Divergent Roles of MLL3 and MLL4 in Hematopoietic Stem Cell Fate Decision

Abstract

Complex gene regulatory mechanisms, orchestrated by transcription factors (TFs) and epigenetic regulators, govern cell fate decisions in normal and malignant hematopoiesis. While prior works have extensively characterized the TF-networks in blood development, we currently have limited understanding of epigenetic regulation, in particular molecular rules underlying cofactor usage, in hematopoietic cell fate decisions. Kmt2c and Kmt2d are homologous genes that encode transcriptional coactivators, MLL3 and MLL4, respectively. They have been shown to coordinate embryonic stem cell fate transitions and are recurrently mutated in leukemias, offering an ideal paradigm to study epigenetic regulation of hematopoiesis. While MLL3 and MLL4 are often believed to play redundant roles in stem cell fate decisions, few has characterized their molecular distinctions. MLL3 and MLL4 each nucleate a chromatin-bound complex called the Complex of Proteins Associated with Set1 (COMPASS) and both deposit H3K4me1 at enhancers. Despite similar biochemical functions, Kmt2c and Kmt2d loss-of-function mutations have been reported with contrasting phenotypes in hematopoiesis. Kmt2d loss impairs HSC self-renewal and enhances myeloid differentiation at the expense of lymphopoiesis, whereas Kmt2c deletions enhance HSC self-renewal and impair differentiation. We hypothesize that MLL3 and MLL4 are recruited by different transcription factors and bind chromatin at both overlapping and distinct enhancers to convey specific hematopoietic fates. To test whether MLL3 and MLL4 are redundant in licensing myeloid differentiation, we transplanted wildtype, Kmt2c/d single knockout, and Kmt2c/d double knockout (DKO) donor cells into lethally irradiated recipients. Surprisingly, DKO recipients showed significant expansion of phenotypic HSCs, MPPs, and pre-granulocyte-monocyte progenitors (pGMs), typically associated with a myeloid differentiation block. DKO mice progressively developed bone marrow failure 9-months after transplantation, indicating that MLL3/4-COMPASS is necessary to sustain hematopoiesis. To dissect how Kmt2c/d deletions shape lineage specification and whether DKO imposes a differentiation block, we performed Cellular Indexing of Transcriptomes and Epitopes by sequencing (CITE-seq) on donor Lin -Kit +Sca1 + (LSK) cells from the transplantation described above. While Kmt2c deletion does not markedly reprogram HSC/MPP identity, Kmt2d deletion results in precocious myeloid differentiation. Most notably, we observed that compound Kmt2c/d deletions fundamentally reprogramed HSCs/MPPs toward a B-lymphoid-like state, instead of imposing a myeloid differentiation block. The data establish a model wherein MLL3 and MLL4 have redundant roles in licensing HSC identity and myeloid potential. However, within the HSC to myeloid differentiation trajectory, MLL3 promotes differentiation while MLL4 promotes self-renewal. To understand how MLL3- and MLL4-COMPASS differentially regulate enhancer accessibility, we performed single-cell transposase-accessible chromatin sequencing (scATAC-seq) on the same donor LSK cells as in the CITE-seq. Compound Kmt2c/d deletions impose a unique epigenomic state on HSCs/MPPs (Figure 1A). Motif enrichment analyses revealed that DKO cells (i) were enriched for PAX5 and EBF1 motifs, consistent with B-lymphoid priming. HSCs (ii) were enriched for AP-1 and NFE2L2 motifs, suggesting that MLL4 may preferentially bind AP-1 and antioxidant response elements associated with HSC identity. Clusters of myeloid progenitors (iii and iv) were populated primarily by Kmt2d-deficient cells and enriched for CEBP and SPI1 motifs, suggesting that MLL3 is preferentially recruited to enhancers bound by myeloid TFs. This study provides several novel insights into epigenetic regulation of hematopoiesis. We have shown that HSC identity and myeloid fate specification are licensed by redundant MLL3/4-COMPASS-dependent TF-cofactor interactions, but once HSCs are established, MLL3- and MLL4-COMPASS engage distinct, antagonistic programs. Furthermore, early B-lymphoid priming is mediated via MLL3/4-COMPASS-independent mechanisms (Figure 1B). Ongoing work will link precise transcriptional codes to divergent cofactor utilization patterns and provide critical insights into how epigenetic regulators are deployed to effect distinct cell fates.