One step at a time

Abstract

Antimicrobial peptides (AMPs) are a crucial part of the innate immune system of eukaryotes and present a possible alternative to common antibiotics. It is therefore of great importance to understand their modes of action. Using a single-molecule approach in combination with high resolution imaging and biofunctional assays we were able to determine the different steps occurring during the action of the α-helical AMP Sushi 1 during bacterial lysis in spatial and temporal resolution in a biologically relevant context. Furthermore, we comment on the use of Sushi 1 as a template for new peptides to learn more about structurefunction relationship of AMPs.

Antimicrobial peptides (AMPs) are key elements of the innate immune system.^1,2 Many of them interact with membranes of bacteria leading to perturbation of the lipid bilayer and eventually to inactivation of the pathogen. Since multi-drug resistant bacteria have emerged due to over-prescription of antibiotics, new classes of antibiotics are needed, and AMPs could present a valuable alternative. Nevertheless, despite this importance and the characterization of more than 1,000 natural AMPs which are usually short (9–54 amino acids), are classified into a handful of main structural groups, and represent one of the oldest defence mechanisms found in all kingdoms of life, there is no consensus on how AMPs function on a molecular scale, let alone a definite relation between the amino acid sequence of an AMP and its functional mechanism.

One extensively studied peptide is the 34-amino-acid-long antimicrobial peptide Sushi 1 (S1) derived from Factor C protein of the horseshoe crab (Carcinoscorpius rotundicauda).^3–9 S1 has an amphipathic α-helical structure and carries a positive net charge of +4 at physiological pH. S1 not only possesses extensive antibacterial properties against Gram-negative bacteria but also binds and neutralizes lipopolysaccharide (LPS) present on the bacterial cell wall. Recently we employed nanoparticle conjugates using new single molecule imaging tools combined with biological functional assays as well as high resolution imaging to elucidate the stepwise mechanism of action of S1 on live Gram-negative bacteria. This allowed us to determine the different steps in the action of S1 during bacterial lysis in a spatial and temporal resolution in a biological relevant context.¹⁰ Although nanoparticle-peptide conjugates could be problematic due to the size difference between label and peptide, the conjugates represent valuable biophysical tools to study antimicrobial peptides on membranes of living bacteria. In order to follow the entire process of antimicrobial action we employed a variety of approaches using real-time such as fluorescence correlation spectroscopy (FCS), total internal reflection (TIRF) microscopy as well as end-point methods (transmission electron microscopy, TEM). We developed a novel fluorescent live bacteria lysis assay and used a fully functional nanoparticlelabeled S1 to observe the process of antimicrobial action at the single-molecule level. We demonstrated that S1 targets the outer and inner membranes, but not the intracellular components. We were able to dissect the mechanism of action at a molecular level and found four distinct steps of the bactericidal process: (1) Binding, mediated mainly by charged residues in the peptide; (2) Peptide association, as peptide concentration increases, evidenced by a change in diffusive behavior; (3) Membrane disruption, during which lipopolysaccharide is not released; and (4) Lysis, by leakage of cytosolic content through large membrane defects.

To further elucidate the relationship between the AMP amino acid composition and their sequence patterns and antimicrobial activity, we designed AMPs either derived from the same amino acid composition as S1 (manuscript in preparation) but different in sequence, or peptides with the same general sequence pattern but varying amino acids (unpublished data). In the first case we designed an S1 derivative, keeping the same amino acid composition, but with a partially randomized sequence introducing a structure which formed an amphiphilic α-helix when interacting with negatively charged lipids. This randomized peptide maintained the antimicrobial activity [S1 and randomized S1, neutralized 2 EU/ml LPS at 125 nM and 500 nM, respectively, as determined by PyroGene kit (Cambrex Inc., USA)] but displayed a different mechanism of action. While experiments with S1 are consistent with a pore forming peptide, the randomized version of S1 seemed to lead to a loss of bacterial membranes and in contrast to S1, was not found in the cytosol (Fig. 1). In the second case we designed a peptide, called V4, de novo.¹¹ This β-hairpin shaped cyclic peptide has a dual core structure of HBHPHBH (B, basic; H, hydrophobic; P, polar residue; respectively) similar to the pattern found in S1. The peptide proved to be effective against bacteria showing that the above pattern can work as a template for AMPs, since the charge provided for the binding to bacterial membranes, its amphiphilic structure leading to aggregation on the membrane, and the hydrophobicity of the nonpolar face of the β-hairpin allowed the integration of the peptide into the membrane followed by membrane disruption. However, the hydrophobicity of V4, based on valine as the hydrophobic amino acid, was so high that the peptide was difficult to dissolve and as such, was perceived to be inadequate in the aqueous phase as an effective AMP.¹² Changes in the hydrophobic amino acids from valine to L-norvaline, L-2-aminobutyric acid and alanine, in descending order of hydrophobicity, increased the peptide solubility in aqueous solution while at the same time decreased their affinity to bacterial membranes to a lesser extent.¹¹ As shown in Figure 2 this results in an optimal hydrophobicity for a particular peptide class, here V4 and its derivatives, with greater solubility and good affinity for bacterial membranes. Interestingly, the change in hydrophobicity of V4 and its derivatives did not lead to an appreciable change in mechanism. In all cases vesicle aggregation is accompanied by leakage, which is the proposed cause of bacterial death due to these peptides.^13,14

Figure 1.

Transmission Electron Microscopy (TEM) images of nanogold labelled (A) “designed-randomized” S1 and (B) S1 on E. coli. While S1 can be found on the inner and outer leaflets of the bacterial membranes and to a minor extent even in the cytoplasm, this is not the case for randomized S1, which seems to detach the membrane but was absent in the cytosol.

Figure 2.

Comparison between the AMP V4 and derived peptides, where the 8 valines in V4 were replaced by either L-norvaline (V4nor), L-2-aminobutyric acid (V4abu), or L-alanine (V4ala). (A) With decreasing hydrophobicity (see lipophilicity index in ref. 11), the solubility (black filled circles) increases, while at the same time the binding affinity (grey empty circles), measured as the fraction of peptides bound to POPG (1-Palmitoyl-2-Oleoylsn-Glycero-3-[Phospho-rac-(1-glycerol)]) vesicles at a lipid concentration of 50 µM, decreases. The inset shows the general structure of the V4 peptide and derivatives, where X_H annotates hydrophobic amino acid. (B) The peptides can aggregate, permeate and disrupt POPG vesicles. The different effects occur at different peptide to lipid ratios (P/L).

The above examples show, not surprisingly, that it is not the composition of amino acids alone but especially the combination with the secondary or supersecondary structure within the peptide that determines the molecular shape and amphiphilic character and consequently the action mechanism of an AMP. But it shows as well that once a structure has been found it can be tuned to a certain extent by replacing different amino acids with amino acids of the same type, i.e., basic, polar or non-polar, with a higher or lesser degree of the amino acid characteristic.^11,15

Biophysical methods and their development are needed to study fundamental processes with high resolution and without ensemble averaging. This is especially important in the field of AMPs, molecules that are active in very low concentrations and can further change their bulk behaviour upon change in concentration. In our opinion, a fruitful way forward for the design of AMPs might be the use of the combination of biochemical and biophysical tools as presented in Leptihn et al.¹⁰ to determine the characteristics of the action mechanism of different natural peptides and then use peptides with the same basic characteristic amino acid patterns to fine tune and hence limit their haemolytic activity by increasing or decreasing the hydrophobicity, polarity and/or charge of the peptides. The combination of different techniques, including computation, biochemical approaches to peptide design and characterization and biophysical tools to investigate functions of peptides on living samples will result in better understanding on how the sequence of a peptide and its structure relates to its mechanism of action.

Addenda to: Leptihn S, Har JY, Chen J, Ho B, Wohland T, Ding JL. Single molecule resolution of the antimicrobial action of quantum dot-labeled sushi peptide on live bacteria. BMC Biol. 2009;7:22. doi: 10.1186/1741-7007-7-22.