The team has led three major studies linking noncoding variants to type 2 diabetes (T2D). The first, Miguel-Escalada et al. (Nature Genetics, 2019), created a 3D chromatin interaction map in human pancreatic islets, enabling linkage of noncoding risk variants to target genes by physical proximity. Over 1,300 enhancer hubs enriched for T2D variants were identified, explaining a substantial portion of heritability. CRISPR validation confirmed these enhancers regulate diabetes-related genes, including previously unlinked ones. Incorporating enhancer-based polygenic risk scores improved T2D prediction, especially in lean, early-onset individuals. This work provided a framework for interpreting noncoding variants in islet gene regulation. In Atla et al. (Genome Biology, 2022), we examined RNA splicing as a regulatory layer, analyzing splicing QTLs (sQTLs) in islets from ~400 donors. Many sQTLs overlapped with T2D risk loci, indicating altered splicing as a contributor to disease mechanisms. A third study, co-led with the Hansen team and published in Nature Metabolism (2024), analyzed β-cell function genetics using GWAS data from ~26,000 individuals. We identified 55 associations at 44 loci influencing insulin secretion dynamics. By integrating genetic data with islet transcriptomic and epigenomic maps, we uncovered 92 candidate effector genes regulating insulin secretion. These findings link genetic control of β-cell function to distinct T2D risk pathways.
In studies to understand why certain cis regulatory elements are vulnerable to diabetes-causing mutations, we discovered that rare enhancer mutations upstream of PTF1A causing pancreatic agenesis disrupt a lead enhancer that activates an entire enhancer cluster (Developmental Cell., 2022). This work sets a clear example of a single enhancer mutation causing monogenic disease, providing an underlying mechanism.
In a separate study we identified a noncanonical regulatory element essential for maintaining proper HNF1A levels, critical for β-cell function. Rather than acting as an enhancer or silencer, this element—named HASTER—functions as a transcriptional stabilizer that buffers HNF1A via a feedback loop. Published in Nature Cell Biology (2022) and reviewed in Nature Reviews Molecular Cell Biology (2024), this work establishes transcriptional stability as a distinct regulatory mechanism separate from traditional activation or repression.
During this period we have also dissected beta cell networks linked to human diabetes. We discovered that the beta cell transcriptional regulator HNF1A, which carries causal variants for monogenic diabetes as well as forpolygenic diabetes susceptibility, controls a broad RNA splicing program in beta cells, and incriminates this transcription-splicing regulatory axis in human diabetes.
Finally, this grant enabled the DecodeDiabetes Study, which is perhaps the largest genomic dataset of patients with clinical diagnosis of monogenic diabetes but no known gene mutations. This work is ongoing and has led to an overall strategy for noncoding variant interpretation, and uncovered insights into the genetic underpinnings of young-onset diabetes,as well as into genetic regulators of diabetes-relevant networks
