Abstract / Summary
Background: This exploratory study aimed to characterize amino acid related features in osteoarthritis (OA). Methods: We analyzed independent public synovial transcriptomic datasets and paired fecal metabolomics and 16S ribosomal RNA profiles from 20 patients with knee OA and 14 controls. Metabolite-sensing genes were intersected with nominally differential genes and prioritized using elastic net, support vector machine recursive feature elimination, and extreme gradient boosting. Results:SLC1A5 (solute carrier family 1 member 5) expression was lower in the discovery cohort GSE89408 (log2 fold change = −0.532, p = 0.0159), but did not meet the false discovery rate (FDR) threshold of 0.05 (FDR = 0.0697). GSE55235 showed a concordant decrease (log2 fold change = −0.695, p = 0.00536; area under the receiver operating characteristic curve = 0.830). Single-cell computational predictions from three OA donors suggested endothelial, smooth muscle, and mast-cell contexts. Tissue-level SLC1A5 expression correlated with inferred M2 macrophage proportions after group adjustment (r = −0.575, FDR = 0.002). Six features with Level 3 putative glutamine-, asparagine-, or glutamate-containing tripeptide annotations differed between groups (q = 0.00245 to 0.00948), whereas the corresponding free amino acids did not reach FDR significance. Bacteroides showed a lower centered log-ratio value in OA (q = 0.00273); none of its six tripeptide correlations remained significant after group adjustment. Conclusions: These observations generate a hypothesis about intestinal amino acid-related features and synovial SLC1A5 expression; they do not establish a direct gut-joint relationship, peptide identity, or a metabolic mechanism.