-S analog without a disulfide bond retains antimicrobial activity, compared to its parent peptide.
The amino acid compositions of AR-1 and AR-1-S are mainly hydrophobic, and include three tyrosines and two tryptophans critical for membrane permeabilization. The peptides have six cationic arginine residues that play key roles in interactions with the negatively charged bacterial membrane. AR-1-S is about two- to fourfold less active against bacterial cells than AR-1, but retains antibacterial activity. Therefore, our biological data imply that the hydrophobic–hydrophilic balance, disulfide bridge, as well as amphipathic β-sheet structure play important roles in peptide activity.
To date, several antimicrobial peptides containing a single disulfide bond have been purified from frog skin.29–33 All these peptides are characterized by a single intramolecular disulfide bridge between two cysteine residues forming a small 68-residue ring at the C-terminal part of the molecule. However, AR-1 has a single disulfide bridge (Cys3-Cys20) forming a large 18-residue ring.6 In this respect, arenicin is a new class of antimicrobial peptide.6
Most antimicrobial activities of peptides with a β-sheet structure manifest in the head of a rigid antiparallel β-pleated sheet structure linked by a β-turn to a hydrophilic tail comprising cationic residues derived from the N- or C-termini of the peptide.34 Similarly, AR-1 has a β-sheet structure with amphipathic characteristics (Figure 6), which displays a hydrophobic potential surface. Hydrophilic side chains in the N- and C-termini and β-turn on one face of the β-sheet protrude to one side, whereas hydrophobic residues on the other face protrude to the other side. The structures of AR-1, tachyplesin-I, and polyphemusin-I are shown in Figure 7.35, 36 On the basis color coding of hydrophobic and hydrophilic residues, we propose that arenicin has common structural characteristics, such as β-sheet, amphipathicity and a cationic hydrophilic tail, to other antimicrobial peptides with disulfide bridges. Since AR-1 has only one disulfide bond forming a large ring, the structure is not as rigid as other peptides with β-sheet structures held rigidly by two or more disulfide bonds. Therefore, antimicrobial activity can be enhanced by adding one or more disulfide bonds to increase structural rigidity. Moreover, to increase bacterial cell selectivity, hydrophilicity can be elevated while retaining the amphipathic structure of arenicin. Determination of the structure-activity relationships of arenicin should aid in the rational design of potent bacterial-selective antimicrobial peptides in the future.
Ju-Un Lee is supported, in part, by the second BK21 (MOE). We thank Korea Basic Science Institute for providing us NMR time (NMR Research Program of Korea Basic Science Institute).