Protein arginine methyltransferase 5 (PRMT5) activates WNT/β‐catenin signalling in breast cancer cells via epigenetic silencing of DKK1 and DKK3

Abstract Protein arginine methyltransferase 5 (PRMT5) activity is dysregulated in many aggressive cancers and its enhanced levels are associated with increased tumour growth and survival. However, the role of PRMT5 in breast cancer remains underexplored. In this study, we show that PRMT5 is overexpressed in breast cancer cell lines, and that it promotes WNT/β‐CATENIN proliferative signalling through epigenetic silencing of pathway antagonists, DKK1 and DKK3, leading to enhanced expression of c‐MYC, CYCLIN D1 and SURVIVIN. Through chromatin immunoprecipitation (ChIP) studies, we found that PRMT5 binds to the promoter region of WNT antagonists, DKK1 and DKK3, and induces symmetric methylation of H3R8 and H4R3 histones. Our findings also show that PRMT5 inhibition using a specific small molecule inhibitor, compound 5 (CMP5), reduces PRMT5 recruitment as well as methylation of H3R8 and H4R3 histones in the promoter regions of DKK1 and DKK3, which consequently results in reduced expression CYCLIN D1 and SURVIVIN. Furthermore, CMP5 treatment either alone or in combination with 5‐Azacytidine and Trichostatin A restored expression of DKK1 and DKK3 in TNBCs. PRMT5 inhibition also altered the growth characteristics of breast cancer cells and induced their death. Collectively, these results show that PRMT5 controls breast cancer cell growth through epigenetic silencing of WNT/β‐CATENIN pathway antagonists, DKK1 and DKK3, resulting in up‐regulation of WNT/β‐CATENIN proliferative signalling.

challenge in the selection of adequate treatment strategies. Triplenegative breast cancers (TNBCs) being the most aggressive type, accounts for 15%-20% of all breast cancers with poor prognosis and reduced survival rate. TNBC is characterized by the absence of receptors for oestrogen (ER), progesterone (PR) as well as human epidermal growth factor receptor 2 (HER2/neu), which renders it hard to treat, and results in dismal clinical outcomes. 3 Currently, chemotherapy, alone or in combination with surgery and/or radiotherapy, is the only available option for treating TNBC; however, these approaches are associated with adverse side effects and lack of efficacy. Hence, investigations focusing on discovering novel biomarkers that can be targeted therapeutically in TNBC are needed.
In the last two decades, studies have clearly demonstrated that a combination of genetic lesions and epigenetic alteration of tumour-suppressor genes and oncogenes contributes to breast carcinogenesis. 4,5 Aberrant epigenetic events resulting in altered post-translational modification of histone and non-histone proteins, DNA methylation and dysregulated miRNA expression play a major role in breast cancer aetiology. 4,5 For example, elevated occupancy of DNA methyltransferase (DNMT) 1 and histone deacetylase (HDAC) in the promoter region of oestrogen receptor leads to hypermethylation of CpG clusters, resulting in oestrogen receptor gene inactivation in breast cancer cells. 6 Reduced expression of tumour suppressor miR200b due to DNMT3A-mediated promoter DNA hypermethylation promotes epithelial to mesenchymal transition and mammosphere formation in TNBCs. 7 As epigenetic modifications are reversible, targeting enzymes that modulate DNA methylation, and methylation and or acetylation of histone and non-histone proteins has become a promising therapeutic strategy.
PRMT5 is an important epigenetic modifying enzyme that catalyses monomethylation and symmetric methylation of arginine residues in both histone and non-histone proteins. 8,9 Symmetric methylation of specific arginine residues of various proteins by PRMT5 is considered as an essential step in biological processes such as DNA replication and repair, gene transcription, Golgi apparatus and ribosome biogenesis, protein biosynthesis and mRNA splicing. 9,10 Furthermore, extensive studies have demonstrated that PRMT5 is involved in cell-cycle control, cell migration and cell reprogramming. 10 Recent studies have shown that elevated levels of PRMT5 are associated with several cancers including mantle cell lymphoma, 11,12 metastatic melanoma, 13 epithelial ovarian cancer, 14 neuroblastoma, 15 germ cell tumours, 16 glioma 17,18 and colorectal cancers. 19 PRMT5 in association with BRG1-and hBRM-based SWI/SNF chromatin remodelling complexes induces H3R8 and H4R3 arginine symmetric methylation of promoter histones, which in turn leads to transcriptional repression of target tumour suppressor genes such as Suppressor of Tumorigenicity 7 (ST7) and Nonmetastatic 23 (NM23). 20 In addition to these tumour suppressor genes, PRMT5 also reduces transcription of RB family tumour suppressor genes including RB1, RBL1 and RBL2 in transformed B-cell lymphocytic leukaemia cell lines. 21 Furthermore, PRMT5 indirectly down-regulates the RB1/RBL2-E2F pathway by enhancing expression of CYCLIN D1 and promoting inactivation of RB1 and RBL1 through CYCLIN D1-CDK4/6 dependent phosphorylation. 22 The role played by PRMT5 in breast carcinogenesis remains underexplored. A prior study by Scoumanne et al. (2009) demonstrated that PRMT5 regulates proliferation of MCF7 cells, and that its knockdown inhibits their proliferation by inducing G1 cell-cycle arrest, indicating that PRMT5 is a key regulator of cell-cycle progression. 23

PRMT5
was also shown to associate with Programmed Cell Death Protein 4 (PDCD4) and reduce its tumour-suppressor activity in MCF7 cells.
Moreover, patients overexpressing both PRMT5 and PDCD4 show poor survival rate compared with those expressing high PDCD4 levels and low levels of PRMT5. 24 In another study by Yang et al. (2015), PRMT5 levels were found to be up-regulated in various breast cancer cells including MCF7, MDA-MB-231, MCF-10A and clinical samples of ductal carcinoma, and that its expression is positively associated with enhanced mortality. 25 More recently, PRMT5 expression was shown to be increased in breast cancer stem cells (BCSCs), and that its knock down reduces proliferation and self-renewal of BCSCs both in vitro and in vivo. 26 The mechanism by which PRMT5 regulates breast cancer stem cell function involves up-regulation of FOXP1. PRMT5 binds to the FOXP1 promoter and induces symmetrical methylation of histone H3R2, which in turn promotes recruitment of the WDR5 subunit of the SET1/MLL methyltransferase complex that is known to methylate H3K4me3, resulting in elevated expression of FOXP1.
These results indicate that PRMT5 plays an important role in maintaining breast cancer stemness. 26 Several reports have shown that dysregulation of WNT signalling is tightly linked to carcinogenesis. 27,28 WNT signalling is activated by binding of WNT ligand to Frizzled receptor and co-receptor, low-density lipoprotein receptor-related protein 5/6 (LRP5/6).
Binding of WNT to its cognate receptor inhibits formation of the cytosolic destruction complex, which is composed of AXIN1, AXIN2, protein-phosphatase-2A (PP2A), GSK3β, casein kinase1 (CK1) and adenomatous polyposis coli (APC). Consequently, cytosolic β-CAT-ENIN levels increase, which then translocates to the nucleus, interacts with T-cell factor and lymphoid enhancer factor (TCF/LEF), and activates transcription of target genes such as c-MYC, CYCLIN D1 and SURVIVIN. [29][30][31][32] WNT/β-CATENIN signalling is known to be activated in 50% of breast cancer patients, 33 and promoter DNA hypermethylation of pathway antagonists such as APC, DKK3, SFRP1 and SFRP2 has been reported in many breast cancer samples. [34][35][36] We have recently shown that PRMT5 activates WNT/β-CATENIN signalling pathway in three different types of non-Hodgkin's lymphoma cell lines, mouse primary lymphoma cell lines and clinical samples through epigenetic silencing of AXIN2 and WIF1. 37 However, the role played by PRMT5 in regulating WNT/β-CATENIN signalling in breast cancer remains unknown. Therefore, we have investigated in the current study the impact of PRMT5 inhibition on WNT/β-CATENIN signalling in triple-negative breast cancer (TNBC) cells. Our findings indicate that elevated levels of PRMT5 promote WNT/β-CATENIN proliferative signalling through transcriptional repression of WNT antagonists, DKK1 and DKK3. ChIP assay confirmed binding of PRMT5 to the promoter region of WNT antagonists, DKK1 and DKK3, and hypermethylation of promoter H3R8 and H4R3. Moreover, PRMT5 inhibition either alone or in combination with Trichostatin A (TSA) and 5-Azacytidine (5-Aza) led to transcriptional derepression of DKK1 and DKK3, and decreased expression of WNT/β-CATENIN target genes, CYCLIN D1 and SURVIVIN. These changes were also accompanied by reduced proliferation, migration, and invasion and enhanced cell death of breast cancer cells.

| Real-Time PCR
Total RNA was extracted using TRIzol reagent (Invitrogen TM ). Briefly, TNBC cells from a T75 flask were removed using 750 μL of TRIzol reagent, transferred to an eppendorf tube and incubated at room temperature for 5 minutes. Next, 300 μL of chloroform was added and samples were mixed vigorously and incubated at room temperature for 10 minutes. Samples were then centrifuged at 4°C for 15 minutes before the upper layer containing RNA was transferred to a new eppendorf tube containing 500 μL of isopropanol. After incubation at room temperature for 10 minutes, the reactions were centrifuged at 4°C for 15 minutes. The RNA pellet was rinsed with 75% ethanol, air-dried before resuspension in 20 μL of nuclease-free water. Next, mRNA was converted into cDNA using High-Capacity cDNA Reverse Transcription Kit as per manufacturer's instructions (Applied Biosystems TM ). Briefly, 2 μg of total RNA was reverse transcribed in a 20-μL reaction mixture containing 2.5 μmol/L random primers, 100 mmol/L dNTP Mix and Taqman reverse transcription reagents. To measure the mRNA levels of WNT target genes and antagonists, real-time PCR was carried out using TaqMan TM Universal PCR Master Mix in a 20-μL reaction as described previously. 11 The following primers and probe sets were used to detect PRMT5 (forward, 5'-CCAGAGCCTTGGAAGCA-3'

| Western blot analysis
Whole-cell extracts were prepared in radioimmune precipitation assay were subjected to Western blot analysis as described previously. 11 Briefly, 20 to 40 μg of total protein were separated using 7-12% SDS-PAGE and transferred to PVDF membrane. The membrane containing

| Chromatin immunoprecipitation (ChIP) assay
Chromatin immunoprecipitation was carried out as described previously. 11

| Transwell migration and invasion assays
Boyden migration assay was carried out using a 24-well plate with polycarbonate transwell inserts of pore size 8.0 μm (BD, Falcon).
Boyden invasion assay was conducted using the BD Bio-Coat Matrigel were fixed by treating them first with 4% formaldehyde, followed by absolute methanol, and then cells were stained with 1% Crystal violet.
Images from 4 random fields were captured using an inverted microscope, and cells were counted using ImageJ software.

| Proliferation assay
Approximately 2 × 10 4 cells were seeded into a 24-well plate, and 24 hours later, cells were treated with or without PRMT5 inhibitor for different time intervals. Next, cells were collected, resuspended in 20-μL solution containing equal volume of culture medium and Trypan blue dye, and counted in order to determine the number of viable cells.

| Flow cytometry
Approximately 6 × 10 4 cells were seeded in a 6-well plate prior to treatment with or without inhibitor for 24 or 48 hours. Next, cells were har-

| ELISA Assay
Levels of secreted antagonists of the WNT/β-CATENIN pathway were detected by ELISA technique using DuoSet ELISA kit (R&D system).
Essentially 100 μL of capture antibody was added to 96-well culture plate and incubated at room temperature overnight. Next, the antibody solu-

| Statistical analysis
The real-time RT-PCR and ChIP experiments were performed at least two times using different biological replicates, and the data obtained were represented as mean ± SD. Statistical validation of the data obtained from multiple samples within different groups was performed by two-way ANOVA analysis using GraphPad Prism 7 software, and P values ≤.05 were considered as statistically significant.

| PRMT5 levels and WNT/β-CATENIN target genes are up-regulated in breast cancer cell lines
Previous studies have shown that PRMT5 levels are elevated in a wide variety of cancer cells including glioblastoma, melanoma, non-small cell lung carcinoma, lymphoma and leukaemia cells. 11,13,18,38 Enhanced PRMT5 expression has also been shown in breast cancer cell lines as well as clinical samples of ductal carcinoma. 25 In light of these findings, we investigated PRMT5 expression in HER2-negative MCF7 cells, and TNBC cell lines, HCC1937 and BT549 ( Figure 1). Real-time RT-PCR revealed that PRMT5 mRNA levels were enhanced by sevenfold (P < 10 −3 ) in MCF7, 9.9-fold (P < 10 −3 ) in HCC1937 and 3.7-fold (P < 10 −3 ) in BT549 cells compared with normal human mammary epithelial cells (HMECs) ( Figure 1A). In accord with this result, Western blot analysis showed that PRMT5 protein expression was increased in breast cancer cell lines compared to normal HMECs ( Figure 1B).  Figure 1C). In agreement with work by Lin et al. (2000), which showed previously that CYCLIN D1 is not expressed in BT549 cells, we were unable to detect CYCLIN D1 mRNA in BT549 cells ( Figure 1C). 33 An initial study by Bartkova et al. (1998) investigated normal HMECs ( Figure 1D).
When we examined expression of secreted frizzled-related proteins (SFRPs), we found that SFRP2 mRNA was unaffected in MCF7 and HCC1937 cells compared with control HMECs ( Figure 1F). This was not the case in BT549 cells, which showed elevated levels of cells did not differ from normal HMECs; however, its level was elevated in HCC1937 (3.4-fold, P < 10 −3 ) and reduced in BT549 cells (6.5-fold, P < 10 −3 ) ( Figure 1F).
As both DKK1 and DKK3 showed significant reduction in their mRNA levels in both TNBCs, HCC1937 and BT549, we evaluated their protein levels in both normal and transformed breast cells.
In accord with our RT-PCR results, Western blot analysis showed that both DKK1 and DKK3 proteins were suppressed in HCC1937 and BT549 compared with normal HMECs ( Figure 1G). In contrast to TNBCs, MCF7 cells had detectable expression of DKK1 protein, which showed altered mobility by SDS-PAGE compared with normal HMECs. In addition, similar to TNBCs, DKK3 protein was significantly suppressed in MCF7 cells compared with normal HMECs ( Figure 1G). Collectively, these results indicate that expression of WNT/β-CATENIN pathway antagonists, DKK1 and DKK3, is altered in HCC1937 and BT549 cell lines, and suggest that PRMT5 might be involved in their regulation.

| PRMT5 epigenetically silences DKK1 and DKK3, and reduces their expression
In light of our recent work, which showed that PRMT5 epigenetically Immunoprecipitation of nucleoprotein complexes was performed using preimmune (PI), anti-PRMT5, anti-H3(Me 2 ) R8 or anti-H4(Me 2 )R3 antibodies, and the purified DNA was used to detect the promoter sequences of DKK1 and DKK3 by real-time PCR using specific primers and probe sets. ChIP assays were repeated two times with three technical replicates. The results are represented as mean ± SD. *** indicates P values < 10 −3 , ** indicates P values < 10 −2 and * indicates P value < 10 −1 5.6-fold (P < 10 −3 ) in the presence of 12 μmol/L of CMP5 in BT549 cells ( Figure 3A). The mRNA levels of DKK1 and DKK3 genes were unaffected in HCC1937 cells ( Figure 3B), suggesting that there might be other mechanisms involved in their regulation. As control, the mRNA levels of PRMT5 were also measured and were found to be unaffected in the presence of CMP5 ( Figure 3A,B). To determine the biological significance of DKK1 and DKK3 transcriptional derepression, we measured the levels of DKK1 and DKK3 proteins in BT549 cells treated with CMP5 ( Figure 3C). ELISA analysis showed that both with 24 μmol/L of 5-Aza first for 24 hours, followed by addition of 9 μmol/L CMP5 for another 24 hours also did not induce significant transcriptional derepression of DKK1 (1.6-fold, P < 10 −3 ) and DKK3 (1.7-fold, P < 10 −3 ) ( Figure 3D).
Deacetylation of histone lysine residues by histone deacetylase enzymes (HDAC) has been shown to be essential for transcriptional silencing of genes. 42 An earlier study by our group showed that PRMT5 can associate with mSin3A/histone deacetylase 2 (HDAC2) to form a transcriptional repression complex with the BRG1-based hSWI/SNF chromatin remodelers. 20 In addition, we have previously shown that PRMT5 associates with an NF-κB transcriptional repres- showed that mRNA level of DKK1 and DKK3 by 1.6-fold and twofold, respectively (data not shown), suggesting that combinational treatment with all three inhibitors was not able to induce further expression of DKK1 and DKK3 in these cells.
As PRMT5-induced H4R3 histone methylation is known to direct binding of DNMT3A, and induce gene silencing, 41 we tested the effect of an initial treatment with CMP5, followed by addition of TSA and 5-Aza in HCC1937 cells. We treated HCC1937 cells with 6 µmol/L of CMP5 for 4 hours, and then added 100 nmol/L of TSA and 10 µmol/L of 5-Aza for an additional 8 hours. Results from this experiment showed that DKK1 and DKK3 expression was not affected ( Figure 3E). Next, we treated HCC1937 cells with CMP5 (6 µmol/L) for 8 hours, followed by an incubation with TSA (100 nmol/L) and 5-Aza (10 µmol/L) for another 16 hours. Realtime RT-PCR analysis showed that the mRNA levels of DKK1 were enhanced by 1.9-fold (P = .023) and those of DKK3 were elevated by 3.9-fold (P < 10 −3 ) ( Figure 3E). Furthermore, when we increased time of incubation with CMP5 to 12 hours, followed by incubation with TSA and 5-Aza for an additional 24 hours, DKK1 transcriptional derepression remained at 1.8-fold (P = .0063), whereas DKK3 mRNA was derepressed by 4.5-fold (P < 10 −3 ) ( Figure 3E).
Similarly, treating CMP5 for 24 hours and then incubating with TSA and 5-Aza for next 24 hours resulted in derepression of mRNA levels of DKK1 and DKK3 by 1.9-fold (P = .0055) and threefold (P < 10 −3 ), respectively ( Figure 3E). Collectively, these results indicate that PRMT5 is an important epigenetic enzyme, which is involved either alone or in combination with other epigenetic enzymes in silencing WNT pathway antagonists, DKK1 and DKK3 in TNBC cells.

| Inhibition of PRMT5 down-regulates WNT/β-CATENIN target genes in HCC1937 and BT549 cell lines
Having found that PRMT5 inhibition can lead to transcriptional dere- F I G U R E 5 PRMT5 inhibition alters its recruitment and symmetric dimethylation of histones, H3R8 and H4R3 on promoter regions of DKK1 and DKK3. ChIP assay was performed to detect recruitment of PRMT5 and enrichment of its epigenetic marks in the promoter region of WNT/-CATENIN antagonists as described in Figure 3. Cross-linked chromatin from either DMSO-or CMP5-treated BT549 (A and B) and HCC1937 (C and D) cells was immunoprecipitated using PI or the indicated immune antibodies. DKK1 and DKK3 promoter sequences of were detected by real-time PCR using specific primers and probe sets. The experiment was repeated two times with three technical replicates, and data in each graph represent the mean ± SD. *** indicates P values < 10 −3 , and ** indicates P value < 10 −2

| PRMT5 inhibition alters its recruitment and H3R8 and H4R3 symmetric methylation in the promoter region of DKK1 and DKK3
Treatment of BT549 with PRMT5 inhibitor resulted in DKK1 and DKK3 transcriptional derepression ( Figure 3A). Therefore, we con-

| PRMT5 inhibition reduces viability of breast cancer cells
To assess the effect of PRMT5 inhibition on viability of MCF7, HCC19 and BT54937 cell lines, we incubated them with increasing amounts of  Figure 6C).

| PRMT5 is required for migration and invasion of TNBC cells, and its inhibition induces apoptosis
Enhanced migration is an intrinsic property of cancer cells, which also helps metastasis, a hallmark of TNBC. 45 The role of PRMT5 in breast cancer cell migration was studied by Boyden chamber assay.
Treatment of MCF7, HCC1937 and BT549 with CMP5 at their respective LC 50 dose, showed that migration of MCF7 and HCC1937 cells was completely inhibited compared to control DMSO-treated cells. However, migration of BT549 cells was reduced by 58.6% (P < 10 −3 ) (Figure 7A,B). To assess if PRMT5 inhibition can impact   Evan et al. (1992) showed that when expressed in the absence of proliferation signals, c-MYC induces cell death. 57 In agreement with this result, a later study by Murphy et al. (2008) showed that distinct c-MYC expression levels regulate proliferation and apoptosis. 58 While low levels of c-MYC protein promote cell proliferation, its elevated expression activates the ARF/p53 tumour-suppressor pathway, which leads to cell death. 58 In our study, we found that PRMT5 inhibition reduces viability of breast cancer cells and induces their death (Figures 6 and 7). derepression more than 1.6-to 1.7-fold ( Figure 3D). PRMT5 is also known to associate with histone deacetylases 2 and 3 in large multisubunit transcriptional repressor complexes. 12 Figure 3E).
ChIP assays showed that PRMT5 is enriched along with its epigenetic marks on the promoters of DKK1 and DKK3 in TNBC cell lines, and that its inhibition with CMP5 results in its decreased binding ( Figure 5). Our previous work showed that CMP5-mediated inhibition of PRMT5 leads to its decreased recruitment to target gene promoters, which is accompanied by reduced symmetric methylation of H3R8 and H4R3. 12,37 In agreement with these findings, our results show a similar trend of decreased PRMT5 recruitment and enrichment of its epigenetic marks in the promoter regions of DKK1 and DKK3 in both HCC1937 and BT549 cells. Therefore, it appears that the PRMT5 catalytic activity is required for its recruitment to target promoters.
Metastasis is a common characteristics of all aggressive breast tumours and contributes to poor prognosis and increased mortality. 45 A previous study by Dey et al. (2013) showed that enhanced activation of WNT/β-CATENIN signalling is positively correlated with increased metastatic potential and overall worse prognosis of breast cancer patients, highlighting the role of WNT/β-CATENIN activation in breast cancer metastasis. 60 Furthermore, inhibition of WNT/β-CATENIN activation has been shown to reduce proliferation and migration of BT549 cells, and induce their death in vitro. 61 As our study showed that PRMT5 in- is also epigenetically suppressed by PRMT5; however, the mechanism by which DKK3 down-regulates WNT signalling remains unclear. 55 Regardless, it appears that PRMT5 targets various WNT/-CATENIN pathway antagonists that operate at different levels, highlighting the complexity and diversity of its mechanism of action in the different cell types. The net outcome of PRMT5-induced epigenetic silencing is to promote growth and survival of cancer cell, which renders its ideal for therapeutic intervention, because its selective inhibition is marked by decreased expression of pro-survival proteins, CYCLIN D1 and SURVIVIN. These molecular changes trigger reduced proliferation, migration and invasion, and increased cell death, which are all desired attributes for killing tumour cells.

| CON CLUS ION
Protein arginine methyltransferase 5 (PRMT5) activity is dysregulated in many aggressive cancers and its enhanced levels are associated with increased tumour growth and survival. The current study was focused on investigating the role of PRMT5 in promoting breast tumorigenesis. The results of the study showed that PRMT5 is overex- PRMT5 inhibition also altered the growth characteristics of TNBC cells including proliferation, migration and invasion, and induced their death.

ACK N OWLED G EM ENTS
We thank staff from the department of Biological and Environmental Sif). Open access funding provided by the Qatar National Library (QNL).

CO N FLI C T O F I NTE R E S T
The authors declare that they have no conflict of interest with the contents of this article.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.