Chromatin Architecture Protein CTCF Plays a Role in Determining 5-Azacytidine Response of Leukemic Cells By Potentially Regulating Hoxb Genes

Myelodysplastic syndromes (MDS) are a group of blood disorders that arise from abnormal proliferation and differentiation of malignant hematopoietic stem and/or progenitor cells (HSPC). Patients with MDS can often progress to more advanced myeloid malignancies, such as acute myeloid leukemia (AML). The DNA methyltransferase inhibitor 5-azacytidine (AZA) is a core treatment modality for many patients with MDS/AML, either as a single agent or in combination with other novel agents (e.g., venetoclax). AZA, a cytidine analogue, directly and irreversibly inhibits DNA methyltransferases when incorporated into DNA during S-phase. Only 50-60% of MDS and 15% of AML patients are responsive to single-agent AZA treatment and the mechanisms of AZA resistance remains unclear. Previous studies have suggested that AZA treatment spares the leukemic stem cell population (LSC), highlighting the need to characterize the LSC to understand mechanisms of therapy resistance. Studies have shown that the binding of the chromatin architecture protein CTCF, to its target sites are perturbed in AML. CTCF is a ubiquitously expressed transcription factor shown to play multiple roles in regulating gene expression. CTCF is most widely known for its role in regulating and stabilizing the three-dimensional genome architecture by facilitating chromatin loops to enable interactions of gene enhancer and promoters. Interestingly, methylation changes due to AZA exposure have been shown to have an effect on CTCF occupancy in AML patient cells.

Given the role AZA plays in DNA demethylation, we performed single cell whole genome bisulfite sequencing (scWGBS) on bone marrow samples of MDS patients prior to AZA treatment to identify any differences in the methylation profile of AZA responders and non-responders. We sorted single cells from the Lin^-CD34⁺CD38^- compartment, which encompasses the LSC population, and performed scWGBS. In line with previously published literature, on examining the average methylation per single cell, we did not see any differences in the global methylation levels between the responders and non-responders. However, we saw differentially methylated regions (DMRs) at specific genomic regions such as the CpG islands, promoter and enhancer binding sites. Using published CHIP-seq datasets, we further identified hypermethylation at 63 transcription factor binding sites (TFBS) in non-responders vs. responders, including several known to play a role in determining chromatin structure of hematopoietic cells: CTCF, Cohesin complex subunits and ZNF143. A CRISPR knockout screen of the 63 transcription factors (TF) in 2 leukemic cell lines (MOLM-13 and SKM-1) identified 9 potential TFs that caused an increase in resistance to AZA in both cell lines. CTCF, ZNF143 and SMC1A were among the 9 TFs identified. Further, knocking down CTCF in leukemic cell lines using shRNA constructs resulted in increase in resistance to AZA as compared to shControl. This validated the role CTCF played in determining AZA response in leukemic cells.

To functionally annotate the DMRs obtained from patient scWGBS, we used GREAT, a previously published tool developed to predict functions of cis-regulatory regions, and identified genes associated with and possibly regulated by factors binding to our regions of interest. The genes from the HoxB cluster were among the top genes identified. HoxB genes have been shown to play a role in hematopoietic stem cell maintenance and regulating hematopoietic differentiation. Additionally, using the ENCODE database, we found CTCF binding sites within the regulatory region of HoxB genes indicating the potential regulation of HoxB by CTCF. In line with this, knockdown of CTCF in leukemic cell lines resulted in reduction in HoxB gene expression, with HoxB4 showing almost 50% reduction. Lastly, knockdown of HoxB genes, similar to CTCF knock down resulted in an increase in resistance to AZA compared to shControl.

Together, these results indicate a potential role of CTCF in determining the response to AZA by regulating HoxB gene expression in leukemic cells. Given the role of CTCF in forming chromatin loops to regulate gene expression, further studies on changes in chromatin structures within the LSC compartment in AZA responder and non-responder patients will provide deeper insights into how CTCF may determine responsiveness to AZA.

Buckstein:Taiho: Honoraria, Research Funding; Takeda: Research Funding; BMS: Honoraria, Research Funding.