The SWI/SNF chromatin remodeling complexes regulate gene transcription by modulating chromatin accessibility. Three SWI/SNF subcomplexes, BAF, PBAF and ncBAF, share common core subunits and are distinguished by unique subunits including the ARID proteins: ARID1A /1B (BAF) and ARID2 (PBAF). Despite SWI/SNF complexes’ prominent role as tumor suppressors in solid tumors, previous studies have shown that they are mostly required for AML leukemogenesis. This research has led to clinical study of small molecule inhibition of the common ATPase subunits for AML. However, dose limiting toxicities may impede development of SWI/SNF-targeted therapies and raise the question whether targeting of specific subcomplexes would be preferred. The functions of specific subcomplexes in normal and malignant hematopoiesis are still being understood. We previously showed loss of ARID2 versus ARID1B has opposing roles in AML leukemogenesis: ARID2-loss impairs while ARID1B loss promotes leukemogenesis, highlighting the importance of understanding the role of different BAF subcomplexes. In this study, we characterize the role of ARID1A in AML leukemogenesis using patient data, mouse models and subcomplex-specific inhibitors.

We used Tempus Lens to identify patients (Pts) with AML who had both Tempus xT (DNA-seq) and xR (RNA-seq) (n=592). Pts with acute promyelocytic leukemia (APL) were excluded. RNA-seq data were normalized to correct for assay/batch effects, quantified as transcripts per million (TPM) and reported as log2(TPM+1). Pts were classified as ARID1A low (n=150) and ARID1A high (n=148) based on top and bottom quartiles of ARID1A expression. We used Cox proportional hazards models to assess real-world overall survival (rwOS), defined as the time from diagnosis to death; data were censored at loss to follow up or the 3-year study cutoff. A comparison of mutational profiles between ARID1A low and ARID1A high AML samples revealed significant enrichment of recurrent gene mutations in FLT3, NPM1, and NRAS (all p<0.001), KMT2A (p=0.014) and KIT (p=0.019) in the ARID1A-high group. In univariate analysis, low ARID1A expression was significantly associated with improved rwOS (HR = 0.56; 95% CI: 0.39 – 0.81; p=0.002). In a multivariate analysis that adjusted for other BAF complex genes (BICRA, BICRAL, ARID1B, ARID2), only low ARID1A expression was significantly associated with improved survival (HR = 0.32; 95% CI: 0.16–0.64; p =
0.001).

We investigated the role of ARID1A in AML leukemogenesis using a retroviral model of MLL-AF9 AML. We first generated transformed leukemic cells from wildtype and Arid1a f/f mice cells and characterized the cellular effect of ARID1A-loss in leukemia cells upon CRE-induced deletion. We found that Arid1a deletion reduced cell proliferation, as shown in BrdU incorporation assay by approximately 50% reduction in cells in S-phase with a concomitant increase in percentage of G1 cells. Likewise, colony forming assays demonstrate drastic reduction of blast colonies in Arid1a KO cells suggesting greatly impaired clonogenic ability. ARID1A-loss also led to a sharp increase in apoptosis by Annexin V. Moreover, we observed increased maturation markers by 3-5-fold increased CD11b and F4/80 surface expression in Arid1a KO cells compared to controls. Together, these data reveal that deletion of Arid1a decreased cell proliferation, increased cell differentiation and death, and impaired clonogenic ability, suggesting that ARID1A is essential for AML cell survival. The severity of ARID1A-loss phenotypes in MLL-AF9 leukemia is similar to those observed with SMARCA4 ATPase knockout and greater than seen for other subcomplexspecific genetic models. Consistent with data from our genetic models, we compared subcomplex-specific versus ATPase inhibitors in human leukemia cells and found that the ARID1A inhibitor showed the most similar effect to that of the ATPase inhibitor, compared to PBAF- and ncBAF-specific inhibitors. Together, these results suggest that targeting ARID1A may achieve similar efficacy as ATPase inhibition with perhaps less toxicity due to a more narrow target range. Studies are ongoing to study the effect of Arid1a-loss on leukemogenesis in vivo as well as to understand the molecular mechanism of ARID1Adependence in MLL-AF9 leukemia by transcriptomics and epigenomics.

VIEW THE PUBLICATION