Matching amino acids membrane preference profile to improve activity of antimicrobial peptides

Abstract

Antimicrobial peptides (AMPs) are cationic antibiotics that can kill multidrug-resistant bacteria via membrane insertion. However, their weak activity limits their clinical use. Ironically, the cationic charge of AMPs is essential for membrane binding, but it obstructs membrane insertion. In this study, we postulate that this problem can be overcome by locating cationic amino acids at the energetically preferred membrane surface. All amino acids have an energetically preferred or less preferred membrane position profile, and this profile is strongly related to membrane insertion. However, most AMPs do not follow this profile. One exception is protegrin-1, a powerful but neglected AMP. In the present study, we found that a potent AMP, WCopW5, strongly resembles protegrin-1 and that the match between its sequence and the preferred position profile closely correlates with its antimicrobial activity. One of its derivatives, WCopW43, has antimicrobial activity comparable to that of the most effective AMPs in clinical use.

Fig. 1. Predicted amphipathy of antimicrobial peptides.

a Secondary amphipathic α-helical model and b primary amphipathic β-strand model of AMPs. For simplicity, the membrane is shown as a simple electrostatic five-slab model. For easier viewing, the exact thickness of the membrane and chirality of the D-form amino acids are omitted and shown in supplementary figures^, (Supplementary Figs. 1 and  2). WCopW5 is more similar to the primary amphipathic β-strand model of AMPs.

Fig. 2. Comparisons of peptide-lipid interactions.

a Isothermal titration calorimetry measurements of peptide interactions with DMPC:DMPG liposomes. ΔH indicates the enthalpy change. ΔS indicates the entropy change. The binding constant (or the association constant K, M–1) is the inverse of the dissociation constant (Kd). The larger binding constants of LWCopW29 and WCopW29 indicate these peptides have high affinity for the artificial bacterial membrane liposomes. An independent experiment yielded the same results (Supplementary Fig. 5). b Addition of DMPC:DMPG liposomes changed the tryptophan fluorescence intensity of the peptides. In the graphs, the 340 nm maxima intensity values were normalized to the initial fluorescence value to allow comparisons of fluorescence intensities. A quencher (water or acrylamide) decreased the fluorescence intensity, while environmental hydrophobicity (lipid) increased the fluorescence intensity. Lipid concentrations are as follows: light blue, 10 μM; blue, 50 μM; dark blue, 100 μM; black, 1000 μM. The red dotted line indicates 10 μM acrylamide; the pink dotted line, 100 μM acrylamide. A high lipid concentration (1000 μM) quenched HLWCopW29-2 and HLWCopW29-4 fluorescence. This effect was comparable to that of acrylamide, indicting larger exposure of tryptophan to water. Independent experiments yielded the same results (Supplementary Fig. 6). c Proton-leakage increases DiSC3(5) fluorescence in S. aureus. Peptides were added at 120 s. The fluorescence intensities of the four analogs were similar at the HLWCopW29-2 and HLWCopW29-4 MIC against OD 0.1 S. aureus (10 μM peptide each). The fluorescence intensities of LWCopW29 and WCopW29 were higher at the LWCopW29 and WCopW29 MICs (2 μM and 1 μM, respectively) (Supplementary Table 2). An independent experiment yielded the same results (Supplementary Fig. 7).

Fig. 3. Antimicrobial peptides advantageous properties of WCopW29.

a Time- and concentration-dependent permeation of the outer and inner membrane (probed with nitrocefin and ONPG degradation, respectively) indicate that WCopW29 permeates both membranes at MIC concentrations. b Scanning electron micrographs showing the time-dependent swelling and peeling off of the membranes of MDR Gram-negative and Gram-positive bacteria exposed to MIC concentrations of WCopW29 over a period of 4 h (Supplementary Figs. 10 and  11). c Antimicrobial kinetics of MDR Gram-negative and Gram-positive bacteria indicate that WCopW29 kills both at MIC concentrations. d Suppression of resistance acquisition by non-resistant Gram-negative and Gram-positive bacteria exposed to MIC and sub-MIC concentrations of WCopW29 for 30 days. Stable antibiotics, which mostly target rarely mutable targets (e.g., colistin for lipopolysaccharide and vancomycin for peptidoglycan pentapeptide), and stereospecific antibiotics, which target readily mutable proteins (e.g., tetracycline for ribosomes and ofloxacin for DNA gyrase) were used as controls.

Fig. 4. WCopW43 is a potential next-generation AMP.

a The MIC50 and MIC90 values for WCopW43 against 30 A. baumannii strains, including 27 carbapenem-resistant strains, indicate that it has a broad spectrum breakpoint MIC against MDR strains, including colistin-resistant strains (Supplementary Table 6). b The antimicrobial kinetics of MDR Gram-negative and Gram-positive bacteria indicate that WCopW43 kills both types at MIC concentrations. c Analysis of non-resistant Gram-negative and Gram-positive bacteria exposed to MIC and sub-MIC concentrations WCopW43 for 30 days indicates emergence of WCopW43 resistance is suppressed. Stable antibiotics and stereospecific antibiotics were used as controls. d Nephrotoxicity indicators in a mouse model. WCopW43 (100 mg/kg) or PBS was subcutaneously injected into BAKB/c mice daily for 3 days. Groups were statistically compared using unpaired t tests (*P < 0.05, **P < 0.01, ns: not significant). e Survival rates among infected mice. BALB/c mice were subcutaneously infected with MDR A. baumannii (1 × 10⁸ cfu/ml), MDR S. aureus (1.5 × 10⁸ cfu/ml) or MDR K. pneumonia (1 × 10⁹ cfu/ml). One hour later, the mice were subcutaneously administered WCopW43 (50 mg/kg) or PBS. f CFU kinetics of the mouse model. From among the mice in the survival experiment, three were randomly selected. One day after infection, tissue was collected, processed and spread to count CFUs. CFU numbers were statistically compared using unpaired t tests.