STAG2 Mutations Alter Cohesin Ring Function and Provide Therapeutic Vulnerabilities in Acute Myeloid Leukemia

Recurrent somatic mutations in core components and modulators of the cohesin ring - a multimeric protein complex that forms a ring structure around DNA and provides spatial genome organization - have been identified across multiple cancer types, including acute myeloid leukemia (AML) and myelodysplastic syndrome (MDS), where they are associated with poor overall survival. Cohesin proteins are involved in sister chromatid cohesion, chromatin organization into loops, transcriptional activation, and DNA damage repair. The mechanisms underlying clonal expansion of these driver mutations are unknown and no therapies have selective efficacy in cohesin-mutant cancers.

We sought to determine the effects of mutations in the most frequently mutated cohesin subunit, STAG2, on cohesin complex composition using immunoprecipitation followed by quantitative mass spectrometry (IP-MS), genetic dependencies of STAG2-mutant cells by genome-wide CRISPR screening, and mutant cohesin association with chromatin using chromatin immunoprecipitation followed by sequencing (ChIP-Seq). Our goal was to understand how these mutations contribute to cellular transformation and to identify possible therapeutic targets.

Applying IP-MS in AML cell lines engineered with different STAG2 mutations, we identified and validated a switch from STAG2- to its paralog STAG1-containing cohesin complexes. In addition, we observed changes in the interaction of the mutant cohesin complex with proteins involved in DNA repair and replication, including PARP1, and RNA-mediated interaction with RNA splicing machinery, including SF3B family members. We next hypothesized that these cohesin-dependent alterations could lead to shifts in genetic dependencies. Using genome-scale CRISPR-Cas9 screens, we identified preferential dependency of STAG2-mutant cells on STAG1, consistent with our proteomics studies. We also found a striking concordance between additional cellular processes highlighted by IP-MS experiments and observed increased dependency of STAG2-mutant cells on DNA damage repair and mRNA processing. Therefore, STAG2 mutations lead to changes in cohesin complex structure and alter interactions with proteins involved in DNA damage, replication, and RNA modification, which become genetic dependencies in this context.

Prompted by this concordance, we evaluated DNA replication, DNA damage and splicing in cohesin-mutant cells. We observed a 4-fold increase in replication fork stalling in STAG2-mutant cells, which was associated with accumulation of double strand DNA breaks and activation of the ATR and ATM DNA damage checkpoints. STAG2-mutant cells demonstrated ~100-fold increased sensitivity to the PARP inhibitor talazoparib, which was consistent across models of other cohesin-mutant subunits. In addition, cohesin-mutant cells showed aberrant splicing and increased sensitivity to treatment with SF3B1 inhibitors E7107 and H3B-8800. In aggregate, genetic or pharmacologic perturbation of DNA damage repair or splicing created a synthetic vulnerability for cohesin-mutant cells in vitro and in vivo.

Finally, we explored how STAG1-containing complexes alter cohesin-mediated genome compartmentalization in cohesin-mutant cells. Using ChIP-Seq, we observed that STAG2 loss leads to a global decrease in cohesin binding to chromatin, including at sites of insulated neighborhood boundaries, with subsequent gene expression changes. Loss of cohesin binding was associated with increased enhancer activity and super-enhancer expansion in STAG2-mutant cells. In addition, we identified changes in the co-localization of the mutant cohesin complex with super-enhancer enriched factors, DNA damage repair and splicing machinery.

These findings are consistent with a model in which wild type and mutant cohesin complexes, defined by their unique composition and patterns of chromatin binding and architecture, have differential abilities to maintain chromatin organization as it relates to spatial organization of super-enhancers, coactivators and transcription factors, as well as DNA damage repair and splicing machinery. Perturbation of any of these components, which have been recently proposed to form phase-separated nuclear bodies, creates vulnerabilities that may be exploited therapeutically with existing drugs in patients with cohesin-mutated malignancies.