The dual function of PRMT1 in modulating epithelial-mesenchymal transition and cellular senescence in breast cancer cells through regulation of ZEB1

Although the involvement of protein arginine methyltransferase 1 (PRMT1) in tumorigenesis has been reported, its roles in breast cancer progression and metastasis has not been elucidated. Here we identified PRMT1 as a key regulator of the epithelial-mesenchymal transition (EMT) in breast cancer. We showed that the EMT program induced by PRMT1 endowed the human mammary epithelial cells with cancer stem cell properties. Moreover, PRMT1 promoted the migratory and invasive behaviors in breast cancer cells. We also demonstrated that abrogation of PRMT1 expression in breast cancer cells abated metastasis in vivo in mouse model. In addition, knockdown of PRMT1 arrested cell growth in G1 tetraploidy and induced cellular senescence. Mechanistically, PRMT1 impacted EMT process and cellular senescence by mediating the asymmetric dimethylation of arginine 3 of histone H4 (H4R3me2as) at the ZEB1 promoter to activate its transcription, indicating the essential roles of this epigenetic control both in EMT and in senescence. Thus, we unraveled a dual function of PRMT1 in modulation of both EMT and senescence via regulating ZEB1. This finding points to the potent value of PRMT1 as a dual therapeutic target for preventing metastasis and for inhibiting cancer cell growth in malignant breast cancer patients.


PRMT1 was able to bring about the characteristics of EMT in breast cancer cells.
To determine the correlation of PRMT1 expression level with the malignancy and metastasis of breast cancer cells, we first compared the PRMT1 level in normal mammary epithelial MCF10A cell and multiple breast cancer cell lines. Western blotting revealed that the PRMT1 expression was apparently higher in cell lines known to be highly metastatic (MDA-MB-231, MDA-MB-435, BT549 and 4T1) than that in breast epithelia cancer cells (MCF7) and normal mammary epithelial cells (MCF10A) (Fig. 1a). We then explored whether PRMT1 can trigger EMT in breast cancer cells. We stably overexpressed PRMT1 in MCF10A cells (designated MCF10A-PRMT1) by using lentivirus infection, in which the PRMT1 expression level was comparable to the endogenous PRMT1 expression in MDA-MB-231 (Fig. 1b). We found that the MCF10A-PRMT1 cells displayed a dramatic change in cell morphology, characterized by the transformation from the cobblestone-like epithelial cells with tight cell-tocell adhesion, into a spindle-shaped fibroblast-like morphology with distinct cellular scattering (Fig. 1c). At the molecular level, the MCF10A-PRMT1 cells exhibited a prominent upregulation of the mesenchymal markers fibronectin and alpha-smooth muscle actin (α -SMA), with a simultaneous downregulation of the epithelial markers E-cadherin and β -catenin, at both mRNA and protein levels (Fig. 1d,e). Similar results were observed in breast epithelia cancer cell MCF7-PRMT1 (Supporting Information Fig. S1a,1b). These changes in molecular markers were further confirmed by inspecting the subcellular localization of the proteins using immunofluorescence staining. As can be seen, MCF10A-PRMT1 cells showed a loss of E-cadherin membranous cell-cell junction staining, while the mesenchymal marker fibronectin was strongly stained (Fig. 1f). In addition, a loss-of-function study in a breast cancer cell line MDA-MB-231 that expresses PRMT1 at a detectable level was performed. The results showed that knockdown of PRMT1 in MDA-MB-231 cells (MDA-MB-231-shPRMT1) augmented the expression of E-cadherin and reduced the expression of vimentin (Fig. 1g). Together, these results implicate that PRMT1 is a novel regulator of EMT in breast cancer cells.

PRMT1 promoted the migration and invasion in breast cancer cells. The EMT progress is typically
accompanied by the loss of the cell-cell contacts and the acquisition of migratory and invasive properties. We next examined if PRMT1 can affect the migration and invasion in breast cancer cells. We showed that ectopic expression of PRMT1 led to increased MMP2 and MMP9 expression (Fig. 2a). Moreover, the activities of mature proenzymes of MMP2 and MMP9 upon cleavage were significantly augmented as measured by gelatin zymography (Fig. 2b). The effect of PRMT1 on cell migration and invasion was assessed by wound healing assay and transwell chamber assay in culture media lacking serum to exclude the effect of cell proliferation. The results demonstrated that PRMT1 overexpression remarkably encouraged the MCF10A migration ability (Fig. 2c,e), whereas knockdown of PRMT1 decreased the migration of MDA-MB-231 cells (Fig. 2d,f). Similarly, the invasion assay confirmed that overexpression of PRMT1 promoted the invasion capacity (Fig. 2g), while loss of PRMT1 reduced the invasion capacity in MDA-MB-231 cells (Fig. 2h).
To verify the in vitro observations, we investigated the effect of PRMT1 on the distant metastasis of breast cancer cells in vivo in nude mice. MDA-MB-231-shPRMT1#1 cells or non-targeting control shRNA were injected Scientific RepoRts | 6:19874 | DOI: 10.1038/srep19874 into the lateral tail veins of female nude mice, and the metastatic lung tumours were examined. Strikingly, we found that the mice injected with MDA-MB-231-shPRMT1#1 cells metastasized much less efficiently than that injected with MDA-MB-231-shCtrl cells, as illustrated by the bioluminescence imaging (Fig. 2i). Noticeably, histologic observation showed that mice bearing MDA-MB-231-shPRMT1#1 cells had no visible metastases, whereas macroscopic lung metastases were found in mice transplanted with MDA-MB-231-shCtrl cells (Fig. 2j). Similar results were observed in in vivo tail vein injection experiment with metastatic murine breast adenocarcinoma 4T1 cells (Supporting Information Fig. S2a-2e). Together, these data clearly demonstrate the critical role of PRMT1 in breast cancer metastasis.
The PRMT1-induced EMT facilitated the acquisition of stem cell-like properties of breast cells. Previous evidence suggests that both disseminated normal mammary epithelial cells and breast cancer cells undergoing the EMT process may acquire stem cell properties of CD44 high /CD24 low antigen phenotype and mammosphere-formation ability 28,29 . To investigate whether PRMT1 can confer the MCF10A cells with the stem cell characteristics, we performed flow cytometry to sort the cells based on CD24/CD44 cell-surface markers to analysis cell populations with the CD44 high /CD24 low expression pattern. We found that MCF10A-PRMT1 cells exhibited a significant increase (P = 0.0013) in the CD44 high /CD24 low population compared with MCF10A-vector cells (Fig. 3a). Also, the mammosphere formation assays demonstrated an increase both in size and number of  mammospheres in MCF10A-PRMT1 cells (Fig. 3b,c). Based on these results, we conclude that PRMT1 is able to induce MCF10A cells to develop stem cell characteristics and self-renewal capability.
Depletion of PRMT1 arrested breast cancer cell growth in G1 tetraploidy and induced cellular senescence. An interesting phenomenon we have noticed in our experiments was that knockdown of PRMT1 in MDA-MB-231 cells not only attenuated the motility and invasion, but also led to a dramatic inhibition of cell growth (Fig. 4a,b). Meanwhile, we did not detect any enhancement in proliferative capacity in MCF10A cells overexpressing PRMT1 (Supporting Information Fig. S2). Analysis of DNA content by flow cytometry detected an elevated accumulation of MDA-MB-231-shPRMT1 cells in G2/M phase with 4N DNA content (Fig. 4c). Moreover, BrdU incorporation was remarkably diminished in PRMT1-depletion cells compared with control cells (Fig. 4d). Previous studies suggested that cells with 4N DNA content can also be arrested in G1 phase (G1 tetraploid cells) 30 , which emerged both in replicative and oncogenic Ras-induced senescence 31,32 . Senescent G1 tetraploid cells exhibit a dramatic downregulation of many G2/M genes, including CDC2, CCNB1 and CCNA2 33,34 . To test the hypothesis that MDA-MB-231-shPRMT1 cells have entered cellular senescence and undergone G1 arrest, we examined the activity of lysosomal senescence-associated β -galactosidase (SA-β -gal) and the expression of p21, CDC2, CCNB1 and CCNA2. The results revealed that both MDA-MB-231-shPRMT1 and MCF7-shPRMT1 cells showed a remarkably intensified SA-β -gal staining ( Fig. 4e and Supporting Information Fig. S3a). Also, depletion of PRMT1 significantly increased the p21 expression, and reduced the CDC2, CCNB1 and CCNA2 expression at both mRNA and protein levels, in contrast to the control cells (Fig. 4f,g and Supporting Information Fig. S3b). Taken together, these data indicate that silencing of PRMT1 can lead to the mitotic errors and trigger the senescence of breast cancer cells, as a result of the downregulation of G2/M genes.

PRMT1 activated ZEB1 expression through catalyzing the H4R3me2as modification at gene's promoter.
To understand the mechanism by which PRMT1 participates in breast cancer EMT and metastasis, we tested whether PRMT1 can activate the expression of the crucial EMT inducers. Interestingly, we found that only the ZEB1 expression was dramatically increased at both mRNA and protein levels, while the levels of other EMT inducers such as Twist, Snail and Slug, remained basically unchanged (Fig. 5a,b) in MCF10A-PRMT1 cells. Meanwhile, knockdown of PRMT1 in MDA-MB-231 cells remarkably decreased ZEB1 expression (Fig. 5c,d).
Since the PRMT1-mediated histone modification of H4R3me2as is linked to transcriptional activation, we intended to determine whether this modification is associated with the active transcription of ZEB1. The results of quantitative chromatin immunoprecipitation (qChIP) assay revealed a prominent elevation in the enrichment of H4R3me2as, together with the presence of PRMT1, at the region between − 327 and − 179 bp of the ZEB1 promoter in MCF10A-PRMT1 cells (Fig. 5e). As expected, less occupancy of this ZEB1 promoter region by H4R3me2as and PRMT1 was observed in MDA-MB-231-shPRMT1 cells (Fig. 5f). These results clearly indicate that PRMT1 facilitated the ZEB1 transcriptional activation by enriching the H4R3me2as modification at the ZEB1 promoter.

ZEB1 was essential for PRMT1-induced EMT, migration, invasion and acquisition of stem-cell-like properties.
To evaluate whether ZEB1 eventually plays a role in the cellular events initiated by PRMT1, we virally transfected MCF10A-PRMT1 cells with two distinct ZEB1 shRNAs to specifically silence the ZEB1 expression. Indeed, we found that knockdown of ZEB1 counteracted the effects of PRMT1 overexpression. Specifically, ZEB1 suppression reduced the levels of mesenchymal cell markers fibronectin and α -SMA that were upregulated by PRMT1 expression (Fig. 6a). Meanwhile, silencing of ZEB1 markedly restored the levels of epithelial cell markers E-cadherin and β -catenin that were downregulated upon PRMT1 expression (Fig. 6a). Our immunofluorescence study further confirmed the changes of E-cadherin and fibronectin expression upon ZEB1 depletion (Fig. 6b). Meanwhile, silencing of ZEB1 strikingly attenuated the PRMT1-induced migratory and invasive abilities (Fig. 6c,d). Furthermore, we determined that loss of ZEB1 expression resulted in a considerable decrease of CD44 high /CD24 low populations in MCF10A-PRMT1 cells (Fig. 6e), indicating that ZEB1 contributed critically to the acquisition of PRMT1-induced stem-cell-like property. These data are in support of our assumption that ZEB1 is necessary for PRMT1-induced EMT, migration, invasion and acquisition of stem-cell-like properties.
To summarize, we describe in this report a novel function of PRMT1 in modulating both EMT and cell growth in breast cancer cells. Specifically, PRMT1 is an important inducer for EMT, and is essential for the acquisition of stem cell properties and maintenance of proliferation in breast cancer cells. Loss of PRMT1 in aggressive breast cancer cells strongly blocks tumour metastasis in vivo, and inhibits breast cancer cells proliferation by inducing cellular senescence. Significantly, PRMT1 executes its function through catalyzing the H4R3me2as modification at a define region of ZEB1 promoter to activate the gene.
MB-231-shPRMT1#1/#2 cells. (g-h) Invasion assays in MCF10A-PRMT1 and MDA-MB-231-shPRMT7#1/#2 cells. Representative images of migrated and invaded cells are shown. Numbers of cells that penetrated through the Transwell chambers in the absence (e,f: migrating cells/field) or presence (g,h: invading cells/ field) of Matrigel were determined. Top panel represents the mean number of cells per field. Cells from five randomly chosen fields were counted, and each experiment was repeated at least three times. (i) Representative bioluminescence images of lung metastases in mice via tail vein injection of indicated cells, and the metastases were quantified by measuring the photo flux (mean of 5 mice). (j) Visible lung metastatic nodules are represented in the graph. Representative H&E stained lung sections are displayed in the lower panel. Scale bars: 100 μ m. Error bars, mean ± SD, **P < 0.01, ***P < 0.001.
Scientific RepoRts | 6:19874 | DOI: 10.1038/srep19874 Discussion PRMT1 is the predominant member of the PRMTs family, responsible for at least 85% of all arginine methylation 35 . The alternative splicing of the human PRMT1 pre-mRNA generates seven protein isoforms (v1-v7) with a unique N-terminal sequence 36 . The PRMT1v1 and v2 mRNA expression was elevated in both breast cancer cell lines and in breast tumour tissues compared to their normal controls, and a strong correlation between PRMT1v1 and poor patient prognosis was established 25,37 . A pervious study emphasized that overexpression of PRMT1v2 isoform promoted motility and invasion in MCF7 breast cancer cells by phosphorylation-dependent downregulation of β -catenin, based on PRMT1v2′ s cytoplasmic localization 38 . In this study, we focused on the function of the most abundant isoform PRMT1v1, which is predominantly localized to the nucleus. We found that PRMT1v1 was essential for the induction of EMT process, deciphering at least in part the action and involvement of PRMT1 in potentiating metastasis in malignant breast cancer. Moreover, PRMT1v1 was a potent regulator of E-cadherin, which is the most predominant hallmark of EMT in maintaining intact cell-cell contacts and preventing cell mobility and invasion. Presumably, the mechanisms by which PRMT1 affects breast cancer progression may vary between PRMT1v1 and PRMT1v2, regarding the distinct characteristics of enzymatic activity, substrate specificity and subcellular localization. These variations may account for the functional difference among the isoforms.
A major insight into the mechanistic action of PRMT1 made in this study is the discovery that the PRMT1-mediated histone H4R3me2as modification at the ZEB1 promoter region is responsible, at least in part, for the induction of EMT, and for the acquisition of stem cell properties and maintenance of proliferation in breast cancer cells. Human PRMT1 is an asymmetric arginine methyltransferase, and the H4R3me2as modification catalyzed by PRMT1 is a typical marker of active chromatin 22 . It has been known that, apart from histone H4, the substrates for PRMT1 cover a wide range of nonhistone proteins. A recent investigation showed that PRMT1 regulated EMT by methylating Twist1 at the arginine residue 34, which is required for E-cadherin repression in non-small cell lung cancer cells 39 . Our data revealed that, among the classic EMT-associated transcription factors (Twist, Snail, Slug and ZEB1) we tested, only the expression of ZEB1 was found to be regulated by PRMT1, and the PRMT1-mediated activation of ZEB1 was crucial for initiation of EMT process in breast cancer cells. Moreover, analysis of the primary amino acid sequences of the E-cadherin repressors revealed that both ZEB1 and ZEB2 harbored one potential PRMT methylation site (RG), whereas none was found in Snail nor in Slug 39 . Nevertheless, whether ZEB1 is a methylation substrate for PRMT1 needs to be further investigated. Besides, PRMT1 was recently identified to methylate Smad6 resulting in activation of TGFβ /Smad signaling pathway, which is also a potent inducer of EMT 40 . Apparently, multiple mechanisms may be involved in the initiation of EMT mediated by PRMT1.
Increasing evidence demonstrates that PRMT1 impacts crucial cellular pathways in cell cycle progression and cellular proliferation. Depletion of PRMT1 results in a marked attenuation in proliferation ability in a number of cancer cell lines, including osteosarcoma, breast, bladder and lung cancer cell lines 18,41,42 . However, the outcome of the anti-proliferation function of PRMT1 suppression has not been well identified. Results from this study first implicate that downregulation of PRMT1 with specific siRNAs induces senescence in breast cancer cells, as assessed by SA-β -gal staining (Fig. 4e and Supporting Information Fig. S3A). Consistent with the results from previous reports 39,41 , no typical sub-G1 peak, a common indicator of apoptosis, was observed in our experiments. Meanwhile, both MDA-MB-231-shPRMT1 and MCF7-shPRMT1 cells exhibited an increased expression of p21 (Fig. 4f,g). The cyclin dependent kinase inhibitors p21 and p16 are commonly upregulated to inhibit the Cdks, and thus to allow the accumulation of dephosphorylated pRb protein, resulting in growth arrest and senescence. Intriguingly, we found in this study that the senescence induced by PRMT1 inhibition was largely dependent on p21, but independent of p16. Indeed, both MCF7 cells (no detectable p16 expression) and MBA-MB-231 cells (deleted p16 expression) retained the ability to undergo senescence. Moreover, we also showed that the MDA-MB-231-shPRMT1 cells possessed the 4N DNA content, accompanied by a dramatic downregulation of several G2/M genes, such as cdc2, cyclin B1 and cyclin A2 (Fig. 4f,g and Supporting Information Fig. S3B). There have been indications that increased expression of p21 and formation of G1 tetraploidy senescence are often closely associated with senescence induced by cancer therapeutic agents in cancer cells 43,44 . Thus, information from this study and from others, implicates a new perspective that downregulation of PRMT1 may probably play an important role in trigger of senescence in cancer cells, although the detailed mechanisms of the action of PRMT1 in connection with p21 or mitotic errors remain to be elucidated. It has been well established that EMT not only facilitates the metastatic dissemination of tumours, but also is associated with cancer therapy resistance 45 , stem cell traits 29 , apoptosis resistance 46 and inhibition of senescence 10 . Several EMT-associated transcription factors like Snail, Twist and ZEB1, have attracted considerable research attention, due to their potentiality to contribute simultaneously to the inhibition of oncogene-induced senescence and to the promotion of EMT program 9 . An interesting finding arising from this study is that, ectopic expression of PRMT1 was able to induce EMT, whereas suppression of PRMT1 provoked senescence in breast cancer cells. We thus speculate that PRMT1 may probably function as an intermediate factor linking the two distinct cellular activities, i.e., EMT and cellular senescence, which play apparently distinct roles in tumorigenesis and cancer progression. Therefore, theoretically, PRMT1 can be targeted either to prevent cancer cell dissemination to develop metastatic lesions, or to eradicate existing metastatic cancer cells by induction of senescence. On the other hand, senescence leaves tumor cells alive and physiologically active in contrast to cell death. Senescent cells within the tumor can produce secreted factors with both tumor-promoting and tumor-suppressing activities 47 . Thus, such a treatment strategy targeted on PRMT1 might need to be combined with chemotherapy to prevent the activation of dormant metastases. However, the mechanistic insights of how PRMT1 overrides senescence and whether expression of ZEB1 can evade senescence induced by loss of PRMT1, etc., await for further investigations.
Cancer stem cells (CSCs) intrinsically contribute to dissemination for metastasis and therapeutic resistance, thus therapeutic strategies specifically targeting CSCs are rationally aimed to improve therapeutic outcome for cancer patients 44 . Data from our previous studies 45,48,49 and from others 45 have shown that EMT induces normal and neoplastic epithelial cells to generate CSC-like properties. In line with these results, we showed in this work that ectopic expression of PRMT1 in MCF-10A cells evidently elevated the percentage of cancer stem cell (CSC)-like cells that exhibit a CD44 high /CD24 low antigenic phenotype and efficient mammosphere-formation ability. CSCs with these characteristics not only constitute a small minority of neoplastic cells within a tumour, and also contribute to seed new tumours at distant sites 50 . Significantly, our work suggests that PRMT1 is a new epigenetic regulator in generation of cancer sternness, and this finding provides a basis for the development of therapeutic strategy for highly aggressive and malignant breast cancers. Plasmids and viral infection. The pcDNA3-PRMT1 expression vector was kindly provided by Dr. Akiyoshi Fukamizu (University of Tsukuba, Tsukuba, Japan). The lentiviral expression plasmid Pwpxld-PRMT1 was constructed from the cDNA sequence of pcDNA3-PRMT1 expression vector. The control, PRMT1 and ZEB1 short hairpin RNA plasmids were constructed in the pDSL-hpUGIP backbone. The lentivirus packaging vectors used were psPAX2 and pMD2.G. Generation of lentivirus in 293T cells and transfection of lentiviral constructs into recipient cell lines were performed following manufacturer's instructions (Invitrogen). The sequences of shRNAs were described in Supplementary Table S1. Stable MCF10A-PRMT1 and PRMT1/ZEB1 knockdown cells were generated by lentivirus infection.

RNA extraction, RT-PCR and real-time PCR analysis. Total RNA was extracted from cells using the
Trizol reagent (TaKaRa, Dalian, China) following manufacturer's instructions. The cDNA was generated using the Reverse Transcription System (Promega). Real-time PCR was carried out on a Roche LightCycler480 using SYBR Green Realtime PCR Master Mix (Roche). The β -actin was used as an internal control. The sequences of PCR primers were listed in Supplementary Table S1.
Immunofluorescence. Indirect immunolabeling was performed as described 52 . Cells were seeded on glass cover-slips in 12-well plates, left overnight before the treatment. Cells were fixed in 1% formaldehyde in culture medium for 10 min at 37 °C, then permeabilized with 0.2% Triton X-100 in PBS for 10min at 4 °C. Cells were then washed twice in PBS and blocked for 1 h with 5% BSA in PBS. Cells were incubated with primary antibodies at 4 °C overnight, washed three times in PBS and incubated with TRITC-conjugated secondary antibodies (Invitrogen) for 1 h at room temperature. Cell nuclei were counterstained with a 500 nM concentration of DAPI (Sigma). Photographs were taken under a confocal fluorescence microscope (Olympus Corporation, Tokyo, Japan) operating at 543nm laser spectrum.
Wound healing, transwell migration and invasion assays. Experiments were performed basically as described previously 48 . For wound-healing assay, cells were seeded at a density of 1 × 10 6 cells/well in 6-well plates. The progress of migration was photographed immediately and during 3 days after wounding (0/12/24/48/72 h) under an inverted microscope. The percentage of migration was calculated as the width of a scratch divided by the initial width of the same scratch times 100. At least five fields were analyzed for each scratch. 5 × 10 4 cells were used for each migration assay, and 5 × 10 5 cells were used for each invasion assay. After cells were incubated for 24h (migration) and 48 h (invasion), those adhered to the lower surface of the membrane were stained with 0.1% crystal violet (Sigma). Randomly selected fields were photographed and five fields per filter were counted using Image J software.