ScienceOpen Preprints · Nov 10, 2026
Age-associated alternative splicing of four kynurenine-pathway enzymes (AFMID, IDO2, KYNU, KYAT1) across human tissues
By Hemming J
Background: The kynurenine pathway converts tryptophan to neuroactive metabolites and NAD⁺, and its activity rises with age. Alternative splicing could change the output of its enzymes without altering gene expression. We examined how splicing of four pathway enzymes, arylformamidase (AFMID), indoleamine 2,3-dioxygenase 2 (IDO2), kynureninase (KYNU) and kynurenine aminotransferase 1 (KYAT1), varies with age across human tissues. Methods: . All 86 annotated exon–exon junctions for the four genes were extracted from GTEx v8 (17,382 samples, 948 donors, 49 testable tissues) and grouped into intron clusters. The primary analysis modelled junction usage on donor age alone with quasi-binomial regression, controlling the false discovery rate (FDR) jointly across 1,158 tests. A secondary analysis added sex, RNA integrity (RIN), ischaemic time and Hardy death-circumstance scale, which are themselves correlated with age. Results: . Age was associated with splicing in 54 of 242 testable tissue × cluster events (FDR < 0.05): 44 in AFMID across 28 tissues, 5 in KYNU and 5 in KYAT1; IDO2 was too weakly expressed to test widely. The dominant pattern was a shift in AFMID from canonical junctions towards exon-skipping isoforms that encode truncated proteins. Skipping of exons 3–10 had a positive age estimate in 38 of 48 tissues, including every non-brain tissue except cultured fibroblasts, and was significant in 17; brain regions showed no increase. In whole blood, KYNU shifted towards a truncated 147-amino-acid isoform (4.7% → 15.0% of reads). In tibial artery, a frameshifting KYAT1 3′ splice site rose from 19% to 34%. After adjustment, 52 of 54 effects kept their direction, with a median 40% reduction in size; 5 remained significant. Conclusions: . Splicing of AFMID, and to a lesser extent KYNU and KYAT1, moves with age towards truncated or non-productive isoforms in many peripheral tissues. Part of this association is shared with post-mortem and RNA-quality variables that co-vary with age, and the true age effect probably lies between the crude and adjusted estimates.