急性淋巴细胞白血病中与化疗反应相关的基因调控网络的整合多组学研究

Acute lymphoblastic leukemia (ALL) is the most common pediatric cancer. While treatment outcomes have improved, drug resistance remains a major clinical concern and the primary cause of relapse in patients, with relapsed patient survival rates at approximately 40%. Although changes in gene expression are known to impact drug resistance in ALL, the gene regulatory elements that control and modulate these genes are frequently undefined, and the role of these elements in anti-leukemic drug resistance remains largely unknown. As a result, a more comprehensive epigenetic effort is needed to profile and validate gene regulatory networks linked to anti-leukemic drug response in ALL. We therefore generated matched B-cell origin ALL patient chromatin accessibility and ex vivo drug sensitivity (pharmacotyping) datasets covering 16 anti-leukemia drugs in primary ALL cells from over 150 patients to generate a compendium of gene regulatory elements associated with ex vivo chemotherapy response. Importantly, ex vivo measurements of chemotherapy response are known to associate with clinical outcomes in ALL patients. We further integrated these data with transcriptomics, tumor-normal whole genome sequencing and three-dimensional chromatin conformation results obtained from a subset of these patient biospecimens. We identified 74,307 drug response elements (DREs) and 2,854 drug response genes (DRGs) using a combination of linear modeling and feature selection techniques (p-adj < 0.05, s-value < 0.005). For this analysis, LC50 drug sensitivity was modeled in both continuous and categorical modes against chromatin accessibility and transcriptomic sequence counts. Although many DREs were specific to a singular drug or common drug family, a subset was associated with response to multiple drugs. DREs harbored 868 somatic mutations, many of which overlapped transcription factor (TF) footprints, with PU.1 among the most common overlapping TF sites. TF footprinting identified 148,732 TF binding events associated with drug response. Common TF footprints that correlated with drug response included EBF family TFs for inotuzumab ozogamicin response and STAT family TFs for trametinib response. On average, 74% of these drug response TF footprints mapped to DREs. Subsequent multi-omic integration of gene expression, chromatin accessibility and TF footprinting with three-dimensional chromatin conformation maps identified over 9000 gene regulatory networks linked to altered drug response. Collectively, this work represents the largest study of chromatin accessibility and TF occupancy impacting ALL chemotherapy drug response and supports an important functional role for many DREs in anti-leukemic drug resistance.