Determining the Mode of Action Involved in the Antimicrobial Activity of Synthetic Peptides: A Solid-State NMR and FTIR Study

Abstract

We have previously shown that leucine to lysine substitution(s) in neutral synthetic crown ether containing 14-mer peptide affect the peptide structure and its ability to permeabilize bilayers. Depending on the substitution position, the peptides adopt mainly either a α-helical structure able to permeabilize dimyristoylphosphatidylcholine (DMPC) and dimyristoylphosphatidylglycerol (DMPG) vesicles (nonselective peptides) or an intermolecular β-sheet structure only able to permeabilize DMPG vesicles (selective peptides). In this study, we have used a combination of solid-state NMR and Fourier transform infrared spectroscopy to investigate the effects of nonselective α-helical and selective intermolecular β-sheet peptides on both types of bilayers. ³¹P NMR results indicate that both types of peptides interact with the headgroups of DMPC and DMPG bilayers. ²H NMR and Fourier transform infrared results reveal an ordering of the hydrophobic core of bilayers when leakage is noted, i.e., for DMPG vesicles in the presence of both types of peptides and DMPC vesicles in the presence of nonselective peptides. However, selective peptides have no significant effect on the ordering of DMPC acyl chains. The ability of these 14-mer peptides to permeabilize lipid vesicles therefore appears to be related to their ability to increase the order of the bilayer hydrophobic core.

Introduction

Antimicrobial peptides (AMPs) constitute new hope to fight pathogenic bacteria resistant against classical antibiotics. Indeed, these short peptides isolated from the innate immune systems of diverse species are able to prevent the invasion of pathogenic microorganisms (1). They possess a wide spectrum of antibacterial activity but are also effective against fungi, viruses, eukaryotic parasites, and tumor cells (2–4). Their primary mode of action is the destabilization of membranes, leading to a depolarization of target cells (5). Despite these advantages, no AMP or derivatives are presently suitable for commercialization because of their side effects and toxicity (6). Therefore, several efforts are devoted to better understand the importance of each parameter of AMPs on their activities, such as hydrophobicity, charges, and secondary structures. These will help to develop more effective and specific antimicrobial agents.

Biophysical techniques are often used to better understand the relationship between each parameter and the ability of AMPs (or derivatives) to destabilize bilayers or vesicles mimicking eukaryotic and bacterial cell membranes (7–10). Bilayers and vesicles used in these studies are often composed of zwitterionic phospholipids, like dimyristoylphosphatidylcholine (DMPC), or negatively charged phospholipids, like dimyristoylphosphatidylglycerol (DMPG), to respectively mimic eukaryotic or bacterial membranes. Biophysical studies on these model membranes allow us to better understand AMP/membrane interactions and to propose different mechanisms used by AMPs to destabilize the membrane (11–14).

We have developed a neutral synthetic peptide, called the base 14-mer peptide, which is amphiphatic under a α-helical structure, like helical AMPs (15). This peptide consists of 10 leucine residues and four phenylalanine residues that are attached to 21-crown-7 ether (Fig. 1). The positions of these crown ethers in the peptide sequence (e.g., 2, 6, 9, 13) confer an amphipathicity to the peptide when helical due to the alignment of all hydrophilic crown ether rings on the same side of the α-helix, whereas hydrophobic leucine residues are on the other side. This neutral peptide is able to permeabilize negatively charged and zwitterionic vesicles (15,16). Using this neutral peptide as a template, we have synthesized cationic peptides by substituting leucine by lysine residue(s). Our results indicated that some of these cationic peptides lost their ability to significantly permeabilize zwitterionic DMPC vesicles and, therefore, are selective toward negatively charged DMPG vesicles, whereas other cationic 14-mer peptides remain active against DMPC vesicles (17) and are therefore nonselective. The position(s) of leucine to lysine substitutions dictate this selectivity. We have also demonstrated that selective peptides do not adopt an amphipathic α-helical structure, like the base 14-mer peptide, but are structured under nonamphipathic intermolecular β-sheets in the presence of both DMPC and DMPG vesicles (17). These findings question the importance of helicity and amphipathicity in the selectivity of AMPs (17).

Figure 1.

(A) Sequences of the neutral base 14-mer peptide and examples of positively charged analogs used in this study. The sequences of other positively charged analogs are designed on the same concept. (B) β-strand diagram and (C) helical wheel projections showing the distribution of amino acid side chains of the base 14-mer peptide. Black circles indicate the leucine to lysine substitution(s).

We have investigated in this work the effects of selective and nonselective 14-mer peptides on phospholipid bilayers. First, solid-state NMR was used to obtain a molecular-level view of how selective and nonselective peptides interact with and destabilize DMPC and DMPG bilayers. More specifically, ³¹P NMR was used to obtain information on the structure and dynamics of lipid headgroups and thus the effects of AMPs on the membrane interface (18). In addition, ²H NMR was used to obtain information on the hydrophobic core of membranes, using phospholipids with deuterated acyl chains (19). Finally, Fourier transform infrared (FTIR) spectroscopy was used to characterize the effect of 14-mer peptides on the hydrocarbon chain conformational order (20).

Material and Methods

Materials

DMPC, DMPG with either protonated or deuterated (d₅₄) acyl chains were purchased from Avanti Polar Lipids (Alabaster, AL) and used without purification. HEPES, EDTA, trifluoroethanol, and deuterium-depleted water were purchased from Sigma-Aldrich (St. Louis, MO). Water used for buffer preparation was distilled and deionized using a Barnstead NANOpurII system (resistivity of 18.2 MΩ/cm; Boston, MA) with four purification columns. Deuterium oxide was obtained from CDN isotopes (Pointe-Claire, QC). All solvents were of reagent grade or high-performance liquid chromatography (HPLC) grade quality, purchased commercially and used without any further purification except for N,N-dimethylformamide (degassed with N₂) and dichloromethane (distilled). FMOC-protected amino acids were purchased from Matrix Innovation (Québec, QC). All other chemicals were of reagent grade.

Peptide synthesis

The 14-mer peptides were prepared by solid-phase synthesis as described previously (17) using Wang resin as solid support and N-Fmoc-protected amino acids. Peptide purity (>80%) was checked by reverse-phase HPLC using an Agilent 1050 chromatograph (Agilent Technologies, Santa Clara, CA), with a gradient of solvents A (90% H₂O/5% CH₃CN/5% isopropanol/0.1% trifluoroacetic acid (TFA)) and B (50% CH₃CN/50% isopropanol/ 0.1% TFA) over 45 min. Final characterization was done using an Agilent time-of-flight mass spectrometer (Agilent Technologies) 6210 with electrospray ionization.

Solid-state NMR

Multilamellar vesicles (MLVs) for solid-state NMR experiments were prepared by codissolving in CHCl₃/MeOH (1/1 v/v) 20 mg of lipids with a suitable amount of peptide in a lipid/peptide molar ratio of 60:1. For ²H experiments on DMPC vesicles, the total amount of lipids was 15 mg DMPC and 5 mg DMPC-d₅₄, whereas it was 18 mg DMPG and 2 mg of DMPG-d₅₄ for ²H experiments on DMPG vesicles. The lipid/peptide samples were dried under a stream of nitrogen and then stored under high vacuum overnight to completely remove any residual solvent.

The dry samples were hydrated and vortex mixed with 80 μL of HEPES 100 mM, EDTA 5 mM (pH 7.4) buffer in deuterium-depleted water (20% of lipids in water w/w). The resulting suspension underwent five freeze (liquid N₂)/thaw/vortex shaking cycles to ensure formation of multilamellar vesicles. For DMPC/DMPC-d₅₄/peptide samples, the thawing was done at 60°C, whereas for DMPG/DMPGd₅₄/peptide samples, the thawing was done at 37°C. The samples were then submitted to a last freeze followed by thawing at room temperature before packing into 4 mm NMR tubes before data acquisition.

For both static ³¹P and ²H NMR experiments, the sample-containing 4 mm NMR tube was inserted into a 5 mm coil of a homebuilt probe. The ³¹P NMR experiments were conducted on a Bruker Avance 300 MHz NMR spectrometer (Bruker Biospin, Milton, ON). Spectra were acquired at a ³¹P frequency of 121.5 MHz using a phase-cycled Hahn echo pulse sequence with gated broadband proton decoupling (21). 4 K data points were recorded and 1200 scans were acquired with a 90° pulse length of 4 μs, an interpulse delay of 30 μs, and a recycle delay of 4 s. The chemical shifts were referenced relative to external H₃PO₄ 85% (0 ppm).

For each experimental spectrum, a simulated spectrum was calculated by taking into account that MLVs are ellipsoidal. The long/short axis ratio (l/s) of ellipsoidal vesicles can be obtained from the S₁ order parameter proposed by Picard et al. (22). This parameter is directly related to the first spectral moment M₁:

$$S_1=\frac{M_1-{\delta }_{iso}}{\delta },$$

where δ_iso is the isotropic chemical shift (measured from magic-angle spinning spectra), and δ is defined by

$$\delta ={\delta }_{//}-{\delta }_{iso}=-2{\delta }_{\perp }-{\delta }_{iso}.$$

The perpendicular chemical shift (δ_⊥) can be obtained by measuring the chemical shift of maximum intensity at the 90° edge of the static ³¹P NMR spectra.

The ²H NMR spectra were recorded on a Bruker Avance 400 MHz NMR spectrometer (Bruker Biospin). Spectra were obtained at 61.4 MHz using a quadrupolar echo sequence (23). The 90° pulse length was 5 μs and the interpulse delay was 60 μs. For DMPG/DMPG-d₅₄ samples, 10,000 scans were acquired using 4 K data points, and the recycle time was set to 500 ms. For DMPC/DMPC-d₅₄ samples, 3200 scans were acquired with the same number of data points and recycle time. A line broadening of 100 Hz was applied to the spectra. The quadrupolar splittings were measured on dePaked spectra (24).

Infrared studies

Multiple lyophilisations in 10 mM HCl were performed on the peptide powder to remove all traces of TFA (25). Dry DMPC (2.5 mg) and peptides in a lipid/peptide molar ratio of 60:1 were then codissolved in chloroform/methanol (1:1) to ensure thorough mixing. The solvent was removed under nitrogen gas, followed by storage under high vacuum overnight to remove all traces of organic solvent. The dry sample was hydrated with 10 μL of HEPES 100 mM, EDTA 5 mM (pH 7.4) buffer with 20% (w/w) lipids in deuterium oxide. The resulting suspension underwent five freeze (liquid N₂)/thaw (37°C)/vortex shaking cycles to ensure the formation of multilamellar vesicles. For the last cycle, the sample was thawed at room temperature. It was then placed between CaF₂ windows, designed by BioTools (Wauconda, IL).

All spectra were recorded with a Nicolet Magna 560 spectrometer (Thermo-Nicolet, Madison, WI) equipped with a nitrogen cooled MCT A detector. A total of 128 scans were averaged at each temperature with a resolution of 2 cm⁻¹. For the recording of the transmission spectra, CaF₂ windows containing samples were inserted into a homemade cell thermoelectrically regulated. All data manipulations were performed with the Grams/AI 8.0 spectroscopic software (Thermo Fisher Scientific, Waltham, MA). The spectra were corrected for CaF₂ contribution by subtraction of a reference spectrum. The 2800–3010 cm⁻¹ spectral regions (corresponding to the carbon-hydrogen stretching vibrations) were baseline corrected using a cubic function. The methylene symmetric stretching frequency was obtained from the center of gravity calculated at the top 10% of the band.

Results

³¹P NMR spectroscopy

We first used ³¹P solid-state NMR to study the effects of selective and nonselective peptides on the polar region of bacterial and eukaryotic-mimic membranes, i.e., DMPG and DMPC bilayers, respectively. The ³¹P NMR static spectra and related spectral parameters are displayed in Fig. 2 and Table 1. The spectra in the absence and presence of 14-mer peptides reveal lineshapes that are typical of fluid systems with axial symmetry (26). In addition, the shape of the DMPC and DMPG ³¹P NMR spectra also indicates that vesicles are ellipsoidal. This behavior is related to the negative diamagnetic susceptibility Δχ of phospholipids (27) that tend to align their long axis perpendicular to the magnetic field. However, the downfield edge intensity is less prominent for ³¹P spectra of DMPC vesicles than of DMPG vesicles, indicating that DMPC vesicles are more ellipsoidal than DMPG vesicles in the magnetic field.

Figure 2.

³¹P NMR spectra of (A) DMPG MLVs in the absence (gray solid line) and presence of two nonselective peptides, K10 (··−··−) and K5K10 (−−); (B) DMPG MLVs in the absence (gray solid line) and presence of two selective peptides, K11 (dash-dotted line) and K4K11 (dashed line); (C) DMPC MLVs in the absence (gray full line) and presence of two nonselective peptides, K10 (dash-dotted line) and K5K10 (dashed line); and (D) DMPC MLVs in the absence (gray solid line) and presence of two selective peptides, K11 (dash-dotted line) and K4K11 (dashed line). The measurements were performed at 37°C and at a lipid/peptide molar ratio of 60:1.

Table 1.

³¹P NMR spectral parameters for DMPG and DMPC membrane vesicles in the absence and presence of nonselective and selective peptides

	l/s (±0.07)	CSA (ppm ± 0.5)		l/s (±0.07)	CSA (ppm ± 0.5)
DMPG	1.06	36.8	DMPC	1.41	46.2
+ nonselective peptides
K3	1.11	−1.6	K3	1.07	−3.2
K5	1.09	−1.4	K5	1.30	−2.3
K10	1.10	−1.6	K10	1.31	−2.4
K12	1.17	−1.7	K12	1.45	−1.7
K3K5	1.11	−1.6	K3K5	1.11	−1.2
K3K10	1.14	−1.2	K3K10	1.24	−3.5
K3K12	1.15	−1.5	K3K12	1.40	−2.1
K5K10	1.09	−1.4	K5K10	1.19	−2.3
K5K12	1.16	−1.1	K5K12	1.37	−3.0
K10K12	1.06	−2.1	K10K12	1.27	−4.5
Mean	1.11	−1.5	Mean	1.25	−2.7
+ selective peptides
K4	1.09	+1.4	K4	1.07	−1.1
K7	1.07	+1.2	K7	1.15	−1.1
K8	1.14	+1.9	K8	1.09	−1.0
K11	1.05	+2.3	K11	1.16	−1.2
K4K7	1.09	+1.8	K4K7	1.17	−1.2
K4K8	1.02	+1.0	K4K8	1.13	−1.1
K4K11	1.04	+1.5	K4K11	1.27	−1.1
K7K8	1.03	+1.1	K7K8	1.12	−0.6
K7K11	1.00	+1.2	K7K11	1.21	−1.4
K8K11	1.12	+0.9	K8K11	1.24	−1.3
Mean	1.07	+1.4	Mean	1.16	−1.1

Measurements were realized at 37°C and at a lipid/peptide molar ratio of 60:1. l/s = long/short axis of vesicles (considered as ellipsoid); CSA = chemical shift anisotropy = $3/2\left({\delta }_{//}-{\delta }_{iso}\right)$.

A small isotropic peak is present in some spectra. Because its intensity is very weak compared to that obtained with AMPs destabilizing membranes by a detergent-like mechanism (28,29), it can be neglected. This indicates that no formation of isotropic nonbilayer phases (such as small fast tumbling micelles or discoids) occurs in the presence of cationic 14-mer peptides. When destabilizing bilayers, these peptides act using a mechanism different of micellization, as previously shown by dynamic light scattering (17).

The presence of selective or nonselective peptides induces different effects on the ³¹P chemical shift anisotropies (CSAs) of DMPG vesicles. Indeed, the results presented in Fig. 2, A and B, indicate a CSA decrease in the presence of nonselective peptides, whereas selective peptides induce a broadening of the ³¹P NMR spectra. Table 1 confirms that all studied nonselective peptides significantly decrease the ³¹PNMR spectra CSA of DMPG vesicles by ∼4%, whereas selective peptides increase it by ∼4%. These results indicate that nonselective and selective 14-mer peptides interact with the lipid headgroups of membrane lipids, inducing a change in headgroup orientation and either an increase or decrease of local motion for nonselective and selective peptides, respectively. However, the interaction between peptides and DMPG vesicles do not induce a significant change in the mean l/s axis ratios, indicating that the morphology of DMPG vesicles is not significantly modified by the presence of cationic 14-mer peptides.

Fig. 2 C shows the ³¹P spectra of DMPC vesicles with or without nonselective 14-mer peptides. As with DMPG vesicles, a decrease of the CSA is noted in the presence of these peptides. Table 1 confirms the decrease of the CSA of ∼3 ppm (6%). This indicates that nonselective peptides increase the local motion and/or change the headgroup orientation. Selective peptides also decrease the CSA of DMPC spectra but to a smaller extent (∼ −1 ppm) (Fig. 2 D). These results indicate that nonselective peptides have more disordering effects on the headgroup of DMPC vesicles than selective peptides.

Fig. 2 D illustrates that the downfield edge intensity of the DMPC ³¹P spectrum is increased significantly in the presence of selective peptides. This result indicates that in the presence of selective peptides, vesicles tend to be more spherical (22), as confirmed in Table 1 (l/s axis ratio of 1.41 without peptides and of 1.07–1.27 in the presence of selective peptides). This effect is less pronounced in the presence of nonselective peptides (Fig. 2 C and Table 1). The ability by which vesicle deformation occurs in the magnetic field is a function of many parameters such as membrane shape, elasticity, curvature, fluidity, and viscosity (30,31). The morphological change of DMPC vesicles in the presence of selective peptide can be explained by changes of the membrane elastic properties, resulting in less deformable vesicles. This decrease in alignment can be explained by a disruption of lipid-lipid interactions, as suggested for other peptides (32). However, we cannot exclude that this deformation could be related to a change of the membrane magnetic susceptibility due to peptide (possessing large positive binding diamagnetic susceptibility due to the peptide bond) binding to vesicles (30,33). Several other antimicrobial peptides have been shown to have an effect on lipid alignment (34–36).

²H NMR spectroscopy

Static ²H NMR experiments were performed to determine the effect of the peptides on the acyl chain order of DMPC and DMPG bilayers. The ²H NMR spectra of DMPC-d₅₄ and DMPG-d₅₄ in the absence and presence of 14-mer peptides are characteristic of fluid lipid membranes (Fig. 3). For some samples, an isotropic peak is observed that may be attributable to nonbilayer structures or residual HDO. From static ³¹P NMR results and previous dynamic light-scattering results (17), nonbilayer structures can be neglected, suggesting that the isotropic peak is due to the presence of residual HDO.

Figure 3.

²H NMR spectra of (A) DMPG MLVs in the absence (gray solid line) and presence of two nonselective peptides, K10 (dash-dotted line) and K5K10 (dashed line); (B) DMPG MLVs in the absence (gray solid line) and presence of two selective peptides, K11 (dash-dotted line) and K4K11 (dashed line); (C) DMPC MLVs in the absence (gray solid line) and presence of two nonselective peptides, K10 (dash-dotted line) and K5K10 (dashed line); and (D) DMPC MLVs in the absence (gray solid line) and presence of two selective peptides, K11 (dash-dotted line) and K4K11 (dashed line). The measurements were performed at 37°C and at a lipid/peptide molar ratio of 60:1.

Fig. 3, A and B, and Table 2 show that nonselective and selective peptides induce an increase of the quadrupolar splittings at both the plateau (Δν_p) and terminal methyl (Δν_m) regions of DMPG vesicles. The effect is however more pronounced at the terminal methyl region (mean increase of 13–17% compared to an increase of 2–3% in the plateau region). These results indicate that 14-mer peptides, whatever their structure and their ability to permeabilize DMPC bilayers, increase the order of the acyl chains of DMPG bilayers with the greatest effect at the center of the bilayers.

Table 2.

Quadrupolar splittings of DMPG-d₅₄ and DMPC-d₅₄ for the plateau (Δν_p) and methyl group (Δν_m) regions

	Δνp (kHz ± 0.3)	Δνm (kHz ± 0.1)		Δνp (kHz ± 0.3)	Δvm (kHz ± 0.1)
DMPG	22.8	2.5	DMPC	24.9	3.0
+ nonselective peptides
K3	+0.7	+0.4	K3	+0.7	+0.4
K5	+0.7	+0.5	K5	+0.7	+0.3
K10	+0.5	+0.4	K10	+1.5	+0.5
K12	+0.6	+0.4	K12	+0.8	+0.3
K3K5	+0.8	+0.6	K3K5	+0.4	+0.2
K3K10	+0.2	+0.4	K3K10	+0.5	+0.4
K3K12	+0.4	+0.4	K3K12	+0.4	+0.3
K5K10	+0.2	+0.5	K5K10	+0.5	+0.4
K5K12	+0.3	+0.4	K5K12	+0.6	+0.4
K10K12	+0.2	+0.3	K10K12	+0.0	+0.2
Mean	+0.4	+0.4	Mean	+0.6	+0.3
+ selective peptides
K4	+0.7	+0.2	K4	−0.2	−0.2
K7	+0.7	+0.4	K7	−0.2	−0.2
K8	+0.9	+0.5	K8	−0.2	−0.2
K11	+0.7	+0.4	K11	+0.4	−0.2
K4K7	+0.7	+0.3	K4K7	−0.2	−0.2
K4K8	+0.2	+0.3	K4K8	−0.6	−0.2
K4K11	+0.2	+0.2	K4K11	+0.0	+0.0
K7K8	+0.9	+0.3	K7K8	+0.2	−0.2
K7K11	+0.4	+0.2	K7K11	+0.0	+0.0
K8K11	+0.9	+0.5	K8K11	+0.0	−0.2
Mean	+0.5	+0.3	Mean	+0.0	−0.1

Nonselective peptides induce the same effect on the acyl chains of DMPC bilayers (Fig. 3 C and Table 2). More specifically, the Δν_p and Δν_m increase on average by 2% and 10%, respectively. However, the quadrupolar splittings of DMPC bilayers do not increase in the presence of selective peptides. Therefore, peptides inducing leakage of DMPC vesicles (i.e., nonselective peptides) increase the order of the DMPC acyl chains, whereas peptides that do not induce leakage have no significant effect. The ²H NMR results indicate that leakage of vesicles induced by 14-mer peptides is accompanied by an increase of the order of the lipid acyl chains.

FTIR spectroscopy

Infrared spectroscopy was also used to investigate the lipid acyl chain order. Fig. 4 shows the wavenumber of the symmetric CH₂ stretching vibration of DMPG and DMPC as a function of temperature. At low temperature, bilayers are in the gel phase with their acyl chains in an all-trans conformation corresponding to lower wavenumbers. The presence of gauche conformers and the decrease of van der Waals interactions in the fluid phase lead to a shift to higher wavenumbers (37). An abrupt change in the frequency of the ν_s(CH₂) mode is noted at the gel-to-fluid transition temperature (T_m), due to the important increase in the number of gauche conformers (38). For both DMPG and DMPC bilayers, a T_m of ∼24.6°C was found, consistent with values reported in the literature (39,40).

Figure 4.

Wavenumber of the CH₂ symmetric stretching vibration as a function of temperature for DMPG (top) and DMPC (bottom) MLVs in the absence and presence of nonselective (left) and selective (right) peptides. The measurements were performed at a lipid/peptide molar ratio of 60:1.

The presence of nonselective 14-mer peptides induces a shift of the phase transition temperature of DMPG bilayers to higher temperatures (+ 8°C; Table 3). An increase of the phase transition temperature of phospholipids is generally observed in the case of electrostatic interactions with proteins and polypeptides (41). More specifically, it has been shown that an increase of T_m is associated with the stabilization of the charges between the protein and the lipid, leading to an increase of the order of the acyl chains and to the stabilization of the gel phase in the presence of protein (42). Nonselective 14-mer peptides also induce a change in the slope of the gel-to-fluid phase transition of DMPG, indicating that nonselective peptides decrease the cooperativity of the lipid chain melting. This is consistent with electrostatic interactions between peptides and lipids, perturbing the lipid network.

Table 3.

Temperature (T_m) and cooperativity index (I_coop) for the phase transitions of DMPG and DMPC bilayers in the presence of nonselective and selective 14-mer peptides

	Tm (°C ± 0.2)	Icoop (°C ± 0.08)		Tm (°C ± 0.2)	Icoop (°C ± 0.08)
DMPG	24.6	1.00	DMPC	24.6	1.00
+ non-selective peptides
K3	+8.5	0.30	K3	−1.8	0.17
K5	+8.7	0.33	K5	−1.7	0.22
K10	+8.7	0.39	K10	−1.6	0.23
K12	+7.5	0.27	K12	−1.7	0.18
K3K5	+7.9	0.28	K3K5	−1.5	0.34
K3K10	+8.7	0.33	K3K10	−1.7	0.25
K3K12	+7.5	0.28	K3K12	−1.6	0.18
K5K10	+8.7	0.36	K5K10	−1.2	0.30
K5K12	+8.5	0.29	K5K12	0.0	0.23
K10K12	+8.3	0.31	K10K12	−1.5	0.30
Mean	+8.3	0.31	Mean	−1.6	0.25
+ selective peptides
K4	+8.6	0.39	K4	−0.1	0.71
K7	+10.0	0.28	K7	−1.0	0.93
K8	(−2.1) +8.8	(0.20) 0.26	K8	−1.0	0.92
K11	+8.8	0.36	K11	−1.1	0.93
K4K7	(−2.0) +8.7	(0.30) 0.24	K4K7	−0.5	0.78
K4K8	(−1.2) +8.8	(0.53) 0.16	K4K8	−1.1	0.98
K4K11	(−1.5) +8.8	(0.41) 0.12	K4K11	−0.2	0.63
K7K8	+8.6	0.35	K7K8	−0.1	0.90
K7K11	+8.6	0.33	K7K11	−0.1	0.87
K8K11	+8.8	0.35	K8K11	−0.5	0.91
Mean	+8.9	0.28	Mean	−0.6	0.86

When a second transition was measured, it is noted between brackets.

Selective peptides also induce an increase of the transition temperature of DMPG bilayers. This result indicates that selective peptides, like nonselective peptides, lead to an increase in conformational order of the DMPG acyl chains. However, for some DMPG/selective peptide samples, a second transition is observed near 23°C (Fig. 4 B and Table 3). This second transition is similar to that observed by Richard et al. (43) when studying the interaction of β-purothionin with DMPG bilayers. The authors suggested that this thermotropic behavior can be explained by the insertion of the hydrophobic residues of the protein into the hydrophobic core of DMPG bilayers, resulting in a decrease of the acyl chain order. The phospholipids close to peptides would have a phase transition temperature lower than those not interacting with the protein. These data also suggest that selective peptides are less homogeneously distributed on the DMPG membranes than nonselective peptides, creating more free-peptide and peptide-containing domains. This could be due to the intermolecular structure adopted by selective peptides (17).

Nonselective 14-mer peptides also have an effect on the thermotropic behavior of DMPC bilayers by highly decreasing the cooperativity of the lipid chain melting (Fig. 4 C). This indicates that nonselective peptides interact with DMPC bilayers in such a way that they induce changes in the hydrophobic core of the bilayers. However, these changes are weaker than on DMPG bilayers suggesting that due to electrostatic interactions in DMPG, nonselective peptides would be more deeply inserted than in DMPC, explaining the larger ordering effect observed by ²H NMR and FTIR.

Compared to other lipid/peptide systems, selective 14-mer peptides have little effects on the thermotropic behavior of DMPC bilayers (Fig. 4 D). Some selective peptides induce a slight decrease of the transition temperature (∼1°C), whereas others induce a slight broadening of the cooperativity. These results are consistent with the ²H solid-state NMR results, indicating that the interaction of selective 14-mer peptides with DMPC bilayers does not induce a significant effect on the bilayer hydrophobic core.

Discussion

We have previously shown that the ability of synthetic 14-mer peptides containing one or two lysine(s) to permeabilize DMPC and DMPG bilayers is related to their secondary structure. Indeed, 14-mer peptides mainly structured as α-helix are nonselective, whereas when adopting intermolecular β-sheets, 14-mer peptides induce only a large release in zwitterionic DMPC vesicles (17). The adoption of one of these two structures is related to the position of lysine residue(s) in the sequence of the 14-mer peptide. The goal of this study was to shed light on the mechanisms of action of selective and nonselective 14-mer peptides on both type of membranes.

Mechanisms of action of nonselective peptides

³¹P solid-state NMR spectra of DMPG and DMPC bilayers in the presence of nonselective 14-mer peptides show no significant isotropic peak, indicating that they induce vesicle leakage via a mode of action different from a detergent-like mechanism, as we previously suggested for the base 14-mer peptide (17,44). Moreover, like the base peptide (30), these peptides decrease the ³¹P CSA of both types of bilayers. A CSA decrease is correlated to an increase of dynamics and/or a change in orientation of lipid headgroups. Previous ¹⁵N NMR studies on the base 14-mer peptide indicated that this perturbation of lipid headgroup network is due to an alignment of the helical peptide parallel to the surface of the phospholipid bilayers (44,45). Because the base 14-mer peptide and nonselective cationic 14-mer peptides adopt helical structure in the presence of bilayers, this suggests that these peptides exhibit a similar orientation within the bilayers, i.e., an in-plane orientation. Moreover, many other helical peptides, like the AMP magainin, inducing the same perturbation on lipid headgroups than the nonselective 14-mer peptides, adopt an in-plane orientation in interaction with membranes (34,46–49).

Adoption of an in-plane orientation could induce a positive curvature of the bilayer due to a lateral expansion in the headgroup region (50). This curvature is favorable to the formation of torus-shaped pores (51). This sinking raft model was also suggested to explain leakage induced by other helical peptides, like delta-lysin (11,52) and the base 14-mer peptide (17,44). In this model, it was also suggested that the membrane binding of the peptides causes a mass imbalance in the local membrane region, inducing an increase in the membrane curvature.

²H solid-state NMR results indicate that nonselective peptides induce an ordering of acyl chains in fluid DMPC and DMPG bilayers. A similar effect was also observed with both types of bilayers in the presence of the base 14-mer peptide (44). These results suggest that under helical conformation, 14-mer peptides induce similar changes on DMPC headgroups and acyl chains whatever their charge. However, this ordering effect on acyl chains is in contrast with previous studies on helical AMPs. Indeed, most helical AMPs that adopt an in-plane orientation at the membrane surface induce a disordering of lipid acyl chains (53). However, previous results also indicate an increase of acyl chain order in the presence of melittin, δ-haemolysin, and cateslytin (54–56). This suggested that the ordering of membrane hydrophobic core was due to a deeper membrane insertion of at least part of the peptide structure. Considering this hypothesis, helical 14-mer peptides could adopt an in-plane orientation with a deeper insertion in the membrane hydrophobic core than most AMPs (which induce an increase of CSA). This difference of penetration depth between helical 14-mer peptides and most AMPs could be due to the large hydrophobicity of 14-mer peptides. Furthermore, our measurements were performed at a high lipid/peptide ratio (60:1).

FTIR results indicate that nonselective peptides induce a broadening of the T_m (without a significant T_m increase) on DMPC bilayers, whereas they increase the T_m of DMPG bilayers without significant broadening, suggesting that they have a greater effect on DMPG than DMPC bilayers. DMPC and DMPG phospholipids only differ by their headgroup nature: zwitterionic in the case of DMPC and negatively charged for DMPG. The difference in ordering could thus be explained by a more significant number of nonselective peptides interacting with DMPG than DMPC bilayers due to the electrostatic attraction between the negative charge of the DMPG headgroup and the positive charge of lysine residues. However, the effects of nonselective peptides on ³¹P and ²H NMR spectra of DMPC and DMPG are similar, indicating no significant differences of interaction. Moreover, Langmuir film measurements indicate that the interaction of 14-mer helical peptides with DMPC and DMPG monolayers is similar (data not shown). This suggests that the difference observed in FTIR could be due to a difference in the penetration depth of peptides in both types of bilayers, with a deeper insertion in DMPG bilayers due to electrostatic interactions.

Mechanisms of action of selective peptides

Compared to helical AMPs, less information is available in the literature about the mechanism by which β-sheet peptides cause membrane leakage. Moreover, most studied β-sheet antimicrobial peptides contain disulfure bridge(s) restraining their structure. Our study is therefore particularly useful because it gives information on the mode of action of linear β-sheet peptides. More specifically, solid-state NMR results indicate that selective peptides induce changes in the ³¹P CSA of DMPC and DMPG bilayers. This suggests that peptides interact with both types of bilayers. The inability of selective peptides to permeabilize DMPC bilayers is therefore not due to an inability to interact with the bilayers. This is confirmed by Langmuir film measurements showing that the selective K4K11 peptide induces an increase of surface area of both DMPC and DMPG monolayers with similar extent (A. Lorin, M. Noël, M.-E. Provencher, N. Voyer, and M. Auger, unpublished results). Because DMPC is zwitterionic, these results indicate that hydrophobic/hydrophilic interactions are sufficient for the interaction of selective 14-mer peptides with bilayers.

However, electrostatic interactions have an effect on the mode of perturbation induced by selective peptides. Indeed, the CSAs of DMPG headgroups increase in the presence of selective peptides, whereas those of DMPC headgroups decrease. This is in contrast with results obtained on nonselective peptides showing that a decrease of CSA is correlated with a permeabilization of DMPC and DMPG bilayers. These differences indicate that selective and nonselective peptides induce different changes on the lipid headgroup network when permeabilizing bilayers. However, the orientation and insertion adopted by β-sheet selective peptides in membranes could be similar to α-helical nonselective peptides, but inducing different responses of the lipid headgroups due to the difference in peptide structure.

Several pieces of information support this hypothesis. First, ²H solid-state NMR and FTIR results indicate that selective peptides induce an ordering of the DMPG acyl chains. Therefore, despite that selective and nonselective 14-mer peptides have different effects on the lipid headgroups of bilayers they permeabilize, they induce an increase of the lipid acyl chain order of these bilayers. Second, comparing our ³¹P NMR results with those from previous solid-state NMR studies on β-sheet structured AMPs indicates that the selective peptides adopt an in-plane orientation at the DMPG membrane surface. Indeed, it was shown that the two β-sheet structured AMPs cateslytin and CDT (a linear analog of tachyplesin I) increase the CSA of negatively charged membranes (56,57), like 14-mer selective peptides on DMPG bilayers. This is reinforced by previous results showing that the β-sheet peptide protegrin, known to be fully immersed and entirely spanning DLPC bilayers, induces a decrease of the ³¹P CSA (58).

Selective 14-mer peptides order the hydrophobic core of DMPG bilayers but not that of DMPC bilayers. The different ordering of zwitterionic and negatively charged bilayers in the presence of a β-sheet peptide was also observed in the case of cateslytin (56). The authors suggested that this difference could be explained by a deeper insertion of aromatic residues of cateslytin into the negatively charged bilayers. On the basis of this hypothesis, we suggest that β-sheet selective 14-mer peptides adopt an in-plane orientation in both types of bilayers. They could therefore cap the DMPC headgroups inducing changes in this membrane region but no change in the membrane hydrophobic core. However, due to the electrostatic interactions between the DMPG negatively charged headgroups and the positively charged lysine residue(s), selective peptides could be inserted more deeply into DMPG bilayers, inducing ordering of acyl chains. This difference of insertion depth could explain the difference of effects on the lipid headgroups, as discussed previously.

The deeper membrane insertion of selective 14-mer peptides into DMPG bilayers could be responsible for the formation of torus-shape pores, as suggested for helical nonselective 14-mer peptides. Formation of such pores was also suggested to explain the leakage induced by β-sheet peptides (13,59–61). More specifically, after initial binding, β-sheet peptides self-assemble and induce formation of peptide-rich domains on the membrane surface. The bilayer is then destabilized due to the asymmetry in mass, charge, or surface tension and transient pores appear according to the sinking raft model. The observation of two phase transitions with some selective peptides also supports this hypothesis. However, based on our results, we cannot conclude that selective peptides induce pore formation. Indeed, Jean-Francois et al. (56) suggested that multimerization of cateslytin peptides at some regions of bilayer surface induce an ordering of lipid acyl chains but without formation of pores. These authors suggested that leakage occurs in membrane defects induced by the thickness difference between free and associated peptides membrane regions. This hypothesis was also proposed to explain leakage induced by two alpha/beta AMPs (62). Therefore, another hypothesis is that the cationic 14-mer peptides could result in the formation of domains of higher order where the permeabilization would be due to defects that would occur at the border of the two domains. Additional experiments are necessary to confirm the mode of action of selective peptides.