A hypermethylation strategy utilized by enhancer-bound CARM1 to promote estrogen receptor α-dependent transcriptional activation and breast carcinogenesis

While protein arginine methyltransferases (PRMTs) and PRMT-catalyzed protein methylation have been well-known to be involved in a myriad of biological processes, their functions and the underlying molecular mechanisms in cancers, particularly in estrogen receptor alpha (ERα)-positive breast cancers, remain incompletely understood. Here we focused on investigating PRMT4 (also called coactivator associated arginine methyltransferase 1, CARM1) in ERα-positive breast cancers due to its high expression and the associated poor prognosis. Methods: ChIP-seq and RNA-seq were employed to identify the chromatin-binding landscape and transcriptional targets of CARM1, respectively, in the presence of estrogen in ERα-positive MCF7 breast cancer cells. High-resolution mass spectrometry analysis of enriched peptides from anti-monomethyl- and anti-asymmetric dimethyl-arginine antibodies in SILAC labeled wild-type and CARM1 knockout cells were performed to globally map CARM1 methylation substrates. Cell viability was measured by MTS and colony formation assay, and cell cycle was measured by FACS analysis. Cell migration and invasion capacities were examined by wound-healing and trans-well assay, respectively. Xenograft assay was used to analyze tumor growth in vivo. Results: CARM1 was found to be predominantly and specifically recruited to ERα-bound active enhancers and essential for the transcriptional activation of cognate estrogen-induced genes in response to estrogen treatment. Global mapping of CARM1 substrates revealed that CARM1 methylated a large cohort of proteins with diverse biological functions, including regulation of intracellular estrogen receptor-mediated signaling, chromatin organization and chromatin remodeling. A large number of CARM1 substrates were found to be exclusively hypermethylated by CARM1 on a cluster of arginine residues. Exemplified by MED12, hypermethylation of these proteins by CARM1 served as a molecular beacon for recruiting coactivator protein, tudor-domain-containing protein 3 (TDRD3), to CARM1-bound active enhancers to activate estrogen/ERα-target genes. In consistent with its critical role in estrogen/ERα-induced gene transcriptional activation, CARM1 was found to promote cell proliferation of ERα-positive breast cancer cells in vitro and tumor growth in mice. Conclusions: our study uncovered a “hypermethylation” strategy utilized by enhancer-bound CARM1 in gene transcriptional regulation, and suggested that CARM1 can server as a therapeutic target for breast cancer treatment.

. CARM1 is required for estrogen-induced gene transcriptional activation.
(A) Box plot showing the expression of CARM1 (FPKM) in a cohort of clinical breast cancer (n=1,102) and normal (n=113) samples from TCGA (The Cancer Genome Atlas).
(D, E) RNA-seq as described in Figure 1A for specific genes were shown as indicated.
(F) Correlation of CARM1's effects on whole transcriptome in response to estrogen between two RNA-seq biological repeats (n=11,816).
(H, I) MCF7 cells as described in (G) were subjected to immunoblotting (IB) (H) and RT-qPCR (I) analysis to examine the levels of indicated protein and mRNA, respectively. Actin was served as a loading control.
(J) MCF7 cells as described in Figure 1E were subjected to RT-qPCR analysis to examine the expression of genes as indicated.
(K) MCF7 cells as described in Figure 1E were subjected to immunoblotting (IB) analysis to examine the protein levels of CARM1. Actin was served as a loading control.
(L) Genomic DNA was extracted from CARM1 knockout cells, followed by PCR using specific primer sets spanning gRNA targeting region (boxed in blue). The resultant PCR products were subjected to Sanger sequencing as shown. Nucleotide being inserted was highlighted in red.
(M) MCF7 cells as described in Figure 1F were subjected to immunoblotting (IB) analysis to examine the protein levels of CARM1 by using anti-specific CARM1 antibodies from two independent vendors as indicated. Actin was served as a loading control.
(N, O) Pol II ChIP-seq signals across specific genes as indicated. Units are mean tags per bin for 16 bins across the transcribed gene region with 2 kb upstream (4 bins of 500 bp each) and 5 kb downstream flanking regions (10 bins of 500 bp each).
(P, Q) Pol II ChIP-seq tag density distribution centered on cognate enhancers of specific genes as indicated (± 4,000 bp).
(R, S) Heat map (R) and box plots (S) representation of the expression levels (FPKM) for genes induced by estrogen and dependent on CARM1 in the cohort of clinical samples, both normal and tumor, as described in (A).

Figure S2. CARM1 is recruited onto ER-bound active enhancers upon estrogen stimulation.
(A) MCF7 cells treated with or without estrogen (E2, 10 -7 M, 1 hr) were subjected to ChIP with anti-CARM1 antibody followed by qPCR analysis with primers specifically targeting enhancer (e) regions as indicated. ChIP signals were presented as percentage of inputs (± s.e.m.).
(B) ChIP-seq tag distribution, including ER, H3K4me1, H3K4me2, H3K4me3, H3K27Ac, P300, MED1, MED12, H3K9me3 and H3K27me3, centered on all of its own sites (left panels) or estrogen-induced CARM1 sites (right panels) (± 4,000 bp). (D, E) Rates of somatic mutations at (D) and in the vicinity of (E) CARM1 methylation sites in human cancers. Rates of somatic mutations at and in the vicinity of (±5 nucleotides) CARM1dependent methylation sites (n=589) were significantly higher than those of arginines in the proteome (P = 1.62e-11 and 7.42e-45, respectively).  (A) MCF7 cells as described in Figure 5F were subjected to RT-qPCR analysis to examine the expression of CARM1.
(B) MCF7 cells as described in Figure 5F were subjected to immunoblotting analysis to examine the expression of MED12 and CARM1. Actin was served as a loading control.
(E, F) MCF7 cells were transfected with control siRNA (siCTL) or two independent siRNAs specific against CARM1 (siCARM1 (1) and siCARM1 (2)) in stripping medium for three days, and treated with or without estrogen (E2, 10 -7 M, 6 hrs), followed by RT-qPCR (E) and immunoblotting (IB) (F) analysis to examine the mRNA and protein levels of MED12, respectively. Actin was served as a loading control.

Figure S6. TDRD3 is required for estrogen-induced gene transcriptional activation.
(A-C) MCF7 cells as described in Figure 6G were subjected to qPCR analysis to examine the expression of selected estrogen-induced genes as indicated.
(D-F) MCF7 cells as described in Figure 6I were subjected to qPCR analysis to examine the expression of selected estrogen-induced genes as indicated. (E, F) T47D (E) and BT474 (F) cells transfected with siCTL or siCARM1 were subjected to colony formation assay.

Supplementary Table Legend
Table S1. Global mapping of CARM1 substrates. The total number of arginine methylation sites (the sum of mono-and asymmetrical di-methylation sites after removing duplicates), and the number of mono-or asymmetrical di-methyl arginine methylation sites for proteins which had arginine methylation sites identified (sheet 1), had arginine methylation sites on which methylation signals decreased at least two-fold (sheet 2) or completely abolished (sheet 3), and had all its methylation sites abolished (sheet 4) were shown. Table S2. MS2 spectrum of arginine-methylated peptides in MED12. MS2 spectrum of methylated arginine residues in MED12 C-terminus as described in Figure 4D were shown as indicated.