Epigenetic regulation of gene expression by MLL3 and MLL4 complexes

The MLL3 and MLL4 complexes monomethylate H3K4 at transcription enhancers throughout the human genome, with estimates ranging from approximately 12,000 to over 20,000 sites, depending on the cell type and developmental stage. Full activation of gene expression through MLL3 and MLL4 complex target enhancers appears to require simultaneous monomethylation of H3K4 by MLL3 and/or MLL4 complexes, and acetylation of H3K27 by the histone acetyltransferase p300/CBP, recruited to enhancers through direct interactions with the MLL3 and MLL4 complexes. KDM6A (also known as UTX), a lysine demethylase that acts as an accessory subunit of MLL3 and MLL4 complexes, facilitates H3K27 acetylation by removing inhibitory methyl groups from H3K27, deposited by the Polycomb repressor PRC2 complex (reviewed in Fagan and Dingwall 2019).

KMT2C (also known as MLL3), the catalytic subunit of the MLL3 complex, contains two closely related plant homeodomain (PHD) zinc finger clusters, with 6-7 zinc fingers, in the N-terminal region, and a single PHD zinc finger near the C-terminus, which are involved in protein-protein interactions. The HMG domain in KMT2C enables DNA binding, while the SET domain provides catalytic activity. KMT2C possesses multiple nuclear receptor (NR) interaction motifs (LLXXL or LXXLL), which are important for recruitment of the MLL3 complex to NR-regulated enhancers (reviewed in Fagan and Dingwall 2019).

While H3K4 monomethylation by MLL3 and MLL4 complexes may not be essential for expression of developmental genes, it is likely important for fine tuning of transcription levels and timing, both during normal development and in cancer. Although a broad dispersion of cancer mutations in the coding regions of the KMT2C and KMT2D genes, as well as the presence of many truncating mutations, imply a tumor suppressor role, activating mutations in the SET domains have also been reported, suggesting that a tumor suppressive vs. oncogenic role is context-dependent (reviewed in Fagan and Dingwall 2019).

KMT2C is frequently mutated in cancer. KMT2C may be important for driving hormone-stimulated proliferation of breast cancer cells that are ESR1-positive and ERBB2-negative. In mice, simultaneous overexpression of Pik3ca and inactivation of the Kmt2c blocks differentiation of the mammary gland and leads to increased stem cell self-renewal through HIF pathway activation (reviewed in Fagan and Dingwall 2019).

Non-small cell lung cancer (NSCLC) is characterized by frequent co-occurrence of mutations in KMT2C and KMT2D (also known as MLL4, the catalytic subunit of the MLL4 complex) (reviewed in Fagan and Dingwall 2019).

DNA damage-induced transcription of TP53 target genes requires both KMT2C and KMT2D. KMT2C is also implicated in TP53-dependent DNA double strand break repair in a transcription-independent manner. KMT2C and KMT2D contribute to maintenance of epithelial cell states by negatively regulating the epithelial-to-mesenchymal transition. KMT2C mutations in lung and breast cancer are frequently found in the first PHD that is involved in the interaction with the BAP1 histone deubiquitinating complex, linked to Polycomb repressor complex-dependent gene silencing (reviewed in Fagan and Dingwall 2019).

Heterozygous germline LOF mutations in KTM2C are associated with Kleefstra syndrome-2 and autism spectrum disorder (reviewed in Fagan and Dingwall 2019).

In mouse, Kmt2c and Kmt2d are implicated in enhancer priming and de novo enhancer activation during embryonic development. This function is not essential for the maintenance of cell identity and self-renewal of embryonic stem cells (ESCs) and somatic cells but is necessary for ESC reprogramming during differentiation and for production of induced pluripotent stem cells (iPSCs) (reviewed in Fagan and Dingwall 2019).

Knockout of KMT2D, the catalytic subunit of the MLL4 complex, in human colon carcinoma cell line HCT116, which already harbors inactivating mutations in both alleles of KMT2C (the catalytic subunit of the MLL3 complex), leads to significant global reduction of H3K4 monomethylation. Knockout studies of KMT2C and KMT2D in HCT116 cells and mouse embryonic fibroblasts (MEFs) implicate at least partially redundant roles of MLL3 and MLL4 complexes in H3K4 monomethylation. Genome-wide ChIP-seq analysis in both HCT116 cells and MEFs showed that ~80% of the MLL4 peaks are enriched at intergenic and intragenic regions, while only ~20% of the peaks map to transcription start sites (TSS). MLL4 binding sites at SAE1 and AP3B1 gene loci in HCT116 cells, and at Nanog and Lefty1 loci in MEFs, are co-occupied by enhancer region markers H3K4me1, EP300, and acetylated H3K27. Many of the genes associated with MLL3/MLL4-bound enhancers in HCT116 cells are implicated in intracellular signaling, while genes associated with MLL3/MLL4-independent enhancers tend to be implicated in regulation of gene expression.

HOXA9, encoded by a target gene of KMT2A (MLL1) and KMT2B (MLL2) complexes, may function as a pioneer factor at de novo enhancers in acute myeloid leukemia (AML) and recruit CEBPA and the MLL3 and MLL4 complexes to enhancers of leukemogenesis-promoting genes.

KMT2D is frequently mutated in cancer (reviewed in Dhar and Lee 2021) and is one of the most frequently mutated genes in non-Hodgkin lymphoma, such as follicular lymphoma and diffuse large B cell lymphoma, where loss-of-function (LOF) of KMT2D appears to be an early event that cooperates with the over-expression of the BCL2 oncogene. Knockout of Kmt2d in mouse B cell progenitors impairs their differentiation and promotes lymphoma development. KMT2D LOF in lymphoma is associated with reduced H3K4me1 mark at enhancers of multiple tumor suppressor genes (reviewed in Fagan and Dingwall 2019).

KMT2D chromatin enrichment sites significantly overlap with TP53 binding sites. Aberrant transcription associated with TP53 mutations in colon carcinoma is dependent on KMT2D-mediated H3K4 monomethylation. KMT2D can be inactivated through phosphorylation by SGK1, a PI3K effector kinase closely related to AKT1. SGK1 is the estrogen-inducible kinase, whose transcription is collaboratively activated by ESR1 and KMT2D. SGK1-mediated phosphorylation on KMT2D on serine S1331 near the second PHD results in downregulation of H3K4 monomethylation at ESR1-target genes, thus constituting a negative feedback loop (reviewed in Fagan and Dingwall 2019).

Heterozygous germline LOF mutations in KMT2D are associated with Kabuki syndrome. Kabuki syndrome patients have a modestly increased predisposition to cancer, in particular lymphoma, Wilms tumor, hepatoblastoma, synovial sarcoma and neuroblastoma. Mice with brain-specific knockout of Kmt2d (Mll4) develop medulloblastoma that shows hyperactivation of Ras and Notch signaling (reviewed in Fagan and Dingwall 2019).

Based on mouse studies, MLL3 and MLL4 complexes play an important role in adipogenesis and myogenesis. Kmt2c KO mice die around birth with no obvious morphological abnormalities in embryonic development, while Kmt2d KO mice show early embryonic lethality around E9.5. Pups with Kmt2d KO in precursors of brown preadipocytes and skeletal myocytes are obtained at the expected Mendelian ratio but display marked reduction in back muscles and die immediately after birth due to breathing malfunction, also showing a decrease in brown adipose tissue mass. In cultured mouse brown preadipocytes, KO of Kmt2d leads to a moderate differentiation defect along with a transient up-regulation of Kmt2c expression, whereas KO of Kmt2c has no effect on adipogenesis, suggesting a more prominent role of KMT2D in development and a partial compensation of KMT2D loss by KMT2C. KO of Kmt2d 3T3-L1 mouse white preadipocytes inhibits adipogenesis, and Kmt2c and Kmt2d are also required for adipogenesis in mouse embryonic fibroblasts. By ChIP-seq, the average length of Kmt2d binding regions is between 350 and 400 bp, and the binding regions change dramatically from the preadipocytes stage to the onset of adipogenesis, but are then Kmt2d-binding regions were largely non-overlapping between brown adipocytes and skeletal myocytes.