Non-Glycolytic Role of the Myeloid Hexokinase 3 (HK3) in Therapy Responses of Myeloid Malignancies

Cancer cells prefer utilizing glycolysis to provide essential components for generating cellular macromolecules. A rate-limiting step in glycolysis is the ATP-dependent phosphorylation of glucose by hexokinases (HK) to yield glucose-6-phosphate (G6P). Four HK isoforms (HK1-4) have been identified in mammalian cells. HK2 expression is frequently increased in cancer cells including leukemias. Recent reports have highlighted a role for non-metabolic functions of glycolytic enzymes in cellular function, including HK2 in supporting acute myeloid leukemia (AML) stemness and therapy resistance. We earlier identified HK3 as a transcriptional target of PU.1 (SPI-1), which is upregulated during differentiation therapy of acute promyelocytic leukemia (APL) cell lines. In contrast to ubiquitously expressed HK1 and HK2 in mammalian cells, we find HK3 gene expression significantly associated with cells of myeloid origin. In primary human CD34+ hematopoietic stem and progenitor cells (HSPCs), we found HK1-3 mRNA levels to be low and similar. While HK1 and HK2 mRNA levels remained stable during CD34+ HSPC differentiation to monocytes and granulocytes, HK3 mRNA levels significantly increased. Based on the published oncogenic role of HK2 and our findings of increased HK3 in APL cells upon differentiation therapy, we further evaluated whether unique roles exist for HK2 and HK3 in myeloid leukemia cells. Gene disruption of HK2 or HK3 function in HL60 and NB4 AML cells resulted in the following phenotypes: a) Loss of HK2, but not HK3, function in AML cell glycolysis as determined by Seahorse analysis, and b) loss of HK3, but not HK2, increased apoptosis and DNA damage. Due to the absence of HK3-specific antibodies, we used CRISPR technology to endogenously HiBiT tag HK2 and HK3 for pull-down and mass spectrometry analysis. Findings indicated that the pro-apoptotic protein BIM directly interacts with HK3 but not HK2. This HK3-BIM interaction was validated by co-immunoprecipitation and proximity ligation assays. Binding of HK3 to BIM resulted in reduced apoptosis upon treatment with all-trans retinoic acid (ATRA). Additionally, the BH3-only proteins such as PUMA and NOXA are significantly enriched in HL60 HK3 loss-of-function cells upon ATRA treatment, based on our RNAseq findings. At the protein level, NOXA expression showed the most prominent induction in HK3-depleted HL60 cells. We found that ectopically expressing Janus kinase 2 (JAK2) V617F or JAK2 Δexon14, reported to be associated with human leukemias, in murine myeloid progenitor cells promoted a substantive increase in Hk3 mRNA compared to control transduced cells. To investigate whether Hk3 might be linked to myeloid cell protection in myeloproliferative neoplasms (MPN) upon JAK2 inhibitor treatment, we determined Hk3 expression under these conditions. Indeed, treating Ba/F3 JAK2 V617F cells with Ruxolitinib resulted in a marked, dose-dependent decrease of Hk3 mRNA expression, whereas Hk1/Hk2 levels did not significantly change. Inhibition of JAK2 can activate BIM expression via the STAT pathway. We hypothesize that increased HK3 levels, promoting increased BIM binding in MPN with JAK2 mutations, may negatively impact JAK2 inhibitor sensitivity. Ongoing studies will analyze STAT-BIM pathway and cell death in HK3 depleted human MPN cells. Our findings provide evidence that HK3 is dispensable for glycolytic activity in myeloid cells, yet promote cell survival under ATRA or JAK2 inhibitor treatment in AML or MPN cells, respectively, possibly through direct interactions with pro-apoptotic BH3-only proteins.