Structure-guided design of a novel, stable, and soluble Cecropin A variant for antimicrobial therapeutic applications

Abstract

Cecropin A is a potent antimicrobial peptide with broad-spectrum activity; however, its clinical application is limited by poor stability, solubility, and bioavailability. In this study, an engineered analog of Cecropin A was developed by incorporating an M cell-targeting peptide (MTP) at the N-terminus, a self-assembling EAK16-II sequence at the C-terminus, and a rigid EAAAK linker to preserve domain integrity. This engineered peptide (Pep A) exhibited significantly improved physicochemical properties, including enhanced solubility (0.930), a higher extinction coefficient (8480 M^− 1cm^− 1), and a more favorable net positive charge (+ 7), all of which may support stronger interactions with microbial membranes and enhanced therapeutic potential. Despite a slightly elevated instability index, the peptide remained within the stable range. The engineered peptide was predicted to be non-toxic, in contrast to its native form, which was identified as toxic. Furthermore, its high solubility upon overexpression facilitates large-scale recombinant production in bacterial systems. Computational modeling using AlphaFold and Robetta confirmed the structural accuracy of the design, with Pep A outperforming other models in compactness and favorable folding. Molecular dynamics simulations further validated the improved stability of Pep A, demonstrating reduced flexibility, consistent hydrogen bonding, and lower solvent exposure. Point mutations at surface-accessible, flexible, and non-critical residues further enhanced its stability. Collectively, these modifications produced a multifunctional Cecropin A analog with improved structural stability, solubility, and bioavailability, establishing Pep A as a promising candidate for advanced antimicrobial and mucosal delivery applications. However, in vitro and in vivo validations are critical for confirming its final applications.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-18067-x.

Introduction

Antimicrobial peptides (AMPs) are a diverse class of molecules that play a crucial role in the innate immune response. They defend against a wide range of diseases and hold significant promise as alternatives or adjuncts to conventional antibiotics, particularly in the fight against multidrug-resistant (MDR) bacterial strains^1–3. Their structural diversity offers advantages over traditional small-molecules, recent advances in peptide engineering have improved their bioavailability, stability, and pharmacokinetic properties^4,5. Among the various classes of AMPs, cecropins have emerged as particularly promising candidates due to their potent antimicrobial activity, though their clinical application still requires improvements in stability and physicochemical properties⁶.

Cecropins are categorized into three principal types‒Cecropin A, B, and D‒and have been isolated from various insect families, including Diptera, Lepidoptera, and Coleoptera⁷. They exert their antimicrobial effects mainly by interacting with microbial membranes, leading to cell disruption. Several mechanisms have been proposed, including the carpet model, which suggests that high concentrations of Cecropins destabilize membranes via micelle-like aggregate formation⁸. Additionally, post-translational modifications such as amidation can enhance their activity^9,10.

Among the three types, Cecropin A has demonstrated broad antibacterial and antifungal activities^8,11. However, its clinical application is limited by poor stability, rapid proteolysis, and low bioavailability¹². To overcome these challenges, peptide engineering strategies—including sequence modifications, stability-enhancing substitutions, and domain fusion—are widely employed^13,14.

Functional domains are especially useful for improving therapeutic properties. For example, M cell-targeting peptides (MTPs) facilitate mucosal delivery¹⁵while self-assembling domains like EAK16-II can protect peptides from enzymatic degradation and enhance delivery efficiency^16–20. Strategies such as L-to-D isomerization and site-directed mutagenesis (SDM) also improve protease resistance and antimicrobial activity. The L-to-D isomerization of AMPs, which increases resistance to degradation by host and microbial proteases, as peptides composed of L-amino acids are generally more susceptible to enzymatic breakdown²¹. Additionally, SDM is used to precisely modify specific amino acids within a peptide or protein sequence. This approach can produce mutant variants with improved stability, enhanced antimicrobial activity, and other favorable functional properties²².

Computational techniques are routinely employed to complement traditional experimental approaches and facilitate the systematic design of peptides and proteins. These tools enable the redesign and optimization of peptide structures, prediction of binding free energies, and exploration of conformational dynamics with greater efficiency and accuracy^23,24. Among these methods, molecular dynamics (MD) simulations offer significant advantages by providing atomic-level insights into the flexibility, stability, and interaction patterns of peptides in various environments²⁵. This allows for predicting structural behavior over time, identifying favorable modifications, and rationalizing peptides with enhanced therapeutic properties.

In this study, we present a novel peptide engineering approach to optimize Cecropin A by simultaneously integrating three distinct strategies that, to our knowledge, have not been combined before. Specifically, we incorporated an M cell-targeting peptide (MTP) at the N-terminus to facilitate targeted delivery, and the self-assembling peptide EAK16-II at the C-terminus to improve structural stability—an approach not previously reported for Cecropin A or similar antimicrobial peptides. Furthermore, we applied site-directed point mutations to fine-tune structural integrity and enhance bioactivity. This unique combination of domain fusion and rational mutagenesis provides a new framework for designing antimicrobial peptides with improved solubility, stability, and therapeutic potential, thereby addressing key limitations that have hindered their clinical translation.

Materials and methods

Retrieval of cecropin A sequence and construction design

The primary sequence and 3D structure of Cecropin A were retrieved from the UniProt database (https://www.uniprot.org/) with an accession number of P01507 and from the RCSB Protein Data Bank (https://www.rcsb.org/) with a PDB code of 2LA2, respectively. Three peptide sequences were used in designing the engineered variant of Cecropin A: MTP with the sequence FYPSYHSTPQRP, an EAAAK linker, and the self-assembling peptide EAK16-II. The MTP was fused to the N-terminus of Cecropin A using the linker, while EAK16-II was attached to the C-terminus using the same linker.

Predicting toxicity, physicochemical properties, and solubility

Peptide toxicity was predicted using the CSM-Toxin server²⁶which employs a deep learning-based natural language model to interpret protein sequences in the context of their biological relevance. Physicochemical properties of the peptides, including length, molecular weight (MW), theoretical isoelectric point (pI), charge, estimated half-life in the mammalian reticulocytes, grand average of hydropathicity (GRAVY), extinction coefficient, aliphatic index, and instability index, were computed using the ProtParam tool²⁷. The solubility upon overexpression was evaluated using the SOLpro server²⁸which applies a machine learning algorithm based on features such as hydrophobicity, amino acid composition, secondary structure, solvent accessibility, and charge distribution²⁹.

Evaluating secondary structure and helical wheel

The secondary structure of peptides was analyzed using the JPred tool³⁰which utilizes the Jnet al.gorithm to provide predictions of secondary structure elements, solvent accessibility, and coil regions within protein sequences. Hydrophobic and hydrophilic interfaces within the secondary structure of the peptides were further analyzed using the NetWheels tool³¹ to generate helical wheel projections, aiding in the visualization of amphipathic α-helices.

Modeling and validating the 3D structure of peptides

The 3D structure of the engineered peptide was modeled using two widely recognized servers for comparative modeling to identify the most accurate structure. The Robetta³² and AlphaFold2³³ servers were employed for initial model generation. Subsequently, the peptide model was refined using the GalaxyRefine server³⁴. The GalaxyRefine server begins by reconstructing all side-chain conformations, followed by iterative structural relaxation through localized molecular dynamics simulations after side-chain repacking, thereby improving the quality of the initial model³⁵.

Following refinement, the resulting 3D structures were compared, and the final model quality was assessed using several validation tools, i.e. ERRAT, PROCHECK, and ProSA-web. The ERRAT server³⁶ evaluated non-bonded atomic interactions by comparing them to a database of well-refined crystallographic structures, helping to identify potential errors or irregularities in atomic contacts. The PROCHECK server³⁷ was employed to analyze the Ramachandran plot, which evaluates the stereochemical quality of the model by examining the distribution of backbone dihedral angles (ϕ and ψ) for each residue. The plot categorizes residues into favored, allowed, and disallowed regions, providing insight into the model’s structural integrity³⁷. Additionally, the ProSA-web server³⁸ was utilized to assess the overall quality of the model by calculating a Z-score that reflects the deviation of the total energy of the structure from values typical of experimentally determined protein structures in the RCSB Protein Data Bank (RCSB PDB).

Molecular dynamics (MD) simulations

MD simulations were conducted using the GROMACS 2022 package³⁹ to analyze the dynamic behavior of the studied peptides in an aqueous environment. The topologies of the peptides, designed for viral proteins, were generated using the OPLS-AA force field⁴⁰. Simulation boxes were solvated with SPC/E water molecules, and appropriate counterions were added to neutralize the system. Subsequently, energy minimization was performed using the steepest descent algorithm until the maximum force dropped below 10 kJ/mol⁴⁰. Periodic boundary conditions were employed in all directions. The simulations were conducted under NPT and NVT ensembles, maintaining a temperature of 310.15 K and a pressure of 1.0 bar. Temperature and pressure were controlled using the V-rescale thermostat and the Parrinello-Rahman barostat, respectively. Intermolecular interactions were calculated using a 1.0 nm cut-off for both short-range electrostatic and van der Waals interactions, while long-range electrostatic interactions were handled using the Particle Mesh Ewald (PME) method. Bond constraints involving heavy atoms were applied using the LINCS algorithm. A 100 ns production simulation was carried out using the leapfrog integration algorithm⁴¹. Molecular interactions and structural trajectories were visualized and analyzed using Discovery Studio software (v. 4.5) and the gmx tools, respectively.

Applying in Silico mutagenesis for residual modifications

Amino acid modifications are a fundamental strategy in peptide design⁴². Research has shown that specific amino acid substitutions can significantly influence the functional and structural properties of peptides²⁰. To assess the effect of point mutations on peptide stability, two computational tools, including CUPSAT and MuPro, were employed⁴³. CUPSAT is a web-based tool that predicts protein or peptide structural stability changes caused by single-point mutations. It generates stability predictions based on solvent accessibility and torsion angle distributions. Users can upload a 3D structure of the protein or peptide in PDB format and specify the mutation site. CUPSAT evaluates whether a mutation stabilizes or destabilizes by calculating the change in Gibbs free energy (ΔΔG). MuPro is an advanced machine learning-based tool to predict stability changes resulting from point mutations⁴⁴. Users can input an amino acid sequence or a structure in PDB format. MuPro provides a ΔΔG value, indicating whether the mutation will likely enhance or reduce protein stability. Its high prediction accuracy stems from integrating support vector machines and neural networks into its algorithm. Together, these tools offer valuable insights into how specific mutations may affect the overall structural stability of peptides.

Results

Designing a modified version of Cecropin A

To design engineered Cecropin A, the selected sequences were strategically combined with the peptide to enhance its stability and solubility. The N-terminal domain was modified by adding the sequence FYPSYHSTPQRPE (MTP). Additionally, the C-terminal segment featured a repetitive AEAEAKAK motif (EAK16-II), characteristic of synthetic peptides designed for self-assembly. The schematic representation of the engineered peptide based on the cecropin A template is shown in Fig. 1.

Fig. 1.

Design of the engineered peptide sequence based on Cecropin A. The construction was generated using three peptide segments: the M cell-targeting peptide (MTP) at the N-terminus, an EAAAK linker to enhance structural flexibility, and a self-assembling peptide sequence at the C-terminus.

Evaluation of toxicity, physicochemical, and solubility properties of peptides

The toxicity, solubility, and physicochemical properties of both the native and engineered peptides are summarized in Table 1. Toxicity analysis using the CSM-Toxin server predicted the native peptide to be toxic, whereas the engineered peptide was identified as non-toxic, underscoring its improved safety profile and potential for therapeutic applications. Physicochemical properties were further evaluated using the ProtParam tool. The engineered peptide exhibited a lower isoelectric point compared to the native peptide (9.82 vs. 10.39) and carried a higher net positive charge (+ 7 vs. +6), changes that may enhance electrostatic interactions with negatively charged microbial membranes and potentially improve antimicrobial efficacy. Its predicted half-life in mammalian reticulocytes was slightly shorter (1.1 h vs. 1.3 h), which could modestly reduce systemic persistence but may also lower the risk of prolonged off-target effects. The engineered peptide also demonstrated a higher predicted absorbance, with an extinction coefficient of 8480 M^− 1 cm^− 1 compared to the native peptide, which could facilitate more sensitive detection and quantification in analytical assays. Hydropathicity analysis showed a more hydrophilic profile for the engineered peptide, as reflected by a lower GRAVY index (-0.533 vs. -0.073), potentially improving solubility and reducing aggregation propensity. The aliphatic index was lower for the engineered peptide (72.00 vs. 108.11), which may indicate reduced thermostability; however, this is less critical for in vivo applications where physiological temperatures are maintained. The instability index increased (30.88 vs. 16.52); however, it remains well below the threshold of 40 defined by Guruprasad et al. (1990) for unstable proteins, thus still falling within the ideal stability range. In peptide engineering, stability is assessed by considering multiple parameters in parallel, such as hydrophobicity, GRAVY score, estimated half-life, and predicted protease resistance; and the observed increase likely reflects a trade-off from sequence modifications aimed at enhancing other critical attributes, including reduced susceptibility to degradation, while maintaining overall stability. Solubility upon overexpression was also slightly lower for the engineered peptide (0.930) compared to the native (0.988), yet still within a favorable range for recombinant production. While these in silico values provide valuable preliminary insights, we acknowledge that solubility, stability, and functional activity under serum-mimicking conditions, physiological pH, and physiological ionic strength can differ from computational predictions. Therefore, future work will experimentally evaluate these properties under biologically relevant conditions to confirm the in silico findings and assess their implications for therapeutic deployment.

Table 1.

The toxicity, physicochemical, and solubility upon overexpression analysis of native and engineered peptides using the CSM-Toxin, protparam, and SOLpro servers, respectively.

Properties	Native peptide	Engineered peptide
Toxicity	Toxic	Non-toxic
Length	37 residues	75 residues
Molecular weight	40.04 kDa	80.05 kDa
Isoelectric point (pI)	10.39	9.82
Charge	+ 6	+ 7
Half-life (mammalian)	1.3 h	1.1 h
Extinction coefficient	5500 M−1 cm−1	8480 M−1 cm−1
GRAVY index	− 0.073	− 0.533
Aliphatic index	108.11	72.00
Instability index	16.52	30.88
Solubility upon overexpression	0.988	0.930

The extinction coefficient is measured in water at 280 nm.

Secondary structure, solvent accessibility, and helical wheel prediction

Prediction of the secondary structure of engineered peptide was evaluated using the JPred tool. The analysis revealed a structure composed of alpha-helical (H) and coil (-) regions, with three to four distinct helical segments interspersed among coil regions (Fig. 2). No extended beta-sheet (E) structures were predicted.

Fig. 2.

JPred secondary structure prediction. The JNETCONF track displays confidence scores (ranging from 0 to 9), with taller black bars indicating high confidence in the assigned structural predictions. Secondary structure predictions from JNETSOL25, JNETSOL0, JNETHMM, JNETPSSM, and JNETJURY identify regions assigned as coils (C), alpha helices (H), and beta strands (B). Alpha-helical areas are highlighted with red horizontal bars, while coils and beta-strands are denoted by the letter’s ‘C’ and ‘B’, accompanied by underlined dashes, respectively. Dashed and dotted lines represent unstructured regions. Black bars indicate predicted solvent accessibility; JNETSOL25 and JNETSOL0 distinguish between buried, exposed, or buried residues within the structure.

As illustrated in Fig. 3, the Schiffer-Edmundson helical wheel model and secondary structure prediction demonstrate that the hydrophobic and hydrophilic residues are arranged on opposite faces of the engineered peptide structure.

Fig. 3.

Schiffer-Edmundson’s helical wheel model represents the amphipathic nature of the engineered peptide, with hydrophobic (yellow) and hydrophilic (colored according to polarity) residues separated on opposite faces.

From prediction to refinement of 3D structures

The 3D structures of the engineered peptide were modeled using AlphaFold and Robetta servers. Subsequently, all generated models were refined using the Galaxy Refine tool to enhance the quality of the 3D structures. Validation results are presented in Table S1. According to the findings, AlphaFold models displayed high predictive accuracy, while Robetta models demonstrated excellent structural quality. For the AlphaFold predictions, refined model 1 was chosen as the final structure, showing a ProSA Z-score of -1.34 (Fig. 4C) and an ERRAT quality score of 98.46% (Fig. 4E). The Ramachandran plot analysis revealed that 98.5% of residues were in favored regions, 1.5% in allowed regions, and none in outlier regions (Fig. 4A).

Fig. 4.

Validation of 3D structures generated by AlphaFold (A,C,E) and Robetta (B,D,F), showing ProSA-web Z-scores (-1.34 and − 1.56), ERRAT quality scores (98.46% and 100%), and PROCHECK Ramachandran plots with 98.5% of residues in favored regions, respectively.

A similar validation process was applied to the Robetta models (Table S1), where model 2 was selected based on its superior evaluation metrics, including a ProSA Z-score of -1.66 (Fig. 4D) and an ERRAT quality score of 100% (Fig. 4F). The Ramachandran plot for this model indicated that 98.1% of residues were in favored regions, 1.9% in allowed regions, and 0% in outlier regions (Fig. 4B). These two high-quality 3D models were subsequently selected for MD simulations and comparative analysis with the native Cecropin A template.

Molecular dynamics (MD) simulations

MD simulations were performed using the GROMACS tool to evaluate the engineered peptides’ structural stability. Figure 5 illustrates the optimized 3D structure representations of Cecropin A, Pep A (AlphaFold), and Pep B (Robetta) after MD simulations. A 100 ns MD simulation was conducted to investigate their dynamic behavior in an aqueous environment.

Fig. 5.

3D Structure representation before and after MD stimulations for Cecropin A (native form), Pep A (the best model of the engineered peptide from AlphaFold), Pep B (the best model of the engineered peptide from Rosetta).

As shown in Fig. 6A, the root means square deviation (RMSD) values for the native peptide exhibited consistent fluctuations during the first 30 ns, followed by a gradual increase to approximately 10 Å, stabilizing around 70 ns. Notably, the RMSD values for Pep A increased significantly during the initial 20 ns, reaching around 25 Å, and then stabilized with minimal fluctuations. In contrast, the RMSD profile for Pep B showed upward fluctuations during the first ~ 40 ns, reaching 20 Å, and remained stable at this level until the end of the simulation.

Fig. 6.

Evaluating changes in RMSD (A), RMSF (B), Rg (C), and SASA (D) of native Cecropin, Pep A (modeled by AlphaFold), and Pep B (modeled by Robetta) during 100 ns MD simulations at 310.15 K and 1 bar.

In addition, root mean square fluctuation (RMSF) analysis was performed to evaluate residue fluctuations. As illustrated in Fig. 6B, the RMSF plot for Pep A generally displayed lower average fluctuations compared to Pep B, particularly around the linker regions (residues 13–17 and 55–59) and the engineered segment (residues 60–75). These findings are consistent with the RMSD results, suggesting that Pep A exhibits higher overall structural stability. Notably, the template segment (residues 18–54) in Pep A exhibited lower RMSF values than in both Pep B and the original Cecropin template, indicating better preservation of the native core structure. In contrast, Pep B showed greater flexibility in its linker and engineered regions, which may contribute to its comparatively higher overall RMSD and potentially less stable conformation.

The Rg plot was analyzed to assess the overall size and shape of the peptides throughout the simulation. According to the results, the Rg value of Pep B decreased from approximately 3 nm to 1.5 nm during the first 36 ns (Fig. 6C) and remained stable at this level until the end of the simulation, indicating that the peptide adopted a more compact conformation. A similar compaction process was observed for both Pep A and the native form, but they achieved compactness more rapidly, within the first 10 and 5 ns, respectively (Fig. 6C). However, the Rg plot of the native form displayed moderate fluctuations over time. Overall, the native form exhibited greater compaction by the end of the simulation compared to Pep A and Pep B. This can be attributed to the absence of additional peptide sequences, such as the linker, EAK16-II, and MTP sequences, present in the engineered peptides, resulting in a smaller overall size and simpler structure. Since Rg and SASA analyses are positively correlated, the native form exhibited the lowest solvent-accessible surface area, consistent with its more compact structure (Fig. 6D).

As shown in Fig. 7A, the native Cecropin exhibited relatively limited motion, ranging from − 2 to 2 on the PC1 axis (range = 4) and from − 1.2 to ~ 0.8 on the PC2 axis (range = 2). The PCA analysis of Pep A indicates that it undergoes minimal large-scale motion or conformational changes, whereas Pep B displays greater dynamic behavior with more pronounced structural fluctuations (Fig. 7B, C). Based on the Rg and SASA analyses, Pep A demonstrated greater compactness and reduced structural fluctuations compared to Pep B.

Fig. 7.

Principal component analysis (PCA) of native Cecropin (A), Pep A (B), and Pep B (C). The PCA results show that Pep A’s motion range closely aligns with Cecropin’s, indicating structural stability, while Pep B exhibits broader ranges along the first two principal components, reflecting greater conformational flexibility. Pep A demonstrates superior structural integrity compared to Pep B.

These findings indicated that Pep A exhibits greater structural stability and reduced solvent exposure compared to Pep B, making it the most stable among the engineered peptides. Time-resolved plots of hydrogen bonds were generated for the native and the two engineered peptides. The native peptide consistently formed fewer hydrogen bonds, averaging around 20 throughout the simulation. In contrast, both Pep A and Pep B formed approximately twice as many hydrogen bonds as in the native (Fig. 8).

Fig. 8.

Time evaluation plots of H-bonds for native Cecropin (colored in navy blue), Pep A (colored in green), and Pep B (colored in light blue) over 100 ns simulations at 310.15 K and 1 bar. Pep A forms a higher average number of H-bonds compared to Pep B, highlighting its superior structural integrity.

Evaluating the effect of in silico mutagenesis on cecropin A

Application of the CUPSAT server

To evaluate the effect of point mutations on the structure and stability of Cecropin A, the CUPSAT server was used to introduce point mutations within the mature peptide sequence. Mutation sites within three distinct regions of Cecropin A, i.e., surface-accessible residues, flexible residues, and non-critical residues, were identified based on key structural and functional features. The sequence segment analyzed was KWKLFKKIEKVGQNIRDGIIKAGPAVAVVGQATQIAK.

As detailed in Table 2, mutations were introduced at the surface-accessible lysine residues at positions 3 (K3) and 10 (K10). These residues exhibited high solvent accessibility and are positioned within the helical regions of the peptide, as supported by torsion angles consistent with alpha-helical structures. K3 displayed a high solvent accessibility of 89.45%, with torsion angles of -53.9° and − 67.1°, indicating it is highly flexible and surface-exposed. Mutational analysis predicted substituting K3 with leucine or glutamine would significantly improve the peptide’s stability. Similarly, K10 exhibited a solvent accessibility of 63.2% and torsion angles of -65.4° and − 58.7°, supporting its location within a helical region. Mutations at K10, particularly to isoleucine, also enhanced stability, suggesting favorable energetic changes (Table 2). These results collectively demonstrate that selective mutation of surface-exposed lysine residues, particularly those in helical segments like K3 and K10, can improve the structural stability of Cecropin A. The findings highlight the importance of solvent accessibility and torsional flexibility in guiding mutation strategies for stability enhancement.

Table 2.

The introduction of mutations at the selected residues in the native peptide using the CUPSAT server and their impact on the peptide stability, solvent accessibility, and torsion angle compatibility.

Mutation position	WT residue	Mutated residue	ΔΔG (kcal/mol)	Effect on stability	Solvent accessibility	Torsion angle compatibility
3	K	A	1.79	Stabilizing	Exposed	Favorable
3	K	V	0.11	Stabilizing	Exposed	Favorable
3	K	L	0.43	Stabilizing	Exposed	Favorable
3	K	I	0.57	Stabilizing	Exposed	Favorable
3	K	M	0.43	Stabilizing	Exposed	Favorable
3	K	E	0.83	Stabilizing	Exposed	Favorable
3	K	R	0.8	Stabilizing	Exposed	Favorable
10	K	V	1.51	Stabilizing	Exposed	Favorable
10	K	I	1.27	Stabilizing	Exposed	Favorable
10	K	W	1.89	Stabilizing	Exposed	Favorable
10	K	F	0.78	Stabilizing	Exposed	Favorable
10	K	D	1.03	Stabilizing	Exposed	Favorable
10	K	H	0.87	Stabilizing	Exposed	Favorable
18	G	L	0.14	Stabilizing	Flexible	Favorable
18	G	Q	0.04	Stabilizing	Flexible	Favorable
18	G	V	0.43	Stabilizing	Flexible	Favorable
18	G	L	0.02	Stabilizing	Flexible	Favorable
18	G	I	0.7	Stabilizing	Flexible	Favorable
22	A	V	1.51	Stabilizing	Buried	Favorable
22	A	I	1.27	Stabilizing	Buried	Favorable
22	A	W	1.89	Stabilizing	Buried	Favorable
22	A	F	0.78	Stabilizing	Buried	Favorable
22	A	K	0.97	Stabilizing	Buried	Favorable
22	A	D	1.03	Stabilizing	Buried	Favorable
22	A	H	0.87	Stabilizing	Buried	Favorable

Mutations introduced at flexible residues, specifically glycine at positions 12 (G12) and 18 (G18), were also evaluated to assess their impact on peptide stability (Table 2). Glycine’s inherent flexibility contributes to the dynamic nature of the peptide, and modifying these sites was aimed at enhancing structural stability. Substituting glycine with more rigid and Stabilizing residues such as alanine, valine, or leucine is predicted to improve the peptide’s structural robustness while potentially modulating its dynamic behavior. These results suggest that targeted replacement of flexible residues with stabilizing amino acids may enhance overall peptide stability without significantly impairing necessary conformational flexibility.

Alanine residues at positions 22 (A22) and 25 (A25) were identified as non-critical for maintaining the core structure of the peptide (Table 2). These sites could accommodate non-disruptive mutations without compromising overall peptide stability. As a result, substitutions at these positions were expected to have minimal structural impact, enabling functional modifications. Additionally, the solvent accessibility of wild-type alanine at position 25 was 81.67%, indicating that it is likely surface-accessible and flexible, further supporting its suitability for mutation without significant structural consequences.

Application of MUpro server

To investigate the impact of specific residues on the stability and structural flexibility of cecropin, we introduced point mutations at four positions using the MUpro server. The substitution of lysine (K) at position 10 with valine (V) resulted in a ΔΔG value of 0.036 kcal/mol, indicating a slight increase in protein stability, with a confidence score of 0.00783. Similarly, replacing glycine (G) at position 12 with leucine (L) yielded a more substantial ΔΔG value of 0.306 kcal/mol, suggesting increased stability, supported by a confidence score of 0.542. These findings demonstrated the potential Stabilizing effects of specific amino acid substitutions in the peptide sequence. At position 22, substituting alanine (A) with isoleucine (I) resulted in a ΔΔG value of 0.147 kcal/mol, indicating a moderate increase in stability with a confidence score of 0.132. Likewise, at position 25, replacing alanine (A) with valine (V) produced a ΔΔG value of 0.163 kcal/mol, reflecting an increase in stability supported by a higher confidence score of 0.57. Collectively, these results highlight the Stabilizing potential of these specific mutations, as predicted by support vector machine (SVM) models based on sequence data.

Discussion

Cecropin A is a promising antimicrobial peptide, but its application is limited by stability, solubility, and bioavailability issues. To address this, different functional segments were incorporated: self-assembling EAK16-II at the C-terminal, MTP sequences at the N-terminal, and a rigid EAAAK linker between domains. The linker forms a stable alpha-helix that separates domains, preventing steric interference and maintaining function⁴⁵. By keeping domains apart, it helps ensure that each domain folds correctly without being sterically hindered or disrupted by neighboring domains⁴⁶.The engineered peptide showed improved physicochemical properties, including predicted non-toxicity, enhanced solubility, and higher net positive charge, which may strengthen interactions with bacterial membranes. Although the instability index increased (30.88), it remained within the stable range (< 40). Notably, the solubility of the engineered peptide upon overexpression was significantly improved, which not only facilitates its formulation for medical use but also supports its efficient production in bacterial systems. The lower aliphatic index (72.00 vs. 108.11) still indicated structural rigidity and adaptability⁴⁷while higher solubility supports recombinant production and eco-friendly large-scale applications⁴⁸.

In peptide engineering, modifying solubility and half-life is key for optimizing pharmacological performance. Short half-life reduces systemic accumulation and toxicity^49,50whereas poor solubility promotes aggregation—sometimes desirable for hydrogel formation but usually detrimental to bioavailability⁵¹. Balancing these properties without compromising activity or inducing immunogenicity remains a major challenge^52,53. The Schiffer-Edmundson helical wheel confirmed an amphipathic structure, supporting interactions with membranes and aqueous environments. In addition to improved physicochemical properties and lack of predicted toxicity, the appended segments may further enhance stability and delivery potential. Previously, it has been reported that MTPs facilitate intestinal transcytosis and mucosal immune responses^54–56. Furthermore, EAK16-II, known for its amphiphilic and self-assembling properties, stabilizes therapeutic agents by forming nanostructures that protect against degradation and enable sustained release. For example, EAK16-II has been used to encapsulate ellipticine, improving its stability, therapeutic efficacy, and cellular uptake^57,58. Additionally, the inclusion of a rigid EAAAK linker maintains structural separation between functional domains, ensuring optimal biological activity⁵⁹.

AMPs such as Cecropin A provide versatile scaffolds for developing multifunctional agents^34,35. Structural modifications (e.g., glycine substitutions, amidation, or D-amino acid incorporation) improve protease resistance and bioactivity¹³while MTPs are widely used in drug delivery and vaccine design due to their potential to target M cells and exploit immune surveillance mechanisms^55,60.

Here, 3D models generated by AlphaFold and Robetta confirmed high structural quality, with most residues in favored Ramachandran regions, ERRAT scores > 90, and acceptable ProSA Z-scores. AlphaFold-derived Pep A was superior across evaluations. MD simulations confirmed structural stability of Pep A and Pep B, with Pep A equilibrating faster, displaying lower residue fluctuations, more stable hydrogen bonding, and compact folding (Rg ~ 14.5Å). Its SASA stabilized after initial rearrangement, while Pep B showed greater fluctuations, indicating reduced stability. PCA further demonstrated Pep A maintained stability comparable to native Cecropin, with limited flexibility and the highest average hydrogen bond count. Collectively, these results identify Pep A as the most stable construct. However, this work has focused on improving peptide solubility and stability through rational sequence modifications. While membrane interaction modelling (e.g., lipid bilayer simulations) would be valuable for confirming retention of membrane-disruptive activity, future studies can incorporate membrane-binding and disruption modelling, as well as experimental assays, to evaluate whether the engineered modifications affect the peptide’s membrane-targeting properties.

To further optimize performance, point mutations were introduced at surface-accessible residues, enhancing stability through favorable energetic contributions. Previous studies support such strategies: tryptophan or phenylalanine substitutions increase protease resistance and membrane affinity^61,62; D-amino acid replacement enhances serum and in vivo stability⁶³; and non-natural amino acids improve cell penetration while reducing cytotoxicity⁶⁴. Charge modifications with lysine or arginine strengthened electrostatic interactions⁶⁵while hydrophobic substitutions optimized amphipathic helices for selectivity and membrane disruption⁶⁶.

In summary, integration of MTP, EAK16-II, and EAAAK into Cecropin A produced Pep A, a multifunctional analog with significantly enhanced solubility, folding, and structural stability, confirmed by computational analyses. These improvements make Pep A as a strong candidate for advanced antimicrobial applications. While this work focused on computational modeling, future studies will incorporate membrane-binding simulations and experimental validation (e.g., TEM, DLS) to confirm biological activity.

Conclusion

This study presents a rationally engineered version of Cecropin A by incorporating M cell-targeting sequences (MTP), a self-assembling EAK16-II domain, and a rigid EAAAK linker to possibly enhance its antimicrobial efficacy, stability, and mucosal delivery. Bioinformatic and molecular dynamics analyses demonstrated improved physicochemical properties, structural stability, and solubility, with Pep A showing the most favorable profile. Structural modeling confirmed proper folding and amphipathic nature, while targeted amino acid substitutions further enhanced protease resistance and functional performance. These results highlight the potential of peptide engineering in improving therapeutic peptides; however, experimental validation is necessary to confirm the computational predictions.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

AMPs

Antimicrobial peptides

Pep A

Peptide A

Pep B

Peptide B

MDR

Multidrug-resistant

MTP

M cell-targeting peptides

SDM

Site-directed mutagenesis

MD

Molecular dynamics

PME

Particle Mesh Ewald

ΔΔG

Gibbs free energy

RMSD

Root means square deviation

RMSF

Root means square fluctuation

SVM

Support vector machine

CPE

Claudin-4-specific peptides

SASA

Solvent-accessible surface area

PCA

Principal component analysis

MW

Molecular weight

pI

Theoretical isoelectric point

GRAVY

Grand average of hydropathicity

PDB

Protein Data Bank

Author contributions

Navid Nezafat, Cambyz Irajie, contributed to the conception, design, and interpretation of literature data. Samaneh Hashemia, wrote the manuscript. Armin Zarei, and Mohammad Sadegh Taghizadeh, Molecular dynamics analysis. Amir Savardashtaki prepared the figures and tables.

Data availability

All the data generated or analyzed during this study are available in the manuscript.

Declarations

Competing interests

The authors declare no competing interests.