Rational design of hyperstable antibacterial peptides for food preservation

Food spoilage and food-borne illnesses are often caused by microbial contamination, leading to significant post-harvest losses (~25% globally). While chemical preservatives are widely used, safety and regulatory concerns motivate a shift toward safer, natural alternatives. Antimicrobial peptides (AMPs) are promising due to their activity and biocompatibility, but their application is limited by issues like toxicity, low stability, poor solubility, and protease susceptibility. Nisin, a widely used peptide preservative, is restricted by loss of activity at neutral/basic pH, thermal instability, and limited spectrum (ineffective against Gram-negative bacteria, molds, and yeasts). The research question is whether rationally designed peptides can combine broad pH stability, thermostability, and antibacterial activity with low cytotoxicity and protease resistance for effective food preservation. The study leverages stable Bowman–Birk inhibitors (BBIs) with a conserved trypsin-inhibitory loop to design chimeric peptides that couple membrane-targeting tails and protease-inhibitory loops, hypothesizing dual mechanisms of action and enhanced robustness for food applications.

AMPs are short (<50 aa), often cationic, and act primarily by membrane destabilization/pore formation, with some targeting intracellular processes (protein, DNA, RNA synthesis). Clinical translation is hampered by stability, toxicity, immunogenicity, and protease susceptibility. Nisin’s limitations in pH and temperature stability and spectrum highlight the need for new peptides suitable for food contexts. Prior work includes isolation/engineering of natural AMPs and de novo designed AMPs with improved specificity and stability. BBIs, plant and animal serine protease inhibitors with high stability (disulfide-bonded loop), have been used in agricultural and therapeutic contexts. Structure-guided insights suggest that cationic and hydrophobic residues drive membrane interactions; protease-inhibitory loops confer intracellular targeting potential. The study builds on these principles, combining rational design with structure-function analysis to engineer robust AMPs for food preservation.

• Peptide design: Analyzed two natural peptides—HVBBI (β-hairpin BBI from Huia versabilis) and SFTI (bicyclic sunflower trypsin inhibitor)—sharing a disulfide-constrained trypsin-inhibitory loop but differing in tail composition and charge. Designed chimeras and variants: HSEP1 (SFTI loop with HVBBI-like tails), HSEP2 (increased hydrophobicity by adding Phe and Gly in the C-tail), HSEP3 (added N-terminal Arg to increase cationicity), and further loop/tail mutants and variants (e.g., K8 mutations, ΔK19, loop replacements, charge/hydrophobicity tuning).
• Peptide synthesis and quality: Synthetic peptides (>95% purity) confirmed by HPLC and ESI-MS; FITC-labeled HSEP2 for localization.
• Antimicrobial assays: Broth microdilution (CLSI modified) in MH broth against seven bacteria (four Gram-positive: Listeria monocytogenes, Bacillus cereus, Staphylococcus aureus, Micrococcus luteus; three Gram-negative: Escherichia coli, Pectobacterium carotovorum, Salmonella typhimurium). Inoculum 5×10^5 CFU/mL; peptide range 0.3–300 µg/mL; MIC via resazurin readout. Tested single peptides and peptide cocktails (HSEP3 with HSEP2-ΔHR or HSEP2-ΔHR,ΔK8 at defined ratios).
• Mechanistic assays: Confocal microscopy with FITC-HSEP2 for membrane localization; LIVE/DEAD staining (SYTO 9/PI) to assess membrane permeability. PI uptake kinetics quantified (excitation/emission 585/620 nm) over 2 h across peptide concentrations.
• Electron microscopy: SEM (glutaraldehyde fixation, ethanol dehydration, gold sputter-coating) and TEM (glutaraldehyde and osmium tetroxide fixation, dehydration, uranyl acetate staining, resin embedding, ultrathin sectioning) to visualize membrane morphology and ultrastructure.
• Trypsin inhibition: BAPNA-based assay at 37 °C; calculated inhibition constants (Ki) via double reciprocal and Dixon plots. Thermostability measured by residual trypsin inhibitory activity after incubation at 95 °C up to 200 min. pH stability by incubating peptides at pH 2.5, 5, 9 for 2 h and assessing residual trypsin inhibitory activity.
• Thermal treatments: Compared antibacterial activity post moist heat (autoclaving at 121 °C, 20 min) and dry heat (95 °C, 30 min) for HSEP3 vs Nisin.
• Molecular dynamics (MD): All-atom simulations with CHARMM36 using GROMACS. Constructed POPE:POPG (3:1) bilayers (200/400/800 lipids) to validate bilayer properties (area-per-lipid, thickness, compressibility). Peptide–membrane simulations used eight HSEP3 peptides bound to the upper leaflet (lipid:peptide 100:2) for 1.5 µs, monitoring bilayer thickness (Dp), area-per-lipid (AL), water permeation, and acyl-chain order parameters (SCD). Three independent replicates; membrane-only control 300 ns.
• Safety assays: Cytotoxicity in human retinal pigment epithelial (ARPE-19) and human intestinal epithelial (HIEC-6) cell lines; viability via WST-1 and MTT assays over peptide dose ranges. Hemolysis of human RBCs (4% suspension) across 4–200 µg/mL peptide, absorbance at 414 nm; Triton X-100 as 100% lysis control.
• Food preservation test: Cooked rice (1 g) dosed with HSEP3; inoculated with 1×10^5 CFU/g of B. cereus, M. luteus, or L. monocytogenes; stored at room temperature up to 6 days. Periodic plating for CFU and resazurin-based viability assessment.

• Rational design and activity: HSEP2 (SVIFGCTKSIPPICFVGFK) improved antibacterial activity versus HSEP1 (MIC against Micrococcus luteus: 6.25 vs 75 µg/mL) with minimal impact on trypsin Ki (HSEP2: 5.8×10^-7 M; HSEP1: 2.3×10^-7 M). Adding N-terminal Arg to create HSEP3 yielded the best activity (M. luteus MIC 1.25 µg/mL; Ki 2.0×10^-7 M). Tail hydrophobicity and cationicity are critical for potency.
• Spectrum (MICs, µg/mL): HSEP3 inhibited 5/7 tested species up to 100 µg/mL: M. luteus 1.25; B. cereus 12.5; L. monocytogenes 50; P. carotovorum 50; Salmonella typhimurium 85; Staphylococcus aureus 150; Escherichia coli 150. HSEP2: M. luteus 6.25; B. cereus 75; P. carotovorum 150; others ≥150.
• Thermostability/pH stability: HSEP3 retained ~50% trypsin inhibitory activity after 200 min at 95 °C (≈30% decrease in first 30 min). At pH 2.5, 5, and 9, HSEP3 maintained >80% trypsin inhibitory activity after 2 h. Antibacterial MICs were unchanged after dry or moist heat exposure.
• Mechanism—membrane effects: Confocal imaging showed peptide localization to bacterial membranes (B. cereus, M. luteus). LIVE/DEAD assays demonstrated increased membrane permeability with HSEP3. PI uptake in B. cereus increased within 5 min, saturating by ~45 min; dose-dependent with threshold near MIC (12.5 µg/mL). SEM revealed corrugated membranes and leakage; TEM showed cytoplasm-devoid regions, homogeneous electron density, membrane integrity loss, and visible pores.
• MD simulations: HSEP3 aggregated on the membrane surface, thinning the bilayer and increasing water permeation/defects. Measured changes: bilayer thickness Dp decreased from 4.12±0.03 nm (peptide-free) to 3.83±0.18 nm (peptide-bound), ~3 Å reduction; AL approximately unchanged (59.84±0.64 vs 60.87±0.75 Å^2). Increased variance in thickness and water density within the hydrophobic core; changes in SCD indicated reduced acyl-chain order. Dominant interactions: cationic residues (Lys/Arg) forming 3–4 H-bonds per peptide (≈6–8 kcal/mol) and hydrophobic contacts. Data support a carpet-like disruption mechanism without stable pore insertion on the simulated timescale.
• Dual-function role of loop and tail: Mutating the trypsin loop lysine K8 in HSEP2 to glycine abolished trypsin inhibition but retained partial antibacterial activity (MIC 37.5 µg/mL vs 6.25 parent), whereas deleting K8 eliminated both trypsin inhibition and antibacterial activity (>150 µg/mL). Deleting C-terminal Lys (ΔK19) reduced antibacterial activity (MIC 25 µg/mL) without abolishing trypsin inhibition (Ki 4.4×10^-7 M). Replacing the loop with a non-trypsin sequence while compensating charge/hydrophobicity (HSEP3-ΔTLCL+, MIC 3.125 µg/mL) preserved good antibacterial activity without trypsin inhibition. These results indicate the loop contributes via both physicochemical effects (charge/hydrophobicity aiding membrane disruption) and intracellular trypsin inhibition; membrane effects are dominant.
• Synergy concept: Peptide cocktails combining HSEP3 with a loop-only trypsin inhibitor (HSEP2-ΔHR) maintained MICs comparable to HSEP3 alone at 3:1 ratio (B. cereus 12.5 µg/mL; M. luteus 1.25 µg/mL). Replacing with an inactive loop (HSEP2-ΔHR,ΔK8) worsened MICs, supporting complementary dual-action design.
• Variant tuning: HSEP3 variants modulating hydrophobicity/polarity/charge yielded differing spectra; HSEP3c and HSEP3d improved activity against M. luteus (3.125 µg/mL) and showed some efficacy against Listeria and Pectobacterium (e.g., HSEP3b, HSEP3c ≤50 µg/mL).
• Safety: Low cytotoxicity in ARPE-19 and HIEC-6 cells: >80% viability up to 160–200 µg/mL (ARPE-19), >70% at 200 µg/mL; low hemolysis (<5%) at 0–200 µg/mL for HSEP2 and HSEP3.
• Food preservation efficacy: In cooked rice inoculated with B. cereus, M. luteus, or L. monocytogenes, HSEP3-treated samples showed no bacterial growth up to 6 days, while controls increased steadily. HSEP3 outperformed Nisin in stability (pH/heat) and antibacterial activity against B. cereus; ~16-fold higher inhibition of B. cereus spore germination versus Nisin.

Rationally designed, disulfide-stabilized peptides that combine cationic, hydrophobic tails for membrane targeting with a protease-inhibitory loop can achieve strong, robust antibacterial activity suitable for food preservation. Data support a dual mechanism: rapid membrane destabilization (carpet model) causing permeability and cell death, complemented by intracellular trypsin inhibition after entry. Tail optimization (increasing hydrophobic residues and positive charge) substantially enhanced activity across species, while loop modifications dissected contributions of charge/hydrophobicity versus specific trypsin inhibition. MD simulations mechanistically corroborate experimental findings, showing peptide aggregation on membranes leading to bilayer thinning, increased water defects, and disrupted acyl-chain order. The peptides maintain activity after extreme thermal and pH challenges, a key advantage over Nisin, and demonstrate low cytotoxicity/hemolysis, indicating a favorable safety profile. Food application testing on cooked rice confirms practical efficacy over multi-day storage. The synergy concept—replacing a fraction of membrane-active peptide with another intracellularly active component—provides a blueprint for designing cocktails with complementary mechanisms to broaden spectrum and reduce resistance risk.

This study delivers a rational design framework for hyperstable antibacterial peptides with dual mechanisms—membrane disruption and intracellular trypsin inhibition—by integrating features from Bowman–Birk inhibitors with tailored cationic/hydrophobic tails. The lead peptide HSEP3 exhibits strong antibacterial activity against multiple foodborne pathogens, retains activity after high-temperature and wide pH exposure, shows low cytotoxicity and hemolysis, and effectively preserves cooked rice by preventing microbial growth up to six days. Compared with the industry standard Nisin, HSEP3 offers superior stability and efficacy against certain targets, including spore germination. The design principles are generalizable to clinical and biomedical contexts and can be combined with combinatorial methods to accelerate discovery. Future work should optimize broad-spectrum activity, systematically explore synergistic peptide/drug combinations targeting multiple intracellular pathways, assess in vivo safety and pharmacodynamics in animal models, and investigate structure–activity relationships that drive species-specific efficacy.

• Species-dependent efficacy: Activity varies across bacterial species; Gram-negative coverage remains limited at practical concentrations.
• Mechanistic scope: MD simulations did not capture insertional pore formation events, potentially due to force field or timescale/peptide concentration limitations; conclusions are consistent with the carpet model but may not encompass all mechanisms.
• Stability assessments: Thermal and pH stability were primarily inferred from residual trypsin inhibitory activity, not comprehensive across all antibacterial readouts in every condition.
• Safety extrapolation: Cytotoxicity and hemolysis assays suggest low toxicity in vitro; however, comprehensive in vivo safety, immunogenicity, and gastrointestinal protease interactions require animal studies.
• Trypsin inhibitor concern: Potential anti-nutritive effects due to digestive serine protease inhibition necessitate dosing windows and in vivo validation analogous to therapeutic index considerations.
• Application scope: Food matrix effects, processing compatibility, and regulatory considerations were not fully explored beyond rice model testing.