Methylation of the chromatin modifier KMT2D by SMYD2 contributes to therapeutic response in hormone-dependent breast cancer
Activating mutations in PIK3CA are frequently found in estrogen-receptor-positive (ER+) breast cancer, and the combination of the phosphatidylinositol 3-kinase (PI3K) inhibitor alpelisib with anti-ER inhibitors is approved for therapy. We have previously demonstrated that the PI3K pathway regulates ER activity through phosphorylation of the chromatin modifier KMT2D. Here, we discovered a methylation site on KMT2D, at K1330 directly adjacent to S1331, catalyzed by the lysine methyltransferase SMYD2. SMYD2 loss attenuates alpelisib-induced KMT2D chromatin binding and alpelisib-mediated changes in gene expression, including ER-dependent transcription. Knockdown or pharmacological inhibition of SMYD2 sensitizes breast cancer cells, patient-derived organoids, and tumors to PI3K/AKT inhibition and endocrine therapy in part through KMT2D K1330 methylation. Together, our findings uncover a regulatory crosstalk between post-translational modifications that fine-tunes KMT2D function at the chromatin. This provides a rationale for the use of SMYD2 inhibitors in combination with PI3Kα/AKT inhibitors in the treatment of ER+/PIK3CA mutant breast cancer.
Co-first author. My contribution was the computational analysis — the ChIP-seq and RNA-seq work behind Figures 2, 3 and 5.
More than 70% of breast cancers depend on the estrogen receptor, and about 40% of the ER+/HER2− ones carry an alteration in the PI3K pathway. That pairing is why alpelisib, a PI3Kα inhibitor, is approved alongside anti-estrogen therapy.
The trouble is that blocking PI3K makes ER signaling stronger, and the tumor escapes through the pathway the other drug is meant to control. Our lab had previously traced how: the PI3K effectors AKT and SGK phosphorylate the chromatin modifier KMT2D at S1331, which keeps it off the chromatin. Inhibit PI3K and that brake comes off — KMT2D is recruited, ER-dependent gene programs switch on, and the tumor grows. The drug creates its own resistance mechanism.
This paper asks what else is controlling that switch.
A second mark, one residue away
Mass spectrometry on immunoprecipitated KMT2D turned up mono- and di-methylation at K1330 — directly adjacent to the S1331 phosphosite. Two regulatory marks on neighbouring residues of a 5,537-amino-acid protein is not a coincidence worth ignoring.
Screening recombinant lysine methyltransferases identified SMYD2 as the enzyme. K1330A and K1330R mutants were not methylated, in vitro or in cells, and knocking down or pharmacologically inhibiting SMYD2 reduced the mark.
The two modifications talk to each other in both directions. AKT1 phosphorylates S1331 with lower catalytic efficiency on peptides that are methylated at K1330 or mutated there. SMYD2 methylates phospho-S1331 KMT2D but cannot touch the phospho-dead S1331A mutant.
What methylation does not do is change KMT2D's enzymatic activity — the K1330A/R mutants methylate H3K4 as well as wild-type. So this is not about the enzyme. It is about where KMT2D goes and what it brings with it.
Where KMT2D goes
ChIP-seq of KMT2D in MCF7 cells, with inducible SMYD2 knockdown, with and without alpelisib, gives the direct test.
Alpelisib increased KMT2D chromatin binding, as the earlier model predicts. Removing SMYD2 attenuated that increase — genome-wide, and specifically at ER and FOXA1 loci, including canonical targets like GREB1 and IGFBP4. Motif analysis of the affected peaks returned ERE, FOXA1, AP-1, GRHL2 and TEAD. CUT&RUN showed SMYD2 and KMT2D co-occupying chromatin at sites enriched for estrogen response and forkhead motifs.
Interaction proteomics filled in the mechanism: components of the INO80 remodeling complex, ESCRT and histone acetyltransferase complexes associate differently with wild-type KMT2D than with the methyl-dead mutant. The mark appears to govern which cofactors come along.
RNA-seq matched. Of the 3,995 genes responding to PI3K inhibition, 1,248 — over 30% — were altered by SMYD2 knockdown. Gene set enrichment put estrogen response among the top signatures negatively regulated by SMYD2 loss, along with proliferation programs, and the canonical ER targets MYC, GREB1, PGR and IGFBP4 were attenuated.
Proving it runs through this one residue
SMYD2 methylates many substrates, so sensitising cells by removing SMYD2 does not establish that KMT2D is the relevant one. The clean experiment is a CRISPR knock-in of K1330A — methyl-dead at that site alone, SMYD2 otherwise intact.
The knock-in cells were more sensitive to alpelisib, to the AKT inhibitor capivasertib and to fulvestrant, and their tumors grew significantly more slowly in mice. By RNA-seq the knock-in phenocopied SMYD2 knockdown: 2,158 shared differentially expressed genes at baseline and 1,564 under alpelisib.
A meaningful part of what SMYD2 does to therapeutic response in these cells runs through a single lysine on KMT2D.
Why it might matter clinically
SMYD2 knockdown or inhibition sensitized MCF7 and T47D cells to alpelisib, capivasertib, fulvestrant and the approved combinations. It held in patient-derived organoids from three ER+/PIK3CA-mutant tumors, where SMYD2 inhibitor plus alpelisib beat either alone. In xenografts, combining SMYD2 knockdown with alpelisib produced durable regression where alpelisib alone only slowed growth.
The patient data points the same way. In TCGA and METABRIC, ER+ tumors with low SMYD2 showed lower ER transcriptional activity. Roughly 70% of breast cancers carry an SMYD2 alteration, mostly copy gain or amplification, and amplification was associated with reduced overall survival.
SMYD2 inhibitors have been modest as single agents. This work suggests the combination is where they belong.
Limits
The mechanism is worked out in cell lines and organoids, with mouse xenografts for the in vivo claim, so the human evidence is correlative. The survival association with SMYD2 amplification is retrospective. And SMYD2 has plenty of other substrates — p53, ERα, EZH2, STAT3, RUNX1 — so the K1330A knock-in shows this axis carries much of the effect without showing it carries all of it.