Microbiome-derived antimicrobial peptides offer therapeutic solutions for the treatment of *Pseudomonas aeruginosa* infections

The study addresses the urgent need for novel therapeutics against multidrug-resistant Pseudomonas aeruginosa, an ESKAPE pathogen notorious for biofilm formation and antimicrobial resistance, especially in cystic fibrosis and chronic wound settings. Microbiome-derived antimicrobial peptides (AMPs), particularly from the rumen microbiome, are promising due to their cationic, amphipathic structures and unique antimicrobial mechanisms. Building on prior discovery and initial characterization of rumen AMPs (Lynronne 1–3 and P15s), the purpose here is to evaluate their in vitro and in vivo efficacy against clinical P. aeruginosa strains, elucidate mechanisms of action via membrane interaction and multi-omics analyses, assess potential for resistance development, and evaluate cytotoxicity and therapeutic index to inform their therapeutic potential.

The introduction reviews increasing multidrug resistance and highlights WHO-priority ESKAPE pathogens. AMPs are presented as a diverse class with broad antimicrobial and immunomodulatory activities and unique modes of action that can complement antibiotics. Prior work (Oyama et al.) identified rumen-derived AMPs (Lynronne 1–3, P15s) effective against MRSA and some P. aeruginosa, with Lynronne 1–3 adopting α-helical amphipathic conformations. The need for agents that inhibit biofilms is emphasized given P. aeruginosa’s role in chronic infections. The review positions microbiome-derived AMPs as promising candidates needing further mechanistic and preclinical evaluation against P. aeruginosa.

- Peptides: Rumen microbiome-derived AMPs Lynronne 1 (19 aa), Lynronne 2 (20 aa), Lynronne 3 (20 aa), and P15s (20 aa). P15s 3D structure predicted by PEP-FOLD and visualized in PyMOL. - Bacterial strains: Multiple clinical P. aeruginosa isolates (e.g., PAO1, LES400, LES431, LES858, C3719, AMT0060-2, AES-1R, NH57388A) with diverse AMR profiles. - Susceptibility testing: MICs determined by modified broth microdilution per ISO 20776-1 in cation-adjusted Mueller Hinton broth (as well as TSB/BHI where applicable); MBCs determined by subculture from MIC wells onto MH agar after 18–24 h at 37 °C. - Time-kill kinetics: Exponential-phase cultures of PAO1 and LES431 exposed to 3× MIC of peptides (Lynronne 1, 2, P15s) and comparator antibiotics (Levofloxacin, Polymyxin B) in MHB; CFU quantified over time (triplicates). - Biofilm assays: 96-well static model. Prevention of attachment: peptides at 3× MIC co-incubated with PAO1/LES431 for 24 h; biomass quantified by crystal violet OD570 and expressed as percent reduction vs control (ANOVA). Disruption of pre-established (24 h) biofilms: peptides added for an additional 24 h followed by CV staining. - Resistance induction: Serial passage (25 days) of PAO1 and LES431 in sub-MIC concentrations of Lynronne 1, 2, 3, P15s; daily MIC determination and passaging from highest growth-permissive concentration. - Cytotoxicity: BEAS-2B (human lung epithelial) and IMR90 (human bronchial fibroblast) cell viability after 48 h peptide exposure (up to 1 mg/mL) measured by resazurin assay; IC50 calculated. - Therapeutic Index: TI computed as IC50 (cell line)/MIC (organism) for relevant combinations. - Membrane permeabilisation: Propidium iodide uptake assays at 4× MIC over time; CTAB as control. - Peptide–lipid interactions: Langmuir monolayer (Langmuir balance) with total lipid extracts from PAO1/LES431 and pure lipids (PG, cardiolipin, PC, PE). Critical pressure of insertion (mN/m) determined. - TEM: PAO1 mid-log cells treated with 3× MIC of Lynronne 1 or P15s for 1 h; fixed and imaged (JEOL JEM1010) for morphological changes. - Transcriptomics (RNA-seq): PAO1 and LES431 treated at MIC for 1 h (n=3 per peptide and untreated); rRNA depletion; libraries prepared with Illumina TruSeq Stranded mRNA (poly-A enrichment omitted); HiSeq 2500 2×125 bp; reads trimmed (Trimmomatic), QC (FastQC/MultiQC), pseudoaligned (kallisto) to PAO1/LES431 references; DESeq2 for differential expression; PCA/PLS-DA/HCA; GO enrichment (ShinyGO). - Metabolomics (FIE-HRMS): Six biological replicates per condition; extraction with chloroform/methanol/water (1:3:1); Exactive HCD mass analyser in positive/negative modes; data log-transformed; PLS-DA and heatmaps via MetaboAnalyst; univariate statistics for key metabolites. - In vivo efficacy: Galleria mellonella model. Determination of PAO1 LD50 and LD; co-injection of bacteria (~10^3 or ~10^4 CFU/larva) with peptides (various mg/kg); survival monitored up to 96 h; Kaplan–Meier and log-rank tests. Separate peptide-only toxicity assessment in larvae.

- Antimicrobial activity: - All four AMPs were active against all clinical P. aeruginosa isolates; overall MIC range 4–512 µg/mL. - Lynronne 1 exhibited the lowest MICs across strains (e.g., 4–64 µg/mL); Lynronne 2 MICs 8–64 µg/mL; Lynronne 3 and P15s less active (32–256 and 32–512 µg/mL respectively). MBCs were at MIC or up to twofold higher. - Time-kill kinetics (3× MIC): - Lynronne 1 and 2 produced ≥3 log10 CFU/mL reductions within 2 h (PAO1) and within 10 min (LES431). - P15s was non-bactericidal against PAO1 within assay duration but caused >8 log10 CFU/mL reduction against LES431. - Levofloxacin bactericidal against both but slower (~3× longer) than Lynronne 1/2; Polymyxin B was >8 log reduction for PAO1 but ineffective against LES431 within the time window. - Biofilm activity: - At 3× MIC, Lynronne 1, 2, and P15s significantly inhibited biofilm attachment/formation for PAO1 (ANOVA P=2.33E-26) and LES431 (P=6.37E-24). - No significant disruption of established (24 h) biofilms at 3× MIC. - Resistance development: - No resistant mutants emerged in PAO1 or LES431 after 25 days of serial passage with Lynronne 1, 2, or P15s. - Cytotoxicity and Therapeutic Index: - IC50 (µg/mL): BEAS-2B: Lyn1 138.9±34.1, Lyn2 1177±309.2, P15s 3337±882.3; IMR90: Lyn1 94.23±21.74, Lyn2 803.2±202.8, P15s 948.9±224.0. - Mean TI: Lynronne 1 = 10.1; Lynronne 2 = 43.0; P15s = 29.7. P15s was least cytotoxic; Lynronne 1 had the lowest TI. - Membrane effects: - All AMPs permeabilised PAO1 and LES431 at 4× MIC within 20 min. Rapid permeabilisation by Lynronne 1/2 (e.g., PAO1: 49–82% at 5 min; LES431: 66–96% at 5 min). P15s slower (PAO1: 16% at 5 min; ~81% by 40 min). - Langmuir monolayer assays showed interactions with bacterial lipids; Lynronne 2 had strong affinity to bacterial extracts. Critical pressures of insertion (mN/m) into pure lipids (PC/PE/PG/Cardiolipin): Lyn1 50.28/48.43/59.51/58.61; Lyn2 55.75/41.82/49.48/39.47; P15s 33.57/33.69/48.74/50.05. - P15s showed notable affinity to PC (eukaryotic-mimicking lipid), despite lower cytotoxicity in cell assays. - TEM: - Lynronne 1-treated PAO1 showed membrane disruption and cytoplasmic leakage; P15s-treated cells showed subtler changes with intact outer membrane after 1 h. - Transcriptomics: - Clear separation of treated vs control by PCA/PLS-DA. In PAO1, many top DEGs associated with catalytic and transporter activity. - Lynronne 1/2 reduced expression of arcA, arcB, arcC (arginine deiminase pathway), suggesting impaired pH homeostasis via arginine metabolism; P15s notably downregulated arcD (arginine-ornithine antiporter). - Reduced expression of cbb3-type oxidase genes (ccoN2, ccoP2) with peptides, consistent with effects on respiration/biofilms; increased expression of NO-reductase subunits (norB/norC) indicative of nitrosative stress response. - Metabolomics: - Distinct metabolite profiles across treatments (PLS-DA). In LES431, significant changes in phosphatidylethanolamines and glycerol; in PAO1, ornithine levels were significantly altered, aligning with ADI pathway transcriptional effects. - In vivo (Galleria mellonella): - PAO1 LD50 ~1.15×10^2 CFU/larva; LD ~1.17×10^4 CFU/larva. - Treatment yielded 100% survival at 24 h for Lynronne 1 (32 mg/kg) and Lynronne 2 (128 mg/kg) at both 10^3 and 10^4 CFU/larva challenges. - P15s at 128 mg/kg did not protect at 10^4 CFU; at 384 mg/kg, 60% survival at 24 h and 10% at 48 h. In infected larvae, P15s treatment associated with melanisation; Lynronne 1/2 showed no melanisation in treated groups.

The findings demonstrate that rumen microbiome-derived AMPs are active against diverse clinical P. aeruginosa isolates, rapidly bactericidal (particularly Lynronne 1 and 2), and capable of preventing biofilm establishment. Mechanistically, these cationic peptides disrupt bacterial membranes, as supported by rapid propidium iodide uptake, TEM evidence for Lynronne 1, and lipid monolayer insertion data. Multi-omics analyses suggest additional impacts on central processes: alterations in arginine metabolism (downregulation of ADI pathway genes for Lynronne 1/2; arcD downregulation for P15s), respiratory chain modulation (reduced ccoN2/ccoP2), and nitrosative stress responses (increased NO-reductase gene expression). The β-sheet P15s appears to act differently from α-helical Lynronne peptides, with slower permeabilisation and strain-dependent killing (potent against LES431 but not PAO1 in time-kill), which may relate to lipid interactions and strain-specific LPS or membrane properties. Although AMPs prevented biofilm formation, they did not eradicate established biofilms at 3× MIC, indicating potential utility in prophylaxis or in combination with biofilm-disrupting treatments or antibiotics. Importantly, no resistance emerged after serial passage, consistent with rapid, multi-target action and lower resistance evolution risk compared with conventional antibiotics. Cytotoxicity profiling indicates a trade-off: Lynronne 1 shows higher antimicrobial potency but lower TI; Lynronne 2 balances strong activity with a higher TI; P15s is least cytotoxic but less efficacious in certain contexts and showed poorer in vivo protection at lower doses. In vivo, Lynronne 1 and 2 provided complete protection in Galleria, supporting their therapeutic potential. Overall, microbiome-derived AMPs, particularly Lynronne 2 (and modified Lynronne 1), are promising candidates, potentially as monotherapies in select contexts or as synergistic partners with antibiotics for difficult P. aeruginosa infections.

This work provides comprehensive preclinical evidence that rumen microbiome-derived AMPs (Lynronne 1, Lynronne 2, Lynronne 3, P15s) are effective against clinical Pseudomonas aeruginosa, with Lynronne 1 and 2 showing rapid bactericidal action, significant anti-biofilm establishment activity, and in vivo efficacy in Galleria mellonella. Mechanistic studies support membrane disruption with additional impacts on arginine metabolism and respiratory processes. Resistance did not emerge during prolonged sub-MIC exposure, and cytotoxicity was acceptable, particularly for Lynronne 2 and P15s with higher therapeutic indices. Future work should focus on: (i) optimizing peptide design (e.g., hydrophobicity/charge distribution, D-amino acid substitution) to reduce cytotoxicity and enhance stability; (ii) evaluating combination therapies with conventional antibiotics and/or physical biofilm-disruption methods; (iii) testing against established biofilms with adjusted dosing/regimens; (iv) expanding in vivo validation to mammalian models and diverse infection sites; and (v) elucidating strain-specific determinants of susceptibility (e.g., LPS and membrane lipid composition).

- Established biofilms were not significantly disrupted at 3× MIC; efficacy was limited to prevention of biofilm formation. - Detailed mechanistic inference from transcriptomics and metabolomics is complex; observed expression changes may reflect general stress responses rather than direct targets. - Time-kill and mechanistic assays focused primarily on two strains (PAO1 and LES431); broader strain coverage for mechanistic work is warranted. - Lynronne 1 exhibited comparatively higher cytotoxicity (lower TI), potentially limiting therapeutic window without optimization. - P15s showed strain-dependent efficacy and slower bactericidal action; in vivo protection required higher doses and was associated with melanisation in infected larvae. - In vivo validation was restricted to the Galleria mellonella model; translation to mammalian systems and pharmacokinetics/pharmacodynamics require further study.