Synthesis, antimicrobial activity, and in silico mechanistic study of linear bacicylin analogs

1. Introduction

The increasing number of microorganisms resistant to many treatments has become a global health issue, and antibiotic-resistant infections are predicted to cause 10 million deaths annually by 2050 if new antimicrobial methods are not developed [1]. Antimicrobial peptides (AMPs) are a viable alternative to traditional antibiotics due to their high activity, low resistance, low toxicity, and quick mode of action [2,3]. They can combat pathogenic microbes resistant to multiple medicines due to the diverse targets of AMPs within bacterial cells, making it more challenging for bacteria to develop effective resistance mechanisms [4].

Wiese et al. isolated a cyclic antibacterial hexapeptide, bacicyclin, from bacteria found in blue mussels [5]. This peptide consists of six amino acids (D-Phe, D-Ala, L-Ile, L-Val, L-Leu, and Gly) and exerts antibacterial action with a MIC of 4.8−7.2 µM against various Gram-positive and Gram-negative bacteria [6,7]. However, the synthesis of bacicyclin is challenging as it is a cyclic peptide, particularly during the cyclisation step, resulting in a low yield.

Therefore, this study investigated the effect of amino acid substitutions on the activity of bacicyclin using a computational approach developed by Meher et al. [8]. Bacicyclin was chosen as the template due to its reported antimicrobial potential, as described by Wiese et al. [5]. In total, 18 analogs of linear bacicyclin were screened using iAMPpred, an online prediction tool for antimicrobial peptides [8], which utilizes a machine learning-based algorithm to predict the likelihood of peptide sequences exhibiting antimicrobial activity based on their physicochemical properties and sequence composition. Six analogs (Figure 1) with high predicted antimicrobial potential were selected for synthesis and further evaluation of their antimicrobial properties.

Close

Figure 1.

Chemical structures of six bacicyclin analogs.

Export to PPT

The linear bacicyclin analogs were synthesized by solid-phase peptide synthesis (SPPS) and their antibacterial properties against a variety of bacteria and fungi species were evaluated. A mechanistic study to understand its mode of action was then conducted on the most effective antimicrobial analogue. This analogue was also compared to linear bacicyclin using molecular dynamics simulations with a membrane model mimicking Gram-positive bacteria 1-palmitoyl-2-oleoyl-phosphatidylethanolamine (POPE):1-palmitoyl-2-oleoyl-phosphatidylglycerol (POPG) 1:3) [9,10] because AMPs exert their effects by disrupting microbial cell membranes causing loss of membrane integrity, ion leakage, and ultimately cell death [11,12].

2. Materials and Methods

The study workflow has been described in Figure 2.

Close

Figure 2.

Research flow diagram.

Export to PPT

3. Results and Discussion

Bacicyclin and its analogs (An-Bas1 – An-Bas-6) were synthesized by the SPPS method with Fmoc chemistry (Figure 2). The first amino acid, Fmoc-Gly-OH, was linked to 2-CTC resin in DCM and basic DIEA, yielding 0.4−0.6 mmol Gly amino acid per gram of resin. The unreacted sites were then capped by the addition of DCM/MeOH/DIEA (80:15:5). The loading resin values for each analogue ranged from 0.40 to 0.61 mmol/g, and thus were categorized as good, as they were in the range of 0.2−0.8 mmol/g. The Fmoc group was deprotected with 20% piperidine in DMF to provide a free amino group for reaction with the second amino acid, Fmoc-amino acid 2-OH, which was then coupled using HATU and HOAt coupling reagents in the presence of DIEA. The mixture of HATU/HOAt efficiently couples diverse peptide types and suppresses any potential racemization [25,26]. The peptides were synthesized by repetitive Fmoc deprotection and coupling before being cleaved from the resin with a TFA/TIS/water mixture (95/0.25/0.25). The yield of the linear peptides was ∼80%.

The peptides were characterized by HR-TOF ESI MS (Table 2, Figures S1-S7) and all HR-TOF ESI spectra contained [M + H]⁺ at m/z, consistent with the calculated molecular [M + H]⁺ ion peak at m/z for each peptide (Table 2). The ¹H-NMR of An-Bas-2 showed six alpha protons with chemical shifts of 3.79, 3.81, 4.18, 4.23, 4.31, and 4.40 ppm, with the ¹³C-NMR spectra showing six carbonyl signals of amide groups at 169.1, 171.0, 171.2, 171.6, 172.2, and 172.5 ppm. In addition, six alpha carbons were confirmed with chemical shifts of 49.1, 52.5, 57.9, 57.5, 51.2, and 47.3 ppm. The ¹H- and ¹³C-NMR data of all the synthetic peptides have been shown in the supplementary materials (Figures S8-S21, Table S1-S7).The activity of the peptides against S. epidermidis, S. aureus, E. coli, and S. typhimurium bacteria, as well as C. albicans fungi, in Table 3 reveals that most synthetic peptides exhibited minimal activity against the tested bacteria and fungi. However, An-Bas-2 demonstrated selective activity against Gram-positive bacteria (S. epidermidis and S. aureus) and the fungal strain C. albicans, with an MIC value of 103.1 µg/mL. Notably, An-Bas-2 showed no activity against Gram-negative bacteria (E. coli and S. typhimurium).

The discrepancy between iAMPpred predictions and experimental results suggests that the computational tool may overestimate activity for certain sequences, resulting in false positives. This limitation is attributed to the model’s reliance on static sequence-based and physicochemical properties, which do not account for the dynamical features critical to antimicrobial activity. In real biological contexts, peptides undergo conformational changes and interact with bacterial and fungal membranes in a highly dynamic manner. For example, their ability to adopt specific structures or align with membrane components can strongly influence their antimicrobial potential. This lack of dynamical features in the predictive model may partially explain why certain peptides predicted to be active by iAMPpred were ineffective in experimental tests. Nevertheless, iAMPpred provided a valuable starting point for identifying candidate peptides and accelerating the design process.

Our findings also underscore the importance of residue composition in determining antimicrobial activity. Among the seven peptides tested, those containing alanine residues were inactive, whereas An-Bas-2, which lacks alanine, displayed significant activity, suggesting that the absence of alanine may play a critical role in the antimicrobial properties of these bacicyclin-based analogs. Future studies could explore replacing alanine with functional residues to enhance the activity of similar peptides.

An-Bas-2 demonstrated selective activity against Gram-positive bacteria and the fungal strain C. albicans, possibly due to similar structural features in Gram-positive bacterial and fungal cells [27]. Both lack an outer membrane, allowing direct peptide interaction with their cell walls and membranes. Additionally, their negatively charged surface components facilitate strong electrostatic interactions with the positively charged An-Bas-2. In Gram-positive bacteria, these interactions involve teichoic and lipoteichoic acids in the peptidoglycan layer, while in fungi, the negatively charged components may include glucans or chitin in the cell wall, as well as phospholipids in the membrane. Nonetheless, the fungal membrane differs from Gram-positive bacteria in its lipid composition, particularly the presence of ergosterol, a sterol unique to fungal membranes. Despite this difference, An-Bas-2 may still effectively disrupt fungal membranes due to its ability to interact with negatively charged lipids or penetrate the fungal membrane structure. This dual activity against Gram-positive bacterial and fungal strains underscores the versatility of An-Bas-2 and its potential as a broad-spectrum AMP.

In contrast, An-Bas-2 showed no activity against Gram-negative bacteria such as E. coli and S. typhimurium, probably due to the presence of an outer membrane in Gram-negative bacteria, which acts as a physical and chemical barrier. The outer membrane contains lipopolysaccharides (LPS) that repel certain antimicrobial peptides, while efflux pumps and porins limit access to the inner membrane [27]. These structural differences may explain the specificity of An-Bas-2 for Gram-positive bacteria and fungi.

Furthermore, An-Bas-2’s superior activity makes it a promising candidate for antimicrobial agents. Cyclic peptide structures have demonstrated enhanced antimicrobial activity compared to their linear counterparts. For example, cyclic bacicyclin is more effective than linear bacicyclin [14]. Incorporating cyclic structures into future designs could further improve the efficacy of antimicrobial peptides.

The in silico mechanistic study of linear bacicylin and An-Bas-2 using Amber 24 involved simulations of a single peptide molecule in a POPE:POPG (1:3) lipid bilayer solvated with 0.15 M KCl solution with explicit TIP3P water molecules. Each simulation was run for 1 µs, with the initial interaction and penetration of linear bacicylin and An-Bas-2 into the lipid bilayer observed within the first 100 ns based on the molecular dynamics snapshots (Figure 3, top: An-Bas-2; bottom: linear bacicylin). The initial interactions of both peptides with the lipid bilayer were facilitated by hydrogen bond formation. For An-Bas-2, hydrogen bonding occurred between the cationic amines at the lysine residue and the N-terminus of the peptide with the anionic head groups of the lipids. In contrast, hydrogen bonding was observed only between the cationic amine at the N-terminus and the anionic lipid head groups for linear bacicylin. The penetration of both peptides into the lipid bilayer was initiated by interactions between the benzene ring of the phenylalanine residues and the lipid heads, as well as the beginning of the lipid tails.

Close

Figure 3.

(a) Molecular dynamics snapshots of An-Bas-2 in the POPE:POPG (1:3) lipid bilayer, (b) Molecular dynamics snapshots of linear bacicylin in the POPE:POPG (1:3) lipid bilayer.

Export to PPT

After the initial penetration of An-Bas-2, its behaviour within the lipid bilayer remained relatively stable from the early stage (0–300 ns) to the middle stage (300–600 ns) with An-Bas-2 successfully penetrating the membrane, primarily facilitated by the phenylalanine residues. However, during the late stage (600–1000 ns), the peptide penetrated deeper into the membrane, driven not only by the phenylalanine residues but also by additional contributions from other hydrophobic residues, specifically valine and leucine. These residues interacted with the lipid tails, enhancing the peptide’s ability to integrate into the hydrophobic membrane core. Linear bacicylin did not exhibit this behaviour. After its initial interaction with the lipid bilayer, its penetration was limited to the membrane surface, as its interactions remained primarily at the interface between the lipid head groups and the aqueous environment. This limited penetration suggests that linear bacicylin has a weaker membrane-disrupting capability compared to An-Bas-2. Moreover, An-Bas-2 demonstrated superior membrane-penetrating properties compared to linear bacicylin. After its initial insertion into the lipid bilayer, An-Bas-2 consistently interacted with both the lipid heads and tails, effectively anchoring within the bilayer. These results highlight An-Bas-2 as a more effective cell-penetrating peptide, aligning with our in vitro experimental findings, where An-Bas-2 exhibited significantly higher antimicrobial activity than linear bacicylin. The combined insights from both simulation and experimental data underscore the importance of hydrophobic interactions and residue composition in enhancing membrane penetration, which directly correlates with antimicrobial efficacy. The deeper insertion and broader interaction profile of An-Bas-2 make it a promising candidate for further development as an antimicrobial agent.

The electron density profiles of the peptides and their corresponding lipid bilayers were determined to evaluate the membrane-penetrating capabilities of An-Bas-2 and linear bacicylin (Figure 4, left: An-Bas-2; right: linear bacicylin). The lipid bilayer electron density for both systems exhibited the expected symmetrical distribution, with peaks near the headgroup region and a dip at the bilayer center (position ∼0 Å), confirming a well-structured bilayer. The electron density profile for An-Bas-2 demonstrates significant penetration into the bilayer, with the maximum density occurring at approximately -1.065 Å relative to the bilayer center, indicating that An-Bas-2 effectively interacts with the hydrophobic membrane core, suggesting its potential for strong membrane-disrupting activity. In contrast, the electron density profile for An-Bas-1 peaks at 0.295 Å relative to the bilayer center, indicating that it primarily interacts with the bilayer interface, exhibiting a limited membrane disruption ability. The comparative analysis highlights that An-Bas-2 penetrates significantly deeper into the lipid bilayer than linear bacicylin, suggesting that An-Bas-2 has a stronger affinity for the hydrophobic region of the membrane, which likely contributes to enhanced antimicrobial activity.

Close

Figure 4.

(a) Electron density profile of An-Bas-2 in the POPE:POPG (1:3) lipid bilayer system, (b) Electron density profile of linear bacicylin in the POPE:POPG (1:3) lipid bilayer system.

Export to PPT

The total number of hydrogen bonds formed was determined to understand the dynamics of the peptide and the POPE:POPG (1:3) lipid bilayer system interactions (Figure 5, left: An-Bas-2; right: linear bacicylin) and per amino acid residue of both peptides (Figures 6 and 7). Generally, hydrogen bonds were formed early in both simulations, with relatively constant hydrogen bond formation in the An-Bas-2/lipid bilayer system until ∼350 ns into the simulation when hydrogen bonds transiently increased for ∼100 ns before returning to their initial state around 800 ns. After 800 ns, the hydrogen bonds significantly increased, indicating deeper peptide membrane penetration, allowing more interactions to occur between the peptide and the bilayer. In contrast, hydrogen bond formation significantly increased after the initial interaction in the linear bacicylin /lipid bilayer system but then remained unchanged for the rest of the simulation. These observations are consistent with the molecular dynamics snapshot analysis (Figure 3). Additionally, the electron density profile analysis supports that An-Bas-2 interacts more effectively with the lipid bilayer. This conclusion is further corroborated by the hydrogen bond calculations with An-Bas-2 forming an average of 3.27 hydrogen bonds compared to 3.15 for linear bacicylin (Figure 5).

Close

Figure 5.

(a) Total hydrogen bonds formed in the An-Bas-2/POPE:POPG (1:3) lipid bilayer system, (b) Total hydrogen bonds formed in the linear bacicylin/POPE:POPG (1:3) lipid bilayer system.

Export to PPT

Close

Figure 6.

Residual total hydrogen bond analysis of the An-Bas-2/POPE:POPG (1:3) for (a) Glycine, (b) Isoleucine, (c) Leucine, (d) Lysine (e) Phenylalanine, (f) Valine.

Export to PPT

Close

Figure 7.

Residual total hydrogen bond analysis of the linear bacicylin/POPE:POPG (1:3) for (a) Glycine, (b) Isoleucine, (c) Leucine, (d) Lysine, (e) Phenylalanine, (f) Valine.

Export to PPT

The residual hydrogen bond calculations for both peptides with the lipid bilayer (Figures 6 and 7) revealed that phenylalanine and lysine (for An-Bas-2) or alanine (for linear bacicyclin) are the primary contributors to the interactions between the peptides and the POPE:POPG (1:3) lipid bilayer. In the molecular dynamics snapshots (Figure 3), valine and leucine in An-Bas-2 were observed to facilitate deeper membrane penetration during the late stage. This was confirmed by the hydrogen bond analysis (Figure 6), which shows an increase in hydrogen bond formation by valine and leucine in the late stage. Additionally, isoleucine and glycine contributed modestly to hydrogen bonding in An-Bas-2, primarily through interactions with the lipid head groups at the bilayer surface. In contrast, alanine introduced in place of lysine in linear bacicyclin (Figure 7) forms moderate hydrogen bonds (average 0.45) but lacks the strong electrostatic interactions observed for lysine in An-Bas-2. Glycine and isoleucine also show higher average hydrogen bond contributions in linear bacicyclin compared to An-Bas-2, indicating a greater reliance on surface-level interactions. Valine and leucine in linear bacicyclin do not exhibit increased hydrogen bonding during the late stages suggesting limited bilayer penetration.

These calculations highlight key differences in how the two peptides interact with the lipid bilayer. Phenylalanine plays a stabilizing role in both systems, forming more hydrogen bonds, whereas lysine dominates hydrogen bonding (average 1.51) in An-Bas-2, reinforcing the importance of electrostatic interactions with the negatively charged lipid head groups for membrane anchoring. Additionally, valine and leucine facilitate deeper bilayer penetration during the late stages of the simulation. Conversely, linear bacicyclin primarily relies on surface-level stabilization, with glycine and isoleucine contributing more prominently, while alanine offers moderate bonding but does not facilitate deeper interactions.

These findings confirm that An-Bas-2 interacts more effectively with the lipid bilayer due to its lysine-mediated electrostatic interactions and deeper penetration facilitated by valine and leucine, whereas linear bacicyclin remains localized at the bilayer interface, relying on surface interactions for stabilization. This dynamic distinction highlights An-Bas-2’s superior interaction profile and shows consistency with the MIC values obtained from the experiment.

This study serves as proof of concept, demonstrating the strength of combining computational predictions, experimental validation, and molecular dynamics simulations for antimicrobial peptide design. While the current approach highlights the feasibility of integrating these methods, real-world applications will require further refinement, including larger datasets, expanded peptide entries, and validation across diverse bacterial targets. Future work should incorporate consensus scoring functions that unify sequence-based predictions, physicochemical properties, and experimental feedback, creating a scientifically interpretable and reliable framework. By addressing these advancements, this multidisciplinary approach has the potential to evolve into a robust, generalizable tool for the rational design of antimicrobial agents to combat antibiotic-resistant pathogens.

4. Conclusions

In conclusion, this study demonstrates the successful design, synthesis, and evaluation of linear bacicyclin analogs, with An-Bas-2 emerging as a promising antimicrobial candidate. An-Bas-2’s enhanced activity stems from its ability to penetrate bacterial and fungal membranes more effectively, facilitated by specific hydrophobic and electrostatic interactions. This work serves as a proof of concept, showcasing the potential of integrating these three approaches for antimicrobial peptide design, but real-world applications will require further refinement and expansion of this framework.

One critical direction for future work is the development of a consensus scoring function that unifies sequence-based predictions, physicochemical properties, and dynamic insights derived from experimental feedback and molecular simulations to provide a more scientifically interpretable and reliable tool for predicting antimicrobial activity, addressing limitations of current models like iAMPpred, which rely on static properties and do not account for dynamical peptide-membrane interactions. Expanding datasets, incorporating feedback loops between predictions and experiments, and testing the framework across diverse bacterial and fungal targets are essential for building a robust and generalizable design strategy.

This study also underscores the importance of residue composition, as exemplified by An-Bas-2’s selective activity, and the potential of cyclic peptide structures for enhanced antimicrobial efficacy. By integrating predictive modelling, experimental validation, and mechanistic simulations, this multidisciplinary approach lays a strong foundation for advancing the rational design of antimicrobial peptides. While this study provides valuable insights, it is ultimately a starting point, and further optimization and validation are needed to translate these concepts into next-generation therapeutics for combating antibiotic-resistant pathogens.