Search

Adult aging is characterized by a progressive deterioration of biological functions at physiological, cellular and molecular levels, but its damaging effects on the transcriptome are not well characterized. Here, by analyzing splicing patterns in ∼1,000 human subjects sampled across multiple tissues, we found that splicing fidelity declines with age. Most prominently, genuine introns fail to be spliced out, manifesting as a broad surge in intron retention, and this is exacerbated by the increase in diverse spurious exon-exon junctions with age. Both of these effects are prominently detected in the majority of human tissues. Collectively, they result in the progressive deterioration of the active transcriptome, wherein functional mRNAs are increasingly diluted with non-functional splicing isoforms. We discuss the concept of “splicing damage” and formulate methods to quantify it. Using these tools, we show that splicing damage increases both with age and with the incidence of diseases. Altogether, this work uncovers transcriptome damage as a critical molecular indicator of human aging and healthspan.

Cardiomyopathy (CMP) is a heritable genetic disorder. Protein-coding variants account for 20-30% of cases. The contribution of variants in non-coding DNA elements that regulate gene expression has not been explored. We performed whole-genome sequencing (WGS) of 228 unrelated CMP families. Besides pathogenic protein-coding variants in known CMP genes, 5% cases harbored rare loss-of-function variants in novel cardiac genes, with NRAP and FHOD3 being strong candidates. WGS also revealed a high burden of high-risk variants in promoters and enhancers of CMP genes in an additional 20% cases (Odds ratio 2.14, 95% CI 1.60-2.86, p=5.26×10−7 vs 1326 controls) with genes involved in α-dystroglycan glycosylation (FKTN, DTNA) and desmosomal signaling (DSC2, DSG2) specifically enriched for regulatory variants (False discovery rate

Gene expression differences between species are driven by bothcisandtranseffects. Whereasciseffects are caused by genetic variants in close proximity to the target gene,transeffects are due to distal genetic variants that affect diffusible elements such as transcription factors. Previous studies have mostly assessed the impact ofcisandtranseffects at the gene level. However, howcisandtranseffects differentially impact regulatory elements such as enhancers and promoters remains poorly understood. Here, we used massively parallel reporter assays to directly measurecisandtranseffects between human and mouse embryonic stem cells at thousands of individual regulatory elements. Our approach revealed thatciseffects are widespread across regulatory elements, and the strongestciseffects are associated with the disruption of motifs recognized by strong transcriptional activators. Conversely, we found thattranseffects are rare but stronger in enhancers than promoters, and can be attributed to a subset of transcription factors that are differentially expressed between human and mouse. While previous studies have found extensive co-occurrence ofcisandtranseffects in opposite directions that stabilize gene expression throughout evolution, we find thatcis-transcompensation is uncommon within individual regulatory elements. Thus, our results are consistent with a model wherein compensatorycis-transeffects at the gene level are explained bycisandtranseffects that separately impact several regulatory elements rather thancis-transeffects that occur simultaneously within a single regulatory element. Together, these results indicate that studying the evolution of individual regulatory elements is pivotal to understand the tempo and mode of gene expression evolution.