Machine learning reveals targets of Gnaphalium hypoleucum DC. flavonoids against rheumatoid arthritis through gut microbiota and anti-inflammation.

A glmBoost + LDA machine learning model demonstrated robust diagnostic performance and identified five key RA-associated genes.

• Differential expression analysis identified 2,676 RA-associated genes.
• The glmBoost + LDA model achieved AUC_train = 0.959 and AUC_val ≥ 0.837.
• The five prioritized key genes were POLB, EGFR, MMP13, VEGFA, and KMT2D.

Molecular docking and dynamics simulations confirmed stable binding of amentoflavone (AF), a primary constituent of GHTFs, to core targets MMP9, MMP13, TOP2A, and ALOX5.

• AF was identified as a primary constituent of GHTFs.
• Stable binding was confirmed through both molecular docking and molecular dynamics simulations.
• The four core binding targets identified were MMP9, MMP13, TOP2A, and ALOX5.

Both GHTFs and AF inhibited proliferation, migration, and nitric oxide release in LPS-stimulated RAW264.7 macrophages in vitro.

• LPS-stimulated RAW264.7 macrophages were used as the in vitro model.
• Both GHTFs and AF suppressed IL-17, TNF-α, and NF-κB signaling pathways.
• AF showed more potent effects than GHTFs, though the difference was not statistically significant (P >0.05).

GHTFs treatment reduced clinical arthritis scores by over 40% in a collagen-induced arthritis mouse model.

• GHTFs treatment reduced clinical arthritis scores by over 40%.
• GHTFs alleviated synovial hyperplasia and preserved collagen volume fraction.
• GHTFs lowered serum levels of TNF-α and IL-1β.
• GHTFs demonstrated superior overall efficacy compared to both AF and methotrexate (P >0.05).

GHTFs treatment modulated gut microbiota composition by enriching beneficial Lactobacillus species and reducing pro-inflammatory taxa.

• GHTFs enriched beneficial Lactobacillus species including L. johnsonii and L. intestinalis.
• GHTFs reduced the abundance of pro-inflammatory taxa.
• GHTFs restored microbial metabolic functions.