Older and Younger Splicing Across 49 Human Tissues

Background. Studies of splicing and age usually count how many events change, and those counts depend heavily on sample size. We asked instead which way each age-linked junction moves — towards use by older or by younger donors — and compared 49 GTEx v8 tissues on that. Version 2.2 adds junctions outside the GENCODE v26 annotation from recount3, classifies junctions against the APPRIS principal junction of their gene, contrasts exonic with intronic per-base coverage, and examines how the circumstances of death shape the age signal. Version 2.4 tests the main findings in independent cohorts.Methods. Annotated junctions were profiled in six decade cohorts (20–79) and classed as older or younger splicing by score tests against null models with sex (FDR < 0.05 within tissue). Junctions were also tested against their gene's principal junction, within each tissue and across tissues with donor-clustered variance, with age-shuffled nulls. recount3 junction counts extended these tests to unannotated junctions in all 49 tissues, with reading-frame and nonsense-mediated decay (NMD) flags. recount3 coverage separated exonic (mature) from intronic (pre-mRNA) signal per gene. The principal-relative, recount3, coverage and brain–body analyses were repeated in donors with a brain sample, who include almost no ventilator cases, and the principal-relative tests also in non-ventilator donors, with Hardy-scale adjustment and in repeated random and age-matched draws; analyses carried forward from version 2.1 use all donors. Version 2.4 adds four independent recount3 cohorts (post-mortem brain; blood, liver and fibroblasts from living donors), splice-motif and uniquely mapped read checks, an aggregate junction burden over every junction read, GC content and qSVA-type degradation covariates.Results. In GTEx, 85% of donors in their 20s but 30% in their 60s died on a ventilator (Hardy scale 0); only 20 of 382 brain donors did. Restricting body tissues to brain donors reduced significant principal-relative associations from 120,523 to 1,612, far more than random or age-matched donor sets of the same size (20,749–27,944 and 8,041–9,895 over three draws). On fixed junction–reference pairs, random and age-matched subsets kept 93–98% of the all-donor age slopes, whereas Hardy-adjusted, ventilator-only, non-ventilator and brain-donor analyses kept 45–55%, mostly with the same sign, and lost the body's majority of increasing alternatives. In brain, testable alternatives decreased relative to their principal reference about twice as often as they increased (31% increasing), and unannotated alternatives more so (12%). In body tissues, alternative splice sites that increased relative to the principal were more often frame-shifting within the coding sequence (all donors 77% vs 47%; brain donors 77% vs 62%, OR 2.0, 95% CI 1.1–3.7, nominal p = 0.018). In brain, frame-preserving exon skips increased and frame-disrupting skips decreased. Gene span, through intron length, was negatively associated with the age slope of an exonic-to-intronic coverage contrast in 12 of 13 brain regions; this persisted without ventilator cases, in 10 of 13 regions with quality and composition adjusted together, and across pseudocount, depth and fixed-gene-set variants, while in body tissues it depended on the donor set. On the same junctions in the same donors, the brain's spread of age effects was 2.0–2.8 times the body's depending on inclusion criteria. Splicing clocks showed weak out-of-fold age correlations (brain median r = 0.21, body 0.04) and no better mean absolute error than a training-fold median. A prospectively specified test linking SLC13 citrate transport to the long-gene association was not supported. In independent cohorts, GTEx brain pairs kept their direction in post-mortem prefrontal cortex (88% of 327; z correlation 0.77), and the long-gene coverage association replicated there (ρ −0.44 against −0.40 in GTEx), whereas in blood, liver and fibroblasts from living donors it was absent or reversed (−0.07 to +0.15). Body-tissue pairs agreed at or below chance in blood, liver and fibroblasts from living donors (33–51%). Counting every junction read, the unannotated share of splicing rose with age in all 13 brain regions (median +2.0% per decade). The brain long-gene association was unaffected by GC content but weakened with degradation components (9 of 13 regions with five, 5 of 13 with ten). SR-gene decay splicing, H2AFY and timing results are carried forward from version 2.1 for all donors only.Conclusions. Two findings lead. First, age-associated splicing differs between brain and body, including within the same donors: in brain, many testable alternatives decrease relative to selected principal references, brain effects on the same junctions are two to three times as variable as body effects and only weakly correlated with them, frame-preserving rather than frame-disrupting forms gain with age, and gene span is negatively associated with an exonic-to-intronic coverage contrast whatever the donor set; the junction directions and the long-gene association replicate in an independent post-mortem brain cohort. Second, age-associated junction use in GTEx body tissues is strongly sensitive to donor selection and procurement-related covariates; about half of the pooled body-tissue age association runs between Hardy categories, and the body-tissue associations do not replicate in samples from living donors. These findings identify reproducible patterns in the analysed outputs but do not isolate ventilation, establish splicing failure or show whether the brain ages faster or slower than the body. SLC13 expression and acetylation-machinery scores do not provide specific support for the proposed metabolic mechanisms.