Zhang, Cheng

Abstract Understanding the function of non-coding genomic sequence variants represents a challenge for biomedicine. Many diseases are products of gene-by-environment interactions with complex mechanisms. This study addresses these themes by mechanistic characterization of non-coding variants that influence gene expression only after drug or hormone exposure. Using glucocorticoid signaling as a model system, we integrated genomic, transcriptomic, and epigenomic approaches to unravel mechanisms by which variant function could be revealed by hormones or drugs. Specifically, we identified cis-regulatory elements and 3D interactions underlying ligand-dependent associations between variants and gene expression. One-quarter of the glucocorticoid-modulated variants that we identified had already been associated with clinical phenotypes. However, their affected genes were ‘unmasked’ only after glucocorticoid exposure and often with function relevant to the disease phenotypes. These diseases involved glucocorticoids as risk factors or therapeutic agents and included autoimmunity, metabolic and mood disorders, osteoporosis and cancer. For example, we identified a novel breast cancer risk gene, MAST4, with expression that was repressed by glucocorticoids in cells carrying the risk genotype, repression that correlated with MAST4 expression in breast cancer and treatment outcomes. These observations provide a mechanistic framework for understanding non-coding genetic variant-chemical environment interactions and their role in disease risk and drug response.

Deleterious variants in dihydropyrimidine dehydrogenase (DPD, DPYD gene) can be highly predictive of clinical toxicity to the widely prescribed chemotherapeutic 5‐fluorouracil (5‐FU). However, there are very limited data pertaining to the functional consequences of the >450 reported no‐synonymous DPYD variants. We developed a DPYD‐specific variant classifier (DPYD‐Varifier) using machine learning and in vitro functional data for 156 missense DPYD variants. The developed model showed 85% accuracy and outperformed other in silico prediction tools. An examination of feature importance within the model provided additional insight into functional aspects of the DPD protein relevant to 5‐FU toxicity. In the absence of clinical data for unstudied variants, prediction tools like DPYD‐Varifier have great potential to individualize medicine and improve the clinical decision‐making process.

Key Points The type III TGF-β receptor is a marker that distinguishes “early” and “late” BFU-Es. TGF-β inhibitors increase early BFU-E cell self-renewal and total erythroblast production.