Synergy between Winter Flounder antimicrobial peptides

Abstract

Some antimicrobial peptides (AMPs) have potent bactericidal activity and are being considered as potential alternatives to classical antibiotics. In response to an infection, such AMPs are often produced in animals alongside other peptides with low or no perceivable antimicrobial activity, whose role is unclear. Here we show that six AMPs from the Winter Flounder (WF) act in synergy against a range of bacterial pathogens and provide mechanistic insights into how this increases the cooperativity of the dose-dependent bactericidal activity and potency that enable therapy. Only two WF AMPs have potent antimicrobial activity when used alone but we find a series of two-way combinations, involving peptides which otherwise have low or no activity, yield potent antimicrobial activity. Weakly active WF AMPs modulate the membrane interactions of the more potent WF AMPs and enable therapy in a model of Acinetobacter baumannii burn wound infection. The observed synergy and emergent behaviour may explain the evolutionary benefits of producing a family of related peptides and are attractive properties to consider when developing AMPs towards clinical applications.

Keywords: Antibiotics; Computational models; Peptides; Pharmacodynamics.

Fig. 1. Winter Flounder AMP bactericidal activity is highly cooperative and can be influenced by synergy.

In vitro pharmacodynamics assays performed for selected WF peptides and combinations thereof, and bactericidal, clinically relevant antibiotics to which each strain remains susceptible: A. baumannii ATCC 17978 in MHB (a, b), A. baumanii AYE in MHB (c, d) or EMRSA-15 NCTC 13616 in RPMI with 5% FBS (e, f) were challenged with increasing concentrations of WF AMPs, WF AMP combinations or clinically relevant antibiotics. Curves shown are fits of averages of three independent repeated experiments (a, c, e) and show the change in bactericidal rate as a function of dose, reported as fractions or multiples of the MIC (×MIC). One-way ANOVA with Tukey post hoc test multiple comparisons for Kappa (b, d, f), highlight the differences in cooperativity between the WF AMPs, combinations thereof, and antibiotics. Mean and SE of three independent repeated experiments are shown. Only significant pairwise comparisons between AMPs or between clinically used antibiotics are shown (comparisons between AMPs and antibiotics are provided in the main text): *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. Selected time-kill curves, used to construct the PD curves in a, are provided in the supplementary material (Supplementary Fig. 1).

Fig. 2. Winter Flounder AMPs are protective when combined in a Galleria mellonella model of Acinetobacter baumannii ATCC 17978 burn wound infection.

Survival curves are plotted for thirty larvae, each treated with classical antibiotics (a), WF2 (pleurocidin) (b), combinations of WF2 and either WF3 or WF1a (c), or WF3 and/or WF4 (d), in each case compared with fifty larvae subjected to burn only or burn plus infection (A. baumannii ATCC 17978) over 96 h. Percentage survival along with significance tests of protection against infection, due to therapy, according to Log-rank (Mantel–Cox) or Gehan–Breslow–Wilcoxon tests are shown in Table 5.

Fig. 3. WF peptides do not adopt ideal α-helix conformations in models of bacterial plasma membranes.

Far-UV circular dichroism spectra of 50 µM Winter Flounder peptides or temporin L in 5 mM Tris buffer at pH with 5 mM SDS micelles (a) or 15 µM in models of Gram-positive (3 mM POPG—b) or Gram-negative (3 mM POPE:POPG (75:25)—c) bacterial plasma membranes at 37 °C. Spectra are representative of three independently repeated experiments. A comparison of H_N to H_A NOESY cross peak linewidths (mean and SD) for WF AMPs and temporin L (TL) in SDS_d25 reveals much greater linewidths for the WF peptides, indicative of intermediate exchange (d). Newly solved structures of the WF AMPs are likely hybrids of at least two conformations adopted in SDS (f) and lower proportions of α-helix conformation are identified relative to previously solved structures of WF2 (2LS9) and, particularly, temporin L (6GS5) (e). Segments are coloured purple for α-helix, blue for 3-₁₀ helix and cyan and white respectively for turns and coils. Note however that Define Secondary Structure of Proteins (DSSP) is incapable of identifying P_II conformation. One-way ANOVA with Bonferroni’s multiple comparisons test: *p < 0.05; **p < 0.01; ****p < 0.0001.

Fig. 4. MD simulation—WF peptides penetrate POPG bilayers more readily than POPE/POPG bilayers.

Sideview snapshots for WF2/pleurocidin after 200 ns in POPG (a) or POPE/POPG (b) bilayers—for clarity only the phosphorus atoms are shown for the lipids. Centre of Mass (COM) is shown as averages of the four peptides in each simulation over the last 100 ns in each of three (WF and pleu-KR) or two (temporin L) replicate 200 ns simulations (c). Mean and SE shown for replicates. Two-way ANOVA with Šídák’s multiple comparisons test: *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

Fig. 5. MD simulation—altered penetration and H-bonding distribution in synergistic combinations of W2 and WF1a inserting into POPE/POPG bilayers.

Centre of mass analysis where each point is one peptide in one of three replicate simulations, the bar is the mean and error is SE (a). Sideview snapshot for WF1a/pleurocidin after 200 ns (c). Lipid acyl chain order parameters for lipids within 4 Å of a peptide shown as averages of the three replicates for POPE (b) or POPG (d) in mixed POPE/POPG bilayers. Hydrogen bonding distribution for WF1a (e, f) or WF2 (g, h) run unmixed (e, g) or as combinations (f, h). Each panel is a sum of four peptides when the peptides are unmixed or two peptides each for the combinations and is representative of n = 3 replicates. Two-way ANOVA with Šídák’s multiple comparisons test: **p < 0.01.

Fig. 6. Ion conductance is sensitive to lipid acyl chain disordering induced by WF AMPs and explains synergistic gain in potency.

Disordering of lipid acyl chains induced by WF AMPs in MD simulations is related to the lowest concentration at which ion conductance is detected by patch-clamp (a Pearson r = 0.7007, p = 0.0111). At these concentrations, only some WF AMPs trigger substantial membrane activity (f DPhPG; g DPhPE/DPhPG) and average MICs for Gram-positive or Gram-negative bacteria are unrelated to these threshold concentrations (b Pearson r = 0.4802; p = 0.1141; red dash = line of identity). At its threshold concentration (7.5 µM) WF2, d, g induces substantial ion conductance in DPhPE/DPhPG bilayers but this is slow (d, h) and precedes bilayer disruption. WF1a (15 µM) is largely inert (c, g). Combining WF2 (3.75 µM) and WF1a (5 µM) ensures substantial ion conductance is induced more rapidly and with less peptide (e, g, h). Time taken to onset of conductance (h) is the average and SE of five independent replicates. One-way ANOVA with Bonferroni’s multiple comparisons test: ***p < 0.001.