Cohesin-STAG2 Mutations Alter Chromatin Structure and Gene Regulation in Human Acute Myeloid Leukemia

Cohesin is a multimeric ring-shaped protein complex involved in shaping the spatial chromatin architecture and regulating enhancer-promoter interactions. Cohesin members are frequently mutated in myeloid malignancies, but the exact epigenetic mechanisms linking them to gene regulation and leukemogenesis are poorly understood. Interestingly, mutations mainly affect the variable cohesin subunit STAG2 (X-linked gene), but not its paralogue STAG1 which implies a pivotal and dominant role of STAG2-cohesin in maintaining the chromatin landscape in normal hematopoiesis. We hypothesized that STAG2 mutations induce specific changes at cohesin-dependent enhancer-promoter elements causing altered gene regulation associated with leukemic transformation. Here, we present the first study directly analyzing transcriptional and epigenetic effects of STAG2 mutations in primary acute myeloid leukemia samples (AML). In 400 AMLs, we found STAG2 mutations in 5.94% samples, whereas STAG1 was never mutated. To study the implications of STAG2 mutations in AML samples, we performed chromatin immunoprecipitation coupled to sequencing (ChIP-seq) targeting cohesin members STAG2, STAG1 and RAD21 as well as the active enhancer/promoter mark H3K27ac. STAG2 mutations were consistently (N=15) associated with complete loss of STAG2 protein as verified by ChIP-seq and western blot. While cohesin occupancy largely remained stable across the genome, suggesting the compensation of STAG2 loss by STAG1, in comparison to AMLs without cohesin mutations we also identified loci with altered cohesin occupancy (RAD21 ChIP-seq) and chromatin activation (H3K27ac). In STAG2 mutant AMLs, decreased RAD21 (2587 peaks, fold change < -2) highly correlated with reduced H3K27ac (P_adj <7.3e-07). Vice versa, gain of RAD21 (2131 peaks, fold change >2) correlated with elevated H3K27ac (P_adj <0.0002). These changes in cohesin-associated enhancers or promoters also showed significant correlation with up- (N=130, P_adj <0.007) or down-regulation (N=133 , P_adj <0.001) of associated transcripts detected by RNA-seq, when comparing STAG2 mutant with cohesin-wildtype AMLs. To address whether these cohesin-dependent epigenetic alterations are also linked to changes in chromatin topology, we performed in situ high-throughput chromosome conformation capture (HiC). In STAG2 mutant AMLs we detected significant weakening of loop interactions (1519 loops, fold change <-1.5) corresponding to short-range contacts (<0.5MB). These loops are partially involved in the potential enhancer-promoter contacts with reduced cohesin and H3K27ac occupancy and dysregulated transcription. To corroborate the findings in AML samples, to verify the potential target genes and to address the leukemogenic potential of STAG2 loss, we additionally performed siRNA-mediated cohesin component depletion in cord-blood derived CD34⁺ primary human hematopoietic stem and progenitor cell (HSPC) cultures. We generated equivalent data sets as for AMLs and detected overlapping transcriptomic and epigenetic effects in HSPCs with STAG2 knockdown (KD) that were not invoked by depletion of STAG1. Especially RAD21 peak signals or genes downregulated in STAG2 mutant AMLs displayed a significant trend towards downregulation in STAG2 KD HSPCs (GSEA: genes: P_adj, _.<1,03e-06 peaks: P_adj.<1,24e-05). Among these genes, several were also associated with the cohesin-dependent epigenetic changes in STAG2 mutant AML and STAG2 KD, including the tumor suppressor gene DACT1, and ITGA9, a regulator of HSC proliferation and differentiation. To address differentiation capacity of HSPCs, we performed colony forming unit assays following KD. We found that STAG2, but not STAG1 depletion interfered with differentiation, specifically erythroid colony formation was affected (3-fold reduction, P_adj.<8.46e-05). Taken together, we show that locus-specific changes in cohesin occupancy, enhancer activity and chromatin looping (resulting from loss of STAG2 and switch to STAG1) correlate with transcriptomic changes involved in leukemogenesis. Our findings corroborate the status of STAG2 as safekeeper of cohesin-associated chromatin architecture and gene regulation in the hematopoietic system.

No relevant conflicts of interest to declare.