Enhancing Therapeutic Potential of Antimicrobial Peptides by Tuning Backbone Conformation

Abstract

Full text embargo up to Dec 31, 2026 The work reported in this thesis involves tuning the backbone conformation of b-hairpin AMPs to reduce their cytotoxicity, which is determined in terms of reduced hemolytic activity, and increasing their proteolytic stability with an end goal of increasing their therapeutic index in vivo against drug resistant bacterial infections. The thesis is titled ‘Enhancing therapeutic potential of antimicrobial peptides by tuning backbone conformation’ and is divided into six chapters. The first chapter introduces the history of infectious diseases and the growing problem of antimicrobial resistance (AMR), focusing on antimicrobial peptides (AMPs) as the most promising alternatives to conventional antibiotics in the fight against AMR – a silent pandemic in the making. These AMPs possess a lot of secondary structural diversity, amongst which the b-hairpin AMPs have been found to be the most potent, with least resistance development against them. But unlike other structural classes, these have one or more disulfide bonds, which, along with the native b-turn, are responsible for stabilizing their structures. Even though this stability aids in broad spectrum antibacterial activity, it also confers high cytotoxicity. Since, these b-hairpin AMPs have an extended conformation, they have low proteolytic stability despite the disulfide bonds, which shortens their half-life in physiological conditions. In the second chapter, we investigate the role of the guanidinium group of arginine residue in AMP toxicity. Extensive studies on arginine-rich cell-penetrating peptides, such as the TAT peptide derived from the HIV-TAT protein, have shown that the guanidinium group can form bi-dentate hydrogen bonds with the charged head groups, such as sulfates and phosphates, on the cell membrane. Then they are easily taken up by the cells either by translocation driven by membrane potential or endocytosis. We fished out 17 naturally occurring potent b-hairpin AMPs from the APD3 database and analyzed the frequency of cationic residues. It was found that they had the highest frequency of arginine residues. Polyphemusin-1 was chosen for this study as it carries a net positive charge of +7 with six arginine residues and is shown to be highly cytotoxic. Arginine (Arg) residues were substituted with ornithine (Orn), which has the same number of methylene groups, carries the same charge as arginine, but is devoid of the guanidinium group in its side chain. CD spectra were acquired, and subsequent hemolytic and antibacterial activity assessments were done. It was found that a single Arg to Orn substitution does not affect the toxicity profile. In contrast, simultaneously, multiple Arg to Orn substitutions in both strands showed significantly decreased toxicity. Since the double disulfide bonds and native turn residues were conserved in the strand-substituted peptide variant, with some increase in therapeutic index, and there were no structural changes observed compared to wild type, it led to the conclusion that multiple guanidinium groups and a stable b-hairpin conformation are responsible for peptide toxicity in such AMPs. In the third chapter, we evaluate the significance of b-turns on the backbone rigidity and overall hydrophobicity of b-hairpin AMPs. Subsequently, their implications on cytotoxicity and antibacterial activity. For this study, we took a reported truncated variant of Arenicin-1 named ALP-1 as the scaffold. As observed by Panteleev et. al., 2016, Journal of Peptide Science, ALP- 1 showed several-fold reduced hemolytic values even with the conserved single disulfide bond as the wild type. Hence, it was the ideal model peptide, onto which two non-natural b-turn motifs [-(D-Ala)-(NMe)(L-Arg)-] and [-(D-Val)-(NMe)(L-Arg)-] were introduced at the reverse turn. These engineered motifs were taken from the study by Lahiri et. al., 2018, Chemical Science, which shows them to nucleate b-hairpin conformation by inducing a type II’ b-turn when placed in the middle of a linear peptide chain due to induction of pseudo-allylic strain. CD and NMR spectroscopy experiments were done to compare the conformation of engineered cyclic variants to the wild type. From the mean residue ellipticity plots and chemical shift indices, it was clear that the mutants with engineered turns show an increased tendency to fold into a distinct b-hairpin conformation. The software Discovery Studio was used to calculate their structures using NOEs from the NMR spectroscopy. It was found that the [-(DVal)-( NMe)(L-Arg)-] motif induced an ideal b-II’ turn that led to a very rigid b- hairpin backbone with restricted movement of the side chains, whereas the [-(D-Ala)-(NMe)(L-Arg)-] motif induced a flexible b-turn with torsion angle values closer to a type II’ turn and had a less rigid backbone conformation. ALP-1 did not show a b-turn and was the most flexible peptide in solution. It was observed that there is a strong correlation between the rigidification of bhairpin backbone and hemolytic activity because the variant with rigid turn motif showed higher toxicity than the one with flexible turn. But since both engineered cyclic variants showed much higher hemolysis than ALP-1, we linearized them by substituting the cysteine residues with threonine, as threonine has the highest propensity to occur in b-sheets, retaining the mutated b-turn motifs. The linear variants showed a structured backbone in the presence of membrane-mimicking environment and were unstructured in solution. Following the same pattern as the cyclic mutants, the linear variants that had the rigid turn motif showed higher hemolysis than the one with flexible turn motif. These peptides have two tryptophan residues, so, Stern-Volmer constants were derived, depicting the extent of tryptophan fluorescence quenching with acrylamide when peptides were incubated with liposomes mimicking both prokaryotic and eukaryotic membranes. The linear variants were found to have better selectivity towards prokaryotic membrane, and the cyclic variant with a rigid b-II’ turn motif embedded the most into the eukaryotic liposomes, indicating its increased toxicity. Structure calculation of the linear variant with [-(D-Ala)-(NMe)(L-Arg)-] turn based on NMR experiments done in presence of 1:1 water : methanol, i.e. a membrane-mimicking environment, found that it had an electrostatic potential surface that was similar to that of ALP- 1, whereas the surface of the cyclic variant with [-(D-Val)-(NMe)(L-Arg)-] turn had an increased amphipathic surface area. This can be directly correlated to the observation that the flexible linear variant with low amphipathicity was the least hemolytic with highest therapeutic index, and the rigid cyclic variant with high amphipathicity was the most hemolytic with reduced therapeutic index amongst all the peptides tested in this study. In the fourth chapter, we aim to make b-hairpin AMP proteolytically stable and determine its efficacy both in vitro and in vivo. In an elaborate study by Edwards et. al., 2016, ACS infectious diseases, it is shown that amongst all potent naturally occurring b-hairpin AMPs tested, Tachyplesin-1 had the best antibacterial activity in vitro. For in vivo efficacy, it is important to make such peptides proteolytically stable. Wild-type double disulfide-bonded Tachyplesin might be susceptible to proteases due to the presence of all natural L-amino acids, and there is a possibility of disulfide scrambling in physiological conditions. Hence, for this chapter’s study, we linearize the wildtype Tachyplesin-1 by substituting all cysteines with threonine residues and incorporating the flexible b-II’ turn motif [-(D-Ala)-(NMe)(L-Arg)-] based on the observation made in the studies with ALP-1. In the quest to increase proteolytic stability, a mirror image of the linear Tachyplesin named as TP-D was synthesized, in which all L-residues were substituted with their corresponding D-amino acids and vice-versa. By incubating TP-D with Proteinase K, it was found to be completely stable till the end of the experiment compared to its L-variant. Its antibacterial activity was determined, and it showed broad-spectrum activity like wild-type. From CD spectroscopy, it was found that TP-D was unstructured in PBS but became structured into a b-hairpin like conformation in presence of E. coli lipopolysaccharide (LPS). Its binding affinity to E. coli LPS was determined using ITC experiment, and it was found that even with loss of both the disulfides, it showed the same affinity to LPS as the wild-type Tachyplesin. Fluorescence microscopy, propidium iodide (PI) uptake assay and scanning electron microscopy experiments showed extensive membrane damage of both gram-negative and gram-positive bacteria by TP-D at 5x MIC (minimum inhibitory concentration). The damage was the highest for Acinetobacter baumannii. TP-D was found effective against multidrug12 resistant ESKAPE pathogens as well. The safety profile assessment of TP-D was done both in vitro and in vivo. It was determined that 20 mg/kg can be used as the treatment dose administered intraperitoneally to mice infected with carbapenem-resistant Acinetobacter baumannii. In vivo experiments showed that 20 mg/kg of TP-D could significantly reduce bacterial load within 8 hours of treatment and successfully alleviate the infection. It was found that the extent of bacterial load reduction in the spleen and lungs harvested from TP-D treated mice were much less than those from the positive control i.e. Polymyxin B treated mice. This indicates that TP-D might be having a high plasma protein binding affinity due to the abundance of positively charged residues in it, which makes it difficult for us to determine its pharmacokinetic parameters. In the fifth chapter, a disruptive approach was taken where we linearize b-hairpin AMP, mould its membrane-induced hairpin structure into a membraneinduced helical conformation, and study its impact on cytotoxicity and antibacterial activity. Several studies have suggested that with increasing secondary structure stability in helical peptides, there is also an increase in proteolytic stability, due to an extensive intra-molecular hydrogen bonding network. So, for this study, we took the potent but highly toxic b-hairpin AMP Polyphemusin-1 and linearized it by substituting all cysteines with threonine residues and introducing the flexible b-II’ turn motif [-(D-Ala)-(NMe)(L-Arg)-] instead of its native turn residues. The linear variant, i.e. PM-aR’-T, was found to be less hemolytic and preserved the broad-spectrum activity as the double disulfide-bonded wild-type. This gave us the linear molecule on which a routine alanine scan was done, and multiple positions were mutated to alanine (high helical propensity) to get a variant that showed helical signature in presence of sodium dodecyl sulfate (SDS) micelles but was unstructured in solution through CD spectroscopy. Since it showed a very similar concentration dependent hemolytic effect as the PM-aR’-T, taking inspiration from our study reported in chapter 2, the N-terminal arginine residues were substituted with ornithine. This reduced hemolytic activity and conserved potency. As a control experiment, the double disulfide bonds were reintroduced into the same sequence, and it was found to be a rigid b-hairpin in solution with very high hemolytic activity. This shows that the flexible backbone conformation that becomes helical in the presence of membrane is responsible for reduced hemolysis. However, this variant did not show an intrinsic helical CD signature in presence of helix-inducing apolar solvents like trifluoroethanol and methanol. Hence, the N-methylation at the turn residues was removed, and D-alanine was substituted with L-alanine to avoid the induction of any reverse turn. Also, two aaminoisobutyric acid residues were introduced at i-1 and (i+3)+1 positions flanking the central four residues of the previous b-turn region because a-aminoisobutyric acid has a very high propensity to nucleate local helical conformation. The final molecule was unstructured in solution but showed a helical signature in presence of apolar solvents as well as bacterial membrane-mimicking SDS micelles. The hemolysis and MIC values were determined for it. Interestingly, it showed no hemolysis even at a concentration as high as 200 μM and had become highly selective towards gram-negative bacteria, which was clear from the PI uptake studies. The same positional mutations were grafted onto the linearized variants of Tachyplesin-1 and Protegrin-1. To our surprise, they also showed membrane induced helicity, non-hemolysis and gram-negative selectivity. In the sixth chapter, all the studies are summarized, followed by materials and methods section and appendix containing other supplementary information.