Machine learning speeds discovery of AMPs to treat ulcerative colitis

Machine learning-based computational approaches accelerate treatment discovery

Ulcerative colitis (UC) is a chronic inflammatory bowel disease characterized by recurrent intestinal inflammation, abdominal pain, and diarrhea. Although current treatments such as 5-aminosalicylic acid, antibiotics, and biologics can control or improve symptoms, many patients experience incomplete responses and side effects. The search for safer and more effective treatments remains a major challenge.

Antimicrobial peptides (AMPs), natural components of innate immunity, have attracted attention as potential therapeutic agents due to their antibacterial and immunomodulatory properties. However, the discovery of novel AMPs has traditionally required labor-intensive screening and experimental validation. In a new study published in electronic gastroenterologyMiao et al. applied machine learning to accelerate discovery of AMPs and identify potential treatment candidates for UC.

Machine learning screens thousands of peptide sequences

The researchers built a machine learning pipeline that combines peptide prediction models and genetic algorithms to screen candidate antimicrobial peptides. By analyzing the structural and physicochemical properties of peptide sequences, the model evaluated over 6,000 potential candidates and ultimately identified 22 promising sequences.

Five peptides were synthesized and experimentally tested. Among them, a peptide called LR, named after its N- and C-terminal residues, showed the most favorable balance between antibacterial activity and low cytotoxicity. In vitro experiments demonstrated that LR exhibits strong bactericidal activity against pathogenic bacteria such as: Escherichia coli and Staphylococcus aureus. Importantly, LR maintained good biocompatibility and exhibited minimal toxicity and low hemolytic activity compared to other candidates.

Lead peptide reduces colitis in animal models

To assess its therapeutic potential, researchers tested LR in a mouse model of dextran sodium sulfate (DSS)-induced colitis. Treatment with this peptide significantly improved disease severity. Key clinical indicators (for exampleweight loss, disease activity index (DAI) and colonic shortening) were significantly improved in mice receiving LR. Histological analysis revealed decreased mucosal damage and decreased infiltration of inflammatory cells in colon tissue. Remarkably, in this model, LR treatment showed a stronger therapeutic effect than both the standard anti-inflammatory drug 5-aminosalicylic acid and the antibiotic ciprofloxacin.

Anti-inflammatory effect and barrier repair

Further mechanistic analysis showed that LR suppressed the inflammatory response. The levels of pro-inflammatory cytokines such as tumor necrosis factor-α (TNF-α) and interleukin-6 (IL-6) were significantly decreased after treatment. At the same time, this peptide helped restore the integrity of the intestinal barrier. Expression of tight junction proteins (In other wordsZO-1, claudin-1, and occlusion) were significantly increased, indicating improved epithelial barrier function. These findings suggest that LR may exert therapeutic effects both by suppressing inflammation and strengthening the intestinal mucosal barrier.

Microbiota regulation emerges as an important mechanism

The study also investigated how peptides affect the gut microbial community. Fecal microbiota sequencing revealed that LR treatment reshaped the microbial composition of colitis mice. Remarkably, the abundance of Akkermansia muciniphila, a beneficial bacterium, increased significantly after AMP treatment. This species is associated with improving intestinal barrier function and reducing inflammation in several intestinal diseases. Further experiments revealed that A. muciniphila Alone, it may partially alleviate the symptoms of colitis, suggesting that microbiota modulation contributes to the peptide’s therapeutic efficacy. Importantly, LR selectively inhibits pathogenic bacteria while sparing them. A. muciniphilahighlighting a favorable antimicrobial profile that preserves the microbiome.

Implications for future treatments

These findings demonstrate how machine learning can streamline the discovery of novel therapeutic peptides. By integrating computational screening and experimental validation, researchers identified stable and selective AMPs with promising anti-inflammatory activity in UC.

Although further studies are needed to assess long-term safety and impact on human disease, this study highlights a new strategy for developing microbiota-friendly treatments for inflammatory bowel disease. As artificial intelligence continues to transform drug discovery, machine learning-based peptide design may open new avenues for treating complex diseases such as ulcerative colitis.