Transcriptional Repression of RASSF2 Impairs Hippo Kinase-Dependent p53 Activation and Tumor Suppression in t(8;21) AML

The t(8;21) chromosomal translocation is a recurring cytogenetic abnormality in acute myeloid leukemia (AML), found in ~10% of AML patients. This translocation results in the stable fusion of the RUNX1 (AML1) and RUNX1T1 genes, and formation of the oncofusion protein RUNX1-ETO (RE). One function of RE in promoting leukemia development is the recruitment of aberrant transcription factor complexes to regulatory regions of RUNX1 target genes known to be critical for myeloid differentiation, such as CEBPA, SPI1, NFE2, and CSF1R . Despite this knowledge, additional RE target genes remain poorly characterized, and the complete molecular mechanism through which RE leads to leukemic transformation remains to be elucidated.

Previously, we used a mouse model of t(8;21) AML to compare gene expression in RE-expressing leukemia cells to the normal mouse hematopoietic progenitor (lin-cKit+) compartment (Lo et al, 2012). Among the most significantly downregulated genes was Ras-association domain family member 2 (Rassf2). RASSF genes are frequently epigenetically silenced in human tumors, making them highly relevant to study in the context of cancer development. Using public gene expression data, RT-qPCR, and western blotting, we found that RASSF2 expression is specifically downregulated in t(8;21) AML cells compared to non-t(8;21) AML cells and normal CD34+ controls. Furthermore, RE knockdown in a t(8;21) AML cell line led to marked upregulation of RASSF2 mRNA. Based on ChiP-seq data analysis this repression appears to be mediated by direct RE binding and HDAC recruitment to the promoter region of both alternative RASSF2 transcriptional start sites in the genomic locus.

To investigate the significance of this repression we used retroviral vectors to re-express RASSF2 in t(8;21) AML cell lines and RE-transduced primary murine hematopoietic cells in vivo and in vitro . Re-expression of RASSF2 significantly improved survival and reduced the total leukemic burden in a primary mouse model of RE-driven AML in vivo . RASSF2 inhibited leukemic transformation and the self-renewal ability of RE-transduced mouse bone marrow cells in a serial replating/colony formation assay, without a noticeable effect on normal hematopoietic progenitors. This inhibition correlated with increased apoptotic cells in the context of RE and RASSF2 co-expression.

Given that RASSF2 is a non-enzymatic scaffolding protein, we hypothesized that it mediates tumor suppressive functions via its Ras-association or SARAH effector domains. We saw no detectable effect of RASSF2 expression on oncogenic Ras signaling in t(8;21) AML cells. This led us to investigate the SARAH domain-mediated interaction of RASSF2 with the pro-apoptotic Hippo kinases, MST1 and MST2. RASSF2 expression in t(8;21) AML cells led to an increased protein amount of both MST1 and MST2 in a SARAH domain-dependent manner. Importantly, use of either a RASSF2 SARAH domain deletion mutant, or Mst1(-/-)Mst2(-/-) mouse bone marrow cells, completely abrogated the ability of RASSF2 to inhibit RE leukemia cell self-renewal in the serial replating/colony formation assay. We found this inhibition to be independent of canonical Hippo pathway signaling, and therefore utilized a proximity-dependent biotin identification approach (BioID, Roux et al, 2012) to map the protein interactome of RASSF2-MST1/2 in a t(8;21) AML cell line. Interestingly, this revealed an interaction between RASSF2 and the protein deacetylase SIRT1. This highlighted a role for RASSF2 in promoting p53 acetylation in response to cellular stress, which is dependent on MST1/2-mediated negative regulation of SIRT1. Overall, our study revealed a critical tumor suppressive role of RASSF2-MST1/2 signaling via the SIRT1-p53 axis in t(8;21) AML.