ASXL1 Mutations Promote Myeloid Transformation Through Inhibition of PRC2-Mediated Gene Repression

Abstract 405

Somatic mutations in ASXL1 have been identified in patients with myeloid malignancies and are associated with worsened overall survival in AML and MDS patients. However the mechanisms of myeloid transformation of ASXL1 mutations had not been delineated. We therefore performed extensive in vitro and in vivo studies to assess the functional implications of ASXL1 mutations in the hematopoietic compartment. Transcriptional and Western blot analysis demonstrated loss of ASXL1 protein in primary leukemia samples with endogenous ASXL1 mutations indicating that these mutations are loss-of-function disease alleles. Further, ASXL1 depletion by shRNA in normal and malignant hematopoietic cells leads to robust upregulation of a set of genes including the posterior HOXA cluster (HoxA5-HoxA13). Increased HoxA gene expression was confirmed in human hematopoietic stem progenitor cells targeted with ASXL1 siRNA and in mice with conditional deletion of Asxl1 in the hematopoietic compartment.

Previous studies in Drosophila had revealed that Asxl forms the polycomb-repressive deubiquitinase (PR-DUB) complex with BAP1, which normally opposes the function of polycomb repressive complex 1 (PRC1) by removing H2AK119 ubiquitination. We verified that wild-type, but not mutant ASXL1 associates with BAP1 in co-immunoprecipitation studies. However, BAP1 depletion in hematopoietic cells did not result in significant changes in HoxA gene expression, suggesting that ASXL1 regulates gene expression in hematopoietic cells independent of its role in the PR-DUB complex. We therefore performed CHIP sequencing for known activating and repressive chromatin marks and histone mass spectrometry to elucidate the genome-wide effects of ASXL1 loss on chromatin state in hematopoietic cells. This allowed us to show that ASXL1 loss resulted in genome-wide loss of the transcriptionally repressive mark H3K27me3 in hematopoietic cells and primary patient samples with ASXL1 mutations. These data were supported by western blot analysis and histone mass spectrometry demonstrating a significant loss of H3K27 trimethylation in ASXL1-mutant cells. Moreover, ASXL1 mutations in primary leukemia samples are characterized by loss of H3K27 trimethylation at the HoxA locus. These data led us to hypothesize that ASXL1 interacts with the PRC2 complex; co-immunoprecipitation studies revealed that ASXL1 associates with members of the PRC2 complex including EZH2 and SUZ12 but not with the PRC1 repressive complex. Importantly, ASXL1 downregulation resulted in loss of EZH2 recruitment to the HOXA locus indicating a role of ASXL1 in recruiting the PRC2 complex to known leukemogenic loci.

We next assessed the effects of ASXL1 loss in vivo by generating a conditional knock-out model of ASXL1 and also by employing shRNA to deplete ASXL1 in hematopoietic cells expressing the NRASG12D oncogene. Consonant with the in vitro data, we observed HOXA9 overexpression with ASXL1 loss/depletion in vivo. Preliminary analysis reveals that conditional, hematopoietic specific ASXL1-knockout (ASXL1^fl/fl Vav-Cre) mice are characterized by progressive expansion of LSK and myeloid progenitor cells in mice less than 6 months of age. After 6 months of age a significant proportion of ASXL1^fl/fl Vav-Cre mice developed leukocytosis, anemia, thrombocytopenia, and splenomegaly; pathologic analysis of tissues revealed a phenotype consistent with myelodysplasia with myeloproliferative features. Moreover, loss of ASXL1 in cooperation with expression of NRasG12D resulted in impaired survival, increased myeloproliferation, and progressive anemia consistent with MPN/MDS in vivo.

Taken together, these results reveal that ASXL1 mutations result in a loss-of-function and suggest a specific role for ASXL1 in epigenetic regulation of gene expression by facilitating PRC2-mediated transcriptional repression of known leukemic oncogenes. Moreover, our in vivo data validate the importance of ASXL1 mutations in the pathogenesis of myeloid malignancies and provide insight into how mutations that inhibit PRC2 function contribute to myeloid transformation through epigenetic dysregulation of specific target genes.