Critical role for MLL3-phd domains in HSC self-renewal and AML suppression Free

MLL3 and MLL4 are homologous histone methyltransferases that nucleate COMPASS (Complex of proteins associated with SET1) complexes. Both proteins have established, seemingly antagonistic roles in hematopoiesis and leukemogenesis. MLL3 promotes HSC differentiation, limits self-renewal capacity and suppresses acute myeloid leukemia (AML) initiation, whereas MLL4 promotes HSC self-renewal and AML maintenance. These functional differences are surprising, given that the proteins have similar domain structures and high homology. Both proteins contain N-terminal PHD domains that mediate interactions with chromatin modifying enzymes (e.g., the BAP1 complex), the DNA methyltransferase TET3 and modified histones (e.g., H4K16ac). MLL3 contains an extra PHD domain not found in MLL4. These domains are often mutated in a variety of human malignancies. Given that PHD mediate chromatin interactions, we chose to investigate the contributions of individual domains to hematopoiesis and leukemogenesis, focusing on MLL3.

To evaluate functions of individual PHD domains in vivo, we developed mice with targeted point mutations that inactivate PHD2 (MLL3^W382L) and PHD5 (MLL3 ^C983F) and PHD7 (MLL3 ^C1097A). These point mutants have been shown to disrupt interactions with chromatin or chromatin binding proteins. In addition, we generated mice with an inactivated SET methyltransferase domain (MLL3^Y4793A). Western blots demonstrated normal protein expression for all mutants except for W382L, which destabilized the MLL3 protein.

We evaluated HSC and progenitor numbers, HSC function in transplantation assays and gene expression and compared effects of each point mutant to the null allele. The PHD2/W382L and PHD5/C983F mutations caused modest expansion of the HSC population and enhanced HSC function, much like the null allele. In contrast, PHD7/C1097A and SET/Y4793A mutations did not enhance HSC function. RNA-Sequencing (RNA-Seq) showed changes in gene expression in C983F and W382L HSCs, MPPs, GMPs that closely resembled complete loss of function. C1097A caused less severe but significant loss-of-function related changes. The SET domain mutation Y4793A had no effect on HSC function. The W382L-associated changes were not surprising, given the loss of protein expression, but the data indicate an important role for PHD5, which binds TET3 and H4K16ac, in HSC differentiation. PHD7, which also binds TET3 and H4K16ac, has a less impactful role, and the SET domain catalytic function is dispensable in HSCs.

We next evaluated leukemia initiation, focusing on leukemias that arise in the setting of cooperating NF1 and p53 mutations, both of which commonly co-occur with MLL3 deletions. We used CRISPR/Cas9 to inactivate NF1 and p53 in HSC/MPPs from point mutant or loss-of-function mice. In activating Nf1 and p53 simultaneously on an Mll3 wildtype background caused T-cell Acute Lymphoblastic Leukemia (T-ALL), whereas Nf1/p53 inactivation on an Mll3 heterozygous null background caused a mixture of T-ALL and Myeloproliferative Neoplasm (MPN). Nf1/p53 inactivation on an Mll3 homozygous null background caused rapid onset AML. All three PHD point mutants phenocopied the null allele, including the milder C1097A allele. These data illustrate a critical role for PHD cluster 5-7 in both restricting HSC self-renewal capacity and suppressing leukemogenesis.

Additional studies are needed to identify proteins that bind PHD5-7 in vivo. While TET3 and H4K16ac are candidates, these partners were primarily evaluated in the context of MLL4. Differences between MLL3 and MLL4 chromatin interactions, particularly at PHD5-6, could account for the distinct functions of these proteins in hematopoiesis.