Therapeutic peptide development is challenging because of the large sequence space and the need to balance potency, selectivity, stability, safety, pharmacokinetics, and manufacturability. Artificial intelligence (AI) has emerged as a useful approach for addressing these challenges, but its contribution to peptide development extends beyond computational prediction and sequence generation. This review examines the progression of AI-driven therapeutic peptide discovery from conventional computational methods and machine learning to deep learning, protein language models, foundation models, generative AI, and structure-aware and multimodal approaches, with particular emphasis on their relevance to therapeutic development. AI-based peptide research has progressed from descriptor-based prediction and structure-guided screening to representation learning, de novo peptide generation, and integrated sequence-structure-chemistry modeling. Experimental studies have demonstrated the biological activity of AI-designed peptides across several therapeutic applications, particularly antimicrobial and target-specific peptides. However, computational peptide production outpaces synthesis, biology, in vivo validation, and developability evaluation. Data heterogeneity, sequence redundancy and leakage, limited negative datasets, uncertainty, inadequate representation of noncanonical chemistry, and discrepancies between predicted and experimentally observed features remain challenges. AI-driven peptide research is moving from computationally assisted discovery toward a more integrated development framework. Future progress will depend on combining sequence, structure, chemistry, target interactions, developability, and experimental data within iterative workflows. The value of AI should therefore be judged not only by predictive performance or sequence novelty but also by its ability to produce synthesizable, biologically active, safe, and experimentally validated peptide candidates with realistic therapeutic potential.
This study aimed to validate the inhibitory effect and mechanism of a yak bone collagen-derived antioxidant peptide (GGGPGPM, named P1) against skin photoaging. A murine skin photoaging model induced by UV irradiation was established and topically treated with hydrogel containing different concentrations of P1. The anti-photoaging effects of P1 were evaluated through skin macroscopic observation, hydration and histological analysis, collagen content determination, and antioxidant enzyme detection. Furthermore, the key protein expression of the Keap1/Nrf2 signaling pathway and the composition of skin microbiota were analyzed by western blotting and amplicon sequencing. After intervention with P1, the collagen content and the activity of antioxidant enzymes in the skin of mice subjected to UV irradiation were significantly improved. Western blot analysis revealed that the high-dose P1 group exhibited 1.75-, 1.90-, and 2.72-fold increases in the expression levels of Nrf2, HO-1, and NQO1 compared to the model group, respectively. Additionally, P1 intervention significantly modulated the skin microbiota structure. At the phylum level, the high-dose P1 intervention elevated the abundance of Firmicutes and reduced the abundance of Actinobacteria, which might be mainly attributed to the rise of Staphylococcus and the decline of Corynebacterium. Specially, there was no significant difference observed between the high-dose P1 group and the blank control group in the composition of skin microbiota. Collectively, P1 might ameliorated UV-induced skin photoaging in mice by activating the Keap1/Nrf2 signaling pathway and restructuring skin microbiota.
Pulmonary fibrosis (PF) is a challenging interstitial lung disease with limited therapeutic options. This study explores the therapeutic effects of alamandine (ALA), a renin-angiotensin system peptide, in reversing established fibrotic progression in an experimental model. Male Wistar rats were divided into four groups (n = 5–6/group): control (CO), ALA-treated (ALA), bleomycin-induced fibrosis (BLM), and bleomycin plus ALA treatment (BA). Fibrosis was induced by intratracheal bleomycin (2.5 mg/kg) on day 0. Subcutaneous ALA treatment (50 µg/kg/day) began on day 10 and continued until day 19. Respiratory mechanics, body weight, Ashcroft score, and lung transforming growth factor-beta (TGF-β) content were evaluated. Plasma RAS peptides were quantified by LC-MS/MS. Bleomycin significantly increased respiratory resistance (CO = 0.104 ± 0.026 vs. BLM = 0.159 ± 0.047 cmH₂O.s/mL in CO, p < 0.008) and Ashcroft score, and reduced body weight gain. ALA treatment from day 10–20 markedly improved body weight gain (p < 0.001), reduced Ashcroft score (BLM = 2.23 ± 0.35 vs. BA = 1.21 ± 0.40, p < 0.0001), and decreased lung TGF-β1 content (BLM = 3.70 ± 1.7 vs. BA = 1.10 ± 0.42 pg/mg protein, p < 0.0015). These results were associated with qualitatively improved respiratory effort, suggesting attenuation of bleomycin-induced fibrogenesis. Despite rapid tissue uptake, late ALA treatment was associated with reduced histological fibrosis, lower TGF-β1 content, and improved selected functional and clinical indicators in rats with established pulmonary fibrosis. These findings suggest a potential therapeutic benefit that warrants further dose- and time-ranging studies.
Abdominal aortic aneurysm (AAA) is a life-threatening vascular disease lacking effective pharmacotherapy. Oxidative stress drives AAA progression, yet conventional antioxidant strategies show limited efficacy. The gut microbiota regulates host redox homeostasis, but its role in AAA remains unclear. Using gut metagenomic data from AAA patients, experimental AAA mouse models and an in vitro oxidative injury model, we investigated the role and underlying mechanisms of the gut microbiota-associatedmetabolite γ-glutamylcysteine (γ-GC) in AAA. The abundance of key γ-GC biosynthesis genes, as well as intestinal and circulating γ-GC levels were significantly reduced in AAA patients, negatively correlating with oxidative stress markers and aneurysm diameter. Oral γ-GC supplementation improved intestinal barrier integrity, reduced plasma lipopolysaccharide (LPS) levels, and suppressed aneurysm expansion and elastic fiber degradation. Furthermore, γ-GC alleviated oxidative stress in plasma and aneurysmal tissues, whilst reducing vascular smooth muscle cell (VSMC) apoptosis, macrophage infiltration, and the expression of pro-inflammatory cytokines (TNF-α, IL-1β, IL-6) and matrix metalloproteinases (MMP2, MMP9). Mechanistically, γ-GC upregulated Polo-like kinase 1(Plk1) expression and inhibited reactive oxygen species (ROS)-induced VSMC apoptosis and pro-inflammatory cytokine expression. Notably, γ-GC modulated the spatial distribution of Plk1 and G3BP1, thereby attenuating ROS-induced G3BP1 condensation and promoting the disassembly of pathological stress granules. The gut microbiota-associatedmetabolite γ-GC is negatively correlated with AAA severity. γ-GC supplementation ameliorates AAA through restoring redox homeostasis and suppressing pathological stress granule formation. While the functional link between Plk1 and G3BP1 condensation requires further investigation, our findings suggest thatγ-GC represents a promising therapeutic candidate for early pharmacological intervention in AAA.
This research aims to identify novel antimicrobial peptides (AMPs) from casein, subsequently conducting an in-depth exploration of their antibacterial mechanism of action. Casein was hydrolyzed by pepsin and fractionated via ultrafiltration, and the peptides were identified using Nano-LC-MS/MS. Then, the candidate antimicrobial peptides were screened based on in silico analysis of their antimicrobial potential, net charge, and solubility. The minimum inhibitory concentration (MIC) against E. coli ATCC 25922 was determined, and killing kinetics was analyzed. The effect on bacterial membrane integrity was assessed through the analysis of nucleic acid leakage, outer membrane permeability, inner membrane permeability, and cell membrane damage. Preliminary safety was evaluated via hemolytic activity and impact on L. acidophilus CICC 6074 and L. reuteri DSM 17938 growth. Two peptides, VRYL and RYL, were successfully identified from bovine casein hydrolysate. Both demonstrated potent inhibitory activity against E. coli ATCC 25922. VRYL exhibited a MIC of 3.5 mg/mL, which was lower than RYL’s MIC of 5.5 mg/mL. This indicates that VRYL possesses stronger inhibitory activity against E. coli compared to RYL. The antimicrobial mechanism of VRYL was analyzed and revealed that it exerts its antimicrobial activity by compromising the integrity of both the outer and inner bacterial membranes. Furthermore, VRYL exhibited favorable preliminary safety with no observable significant hemolytic activity. The peptide VRYL possesses high antimicrobial activity and favorable preliminary safety characteristics. These findings suggest that VRYL holds substantial promise for utilization in the food industry and as a natural antimicrobial agent.
