Abstract
C/EBPα is an essential transcription factor involved in regulating the expression or function of certain cell-cycle regulators, including in breast cancer cells. Although protein arginine methyltransferases have been shown to play oncogenic roles in a variety of cancers, little is known about the role of arginine methylation in regulating the antiproliferation activity of C/EBPα. Here, we report that the protein arginine methyltransferase 1 (PRMT1) is overexpressed in human breast cancer and that elevated PRMT1 correlates with cancer malignancy. RNA-sequencing analysis revealed that knockdown of PRMT1 in breast cancer cells is accompanied by a decrease in the expression of pro-proliferative genes, including cyclin D1. Furthermore, tandem affinity purification followed by mass spectrometry identified PRMT1 as a component of the C/EBPα complex. C/EBPα associated with and was methylated by PRMT1 at three arginine residues (R35, R156, and R165). PRMT1-dependent methylation of C/EBPα promoted the expression of cyclin D1 by blocking the interaction between C/EBPα and its corepressor HDAC3, which resulted in rapid growth of tumor cells during the pathogenesis of breast cancer. Inhibition of PRMT1 significantly impeded the growth of cancer cells from patients with triple-negative breast cancer. This evidence that PRMT1 mediates C/EBPα methylation sheds light on a novel pathway and potential therapeutic target in breast cancer.
This study provides novel mechanistic insight of the role of the arginine methyltransferase PRMT1 in breast cancer pathogenesis.
Introduction
Protein arginine methylation catalyzed by the protein arginine methyltransferase (PRMT) family is a major posttranslational modification that regulates multiple cellular processes (1–3). PRMT1 is thought to perform 85% of PRMT activity in mammalian cells (4). Many nonhistone substrates of PRMT1 are involved in biological processes such as transcriptional regulation and cell signaling (1, 5), and methylation of histone H4 at arginine 3 (H4R3me2a) by PRMT1 activates gene regulation (6, 7). PRMT1-mediated arginine methylation of FOXO1 enhances its transactivation function by stabilizing the FOXO1 protein (8). PRMT1 also functions as a coactivator of RUNX1 to activate its target genes by disrupting the interaction between RUNX1 and SIN3A (9). Given the extensive substrates of PRMT1 in cells, aberrant expression of PRMT1 has been implicated in the pathogenesis of several diseases, including cancer (3). PRMT1 upregulation (10–12) or aberrant splicing (13) has been observed in many types of malignancies. However, the function of PRMT1 in breast cancer and the mechanism of its effect on gene transcription are still incompletely understood.
CCAAT/enhancer binding proteins (C/EBP) are a family of basic leucine zipper (bZIP) DNA-binding proteins that regulate the transcription of several tissue-specific genes and some growth-related genes (14). C/EBPα is the founding member of the C/EBPs family and contains a C-terminal bZIP and N-terminal transactivation domain (TAD1–3; refs. 15, 16). C/EBPα has two distinct isoforms, p42 and p30, with varying N-terminus length due to translation from different initiation codons in a single mRNA. C/EBPα-p30 lacks the major transactivation domain (1–120 aa) and can neutralize the transcriptional activity of C/EBPα-p42 (17, 18). C/EBPα is a strong inhibitor of cell proliferation and has been implicated as a tumor suppressor in various malignant tumors. The antimitotic function of C/EBPα depends on protein–protein interactions and regulation of its binding proteins, such as p21 (19), CDK4/6 (20), and E2F (21, 22). C/EBPα also functions as a transcriptional factor to activate or repress some growth-related genes, such as N-Myc (23), Sox4 (24), and BMI1 (25), to modulate cell proliferation.
A key component of C/EBPα regulation is posttranslational modifications (PTM). C/EBPα is decorated with various PTMs, including phosphorylation (26–29), SUMOylation (30–32), and acetylation (33). Most of these modifications repress C/EBPα function by interfering with its DNA-binding function or blocking interactions with partners. However, little is known about the role of arginine methylation in regulating the antiproliferation activity of C/EBPα.
Here, we report that C/EBPα directly interacts with and is arginine methylated by PRMT1; methylated C/EBPα impairs the binding of HDAC3 to derepress the expression of cyclin D1 followed by cell growth in breast cancer cells. PRMT1 is significantly upregulated in breast carcinoma samples, and PRMT1 knockdown impairs the proliferation capability of breast cancer cells. Importantly, a specific inhibitor of PRMT1 profoundly decreases the growth of cancer cells from patients.
Materials and Methods
Cell cultures
MDA-MB-231 (catalog no. 3111C0001CCC000014), MDA-MB-435 (catalog no. 3111C0001CCC000351), MDA-MB-468 (catalog no. 3111C0001CCC000249), MCF7 (catalog no. 3111C0001CCC000013), SK-BR-3 (catalog no. 3111C0001CCC000085), HEK293T (catalog no. 3111C0001CCC000091), HeLa (catalog no. 3111C0001CCC000011), BT549 (catalog no. 3111C0001CCC000336), and MCF10A (catalog no. 3111C0001CCC000406) cell lines were purchased from Chinese National Infrastructure of Cell Line Resource. Before the experiments, all the cell lines were authenticated on cell micrograph compared with the cell lines on ATCC. HEK293T cells showed 90% transfect efficiency with GFP-tag plasmid. Mycoplasma contamination was detected by the EZ-PCR Mycoplasma Test Kit (catalog no. 20-700-20) as described previously (34). MDA-MB-231, MDA-MB-435, MDA-MB-468, MCF7, SK-BR-3, HEK293T, HeLa, and BT549 cells were grown in DMEM supplemented with 10% FBS and 100 U/mL penicillin–streptomycin. MCF10A cells were maintained in DMEM/F12 (1:1) medium supplemented with 5% horse serum, 0.5 μg/mL hydrocortisone, 10 μg/mL insulin, 20 ng/mL EGF, 0.1 μg/mL cholera toxin, 100 U/mL penicillin–streptomycin.
Primary breast cancer cells (TNBC-1 and TNBC-2) were grown in DMEM/F12 (1:1) medium supplemented 5% FBS, 0.4 μg/mL hydrocortisone, 5 μg/mL insulin, 10 ng/mL EGF, 10 ng/mL cholera toxin, 25 μg/mL adenine, 5 μmol/L Rock inhibitor, and 100 U/mL penicillin–streptomycin. These cells were treated with furamidine dihydrochloride (SML1559, Sigma), a specific inhibitor of PRMT1 (35) at concentration of 0, 20, 40, and 60 μmol/L for 6 days.
Antibodies
Antibodies for PRMT1 (07-404) and asymmetric dimethyl-arginine (ASYM25, anti-Rme2, 09-814) were from Millipore; antibody for C/EBPα (2295) was from Cell Signaling Technology; antibody for cyclin D1 (ab134175) was from Abcam; antibodies for actin (sc-47778), GAPDH (sc-166574), and GST (sc-138) were from Santa Cruz Biotechnology; antibody for FLAG (F3165) was from Sigma; antibodies for FLAG (PM020), Myc (M047-3, and 562), and GFP (598) were from MBL; antibody for HDAC3 (A2139) was from Abclonal.
Plasmids
Expression plasmids of C/EBPα-pcDNA6, C/EBPβ-pcDNA6, C/EBPδ-pcDNA6, C/EBPγ-pcDNA6, HDAC1-pcDNA6, HDAC2-pcDNA6, and HDAC3-pcDNA6 were described previously (36). HDAC1 and HDAC3 coding region were further cloned into Myc-tagged pCMV-3tag7 vector. C/EBPα was cloned into pCMV-3tag7 and pGEX4T-1 vectors. Deletion mutants of C/EBPα (ΔTAD1, 1-120, ΔbZIP, p30, bZIP) were amplified by PCR using C/EBPα-pcDNA6 as template and inserted into pcDNA6. Expression vectors of C/EBPα deletions ΔTAD2 and ΔTAD3 were generated using Seamless Cloning kit (Biomed). PRMT1 cDNA was amplified by PCR from HEK293T cells cDNAs and cloned into pcDNA6 and pCMV-tag3B vector. PRMT1 mutant (E153Q) and C/EBPα mutants (R35K, R156K, R165K, R35/156/165K, R156F, R35/156/165F, R289A) were generated by PCR-based site-directed mutagenesis. Cyclin D1 promoter, extending from −808 to +133 relative to the transcription start site, was cloned into the pGL3-basic luciferase reporter vector, and was used as templates to subclone and generate a series of promoter deletions of cyclin D1 (−304/+133, −113/+133, −8/+133, −113/+15). Promoter of cyclin D1 with disrupted potential C/EBPα-binding site was constructed using PCR-based site-directed mutagenesis. The primers were listed in Supplementary Table S1.
Reporter gene assay
Reporter gene assays were performed as described previously (37).
PRMT1 knockdown (shPRMT1) cell lines
PRMT1 shRNAs were cloned into pLKO.1 vector, and control shRNA was a hairpin designed against GFP. HEK293T cells were transfected with indicated pLKO.1 plasmid and lentiviral constructs (psPAX2 and pMD2.G). Lentiviruses were collected 48 hours posttransfection. MDA-MB-231 cells were infected using lentivirus expressing shPRMT1 or shGFP with 8 μg/mL polybrene. Infected cells were selected with 1.5 μg/mL puromycin 36 hours after infection. Multiple monoclonal cultures were screened for shPRMT1 by Western blotting and RT-PCR analysis. The primers were listed in Supplementary Table S2.
Colony formation assay
Cells were seeded in 35-mm dishes at a density of 103 cells/dish. After continued culturing for about 10 days, cells were washed by PBS twice, then stained with 0.05% crystal violet. Cells were then air dried and photographed.
MTS assay
Cells were seeded at a density of 3 × 103 cell/well in quintuplicate in 96-well plates. Proliferation was assessed using CellTiter 96 AQueous One Solution Cell Proliferation Assay (MTS). Briefly, solution reagent was added to each well and incubated at 37°C for 1 hour at various time points. Subsequently, cell proliferation was assessed by measuring the absorbance at 490 nm. Wells without cells (only medium) were used as blanks.
Anchorage-independent growth assay
Soft-agar assays were constructed in 35-mm dishes. The base layer of cell growth matrix consisted of 2 mL of 1 × DMEM, 10% FBS, and 0.6% low melting point agarose. Plates were chilled at 4°C until solid. Upon solidification, the top layer was poured, consisting of 3 × 103 shPRMT1 or shGFP MDA-MB-231 cells suspended in 2 mL of 1 × media and 0.35% low melting point agarose. A further 1 mL 1 × media without of agarose was added when needed. Cells were cultured for 4 weeks, colonies were stained with 0.005% crystal violet and counted (>200 μm in diameter) in five random 1-cm2 areas.
Wound-healing assay
Cells were plated in growth medium in 6-well plates. After confirming the formation of a complete monolayer, the cells were wounded by scratching lines with a standard 200 μL plastic tip. Cell migration throughout the wound area was observed and measured at different time points.
Xenograft assay
Four-week-old female BALB/c nude mice were purchased and maintained at animal center for 1 week prior to injection to allow the mice adjusting the new environment. The assay was performed according to ethical approval. A total of 5 × 105 shPRMT1 or shGFP MDA-MB-231 cells suspended in 100 μL PBS were injected subcutaneously into nude mice (8 mice for each group). The size of the resultant tumors was measured every 5 days for 45 days. On day 45 after implantation, the tumors were harvested and tumor weights were measured. The tumor volume was calculated with the formula V = L × W2/2 (V, volume; L, length; W, width of tumor). All studies involving mice were approved by the Animal Care and Use Committee of Chinese Academy of Medical Sciences.
Flow cytometry assay
For cell-cycle analysis, 2 × 106 cells were trypsinized, washed twice in ice-cold PBS, and fixed in cold 75% alcohol at 4°C overnight. After that, the cells were suspended in PBS with 100 μg/mL RNAase A for 30 minutes at 37°C, stained with 50 μg/mL propidium iodide (PI) for 15 minutes in the dark at room temperature, measured for DNA content with Accuri C6 (BD Biosciences). A total of 20,000 cells were counted and all analyses were performed in triplicate.
Western blotting, coimmunoprecipitation, and GST-pulldown
Western blotting, coimmunoprecipitation (co-IP), and GST-pulldown assays were performed as described previously (37).
In vitro methylation assay
FLAG-C/EBPα, PRMT1-WT, and PRMT1-E153Q were expressed and purified from HEK293T cells with anti-FLAG M2 agarose beads. Purified C/EBPα was incubated with PRMT1-WT or PRMT1-E153Q in buffer [50 mmol/L Tris-HCl (pH 8.0), 1 mmol/L phenylmethylsulfonylfluoride, and 0.5 mmol/L DTT] supplemented with S-adenosyl-L-[methyl-3H]methionine at 37°C for 1 hour. Reactions were stopped by adding 5 × SDS-PAGE loading buffer and heating. Samples were analyzed by SDS-PAGE and fluorography.
Electrophoretic mobility shift assay
Electrophoretic mobility shift assay (EMSA) were performed as described previously (36). Briefly, oligonucleotides from cyclin D1 promoter (+62/+93, GTTTTGTTGAAGTTGCAAAGTCCTGCAGCCTC) were labeled with biotin at 5′-end. C/EBPα proteins were purified from HEK293T cells transfected with FLAG-C/EBPα-pcDNA6. For specific competition, unlabeled oligonucleotide was added to the binding reaction mixtures at a 200-fold molar excess. For supershift assays, 2 μg of specific anti-C/EBPα antibody was used to the reactions.
qRT-PCR and chromatin immunoprecipitation
qRT-PCR and chromatin immunoprecipitation (ChIP) assays were performed as described previously (37). The primers were listed in Supplementary Tables S3 and S4.
Mass spectrometry for C/EBPα-associated proteins
About 4 × 108 HeLa cells overexpressed with C/EBPα were harvested for assays. FLAG-tagged C/EBPα proteins were immunoprecipitated using anti-FLAG M2 beads. The beads were washed 8 times using RIPA buffer and eluted with 3 × FLAG peptide. Elutes were resolved on SDS-PAGE, stained with Coomassie brilliant blue, and bands were excised and digested with chymotrypsin and subjected to LC/MS-MS sequencing and data analysis.
Mass spectrometry for C/EBPα methylation
Expression plasmids of C/EBPα and PRMT1 were transfected into HEK293T cells. C/EBPα was purified using anti-FLAG M2 beads and eluted with 3 × FLAG peptide. Elutes were resolved on SDS-PAGE and stained with Coomassie brilliant blue. The band of C/EBPα was excised and subjected to LC/MS-MS analysis.
IHC analysis
Tissue microarray containing 30 normal breast tissues (N), 10 fibroadenoma (FA), 9 fibrocystic mastopathy (FM), 7 mild dysplasia (MD), 4 severe dysplasia (SD), 10 ductal carcinaoma in situ (DCIS), 2 invasive lobular carcinoma (ILC-I), 3 invasive ductal carcinoma (IDC-I), 23 IDC-II, and 5 IDC-III were purchased from Superbiotek. IHC was performed using anti-PRMT1 (1:500) antibody. Each sample was assigned with a score according to the intensity of the staining (0 = no staining, 1 = weak staining, 2 = moderate staining, and 3 = strong staining) and the proportion of the stained cells (0 = 0%, 1 = 1%–25%, 2 = 25%–50%, 3 = 50%–75%, 4 = 75%–100%). The stained tissues were scored by three individuals blinded to the clinical parameters.
Patient samples
This study was approved by The Clinical Research Ethics Committee of the Institute of Basic Medical Sciences, Chinese Academy of Medical Sciences. The written informed consent was obtained from the participants. Primary breast cancer tissues and their corresponding adjacent noncancerous tissues were obtained from patients treated at Peking Union Medical College Hospital after receiving their written informed consent. Specimens were snap-frozen in liquid nitrogen at time of surgery and stored at −80°C.
RNA-seq
Total RNAs isolated from shPRMT1 (sh1 and sh2) or shGFP MDA-MB-231 cells using TRIzol reagent (Invitrogen). RNA integrity and concentration were determined using Nanodrop and agarose gel electrophoresis. The purified RNA samples, with RIN (RNA Integrity Number) over 8.0, determined by Agilent 2100 Bioanalyzer (Agilent) were sequenced at ThorGene corporation, who constructed their digital gene expression libraries and sequenced by means of Illumina HiSeq 2500 platform to obtain the expression libraries of 150-nt read length. Independent triplicate cultures were sampled. The clean reads were mapped against human genome hg19 with less than two-base mismatching, using Tophat (version 2.1.1b). The normalized expression values for each gene were calculated by FPKM (expected number of Fragments per Kilo base of transcript sequence per Million base pairs sequenced). Differential expression gene was determined by Cuffdiff package (v2.2.1). The transcript levels of genes having a P value of less than 0.05 were significantly differential between two samples. RNA-seq data files are available from the NCBI Gene Expression Omnibus (http://www.ncbi.nlm.nih.gov/geo/) database (accession number GSE121168).
Statistical analysis
The two-tailed, Student t test was used to compare two groups. All data are shown as the mean with SDs from at least three independent experiments. Probabilities of P > 0.05 were considered as not significant (#), P < 0.05 as significant (*), and P < 0.01 as highly significant (**). Survival analysis was conducted using the Kaplan–Meier method. P < 0.05 indicated significant difference.
Results
Elevated PRMT1 correlates with malignancy of breast cancer
To define the clinical relevance of PRMT1 in breast cancer, we analyzed the correlation of PRMT1 expression and breast cancer patients' clinical behaviors using Kaplan–Meier survival analysis with an online tool. Patients whose tumors expressed higher levels of PRMT1 had significantly worse overall survival duration (Fig. 1A). Moreover, we analyzed a published clinical dataset (GSE29044), and statistical analysis revealed significantly positive correlations of the expression of PRMT1 and the pro-proliferative gene cyclin D1 (Fig. 1B). We then determined the expression of PRMT1 in breast tumor samples (T) and adjacent nontumor (N) tissues from 8 patients (including four groups, Erb-B2, Basal, Luminal-B, and Luminal-A). Western blotting showed that the expression levels of PRMT1 were high relative to the adjacent nontumor tissue in 7 of 8 cancer samples (lanes 1, 5, 7, 9, 11, 13, and 15 of Fig. 1C). Furthermore, a breast cancer tissue microarray containing 103 samples from normal tissues, benign tumors, and malignant tumors excised prior to treatment was employed to detect PRMT1 expression by IHC staining (Fig. 1D; Supplementary Fig. S1; Supplementary Table S5). Significantly greater PRMT1 staining was observed for malignant tumors, while weaker IHC staining was observed for normal and benign tumor tissues (Fig. 1E). The data revealed that the expression of PRMT1 is elevated in breast cancer tissues and that the level of its expression is highest in grade III of invasive ductal carcinoma.
PRMT1 correlates tumor grade in breast cancer. A, Kaplan–Meier survival analysis of the correlation between PRMT1 expression levels and survival time in patients with breast cancer using the online tool (http://kmplot.com/analysis/; P < 0.05). B, Correlation analysis of public dataset (GEO: GSE29044) for the mRNA expression of PRMT1 and cyclin D1 by R programming. C, Representative Western blot analysis showing 8 pairs of breast cancer (T) and adjacent noncancer tissues (N). Anti-PRMT1 antibody and anti-actin were used as control (top). The panels were scanned with ImageJ, a software for scanning grayscale, and the relative density of PRMT1 relative to actin is presented (bottom). D and E, IHC staining of breast tissue array for PRMT1 expression. D, Representative images in each group are shown. E, PRMT1 score shown is the mean ± SD, and statistical significance was determined by two-tailed t test. #, P > 0.05; *, P < 0.05; **, P < 0.01. F, Comparing endogenous expression of PRMT1 in human breast cancer cell lines by Western blot analysis. Breast cell line (MCF10A) and GAPDH served as a control and reference gene, respectively (top). The panels were scanned as in C and the relative expression of PRMT1 relative to MCF10A without EGF is presented (bottom).
PRMT1 correlates tumor grade in breast cancer. A, Kaplan–Meier survival analysis of the correlation between PRMT1 expression levels and survival time in patients with breast cancer using the online tool (http://kmplot.com/analysis/; P < 0.05). B, Correlation analysis of public dataset (GEO: GSE29044) for the mRNA expression of PRMT1 and cyclin D1 by R programming. C, Representative Western blot analysis showing 8 pairs of breast cancer (T) and adjacent noncancer tissues (N). Anti-PRMT1 antibody and anti-actin were used as control (top). The panels were scanned with ImageJ, a software for scanning grayscale, and the relative density of PRMT1 relative to actin is presented (bottom). D and E, IHC staining of breast tissue array for PRMT1 expression. D, Representative images in each group are shown. E, PRMT1 score shown is the mean ± SD, and statistical significance was determined by two-tailed t test. #, P > 0.05; *, P < 0.05; **, P < 0.01. F, Comparing endogenous expression of PRMT1 in human breast cancer cell lines by Western blot analysis. Breast cell line (MCF10A) and GAPDH served as a control and reference gene, respectively (top). The panels were scanned as in C and the relative expression of PRMT1 relative to MCF10A without EGF is presented (bottom).
Next, we detected the expression of PRMT1 in multiple breast cancer cell lines. The MCF10A cells, derived from nontumorigenic breast epithelial cells, were cultured in growth medium with EGF or without EGF, and Western blotting showed that EGF was responsible for PRMT1 expression (lane 1 vs. lane 2, Fig. 1F). Consistent with literature (38), PRMT1 levels in MCF7, MDA-MB-435, and SK-BR-3 were increased relative to MCF10A (−EGF), but similar to the MCF10A cells (+EGF), the levels of PRMT1 expression were higher in MDA-MB-231, -468, and BT-549 cells than that in MCF7 and SK-BR-3 cells (Fig. 1F). Given that MDA-MB-231, -468, and BT-549 cells are more aggressive than MCF7 and SK-BR-3 cell lines, the result suggested that the expression of PRMT1 correlated with the malignance of breast cancer.
PRMT1 regulates the expression of cell-cycle genes in MDA-MB-231 cells
To gain insight into the mechanism by which PRMT1 regulates breast cancer cell growth, we employed MDA-MB-231, an aggressive breast cancer cell line, to establish two stably transfected clones in which PRMT1 was knocked down by cotransduction of the cells with lentivirus encoding shRNAs specific for PRMT1, designated sh1 and sh2. RNA sequencing was performed to systematically determine the differential gene expression profile in PRMT1-knockdown MDA-MB-231 cells. A total of 1,493 differentially expressed genes (DEG) were identified in both sh1 and sh2 cells compared with shGFP cells (P < 0.05), including 932 downregulated genes (Fig. 2A and B) and 561 upregulated genes (Supplementary Table S6). Top 20 pathways for the genes downregulated in PRMT1-knockdown cells were shown in GO enrichment analyses, in which the cell-cycle pathway was most significant (Fig. 2C). A bioinformatics analysis with Enrichr (39) was carried out and revealed that cell-cycle genes were enriched in the downregulated genes provided in Supplementary Table S6, including CCNA2, CCNB1, CCND1, and CCNE2, CDK6, TP53, RBL1, CDC20, and CDC23 genes (Fig. 2D). Representative data from Integrative Genomics Viewer 2.0 (IGV2.0) showed that the expression of cyclin D1 was decreased in sh1 and sh2 cells compared with shGFP cells, with confirmation of the downregulation of PRMT1 and no changes in PRMT5 and CARM1/PRMT4 (Fig. 2E). We next focused on cyclin D1 due to its crucial role in cell proliferation. Western blotting and real-time RT-PCR assays revealed that the mRNA and protein expression of cyclin D1 were decreased by PRMT1 knockdown in sh1 and sh2 cells (Fig. 2F and G). To verify the role of PRMT1 in the transcription level of cyclin D1, a luciferase reporter assay was employed to examine the effect of PRMT1 on cyclin D1 promoter activity driven by a DNA fragment of -806/+133 of the gene. The reporter assay showed that cyclin D1 transcriptional activity in MDA-MB-231 cells was induced by ectopically expressed PRMT1 but not by overexpression of the catalytically inactive mutant PRMT1-E153Q (Fig. 2H). Accordingly, we also conducted the reporter assay in PRMT1 knockdown cell lines. Consistently, knocking down PRMT1 repressed the activity of the promoters (Fig. 2I). ChIP assays were performed by using anti-FLAG antibody from FLAG-PRMT1-WT or -E153Q–transfected MDA-MB-231 cells, and the result showed that both ectopic WT-PRMT1 and E153Q-PRMT1 occupied on the promoter of cyclin D1 (Fig. 2J). Our findings suggested that PRMT1 regulates the expression of cell-cycle genes dependent on its enzyme activity, but not on its occupancy of the promoter.
PRMT1 knockdown affects the expression of cell-cycle genes. A, The Venn diagram of overlapping downregulated genes in two independent shPRMT1 MDA-MB-231 cell lines (sh1 and sh2) compared with shGFP control. B, Heatmap of the downregulated genes (852) overlapping between sh1 and sh2. Each column represents one indicated sample, and each row indicates a transcript. Expression values are depicted as a ratio relative to cells expressing shGFP and are represented as a red–blue color scale. P < 0.05. C, GO analysis of shPRMT1 downregulated genes showing the top 20 significant pathways in PRMT1-knockdown cells. P < 0.05. D, The bioinformatics analysis with Enrichr was carried out. Enriched terms, columns; input genes, rows; and the cells in the matrix indicate if a gene is associated with a term. E, IGV browser views of RNA-seq reads covering indicated genes in shPRMT1 (sh1 and sh2) and shGFP MDA-MB-231 cells. The x-axis indicates the genomic location, and y-axis represents the normalized scale of RNA-seq reads. PRMT5, CARM1, and GAPDH were used as negative controls, GDF15 as a positive control. F and G, The expression of indicated genes was measured by Western blot (F) and RT-qPCR (G) analysis in shPRMT1 (sh1 and sh2) or shGFP MDA-MB-231 cells. H, The promoter activity of cyclin D1 was tested using luciferase reporter assay in HEK293T cells. Cells were transfected with cyclin D1 promoter-luciferase reporter, Renilla, and PRMT1-WT, -E153Q, or vector control. I, Reporter gene assay showing a decreased luciferase activity in PRMT1 knockdown cells (sh1 and sh2) or shGFP cells transfected with reporter construct carrying cyclin D1 promoter and Renilla. Each reporter plasmid was transfected at least three times, and each sample was assayed in triplicate. J, ChIP-qPCR analysis of specific binding of PRMT1-WT or -E153Q to the promoter (−10/+100) of cyclin D1 in MDA-MB-231 cells. −3400/−3300 region of the gene served as a negative control. IgG was used as an IP control. Data of 2G-J are the mean value ±SD from at least three independent experiments (#, P > 0.05; **, P < 0.01).
PRMT1 knockdown affects the expression of cell-cycle genes. A, The Venn diagram of overlapping downregulated genes in two independent shPRMT1 MDA-MB-231 cell lines (sh1 and sh2) compared with shGFP control. B, Heatmap of the downregulated genes (852) overlapping between sh1 and sh2. Each column represents one indicated sample, and each row indicates a transcript. Expression values are depicted as a ratio relative to cells expressing shGFP and are represented as a red–blue color scale. P < 0.05. C, GO analysis of shPRMT1 downregulated genes showing the top 20 significant pathways in PRMT1-knockdown cells. P < 0.05. D, The bioinformatics analysis with Enrichr was carried out. Enriched terms, columns; input genes, rows; and the cells in the matrix indicate if a gene is associated with a term. E, IGV browser views of RNA-seq reads covering indicated genes in shPRMT1 (sh1 and sh2) and shGFP MDA-MB-231 cells. The x-axis indicates the genomic location, and y-axis represents the normalized scale of RNA-seq reads. PRMT5, CARM1, and GAPDH were used as negative controls, GDF15 as a positive control. F and G, The expression of indicated genes was measured by Western blot (F) and RT-qPCR (G) analysis in shPRMT1 (sh1 and sh2) or shGFP MDA-MB-231 cells. H, The promoter activity of cyclin D1 was tested using luciferase reporter assay in HEK293T cells. Cells were transfected with cyclin D1 promoter-luciferase reporter, Renilla, and PRMT1-WT, -E153Q, or vector control. I, Reporter gene assay showing a decreased luciferase activity in PRMT1 knockdown cells (sh1 and sh2) or shGFP cells transfected with reporter construct carrying cyclin D1 promoter and Renilla. Each reporter plasmid was transfected at least three times, and each sample was assayed in triplicate. J, ChIP-qPCR analysis of specific binding of PRMT1-WT or -E153Q to the promoter (−10/+100) of cyclin D1 in MDA-MB-231 cells. −3400/−3300 region of the gene served as a negative control. IgG was used as an IP control. Data of 2G-J are the mean value ±SD from at least three independent experiments (#, P > 0.05; **, P < 0.01).
C/EBPα associates with and is methylated by PRMT1
C/EBPα has a well-established role as a repressor of the pro-proliferative E2F target genes and has been shown to play a tumor suppressive role in breast cancer (40, 41). Interestingly, our PRMT1 knockdown RNA-seq data revealed an enrichment of E2F-binding sites in the promoters of downregulated genes (Supplementary Fig. S2A; Supplementary Table S6). We hypothesized that PRMT1 might directly regulate C/EBPα activity and we first performed a tandem affinity purification combined with mass spectrometry from FLAG-tagged C/EBPα-overexpressing HeLa cells. We indeed identified PRMT1 as a potential partner of C/EBPα in this setting (Fig. 3A; Supplementary Table S7). To confirm the association between C/EBPα and PRMT1, we then performed a co-IP assay with HEK293T cells cotransfected with FLAG-PRMT1 and Myc-C/EBPα. C/EBPα blotted with anti-Myc was found to coexist with PRMT1 IPed by anti-FLAG (Fig. 3B). In addition, pulldown assays showed that FLAG-PRMT1 was pulled down by purified GST-C/EBPα but not by GST alone (Fig. 3C), which indicated that PRMT1 interacts with C/EBPα. Most importantly, in vitro methylation assays showed that recombinant C/EBPα was methylated by PRMT1 based on 3H labeling but not by PRMT1-E153Q, a catalytically inactive mutant of PRMT1 (Fig. 3D). Furthermore, a commercial antibody against asymmetric dimethyl-arginine (ASYM25, Rme2) was employed to detect the methylation of C/EBPα. There was a strong signal for FLAG-C/EBPα immunoprecipitated from cells but not for C/EBPα purified from E. coli (Supplementary Fig. S2B), and the methylation signal was diminished in the presence of a PRMT1-specific inhibitor (Fur, furamidine dihydrochloride; Supplementary Fig. S2C), which confirmed its specificity for arginine methylation. Knockdown of PRMT1 significantly inhibited arginine methylation of C/EBPα in cells (Fig. 3E).
C/EBPα interacts with and is methylated by PRMT1. A, Immunopurification and mass spectrometry analysis of C/EBPα binding proteins. Cell extracts from HeLa cells ectopic expressing FLAG (Ctrl) or FLAG-C/EBPα were purified with anti-FLAG M2 beads and eluted with FLAG peptide. The elution was resolved by SDS-PAGE and silver-stained, the protein bands were excised and analyzed by mass spectrometry. B and C, The interaction between PRMT1 and C/EBPα was detected by Co-IP and GST-pulldown assays. B, FLAG immunoprecipitates from HEK293T cells expressing FLAG-PRMT1 and Myc-C/EBPα were analyzed by Western blotting with anti-FLAG and anti-Myc antibodies. C, GST-pulldown assay was performed using FLAG-PRMT1 containing HEK293T whole-cell lysates incubated with GST or GST-fused C/EBPα, followed by Western blotting with anti-FLAG and anti-GST antibodies. D, Autoradiography of in vitro methylation assay using purified C/EBPα and PRMT1-WT or PRMT1-E153Q from HEK293T cells ectopically expressed with indicated plasmids. Total amounts of C/EBPα and PRMT1 are shown by Coomassie Brilliant Blue (C.B.B) staining. E, Detection of C/EBPα methylation by Western blotting using anti-Rme2. C/EBPα was immunoprecipitated with anti-FLAG M2 beads from PRMT1-knockdown MDA-MB-231 cells expressed FLAG-C/EBPα. F, Schematic drawing of the wild-type and truncated FLAG-C/EBPα fragments. G, Domain mapping of C/EBPα regions methylated by PRMT1. FLAG-tagged C/EBPα delete truncations were expressed in HEK293T cells and purified by anti-FLAG M2 agarose. Methylation of these truncations was analyzed by Western blotting using anti-Rme2. H, Mass spectrum analysis of the chymotryptic C/EBPα peptide mixture to identify methylated sites in vivo. FLAG-tagged C/EBPα was purified from HEK293T cells cotransfected with C/EBPα and PRMT1. Red, methylated arginine of C/EBPα. I, Western blotting assay by anti-Rme2 with a series of arginine mutants of FLAG-C/EBPα purified from HEK293T cells transfected with indicated vectors. Less exposed means exposing shorter time.
C/EBPα interacts with and is methylated by PRMT1. A, Immunopurification and mass spectrometry analysis of C/EBPα binding proteins. Cell extracts from HeLa cells ectopic expressing FLAG (Ctrl) or FLAG-C/EBPα were purified with anti-FLAG M2 beads and eluted with FLAG peptide. The elution was resolved by SDS-PAGE and silver-stained, the protein bands were excised and analyzed by mass spectrometry. B and C, The interaction between PRMT1 and C/EBPα was detected by Co-IP and GST-pulldown assays. B, FLAG immunoprecipitates from HEK293T cells expressing FLAG-PRMT1 and Myc-C/EBPα were analyzed by Western blotting with anti-FLAG and anti-Myc antibodies. C, GST-pulldown assay was performed using FLAG-PRMT1 containing HEK293T whole-cell lysates incubated with GST or GST-fused C/EBPα, followed by Western blotting with anti-FLAG and anti-GST antibodies. D, Autoradiography of in vitro methylation assay using purified C/EBPα and PRMT1-WT or PRMT1-E153Q from HEK293T cells ectopically expressed with indicated plasmids. Total amounts of C/EBPα and PRMT1 are shown by Coomassie Brilliant Blue (C.B.B) staining. E, Detection of C/EBPα methylation by Western blotting using anti-Rme2. C/EBPα was immunoprecipitated with anti-FLAG M2 beads from PRMT1-knockdown MDA-MB-231 cells expressed FLAG-C/EBPα. F, Schematic drawing of the wild-type and truncated FLAG-C/EBPα fragments. G, Domain mapping of C/EBPα regions methylated by PRMT1. FLAG-tagged C/EBPα delete truncations were expressed in HEK293T cells and purified by anti-FLAG M2 agarose. Methylation of these truncations was analyzed by Western blotting using anti-Rme2. H, Mass spectrum analysis of the chymotryptic C/EBPα peptide mixture to identify methylated sites in vivo. FLAG-tagged C/EBPα was purified from HEK293T cells cotransfected with C/EBPα and PRMT1. Red, methylated arginine of C/EBPα. I, Western blotting assay by anti-Rme2 with a series of arginine mutants of FLAG-C/EBPα purified from HEK293T cells transfected with indicated vectors. Less exposed means exposing shorter time.
To determine which region of C/EBPα is methylated by PRMT1, we purified a series of C/EBPα deletion mutants from transfected HEK293T cells (Fig. 3F). Western blotting with Rme2 showed that full-length C/EBPα (Fig. 3G, lane 1), and ΔbZIP (1–281 aa), were methylated in IPed C/EBPα (Fig. 3G, lane 2), but no signals were detected for the fragments of 1–120 aa and bZIP (282–358 aa; Fig. 3G, lanes 4 and 5). A weak signal was detected for the p30 fragment (120–358 aa), suggesting that arginine methylation by PRMT1 mainly occurs in the 120–281 aa region. Next, we carried out immunoprecipitation of C/EBPα with anti-FLAG from transfected HEK293T cells (Supplementary Fig. S2D), followed by mass spectrometric analysis to identify the arginine methylation sites on C/EBPα. Three arginine residues (R35, R156, and R165) of C/EBPα were found to be monomethylated (Supplementary Fig. S2E–S2G), while only R156 was dimethylated (Fig. 3H). Western blotting with Rme2 showed that the methylation signal decreased markedly when these two or three arginine residues were individually substituted with lysine (R2K, R156/165K; R3K, R35/156/165K), and each of the single mutants R35K, R156K, or R165K also displayed a decreased signal at least to some extent (Fig. 3I). These results demonstrated that C/EBPα could be methylated by PRMT1.
C/EBPα represses the transcription of cyclin D1 via HDAC3
Because of the inhibitory effect of C/EBPα on cell growth, an MDA-MB-231 cell line stably expressing C/EBPα could not be constructed. Transiently expressing C/EBPα in MDA-MB-231 cells significantly inhibited cell growth according to colony formation assays (Fig. 4A) and MTS assays (Fig. 4B). FACS assays indicated that ectopic C/EBPα induced G1–S arrest in MDA-MB-231 cells compared with the vector control (Fig. 4C). Western blotting showed that ectopically expressed C/EBPα repressed cyclin D1 expression (Fig. 4D). Knockdown of C/EBPα was accompanied by activation of cyclin D1 at the mRNA level (Supplementary Fig. S3A). Reporter assays showed that ectopic C/EBPα markedly reduced cyclin D1 promoter activity in HEK293T cells (Fig. 4E). To further analyze the domains of C/EBPα that were implicated in repressive activity, we constructed a series of deletion mutants of C/EBPα in which the TAD1, TAD2, TAD3, or bZIP domain was removed individually or in combination. HEK293T cells were then transfected by the cyclin D1 reporter along with the indicated C/EBPα truncations or an empty vector as a control. The result showed that removal of TAD1 or TAD3 did not repress cyclin D1 to the same extent as full-length C/EBPα, while TAD2 did (Fig. 4F), implying that TAD1 and TAD3 are more essential than TAD2 for repression of cyclin D1 activity. Moreover, the bZIP deletion of C/EBPα abolished its repressive activity (Fig. 4F), which indicated that the DNA binding activity of C/EBPα is essential for repression of cyclin D1. To identify the response element for C/EBPα on the cyclin D1 promoter, a series of 5′ truncated promoter-luciferase constructs were cotransfected with C/EBPα (Fig. 4G, right), and then the luciferase activities were measured. We found that the −808/+133, −304/+133, −113/+133, or −8/+133 fragment but not the −113/+15 fragment of the promoter was repressed by ectopically expressed C/EBPα (Fig. 4G, left), which indicated that there is a critical response element for C/EBPα-mediated repression within the +15/+133 region of the cyclin D1 gene. Analyzing the DNA sequence in the +15/+133 region revealed a consensus C/EBPα binding site, AGTTGCAAA, at +72/+80 in the cyclin D1 gene regulatory region, which is identical with the respective sequences of the cyclin D1 genes in mouse and in rat as well (Supplementary Fig. S3B). Mutation of this site to AGTGTCCGA in a reporter driven by the −8/+133 promoter eliminated C/EBPα-mediated transcriptional repression (Fig. 4H). Furthermore, EMSA was performed by using synthesized 32-mer oligonucleotides (+62/+93 of the cyclin D1 gene) labeled with biotin at the 5′end as probe. As a specific competitor, the unlabeled probe eliminated the specific band (S), and adding anti-C/EBPα antibody to the incubation medium clearly elicited a novel supershift band (SS) at the top of lane 5 (Fig. 4I). ChIP data confirmed that C/EBPα could efficiently target the region (−10/+100) downstream from the transcriptional start site of the cyclin D1 gene (Fig. 4J). The results demonstrated that C/EBPα can directly bind to the promoter of the cyclin D1 gene, which is critical for C/EBPα-mediated negative regulation.
C/EBPα downregulates cyclin D1 transcription by recruiting HDAC3. A and B, Effect of C/EBPα overexpression on cell proliferation was measured by colony formation assay (A) and MTS assay (B) in MAD-MB-231 cells overexpressed with vector (Ctrl) or C/EBPα plasmids. C, Cell-cycle profile analysis by flow cytometry in MAD-MB-231 cells transfected with vector (Ctrl) or C/EBPα expression construct. The percentage of cells in G1, S, and G2–M phases is shown. D, The expression of the indicated proteins in MAD-MB-231 cells transfected with C/EBPα expression plasmid was analyzed by Western blotting assay. E, Reporter gene assays were performed in HEK293T cells transfected with luciferase reporter plasmid containing cyclin D1 promoter and different amounts of plasmids expressing C/EBPα. F, The effects of truncated C/EBPα on reporter gene assays. The expression of truncations of C/EBPα is shown at bottom. G, Reporter gene assays were performed to map the cis-acting elements of C/EBPα in cyclin D1 promoter. HEK293T cells were cotransfected with C/EBPα or control plasmid, together with the indicated cyclin D1-luciferase reporter plasmids. Left, schematic diagrams of the cyclin D1-luciferase reporter constructs. H, Luciferase activities were measured in HEK293T cells transfected with reporter vectors carrying wild-type or site-mutated cyclin D1 promoter with an empty vector (Ctrl) or C/EBPα-expressing plasmid. I, EMSA assay using biotin-labeled probe containing a potential C/EBPα-binding site from cyclin D1 promoter was performed with purified C/EBPα from HEK293T cells ectopic expressing FLAG-C/EBPα. S, shift band of C/EBPα-DNA complex; SS, supershifted band with C/EBPα antibody; free probe, migration of free probe. ChIP-qPCR analysis of C/EBPα (J) or HDAC3 (K) recruitment on indicated region of cyclin D1 promoters in MDA-MB-231 cells transfected with FLAG-C/EBPα or HDAC3 plasmid. IgG was used as an IP control. Data of C, E–H, J, and K are the mean ± SD from at least three independent experiments. #, P > 0.05; *, P < 0.05; **, P < 0.01.
C/EBPα downregulates cyclin D1 transcription by recruiting HDAC3. A and B, Effect of C/EBPα overexpression on cell proliferation was measured by colony formation assay (A) and MTS assay (B) in MAD-MB-231 cells overexpressed with vector (Ctrl) or C/EBPα plasmids. C, Cell-cycle profile analysis by flow cytometry in MAD-MB-231 cells transfected with vector (Ctrl) or C/EBPα expression construct. The percentage of cells in G1, S, and G2–M phases is shown. D, The expression of the indicated proteins in MAD-MB-231 cells transfected with C/EBPα expression plasmid was analyzed by Western blotting assay. E, Reporter gene assays were performed in HEK293T cells transfected with luciferase reporter plasmid containing cyclin D1 promoter and different amounts of plasmids expressing C/EBPα. F, The effects of truncated C/EBPα on reporter gene assays. The expression of truncations of C/EBPα is shown at bottom. G, Reporter gene assays were performed to map the cis-acting elements of C/EBPα in cyclin D1 promoter. HEK293T cells were cotransfected with C/EBPα or control plasmid, together with the indicated cyclin D1-luciferase reporter plasmids. Left, schematic diagrams of the cyclin D1-luciferase reporter constructs. H, Luciferase activities were measured in HEK293T cells transfected with reporter vectors carrying wild-type or site-mutated cyclin D1 promoter with an empty vector (Ctrl) or C/EBPα-expressing plasmid. I, EMSA assay using biotin-labeled probe containing a potential C/EBPα-binding site from cyclin D1 promoter was performed with purified C/EBPα from HEK293T cells ectopic expressing FLAG-C/EBPα. S, shift band of C/EBPα-DNA complex; SS, supershifted band with C/EBPα antibody; free probe, migration of free probe. ChIP-qPCR analysis of C/EBPα (J) or HDAC3 (K) recruitment on indicated region of cyclin D1 promoters in MDA-MB-231 cells transfected with FLAG-C/EBPα or HDAC3 plasmid. IgG was used as an IP control. Data of C, E–H, J, and K are the mean ± SD from at least three independent experiments. #, P > 0.05; *, P < 0.05; **, P < 0.01.
To further elucidate the cofactors recruited by C/EBPα to mediate transcriptional inhibition of cyclin D1, we focused on the HDAC family of corepressors based on our previous report that the recruitment of HDAC3 by C/EBPα at its response element in the SIRT7 promoter is sufficient and prerequisite for repression of the SIRT7 gene (36). We found that transfection of HDAC1, HDAC2 or HDAC3 reduced cyclin D1 promoter activity in HEK293T cells (Supplementary Fig. S3C) and that cotransfection with C/EBPα led to a more effective repression of the cyclin D1 promoter compared with transfection of C/EBPα alone (Supplementary Fig. S3D). Importantly, ChIP assays further proved that HDAC3 can be recruited to the promoter of cyclin D1 (Fig. 4K). Taken together, these data suggest that C/EBPα directly represses cyclin D1 gene expression, at least, in part, by recruiting HDAC3 to its promoter.
PRMT1-mediated methylation negatively regulates the transcriptional activity of C/EBPα
On the basis of above results, we hypothesized that enhanced PRMT1 could disturb the transcriptional repressive activity of C/EBPα in breast cancer cells.
To confirm this hypothesis, we first transfected expression vectors encoding C/EBPα and/or PRMT1 in HEK293T cells and found that C/EBPα-mediated repression of the cyclin D1 promoter was impaired by coexpressing PRMT1 (Fig. 5A). Colony formation assay showed that knock down of C/EBPα promoted the growth of MDA-MD-231 breast cancer cells, importantly, knock down of C/EBPα partially rescued the effect of PRMT1 knocking down in sh1 cells (Fig. 5B). When arginine 156 was mutated to phenylalanine (R156F), a mimic of methylated arginine (42), the repressive activity of C/EBPα on the cyclin D1 promoter was reduced compared with wild-type C/EBPα, while the R156K mutant retained repressive activity (Fig. 5C). The combined mutant C/EBPα-3RF (R35F, R156F, and R165F) further disturbed C/EBPα-mediated repression of the cyclin D1 reporter (Fig. 5D, lane 4), especially compared with 3RK (R35K, R156K, and R165K;Fig. 5D, lane 6). Cotransfection with PRMT1 had no effect on either 3RF or 3RK mutant (Fig. 5D). We next examined the growth-inhibitory effects of the C/EBPα mutants in MDA-MB-231 cells. The cells transfected with the R156F mutant grew faster than those transfected with wild-type C/EBPα, while cells transfected with the R156K mutant did not (Fig. 5E). The data indicated that PRMT1 mediates methylation of C/EBPα on R156 to disturb the transcriptional repressive activity of C/EBPα on the cyclin D1 gene in breast cancer cells.
PRMT1-mediated C/EBPα methylation promotes the disassociation of HDAC3 from C/EBPα. A, Overexpression of PRMT1 impaired the inhibitory activity of C/EBPα in HEK293T cells by luciferase reporter gene assay. B, Colony formation assay in MDA-MB-231 cells double knocking down of PRMT1 and C/EBPα. C, Luciferase activities were measured using reporter assay in HEK293T cells cotransfected with expression vectors of wild-type C/EBPα (WT) or mutated C/EBPα (R156K and R156F) together with cyclin D1 reporter. D, Reporter gene assay in HEK293T cells transfected with expression plasmids of wild-type (WT) or mutated C/EBPα (3RF, 3RK), or together with or without PRMT1. E, Colony formation assay in MDA-MB-231 cells transfected with wild type C/EBPα (WT) or mutated C/EBPα of R156F and R156K (left). F, Representative Western blot analysis showing 8 pairs of breast cancer (T) and adjacent noncancer tissues (N). Anti-C/EBPα antibody and GAPDH were used as control (top). G, C/EBPα bound to a specific sequence within cyclin D1 promoter. EMSA assay using biotin-labeled probe from cyclin D1 promoter was performed with purified C/EBPα from HEK293T cells overexpressing wild-type (−WT) or mutated C/EBPα (−3RK, −3RF). S, shift band of C/EBPα-DNA complex; SS, supershifted band with C/EBPα antibody; free probe, migration of free probe. H, Domain mapping of C/EBPα region interaction with HDAC3. Myc-HDAC3 was coexpressed with FLAG-C/EBPα or its deletions in HEK293T cells and immunoprecipitated by anti-Myc antibody conjugated to agarose. Coprecipitated C/EBPα deletions were detected by Western blotting using anti-FLAG antibody. Bottom, input. I, Co-IP assay was performed using HEK293T cells transfected with FLAG-C/EBPα-WT or C/EBPα-3RF and Myc-HDAC3 plasmids. IPed without (Input; bottom) or with anti-FLAG (top) and then immunoblotting with anti-HDAC3 or anti-FLAG individually. J, Co-IP assay of C/EBPα and HDAC3 proteins from HEK293T cells cotransfected with FLAG-HDAC3, Myc-C/EBPα, with PRMT1-WT or PRMT1-E153Q expressing vectors. IPed without (Input; bottom) or with anti-Myc (top) and then immunoblotting with anti-HDAC3, anti-Myc, or anti-FLAG for PRMT1. K, ChIP-reChIP assay showing that C/EBPα-mediated HDAC3 recruitment to the promoter of cyclin D1 is arginine methylation dependent. MDA-MB-231 cells were cotransfected with FLAG- C/EBPα-WT, -R156F/K, or -3RF and Myc-HDAC3. Anti-FLAG antibody was used as the initial ChIP (1st) to gain the C/EBPα-associated chromatin fragments. Then, these fragments were subjected to reChIP (2nd) using anti-Myc antibody. IgG was used as a ChIP control. Data of A–D and J are the mean ± SD from at least three independent experiments. #, P > 0.05; **, P < 0.01.
PRMT1-mediated C/EBPα methylation promotes the disassociation of HDAC3 from C/EBPα. A, Overexpression of PRMT1 impaired the inhibitory activity of C/EBPα in HEK293T cells by luciferase reporter gene assay. B, Colony formation assay in MDA-MB-231 cells double knocking down of PRMT1 and C/EBPα. C, Luciferase activities were measured using reporter assay in HEK293T cells cotransfected with expression vectors of wild-type C/EBPα (WT) or mutated C/EBPα (R156K and R156F) together with cyclin D1 reporter. D, Reporter gene assay in HEK293T cells transfected with expression plasmids of wild-type (WT) or mutated C/EBPα (3RF, 3RK), or together with or without PRMT1. E, Colony formation assay in MDA-MB-231 cells transfected with wild type C/EBPα (WT) or mutated C/EBPα of R156F and R156K (left). F, Representative Western blot analysis showing 8 pairs of breast cancer (T) and adjacent noncancer tissues (N). Anti-C/EBPα antibody and GAPDH were used as control (top). G, C/EBPα bound to a specific sequence within cyclin D1 promoter. EMSA assay using biotin-labeled probe from cyclin D1 promoter was performed with purified C/EBPα from HEK293T cells overexpressing wild-type (−WT) or mutated C/EBPα (−3RK, −3RF). S, shift band of C/EBPα-DNA complex; SS, supershifted band with C/EBPα antibody; free probe, migration of free probe. H, Domain mapping of C/EBPα region interaction with HDAC3. Myc-HDAC3 was coexpressed with FLAG-C/EBPα or its deletions in HEK293T cells and immunoprecipitated by anti-Myc antibody conjugated to agarose. Coprecipitated C/EBPα deletions were detected by Western blotting using anti-FLAG antibody. Bottom, input. I, Co-IP assay was performed using HEK293T cells transfected with FLAG-C/EBPα-WT or C/EBPα-3RF and Myc-HDAC3 plasmids. IPed without (Input; bottom) or with anti-FLAG (top) and then immunoblotting with anti-HDAC3 or anti-FLAG individually. J, Co-IP assay of C/EBPα and HDAC3 proteins from HEK293T cells cotransfected with FLAG-HDAC3, Myc-C/EBPα, with PRMT1-WT or PRMT1-E153Q expressing vectors. IPed without (Input; bottom) or with anti-Myc (top) and then immunoblotting with anti-HDAC3, anti-Myc, or anti-FLAG for PRMT1. K, ChIP-reChIP assay showing that C/EBPα-mediated HDAC3 recruitment to the promoter of cyclin D1 is arginine methylation dependent. MDA-MB-231 cells were cotransfected with FLAG- C/EBPα-WT, -R156F/K, or -3RF and Myc-HDAC3. Anti-FLAG antibody was used as the initial ChIP (1st) to gain the C/EBPα-associated chromatin fragments. Then, these fragments were subjected to reChIP (2nd) using anti-Myc antibody. IgG was used as a ChIP control. Data of A–D and J are the mean ± SD from at least three independent experiments. #, P > 0.05; **, P < 0.01.
To investigate the molecular mechanisms by which methylation of C/EBPα by PRMT1 regulates C/EBPα repressive activity on cyclin D1, we first detected the expression of C/EBPα in the samples of breast cancer as described in Fig. 1C and found that the expression of C/EBPα was not reduced in tumor samples (T) compared with adjacent normal tissues (N; Fig. 5F). EMSA was then performed by using purified WT- and 3RF-C/EBPα from transfected HEK293T cells. The mimic of arginine methylated C/EBPα (3RF) did not affect its DNA-binding activity on the cyclin D1 promoter comparing with the WT-CEBPα, implying that the recruitment of cofactors might be affected (Fig. 5G). We found that HDAC3 interacted with C/EBPα in the 118–281 aa region (Fig. 5H), in which the methylation sites R156/165 modified by PRMT1 are located. Furthermore, compared with WT-C/EBPα, ectopic 3RF-C/EBPα disassociated with HDAC3 (Fig. 5I). Compared with WT-PRMT1, the presence of E153Q-PRMT1 resulted in an increased association of HDAC3 and C/EBPα (Fig. 5J). Most intriguingly, ChIP/reChIP assays demonstrated that C/EBPα cooccupied the promoter of cyclin D1 with HDAC3; the R156F or 3RF mutant of C/EBPα diminished the cooccupancy with HDAC3, while R156K did not affect the occupancy (Fig. 5K). These data indicated that HDAC3 preferentially interacts with unmethylated C/EBPα and that this interaction is weakened once C/EBPα is methylated by PRMT1, subsequently leading to attenuation of the transcriptional repressive activity of C/EBPα on the cyclin D1 promoter.
Loss of function of PRMT1 inhibits cell growth and tumorigenesis
We found that both mRNA and protein expression of PRMT1 were much lower in sh1 and sh2 cells than in cells mock-transfected with shRNA against GFP (shGFP; Fig. 6A). Knockdown of endogenous PRMT1 in MDA-MB-231 cells significantly decreased cell growth in the colony formation assay and MTS assay (Fig. 6B and C). The FACS assay indicated that PRMT1 knockdown resulted in G1–S arrest compared with the shGFP control (Fig. 6D). Knockdown of PRMT1 in MDA-MB-231 cells attenuated cell migration ability in the wound-healing assay (Fig. 6E) and decreased anchorage-independent growth of cells in the soft-agar colony formation assay (Fig. 6F). To assess the role of PRMT1 in breast tumorigenesis in vivo, we implanted shPRMT1 or shGFP MDA-MB-231 cells in female BALB/c nude mice. The results indicated that PRMT1 knockdown led to dramatic reductions of tumor volume and weight (Fig. 6G). Overall, these results suggest that PRMT1 plays an oncogenic role in breast cancer cells. To explore the possibility of using a PRMT1 inhibitor to treat breast cancer, we selected two triple-negative breast cancer cell lines (TNBC1 and TNBC2) for treatment of Fur, a specific inhibitor of PRMT1 (35), and found that cancer cell growth was significantly inhibited by the inhibitor in a concentration-dependent manner (Fig. 6H). This result indicates that PRMT1 inhibitor is a potential drug for treatment of human breast cancer.
Knockdown or inhibitor of PRMT1 restrains cell growth and tumorigenesis of breast cancer cells. A, RT-PCR and Western blotting analysis were performed to validate the knockdown efficiency of PRMT1 in MDA-MB-231 cells (sh1 and sh2). RT-PCR, relative PRMT1 expression levels were normalized to GAPDH and are represented as percentage of shGFP (top). Western blotting, GAPDH served as internal reference (bottom). B, Colony formation assay shown in shPRMT1 (sh1 and sh2) or shGFP MDA-MB-231 cells. C, MTS assays shown (sh1 and sh2) in shPRMT1 or shGFP MDA-MB-231 cells. D, Cell-cycle analysis of MDA-MB-231 cells with PRMT1 knockdown (sh1 and sh2). The percentage of cells at different phases (G1, M, and G2–M) is shown. E, Wound-healing assays were performed to value the migration ability of MDA-MB-231 cells with PRMT1 knockdown (sh1 and sh2; n = 6). Images present the wound closure rate at different time points (0, 12, 24 hours). F, Soft-agar assays show anchorage-independent growth ability of PRMT1 knockdown cells (sh1 and sh2) compared with shGFP cells. G, Xenograft model in athymic mice injected with shPRMT1 (sh1 and sh2) or shGFP MDA-MB-231 cells (n = 8). Tumors were excised from the mice and weighed. H, Proliferation ability was assessed in primary breast cancer cells treated with increasing concentrations of PRMT1 inhibitor (0, 20, 40, or 60 μmol/L) for 6 days. I, Schematic of C/EBPα methylated by PRMT1 to regulate cyclin D1 expression via preventing HDAC3 recruitment in human breast cancer. Data of D and F are the mean ± SD from at least three independent experiments. **, P < 0.01. Data of E and G are the mean ± SD, and statistical significance is shown.
Knockdown or inhibitor of PRMT1 restrains cell growth and tumorigenesis of breast cancer cells. A, RT-PCR and Western blotting analysis were performed to validate the knockdown efficiency of PRMT1 in MDA-MB-231 cells (sh1 and sh2). RT-PCR, relative PRMT1 expression levels were normalized to GAPDH and are represented as percentage of shGFP (top). Western blotting, GAPDH served as internal reference (bottom). B, Colony formation assay shown in shPRMT1 (sh1 and sh2) or shGFP MDA-MB-231 cells. C, MTS assays shown (sh1 and sh2) in shPRMT1 or shGFP MDA-MB-231 cells. D, Cell-cycle analysis of MDA-MB-231 cells with PRMT1 knockdown (sh1 and sh2). The percentage of cells at different phases (G1, M, and G2–M) is shown. E, Wound-healing assays were performed to value the migration ability of MDA-MB-231 cells with PRMT1 knockdown (sh1 and sh2; n = 6). Images present the wound closure rate at different time points (0, 12, 24 hours). F, Soft-agar assays show anchorage-independent growth ability of PRMT1 knockdown cells (sh1 and sh2) compared with shGFP cells. G, Xenograft model in athymic mice injected with shPRMT1 (sh1 and sh2) or shGFP MDA-MB-231 cells (n = 8). Tumors were excised from the mice and weighed. H, Proliferation ability was assessed in primary breast cancer cells treated with increasing concentrations of PRMT1 inhibitor (0, 20, 40, or 60 μmol/L) for 6 days. I, Schematic of C/EBPα methylated by PRMT1 to regulate cyclin D1 expression via preventing HDAC3 recruitment in human breast cancer. Data of D and F are the mean ± SD from at least three independent experiments. **, P < 0.01. Data of E and G are the mean ± SD, and statistical significance is shown.
Discussion
In contrast to C/EBPβ, there have been few reports of PTMs of C/EBPα. Regarding C/EBPα, the early discovered PTMs mainly included phosphorylation (26–29), SUMOylation (30–32), and acetylation (33). Although C/EBPα is expressed in a number of tissues, its function has largely been characterized in adipocytes and the hematopoietic system (16). Here, we demonstrated that the transcriptional factor C/EBPα interacts with and is methylated by PRMT1. Methylated C/EBPα loses its repressive function on cellular growth and cyclin D1 transcription via interfering with the recruitment of HDAC3 to the cyclin D1 promoter in breast cancer. The schematic model is summarized in Fig. 6I. This report is the first to demonstrate that C/EBPα arginine methylation can regulate its function.
C/EBPα contains three independent transactivation elements named TAD1, TAD2, and TAD3, which can cooperate with one another to achieve protein function (15, 43). TAD1 and TAD2 interact with the basal transcriptional apparatus (44, 45). TAD3 contains a transcriptional regulatory domain that can inhibit the activity of an adjacent transcriptional activation domain (15). In addition, TAD3 plays an important role in modulating C/EBPα function, and it recruits cofactors or allows PTMs to enhance, repress, or even convert the activity of C/EBPα. We demonstrated that C/EBPα inhibits the transcription of cyclin D1 and that the TAD1 and TAD3 domains but not TAD2 are required for this activity. HDAC3 binding to the TAD3 domain participates in the repressive process mediated by C/EBPα.
Deregulation of C/EBPα can occur at the posttranscriptional level through PTMs. ERK1/2-mediated phosphorylation of C/EBPα at Ser 21 inhibits its activity in FLT3-activated human AML (46). C/EBPα phosphorylation on Ser 193 is required for the growth-inhibitory function in hepatic carcinoma (29). We found that elevated PRMT1 in breast cancer is a novel regulator of C/EBPα transcriptional activity and that methylated C/EBPα results in loss of control of the cell cycle in breast cancer cells.
C/EBPα modulates cell growth by regulating the expression of several proliferation-related genes. Our data showed that overexpression of C/EBPα leads to a decrease in the expression levels of cyclin D1 in breast cancer cells. A previous study showed that C/EBPα reexpression in gastric carcinoma cells leads to a reduction of cyclin D1 (47), but the mechanism has not been described in detail. We identified one consensus C/EBPα binding site in the cyclin D1 promoter. Cyclin D1 is a direct target gene of transcriptional repression by C/EBPα and that inhibition of cyclin D1 is at least partially required for C/EBPα-mediated growth arrest. Previous reports from our lab and others have shown that C/EBPα causes repression of target genes through the recruitment of an HDAC-dependent repression complex (36, 48). This study revealed that HDAC3 is involved in C/EBPα-mediated repression of cyclin D1.
PRMT1 has been implicated in various types of human tumors, including breast cancer (1, 3), however, the underlying mechanisms have not been completely elucidated. In this report, we find that PRMT1 is dramatically upregulated in primary breast cancer samples, and patients with higher PRMT1 expression show a higher malignancy grade. Silencing PRMT1 in MDA-MB-231 cells leads to significant cell-cycle arrest and a reduced ability of transformation and migration, suggesting that PRMT1 plays a key role in breast tumorigenesis. On the basis of RNA-seq data, we determined that PRMT1 knockdown in breast cancer cells can activate or repress several proliferation-related genes, including cyclin D1. We showed that PRMT1 methylates the tumor suppressor C/EBPα, and methylation of C/EBPα at least partially alleviates its inhibitory effects on its target gene, cyclin D1.
In breast cancer, CARM1, another protein arginine methyltransferase, methylates chromatin remodeler BAF155 to enhance tumor progression and metastasis (49). Here, we have shown that C/EBPα can be specifically methylated by PRMT1 and that this methylation functions to impair the transcriptional repressive activity of C/EBPα by decreasing its interaction with an HDAC3-containing corepressor complex. These findings further confirm the oncogenic role of PRMT1 and the tumor suppressor function of C/EBPα during breast tumorigenesis, and thus potential drugs targeting PRMT1-C/EBPα pathway may have utility as breast cancer therapeutics.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Authors' Contributions
Conception and design: Y. Zhang, L.-M. Liu, M.-B. Cheng
Development of methodology: L.-M. Liu, W.-Z. Sun, X.-Z. Fan, M.-B. Cheng
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): L.-M. Liu, W.-Z. Sun, Y.-L. Xu
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): Y. Zhang, L.-M. Liu, X.-Z. Fan, M.-B. Cheng
Writing, review, and/or revision of the manuscript: Y. Zhang, L.-M. Liu, W.-Z. Sun, M.-B. Cheng
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): Y. Zhang
Study supervision: Y. Zhang
Acknowledgments
Y. Zhang received grants from National Natural Science Foundation of China (91519301, 31871310) and CAMS Initiative for Innovative Medicine (2016-I2M-3-002). This work was supported in part by the CAMS Initiative for Innovative Medicine (2017-I2M-3-009 to M. Cheng).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
References
Supplementary data
PRMT1 methylated C/EBPα