Abstract 6043: Chromatin remodeling in patient derived colorectal cancer models

Abstract Patient-derived models of cancer (PDMCs) are becoming the gold standard preclinical models for therapeutic development and precision oncology. However, how tumor cells evolve in PDMC and the implication on therapeutic response remains largely unclear. Herein, we leveraged ATAC-seq to compare the chromatin accessibility landscapes of matched sets of colorectal cancer (CRC) patient-derived organoids (PDOs), patient-derived xenografts (PDXs), PDO-derived PDX (PDOXs), and original patient tumors (PTs), supplemented by mRNA-seq for the transcriptomics information. In the study, nine surgically resected CRC tumor specimens were recruited at Duke. For a PT-PDMCs set, part of the original patient tumor was prepared for sequencing, and the remainders from the same specimens were subjected to derive PDXs and PDOs. Throughout the study, eight PDOs were generated, and three subcutaneous PDXs grew. Besides, we also developed six patient-derived xenografts through organoids. In total, we successfully established 6 matched PT-PDO-xenograft sets, including 3 sets that have both conventional PDXs and PDOXs. The global chromatin accessibility derived from PDMCs and PTs ATAC-seq libraries presented broad congruence (79.6% peaks remained unchanged), suggesting that PDMCs mostly retained their epigenetic identity of patient tumors. Nevertheless, paired differential analysis identified alterations in chromatin accessibility between PTs and PDMCs. We discovered two principal remodeling axes. The first one separates PT from all PDMCs, and the second separate in vitro and in vivo models. Interestingly, PDOXs are much more similar to PDXs than to PDOs, indicating that the PDMCs environment is the driving force for chromatin remodeling. Single cell multiome sequencing (10X Genomics) also agreed that the PDOXs are more similar to PDXs compared to PDOs. Based on our findings, PDOXs may be a good substitution for conventional PDXs. Between in vitro and in vivo models, bivariate genomic footprinting analysis was used to further identify transcription factors (TF) that drive chromatin remodeling differentially. Our data suggested KLF14 and EGR2 have enriched motif-binding footprints in PDOXs from all patient cases. Their silencing enhanced CRC growth, which partially explains why PDXs tend to grow slower than PDOs. Downstream genes of EGR2 and KLF14 have also been analyzed. One of the targets, EPHA4, has been reported it would influence the drug sensitivities. Collectively, patient CRC cells undergo both common and distinct chromatin remodeling in PDOs and PDXs/PDOXs, driven by their respective microenvironments. The datasets allow researchers to look into chromatin accessibility, gene expression, transcriptional factor binding, and pathways that are altered in PDMC compared to original tumors, which will shed light on their predictability for therapeutic profiling. Citation Format: Kun Xiang, Ergang Wang, Qiang Huang, David Hsu, Xiling Shen. Chromatin remodeling in patient derived colorectal cancer models [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 6043.