Regulatory and epigenetic landscapes reveal biological processes, disease mechanisms and diagnostic markers

National Human Genome Research Institute

Investigators

Linked publications, trials & patents

Abstract

My program advances knowledge of gene regulation, reveals disease mechanisms, and develops genomic medicine tools. The unifying theme is that noncoding regulation explains how disease processes unfold, and how they can be targeted. I. Gene Expression and Epigenetic Mechanisms Isoform Expression as Cancer Drivers Mutations drive cancer, but so do isoform-level changes. In melanoma, we profiled 538 kinase genes and 3,040 isoforms across TCGA tumors (Holland et al. 2022). Isoform ratios distinguished clinical subgroups invisible to gene-level analysis, separating BRAF from RAS mutants and uncovering phenotypes such as immune infiltration and apoptosis loss. Functionally, isoform-specific roles of SLK altered apoptosis and localization. This work established isoform expression profiling as a strategy for therapeutic target identification. Somatic Mutations and Splicing Alteration We extended this to ask how mutations shape splicing globally. Our iSoMAs framework (Tan et al. 2025, PLoS Comp Biol) integrates PCA with mutation data to map somatic mutation–splicing associations across 9,738 tumors. We identified 908 genes linked to altered isoform use, spanning oncogenes, tumor suppressors, RNA binding proteins, and splicing factors. This scalable approach reframes mutations as network-level regulators of isoform landscapes, advancing both biology and clinical interpretation. Latent Splicing Mechanisms We also uncovered a quality-control system, the Suppression of Splicing (SOS) mechanism, which prevents activation of latent splice sites that introduce premature stops. In collaboration with Dr. Ruth Sperling, we identified nucleolin (RNA Biol. 2022) and ALYREF (bioRxiv 2025) as factors in this pathway, showing initiator-tRNA–guided protection of reading frames. This expands our mechanistic foundation for splicing fidelity, with direct implications for stress and cancer biology. Expression Analysis in Neurodevelopment Extending our transcriptomic focus, we collaborated on ADHD (Sudre et al. 2023). We identified cortico-striatal dysregulation with downregulation of glutamatergic pathways, reinforcing neurobiological models of the disorder and demonstrating the versatility of our regulatory genomics approaches. II. Disease Mechanisms and Phenotypic Expression CIMP and Epigenetic Subtypes We first defined CpG Island Methylator Phenotype (CIMP) across 14 cancer types (Sanchez-Vega et al. 2015), later revealing its presence in ovarian tumors (Kolbe et al. 2012) and now addressing its heterogeneity in colon cancer. This body of work reframes CIMP as an umbrella of distinct subtypes, shifting focus from mutations to landscapes of epigenetic regulation. Ovarian Cancer Epigenetics and the Megacomplex Building on this, we showed that aggressive serous ovarian cancers lack CIMP+ but exhibit intermediate methylation. RNA-seq pointed to an estrogen-responsive enhancer program, and with ChIP-seq we defined the Megacomplex: ERα, FOXA1, PITX1, and MED1 co-occupying >2,000 super-enhancers (Jaiswal et al. 2025, Cancer Lett.). This enhancer-driven architecture sustains estrogen signaling independent of hormone response, explaining clinical resistance to anti-hormonal therapy. Combined inhibition with Fulvestrant and JQ1 dismantled the Megacomplex and triggered apoptosis. This work establishes enhancer architecture as a therapeutic target in ovarian cancer. Super-Silencers and Tumorigenesis In parallel, we demonstrated that super-silencers manifest as broad H3K27me3-dense domains, to maintain lineage-specific programs. In B-cell lymphoma, ~60% of these regions are lost and converted to super-enhancers, activating oncogenes such as BACH2 (Huang et al. 2025, Nat Commun). This work shows that losing genome repression is as oncogenic as gaining genome activation, advancing models of epigenetic instability. III. Advances in Genomic Medicine DNA Methylation Biomarkers for Cancer Detection Our discovery of ZNF154 as the first pan-cancer methylation biomarker (Sanchez-Vega et al. 2013; Miller et al. 2020, 2021) positioned methylation as a universal language of cancer detection. We have since developed multi-marker panels with high sensitivity and specificity, implemented on both capture arrays and qPCR platforms. Marker Combinations and Clinical Utility Extending this work (Funderburk et al. 2023, Cancers), we demonstrated that combining pan-cancer biomarkers improved accuracy detecting 14 tumor types. Current work expands this to 47-marker panels that classify tissue of origin alongside pan-cancer detection. These assays represent a translation of our mechanistic insights into clinical tools for early cancer detection.