PRMT5 is upregulated in breast cancer cells under hypoxia

PRMT5 has been reported to be upregulated in several tumor types [Kim and Ronai, 2020; Sapir and colleagues, 2021]. However, the influence of hypoxia upon PRMT5 gene regulation remains unexplored. To address this caveat, we first analyzed a publicly available patient microarray data obtained from TCGA (GSE76250) where in the samples were stratified as hypoxia low or hypoxia high based on hypoxia signature gene expression [7]. Among all the PRMTs, we found PRMT5 to be the most significantly upregulated gene under hypoxia. (Fig 1A and 1B). Additionally, we performed a Spearman’s correlation test to observe the correlation of PRMT5 gene expression with 10 classical hypoxia gene markers (P4HA1, ENO2, NEDD4L, HK2, LDHA, FOXO3, SLC2A1, BHLHE40, PGGK1, BNIP3L) using TCGA breast cancer tumor gene sets (BRCA Tumor). We observed a significant positive correlation (R = 0.42) between the gene expression of PRMT5 and hypoxia markers (Fig 1C) indicating that hypoxia regulates PRMT5 gene expression in human breast cancers. Next, to validate PRMT5 expression under hypoxia, we performed quantitative RT-PCR (qRT-PCR) and immunoblotting to understand PRMT5 gene and protein expression in two breast cancer cell lines MCF7 and MDA-MB-231 subjected to hypoxia. We found a significant upregulation in PRMT5 gene (Fig 1D) and protein expression (Fig 1E) under hypoxia in both MCF7 and MDA-MB231 breast cancer cell lines. PRMT5 symmetrically dimethylates H3R8 and H4R3 [19,22,30]. Hence, we performed immunoblotting analysis to check the expression of PRMT5-mediated histone modifications H3R8me2s and H4R3me2s under hypoxia. We observed an increase in PRMT5-mediated histone marks H3R8me2s and H4R3me2s under hypoxia in both MCF7 and MDA-MB-231 cell lines (Fig 1E) indicating that PRMT5 upregulation could have functional implication via histone modifications under hypoxia. To extend our observations to clinical samples, we checked the PRMT5 transcript and protein levels along with the PRMT5-mediated modifications H4R3me2s and H3R8me2s in patient-derived breast cancer cell line BC8322, subjected to hypoxia. We observed a significant increase in PRMT5 mRNA (Fig 1F) as well as protein expression along with the corresponding increase in histone marks H4R3me2s and H3R8me2s under hypoxia (Fig 1G). Next, we performed IHC-F to detect PRMT5 expression in tumor sections derived from breast cancer patients. CA9 was used as a marker for hypoxia-positive tumor regions. Our immunostaining analysis showed a positive correspondence of CA9 and PRMT5 expression. Evidently, we found that regions with negative CA9 staining, indicative of non-hypoxic regions, showed reduced PRMT5 expression (Figs 1H and S1A). We also observed a significant positive correlation between PRMT5 and CA9 fluorescent staining indicating that PRMT5 is upregulated under hypoxia (Fig 1I) suggesting that PRMT5 upregulation is clinically relevant and can have implications in governing tumor progression. Together, these results indicated that PRMT5 is upregulated in breast cancer cells under hypoxia.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. PRMT5 is upregulated under hypoxia in breast cancer cells and clinical patient samples.

A) Graph showing Log2 signals of all PRMTs in breast cancer patient microarray data (GSE76520) stratified as Hypoxia low vs. Hypoxia high. B) PRMT5 mRNA expression in breast cancer patient microarray (GSE76520) stratified as Hypoxia low vs. Hypoxia high. C) Spearman’s correlation analysis of PRMT5 mRNA expression with 10 hypoxia marker genes. D) PRMT5 mRNA expression in MCF7 and MDA-MB-231 breast cancer cell lines subjected to hypoxia. E) Immunoblot showing PRMT5, H4R3me2s, and H3R8me2s expression under hypoxia in MCF7 and MDA-MB-231 cell lines. F) PRMT5 mRNA expression in patient derived cell line BC8322 subjected to hypoxia. G) Immunoblot showing PRMT5, H4R3me2s and H3R8me2s expression under hypoxia in in patient derived cell line BC8322. H) IHC-F analysis showing PRMT5 and CA9 expression in tumor sections of breast cancer patients. I) Pearson correlation analysis between PRMT5 and CA9 fluorescence intensities calculated from IHC-F. Scale bar 20 µm. Error bars, mean ± SEM; two-tailed t test, *p < 0.05, **p < 0.01, n = 3 biological replicates. Numerical data of (A–C), (F), (H), (I) available in S1 Data, sheet “Figure 1”.

https://doi.org/10.1371/journal.pbio.3003444.g001

CTCF is responsible for PRMT5 upregulation under hypoxia

Following our observation that PRMT5 is upregulated under hypoxia, we next wanted to identify the factors responsible for regulation of PRMT5 expression under hypoxia. Since we observed a significant increase in PRMT5 mRNA expression, we hypothesized that PRMT5 is transcriptionally regulated under hypoxia. To understand the mechanism of transcriptional regulation of PRMT5, we generated 4 sequential promoter deletion constructs spanning the PRMT5 promoter up to −2000 bp upstream of the TSS with 500 bp deletions in each construct. These promoter constructs were cloned upstream of firefly luciferase (Fluc) gene in a pGL3-basic vector (Fig 2A). Following this, we performed luciferase assay in both MCF7, and MDA-MB231 cells subjected to hypoxia. The luciferase construct harboring the full-length promoter (+100 to 2,000 bp) segment and the 1,500 bp segment (+100 to −1,500 bp) upstream of the transcription start site showed a significant increase in luciferase activity under hypoxia in both MCF7 and MDA-MB-231 cell lines (Figs 2B and S2A). This observation was indicative of the fact that a regulatory region present between −1,000 and 2,000 bp of the PRMT5 promoter was responsible for regulating PRMT5 expression under hypoxia. We then made use of publicly available ChIP-seq datasets from ChIP-Atlas to understand which transcription factors might potentially bind between the −1,000 and −2,000 bp region of the PRMT5 promoter and regulate PRMT5 expression under hypoxia. We identified a significant CTCF-binding site within the regulatory region (−1,000 to −2,000) at −1,175 bp position of the promoter which also had a significant binding score in the analysis performed in JASPAR (S2B and S2C Fig). Following this, we verified CTCF binding at PRMT5 promoter under hypoxia using ChIP-Seq data in MCF7 cell line [31] (Fig 2C).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. CTCF is responsible for upregulation of PRMT5 under hypoxia in breast cancer cells.

A) Representation of PRMT5 promoter luciferase construct cloned upstream of F-Luc gene in pGL3 basic vector. B) Luciferase assay showing an increase in luciferase activity under hypoxia in MDA-MB-231 cells. C) Immunoblot showing decrease in PRMT5 expression upon CTCF KD under hypoxia in MCF7 cells. D) MCF7 CTCF ChIP-seq track showing CTCF enrichment at PRMT5 promoter under hypoxia. E) Schematic representation of CTCF binding site at PRMT5 promoters along with the primer positions. F) MeDIP-qPCR showing decrease in DNA methylation at PRMT5 promoter in MCF7 cells normoxia vs. hypoxia. G) CTCF Chip qPCR showing enrichment in CTCF binding at PRMT5 promoter in MCF7 cells normoxia vs. hypoxia. H) Schematic representation of the dCAS9-DNMT3A epigenetic system targeting CTCF binding site of PRMT5 promoter. I) MeDIP qPCR showing enrichment of DNA methylation at PRMT5 promoter under hypoxia post dCAS9-DNMT3A sgCTCFbs vs. sgControl transfections. J) CTCF ChIP qPCR showing reduction in CTCF binding at PRMT5 promoter under hypoxia post dCAS9-DNMT3A sgCTCFbs vs. sgControl transfection. K) Immunoblot showing decrease in PRMT5 expression upon transfection with dCAS9-DNMT3A sgCTCFbs vs. sgControl. L) Luciferase assay showing decrease in luciferase activity in promoter construct harboring a mutated CTCF binding site. Error bars, mean ± SEM; two-tailed t test, one-way ANOVA. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, n = 3 biological replicates. Numerical data of (B), (F–G), (I–J), (L) available in S1 Data, sheet “Figure 2.” Fig 2H, Created in BioRender. Shukla, S. (2025) https://BioRender.com/74kzhl6.

https://doi.org/10.1371/journal.pbio.3003444.g002

Firstly, to establish the role of CTCF in regulating PRMT5 gene expression under hypoxia, we performed shRNA-mediated knockdown of CTCF in both MCF7 and MDA-MB-231 cells. Our immunoblotting analysis revealed a significant reduction in PRMT5 protein expression upon CTCF knockdown under hypoxia in both MCF7 and MDA-MB-231 cell lines (Figs 2D and S2D). DNA methylation is known to regulate gene expression [32–34] and since CTCF binding is also sensitive to DNA methylation [4,15,35], we performed Methylation-Dependent Immunoprecipitation (MeDIP)-qRT-PCR to analyze the DNA methylation profile at the CTCF-binding site of the PRMT5 promoter (Fig 2E) to understand the role of promoter DNA methylation in regulating PRMT5 expression under hypoxia. We observed a significant decrease in DNA methylation at the CTCF-binding site of the PRMT5 promoter (Figs 2F and S2E) under hypoxia indicating that reduced DNA methylation under hypoxia could possibly allow the binding of CTCF to regulate PRMT5 gene expression. Next, to confirm the binding of CTCF at the PRMT5 promoter, we performed CTCF ChIP-qPCR assay and observed a significant increase in CTCF occupancy at the CTCF-binding site of the PRMT5 promoter in both MCF7 and MDA-MB-231 cell lines (Figs 2G and S2F) suggesting that reduced DNA methylation at the CTCF-binding site allows CTCF binding on PRMT5 promoter under hypoxia. In order to further verify the role of DNA methylation in regulating PRMT5 expression under hypoxia, we made use of the dCAS9-DNMT3A epigenetic modifier system to preferentially methylate the CTCF-binding site of the PRMT5 promoter (Fig 2H). The dCAS9-DNMT3A construct with the CTCF-binding site targeting sgRNA was transfected into MCF7 and MDA-MB-231 cells under hypoxia following which we analyzed for DNA methylation profile, CTCF binding, and PRMT5 expression. Firstly, we observed a significant increase in DNA methylation (Fig 2I) and a concomitant decrease in CTCF binding at the CTCF-binding site (Fig 2J) of the PRMT5 promoter upon transfection with the dCAS9-DNMT3A construct targeting the CTCF-binding site of the PRMT5 promoter under hypoxia. This suggested that decrease in methylation under hypoxia is necessary for CTCF binding at the PRMT5 promoter. Expectantly, our immunoblot analysis revealed a significant decrease in PRMT5 protein expression (Fig 2K) indicating that under hypoxia, decrease in DNA methylation allows CTCF binding at the PRMT5 promoter thereby regulating PRMT5 expression. Furthermore, we mutated the CTCF-binding site present in the promoter-luciferase construct (+1 to −1,500 bp) (S2G Fig) and performed luciferase assay under hypoxia to establish the role of CTCF-mediated transcriptional upregulation of PRMT5. In corroboration with our previous results, we observed a significant reduction in luciferase activity in the construct harboring mutant CTCF binding site under hypoxia (Figs 2L and S2H) suggesting that CTCF is responsible for PRMT5 upregulation under hypoxia. Put together, our observations suggested that reduction in DNA methylation at the CTCF-binding site of the PRMT5 promoter allowed CTCF binding and regulation of PRMT5 expression under hypoxia.

PRMT5 regulates EMT and invasion of breast cancer cells under hypoxia

Metastasis is responsible for more than 90% of cancer-related mortalities [36–38]. Therefore, it becomes imperative to improve our understanding of the molecular mechanisms governing the complex process of metastatic cascade in order to tackle its adverse effects. Hypoxia is known to induce EMT and invasion of cancer cells [4]. However, there have been no reports with respect to the role of PRMT5 in regulation of EMT and invasion under hypoxia. Therefore, we envisaged to understand the role of PRMT5 in regulating EMT and invasion under hypoxia.

Firstly, we performed shRNA-mediated PRMT5 knockdown under hypoxia in both MCF7 and MDA-MB231 cell lines to check the effect of PRMT5 on EMT induction. We performed qRT-PCR to check the transcript levels of the mesenchymal marker genes VIM (Vimentin) and CDH2 (N-Cadherin) along with the epithelial marker gene CDH1 (E-Cadherin) and the EMT-regulating transcription factor SNAI1 (Snail) in both MCF7 and MDA-MB-231 cell lines. We observed a significant decrease in the transcript levels of the mesenchymal markers VIM and CDH2 along with EMT transcription factor SNAI1 (Figs 3A and S3A). Epithelial marker CDH1 showed a significant increase in transcript levels upon PRMT5 downregulation under hypoxia in both MCF7 and MDA-MB-231 cell lines (Figs 3A and S3A). We then analyzed the protein levels of epithelial markers E-cad and Cytokeratin-18 in MCF7, and Cytokeratin-18 in MDA-MB-231. Protein levels of mesenchymal marker Vimentin along with EMT-regulating transcription factor Snail were also analyzed in both MCF7 and MDA-MB-231 cell lines. Our immunoblot analyses showed an increase in the epithelial marker E-Cad and Cytokeratin 18 upon PRMT5 downregulation under hypoxia in both MCF7 and MDA-MB-231 cell lines whereas the mesenchymal marker Vimentin along with the transcription factor Snail showed decreased protein expression upon PRMT5 downregulation under hypoxia in both MCF7 and MDA-MB-231 cell lines (Figs 3B and S3B). Therefore, these results suggested that PRMT5 somehow regulated EMT induction under hypoxia. As observed earlier, PRMT5-mediated histone modifications H3R8me2s and H4R3me2s were found to be upregulated under hypoxia. Therefore, we hypothesized if obliterating these histone modifications was sufficient to curb EMT induction under hypoxia. To test this, we used GSK591 which is a substrate competitive inhibitor of PRMT5 that exerts its effect by blocking the ability of PRMT5-MEP50 complex to methylate target proteins [39,40]. We treated MCF7 and MDAMB-231 cells with 5 μM GSK591 under hypoxia [21] and found a significant reduction in the histone marks H3R8me2s and H4R3me2s (Figs 3C and S3C) indicating that GSK591 specifically inhibits PRMT5 activity. Along with this, as observed earlier, our q-RT-PCR results showed a significant decrease in the transcript levels of the mesenchymal markers VIM and CDH2 along with EMT transcription factor SNAI1 whereas the Epithelial marker CDH1 showed a significant increase in transcript levels (Figs 3D and S3D). We also found that the epithelial markers E-cad and Cytokeratin 18 showed an increase in protein expression upon PRMT5 activity inhibition under hypoxia, whereas the mesenchymal marker Vimentin along with the transcription factor Snail showed decreased protein expression upon PRMT5 activity inhibition under hypoxia in MCF7 and MDA-MB-231 cell lines (Figs 3E and S3E). Together, these observations suggested that PRMT5-mediated histone modifications (H3R8me2s/H4R3me2s) could be one of the major players involved in regulation of EMT induction under hypoxia.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. PRMT5 regulates EMT and invasion under hypoxia.

A) qRT-PCR depicting mRNA expression of EMT markers in shCTRL vs. shPRMT5 MCF7 cells under hypoxia. B) Immunoblot showing protein level of EMT markers in shCTRL vs. shPRMT5 MCF7 cells under hypoxia. C) Immunoblot showing decrease in histone marks HR38me2s and H4R3me2s upon vehicle (DMSO) vs. 5 µM GSK591 treatment in MCF7 cells under hypoxia. D) qRT-PCR depicting mRNA expression of EMT markers upon vehicle (DMSO) vs. 5 µM GSK591 treatment in MCF7 cells under hypoxia. E) Immunoblot showing protein level of EMT markers upon vehicle (DMSO) vs. 5 µM GSK591 treatment in MCF7 cells under hypoxia. F) Matrigel invasion assay and its respective quantification (right side) in shCTRL vs. shPRMT5 PRMT5 KD MCF7 cells under hypoxia. G) Matrigel invasion assay and its respective quantification (right side) upon vehicle (DMSO) vs. 5 µM GSK591 treatment in MCF7 cells under hypoxia. Scale bar 2 µm. Error bars, mean ± SEM; two-tailed t test, one-way ANOVA. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, n = 3 biological replicates. Numerical data of (A), (D), (F–G) available in S1 Data, sheet “Figure 3”.

https://doi.org/10.1371/journal.pbio.3003444.g003

Induction of EMT subsequently renders epithelial cells more invasive which initially invade through the adjacent basement membrane to initiate the first step of the metastatic cascade (Cao and colleagues, 2023). Therefore, we intended to study PRMT5-mediated acquisition of invasive potential of breast cancer cells under hypoxia. We performed Matrigel invasion assay and observed a significant decrease in the invasive potential of both MCF7 and MDA-MB-231 cell lines upon PRMT5 downregulation under hypoxia (Figs 3F and S3F). Additionally, treatment of MCF7 and MDA-MB-231 cell lines with 5 μM GSK591, similarly decreased the invasive potential of both breast cancer cell lines under hypoxia (Figs 3G and S3G). In summary, our findings demonstrated that PRMT5-mediated histone modifications could possibly regulate EMT and invasion of breast cancer cells under hypoxia.

PRMT5 regulates TCF3 splicing under hypoxia

The primary goal of our study was to investigate the role of PRMT5-mediated regulation of alternative splicing via modulation of histone modifications. Since, we observed a positive correlation between reduction of PRMT5-mediated histone methylation and EMT induction upon GSK591 treatment under hypoxia, we speculated the involvement of PRMT5-mediated histone methylation in regulation of EMT. Our next step was to investigate how PRMT5-mediated histone methylation could regulate alternative splicing thereby impacting EMT induction under hypoxia.

Firstly, to understand the effect of PRMT5-mediated changes in alternative splicing events under hypoxia, we performed shRNA-mediated knockdown of PRMT5 under hypoxia in MDA-MB-231 cells followed by RNA-Seq. RNA-Seq analysis was performed and various splicing events were classified. Among the total 6,928 significant splicing events identified (FDR < 0.05 and |PSI| > 10%), 54% were cassette exon events (skipped exons), 32% were mutually exclusive exons (MXE), 4% were retained introns, and 4% and 6% were alternative 5′ splice site and alternative 3′ splice site selection events, respectively (Fig 4A). Mutually exclusive exon splicing events are often complex which produce functional transcripts of varying functions [41]. Unlike cassette exon events which can disrupt the protein sequence, MXE often results in diversification of protein function from a single transcript often implying context-dependent tight and complex regulation [42,43]. In the context of tumor progression, mutually exclusive splicing events such as PKM (metabolism) and FGFR (EMT) display how splice switching often favors aggressive tumor progression [14,16,44–46]. Therefore, in order to understand PRMT5-mediated regulation of alternative splicing in promoting EMT under hypoxia, we chose to further analyze MXE events (Fig 4B). Gene Ontology (GO) analysis of the genes undergoing MXE splicing events (FDR < 0.05 and |ΔPSI| > 10%) upon PRMT5 knockdown under hypoxia revealed regulation of apical junctions and epithelial–mesenchymal transition (EMT) to be among the top enriched categories (Fig 4C). This further suggested that PRMT5-mediated changes in alternative splicing significantly impacted EMT processes under hypoxia.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. PRMT5 regulates alternative splicing landscape under hypoxia.

A) Pie chart showing distribution of different types of significant AS events (FDR < 0.05) in shCTRL vs. shPRMT5 MDA-MB-231 cells under hypoxia. B) Distribution of ΔPSI of significant AS events in shCTRL vs. shPRMT5 MDA-MB-231 cells under hypoxia. C) Gene ontology analysis of genes showing significant MXE events upon PRMT5 KD under hypoxia. D) Venn diagram showing TCF3 and FOXM1, as novel alternatively spliced targets of PRMT5 obtained by overlapping EMT transcription factors gene set vs. gens undergoing MXE event in shCTRL vs. shPRMT5 MDA-MB-231 cells under hypoxia. E) Schematic depiction of TCF3 alternative splicing. F) qRT-PCR showing change in exon 18A/18B ratio upon vehicle (DMSO) vs. 5 µM GSK591 treatment in MCF7 cells. G) Immunoblot depicting protein expression of E12 and E47 isoform in MCF7 cells, normoxia vs. hypoxia. H) qRT-PCR showing an increase in exon 18A/18B ratio in shCTRL vs. shPRMT5 MCF7 and MDA-MB-231 cells under hypoxia. I) qRT-PCR showing a decrease in 18A/18B ratio upon PRMT5 OE in shPRMT5 MCF7 cells under hypoxia. Error bars, mean ± SEM; two-tailed t test, one-way ANOVA. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, n = 3 biological replicates. Numerical data of (A–B) (F) (H–I) available in S1 Data, sheet “Figure 4”.

https://doi.org/10.1371/journal.pbio.3003444.g004

EMT transcription factors are critical regulators of EMT induction and are essential for the initiation and progression of cancer cells to a more aggressive and metastatic phenotype [47–49]. Therefore, we overlapped our dataset of genes showing MXE events with a dataset of EMT transcription factors [50] with the aim to identify EMT factors undergoing PRMT5-mediated alternative splicing under hypoxia. Among all the EMT-regulating transcription factors, TCF3 and FOXM1 showed significant splicing changes upon PRMT5 KD under hypoxia (Fig 4D). To delineate the detailed mechanism of how PRMT5 regulated alternative splicing under hypoxia, we further selected TCF3 as our model gene system.

Transcription Factor 3 (TCF3) is a member of the basic-helix-loop-helix (bHLH) transcription factors that has been previously shown to repress E-cad expression in differentiated cells thereby promoting invasive phenotype [51]. Additionally, TCF3 gene has been shown to promote cancer invasiveness and stemness in different cancer types [52–54]. So far, there have been no reports suggesting how TCF3 splicing is affected in tumor hypoxia and its role in orchestrating EMT and invasion. Therefore, we hypothesized that PRMT5 regulates TCF3 splicing via modulating the histone methylation thereby promoting EMT and invasion under hypoxia.

TCF3 alternative splicing event involves exon 18A and exon 18B. Inclusion of exon 18A results in the production of the TCF3-18A (E12) isoform whereas, inclusion of 18B leads to the production of the proinvasive TCF3-18B (E47) isoform (Fig 4E) [51]. Our RNA-Seq analysis displayed increased inclusion of exon 18A (|ΔPSI| > 10% and FDR < 0.05) upon PRMT5 KD under hypoxia. Since these exons are mutually included in the final transcript, we wanted to analyze how exon 18A inclusion was regulated with respect to exon 18B under hypoxia. Firstly, to understand TCF3 splicing pattern and its dependency on PRMT5 in breast cancer hypoxia, we performed qRT-PCR analysis with or without GSK591 treatment using primers designed specifically against exons 18A and 18B. We sequence verified the specificity of these primers and confirmed that they amplified the intended exons only. We observed that, under hypoxia, TCF3 exon 18A/18B ratio was significantly downregulated but was rescued back upon PRMT5 activity inhibition by GSK591 indicating that exon 18A inclusion was significantly downregulated under hypoxia in both MCF7 and MDA-MB-231 cells and that PRMT5 was involved in exon 18A exclusion under hypoxia (Figs 4F and S4A). Additionally, the effect of GSK591-mediated inhibition of PRMT5 activity on TCF3 splicing was also an indication of the fact that PRMT5-mediated histone methylation could directly be involved in dictating TCF3 splicing outcomes. In order to verify the effect of splicing at the protein level, we performed immunoblotting to detect the change in expression of the TCF3 isoforms. We made use of an antibody which would detect the common region present in both the isoforms. Under hypoxia, we observed a decrease in the E12 (TCF3-18A) isoform with a concomitant increase in E47 (TCF3-18B) isoform under hypoxia (Fig 4G). This observation suggested that the splicing shift resulting in the increase of the pro-invasive E47 isoform under hypoxia was an important event and could have a decisive effect on the regulation of EMT and invasion under hypoxia. Next, we performed shRNA-mediated knockdown of PRMT5 in both MCF7 and MDA-MB-231 cell lines under hypoxia to verify PRMT5-mediated regulation of TCF3 splicing. Our qRT-PCR analysis showed an increase in exon 18A/18B ration upon PRMT5 knockdown under hypoxia (Fig 4H). To rescue this effect, we overexpressed PRMT5 in PRMT5 downregulated cells (S4B Fig) following which we performed qRT-PCR analysis to check if replenishing PRMT5 expression overturned the splicing event. As expected, overexpression of PRMT5 under hypoxia led to a significant decrease in the exon 18A/18B ratio indicating that PRMT5 restricted 18A inclusion under hypoxia (Fig 4I). Together, these results showed that PRMT5-mediated symmetric arginine histone dimethylation regulated TCF3 splicing event under hypoxia by causing exon 18A exclusion.

PRMT5-mediated H4R3me2s modification is essential and sufficient to regulate TCF3 splicing under hypoxia

Until now, we delineated that PRMT5-mediated histone methylation regulated TCF3 splicing under hypoxia. We next wanted to systematically deduce the mechanism of how PRMT5-mediated symmetric arginine dimethylation of histones affected the TCF3 splicing outcome under hypoxia. We analyzed the TCF3 splicing locus and identified an evolutionarily conserved region located in the intron between exon 18A and 18B (Fig 5A) which has been previously shown to be involved in TCF3 splicing event [55]. Therefore, we investigated if the ICR could potentially be a site which could be utilized by PRMT5 for exerting its effect on TCF3 splicing regulation. To check this, we performed H3R8me2s and H4R3me2s ChIP-qPCR analysis and checked for the enrichment of these PRMT5-mediated histone marks at the TCF3-ICR region in both MCF7 and MDA-MB-231 cells under hypoxia. We observed a significant increase in both the histone marks, H3R8me2s and H4R3me2s at the TCF3-ICR region under hypoxia (Figs 5B and S4B). This result certainly was in corroboration with our previous observation that PRMT5-mediated histone methylation indeed played a role in dictating alternative splicing of TCF3 under hypoxia. To establish that PRMT5-mediated histone methylation at the TCF3-ICR region is absolutely essential to regulate alternative splicing, we employed a dCAS9 epigenome modification system. We, for the first time generated a dCAS9-PRMT5 construct containing the full-length PRMT5 gene downstream of dCAS9 which can be used to regulate symmetric dimethylation of histones at target regions (S4C and S4D Fig). We made use of this dCAS9-PRMT5 construct to target the TCF3-ICR region and checked the effect of local histone symmetric dimethylation event on TCF3 splicing (Fig 5C). Firstly, we performed shRNA-mediated knockdown of PRMT5 to deplete endogenous PRMT5 protein following which the cells were transfected with the dCAS9-PRMT5 construct targeting the TCF3-ICR region to specifically modify the local histone symmetric dimethylation mark at the TCF3 splicing locus. We then verified if the dCAS9-PRMT5 construct was able to induce symmetric dimethylation of histones at H4R3 and H3R8 positions at the desired position of the splicing locus. Our ChIP-qPCR analysis showed that targeting dCAS9-PRMT5 at the TCF3-ICR region was able to catalyze both H4R3me2s and H3R8me2s modifications at the TCF3-ICR region (Fig 5D). Having established that the dCAS9-PRMT5 construct was specifically able to methylate the histones at the target site, we then used the construct to check if local modifications of H4R3me2s and H3R8me2s at the TCF3-ICR region was sufficient to alter TCF3 splicing event. We performed qRT-PCR analysis following dCAS9-PRMT5 transfection in PRMT5 downregulated cells to check the change in TCF3 alternative splicing pattern. We observed a significant decrease in the 18A/18B ratio upon targeting dCAS9-PRMT5 to the TCF3 splicing locus (Fig 5E) indicating that local histone arginine symmetric dimethylation catalyzed by PRMT5 at the TCF3 splicing locus is sufficient to alter the alternative splicing pattern of TCF3. Put together, these results suggested that PRMT5-mediated histone symmetric dimethylation is an important and a necessary event in the regulation of TCF3 splicing.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. PRMT5-mediated histone methylation regulates TCF3 alternative splicing under hypoxia.

A) Genome track from UCSC genome browser showing TCF3 Intronic Conserved Region (ICR). B) H3R8me2s and H4R3me2s ChIP qPCR depicting increased histone symmetric arginine dimethylation at TCF3-ICR region in MCF7 cells, normoxia vs. hypoxia. C) Schematic representation of the dCAS9-PRMT5 epigenome editing vector targeted at the TCF3-ICR region. D) H3R8me2s and H4R3me2s ChIP qPCR depicting increased histone symmetric arginine dimethylation at TCF3-ICR region after transfection with dCAS9-PRMT5-EV vs. dCAS9-PRMT5-sgICR in MCF7 shPRMT5 cells. E) qRT-PCR showing decrease in exon 18A/18B ratio after transfection with dCAS9-PRMT5-EV vs. dCAS9-PRMT5-sgICR in MCF7 shPRMT5 cells. F) MeDIP qPCR showing increase in DNA methylation at TCF3-ICR region in MDA-MB-231 cells, normoxia vs. hypoxia. G) DNMT3A ChIP qPCR showing change in DNMT3A binding at TCF3-ICR region in MDA-MB-231 cells treated with DMSO (Nx vs. Hx) or 5 µM GSK591(Hx). H) MeDIP qPCR showing decrease in DNA methylation at TCF3-ICR region in MDA-MB-231 cells after treatment with DMSO vs. 5 µM GSK591under hypoxia. I) DNMT3A ChIP qPCR showing change in DNMT3A binding at TCF3-ICR region after transfection with dCAS9-PRMT5 construct in PRMT5-depleted MCF7 cells under hypoxia. J) MeDIP qPCR showing increase in DNA methylation at TCF3-ICR region after transfection with dCAS9-PRMT5 construct in PRMT5-depleted MCF7 cells under hypoxia. K) MeCP2 ChIP qPCR depicting change in MeCP2 binding at TCF3-ICR region in MDA-MB-231 cells treated with DMSO (Nx vs. Hx) or 5 µM GSK591(Hx). L) MeCP2 ChIP qPCR depicting change in MeCP2 binding at TCF3-ICR region after transfection with dCAS9-PRMT5 construct in PRMT5-depleted MCF7 cells under hypoxia. Error bars, mean ± SEM; two-tailed t test, one-way ANOVA. *p < 0.05, **p < 0.01, ***p < 0.001, ***p < 0.0001, n = 3 biological replicates. Numerical data of (B), (D–L) available in S1 Data, sheet “Figure 5.” 5C, Created in BioRender. Shukla, S. (2025) https://BioRender.com/jcru48k.

https://doi.org/10.1371/journal.pbio.3003444.g005

Next, we aimed to investigate the mechanism by which PRMT5-mediated histone methylation resulted in TCF3 exon 18A exclusion under hypoxia. Interestingly, we identified a CpG island within the TCF3-ICR region (S4E Fig) which we hypothesized could facilitate DNA methylation and subsequently influence TCF3 splicing. Therefore, we first performed MeDIP-qPCR analysis in both MCF7 and MDA-MB-231 cells to understand the DNA methylation pattern in hypoxia at the TCF3-ICR region. We observed a significant increase in DNA methylation at the TCF3-ICR region under hypoxia (Figs 5F and S4F) indicating that PRMT5-mediated symmetric arginine dimethylation of histones was somehow responsible for subsequent DNA methylation at the TCF3-ICR region. Next, we wanted to understand how PRMT5-mediated histone methylation was responsible for DNA methylation at the ICR region. PRMT5-mediated H4R3me2s modification is known to recruit DNMT3A [30]. However, there were no reports suggesting the involvement of such mechanism in the regulation of alternative splicing. Since we observed an enrichment of H4R3me2s mark at the TCF3-ICR along with increased DNA methylation, we wanted to check if PRMT5-mediated histone arginine symmetric dimethylation of H4R3 (H4R3me2s) could recruit DNMT3A and subsequently methylate the DNA. To check this, we first performed DNMT3A ChIP-qPCR analysis in both MCF7 and MDA-MB-231 cells under hypoxia with or without GSK591 treatment to simultaneously establish the importance of PRMT5-mediated histone arginine symmetric dimethylation in recruitment of DNMT3A to the target locus. We found a significant enrichment of DNMT3A at the TCF3-ICR region under hypoxia which was significantly downregulated upon GSK591 treatment (Figs 5G and S4G). In correspondence with the decrease in DNMT3A recruitment upon GSK591 treatment, we also observed a significant downregulation in the DNA methylation at TCF3-ICR region upon ablation of PRMT5 methyltransferase activity (Figs 5H and S4H) indicating that under hypoxia, PRMT5-mediated H4R3me2s mark recruits DNMT3A to the TCF3 splicing locus and is responsible for DNA methylation. In addition, we made use of the dCAS-PRMT5 construct to analyze the importance of PRMT5-mediated histone methylation in recruiting DNMT3A followed by subsequent DNA methylation at the TCF3-ICR region. We depleted the endogenous PRMT5 under hypoxia in MCF7 cells and targeted the dCAS-PRMT5 to the ICR region and found that DNMT3A recruitment was rescued upon dCAS9-PRMT5 treatment (Fig 5I). Correspondingly, we also observed that the DNA methylation at the ICR region also was rescued (Fig 5J) indicating that the first step of PRMT5-mediated histone methylation is crucial for the recruitment of DNMT3A and subsequent DNA methylation at the TCF3-ICR region. Together, these results show that PRMT5 symmetrically dimethylates H4R3 under hypoxia which in turn recruits DNMT3A which further facilitates DNA methylation of the TCF3-ICR region. We also observed that abrogation of the histone mark (H4R3me2s) using GSK591 was sufficient to inhibit DNMT3A recruitment and DNA methylation at the TCF3-ICR region.

DNA methylation at the splicing locus attracts DNA-binding proteins resulting in altered RNA Pol II kinetics causing differential alternative splicing outcomes [15,46]. Methyl-CpG-Binding Protein 2 (MeCP2) preferentially binds to methylated DNA sequences and is known to dictate alternative splicing outcomes [56,57]. Since we observed an increase in DNA methylation at the TCF3-ICR region, we wanted to check if MeCP2 was able to bind at the TCF3-ICR region and regulate TCF3 alternative splicing event under hypoxia. Hence, we performed MeCP2 ChIP qPCR analysis with or without GSK591 treatment under hypoxia to check if there was any MeCP2 binding as a result of enriched DNA methylation at the TCF3-ICR. Certainly, we observed a significant enrichment of MeCP2 at the TCF3-ICR region under hypoxia which was abolished upon treatment with the PRMT5 inhibitor GSK591 (Figs 5K and S4I). Once again, we observed that targeting the ICR region under hypoxia in PRMT5-depleted MCF7 cells using the dCAS9-PRMT5 construct resulted in rescue of MeCP2 binding under hypoxia (Fig 5L). In order to further validate the role of MeCP2 in regulating TCF3 splicing, we performed shRNA-mediated MECP2 knockdown under hypoxia (S4J Fig) and observed changes in TCF3 splicing pattern. Our qRT-PCR results showed a significant increase in exon 18A/18B ratio upon MeCP2 downregulation under hypoxia (S4K Fig) indicating that MeCP2 is essential for exon 18A exclusion under hypoxia. Collectively, these results indicated that PRMT5-mediated H4R3me2s modification at the TCF3-ICR region recruited DNMT3A which subsequently led to DNA methylation at the TCF3-ICR region under hypoxia. Binding of MeCP2 to the methylated DNA at the TCF3-ICR resulted in the exclusion of exon 18A under hypoxia.

MeCP2 binding at the methylated TCF3-ICR region affects RNA Pol II kinetics under hypoxia

Since we demonstrated that MeCP2 binds to the methylated TCF3-ICR region, we next checked its role in altering RNA pol II kinetics, thereby affecting TCF3 splicing under hypoxia. We performed RNA pol II ChIP-qPCR analysis with or without GSK591 treatment in both MCF7 and MDA-MB-231 cell lines under hypoxia to understand the change in RNA pol II occupancy levels at the TCF3-ICR region and its dependence on PRMT5-mediated histone arginine symmetric dimethylation. We performed RNA pol II ChIP qPCR and compared the enrichment of RNA pol II at the ICR with respect to sites upstream and downstream of the ICR region. This comparison is an indicative of the change in RNA pol II kinetics and suggests that a site showing more RNA pol II occupancy is likely because of RNA pol II pausing [15,46,58]. We observed a significant increase in RNA pol II occupancy at the TCF3-ICR region compared to sites upstream and downstream of the ICR region under hypoxia. The RNA pol II occupancy further decreased at the ICR region upon GSK591 treatment in both MCF7 and MDA-MB-231 cell lines (Figs 6A and S5A). These results suggested that binding of MeCP2 at the methylated TCF3-ICR region affected RNA pol II kinetics and that PRMT5-mediated histone methylation is critical in initiating the cascade of events which follow. Additionally, our dCAS9-PRMT5-mediated targeting of ICR region in PRMT5-depleted MCF7 cells under hypoxia showed an increase in RNA pol II occupancy at the TCF3-ICR region (Fig 6B) strengthening our observation that PRMT5-mediated histone methylation at the ICR region initiates a cascade of events which eventually regulate TCF3 splicing. Further, to verify that increase in RNA pol II occupancy at the TCF3-ICR is dependent on MeCP2, we performed shRNA-mediated knockdown of MECP2 in MCF7 cells under hypoxia followed by RNA Pol II ChIP-qPCR. Our ChIP-qPCR analysis showed a significant reduction in RNA pol II occupancy at the TCF3-ICR region upon MeCP2 downregulation under hypoxia (Fig 6C) which suggested that RNA pol II elongation rate is affected by MeCP2 binding and could be a critical factor in orchestrating the TCF3 splicing event.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Change in RNA pol II kinetics regulates TCF3 splicing under hypoxia.

A) RNA pol II ChIP qPCR demonstrating change in RNA pol II occupancy at TCF3-ICR region in MCF7 cells treated with DMSO (Nx vs. Hx) or 5 µM GSK591(Hx). B) RNA pol II ChIP qPCR showing change in RNA pol II occupancy at TCF3-ICR region after transfection with dCAS9-PRMT5-EV vs. dCAS9-PRMT5-sgICR in PRMT5-depleted MCF7 cells. C) RNA pol II ChIP qPCR showing change in RNA pol II occupancy at TCF3-ICR region shCTRL (Nx vs. Hx) and shMECP2 (Hx) MCF7 cells. D) qRT-PCR showing increase in exon 18A/18B ratio in shPTBP1, MCF7 and MDA-MB-231 cells under hypoxia. E) PTBP1 PAR-CLIP qPCR showing change in PTBP1 binding on RNA of TCF3 at ICR region in MCF7 cells after treatment with DMSO vs. 5 µM GSK591under hypoxia. F) PTBP1 PAR-CLIP qPCR showing change in PTBP1 binding on RNA of TCF3 at ICR region in after transfection with dCAS9-PRMT5-EV vs. dCAS9-PRMT5-sgICR in PRMT5-depleted MCF7. G) PTBP1 PAR-CLIP qPCR showing change in PTBP1 binding on RNA of TCF3 at ICR region in shCTRL (Nx and Hx) vs. shMECP2 (Hx) MCF7 cells. H) Immunoblot depicting reduction in hnRNPH1 levels under hypoxia in MCF7 cells. I) Schematic showing the CRISPR/Cas9 strategy used for deletion of TCF3-ICR regulatory region which is important for PRMT5-mediated dictation of TCF3 alternative splicing under hypoxia. J) qRT-PCR showing increase in exon 18A/18B ratio in ICR regulatory region mutated MCF7 cells under hypoxia. Error bars, mean ± SEM; two-tailed t test, one-way ANOVA. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, n = 3 biological replicates. Numerical data of (A–G) available in S1 Data, sheet “Figure 6”.

https://doi.org/10.1371/journal.pbio.3003444.g006

In summary, we so far demonstrated that the ICR present between exons 18A and 18B serves as a site of control for TCF3 splicing event. PRMT5-mediated H4R3me2s modification at the TCF3-ICR region increases under hypoxia which then recruits DNMT3A. Recruitment of DNMT3A subsequently results in increased DNA methylation at the TCF3-ICR region which serves as a binding site for MeCP2. Binding of MeCP2 to the methylated DNA results in slower RNA Pol II elongation rate which then becomes a critical factor in determining the TCF3 splicing event.

Increased RNA Pol II occupancy at the TCF3-ICR region recruits PTBP1 to affect TCF3 splicing under hypoxia

Our observations so far suggested that under hypoxia, PRMT5-mediated symmetric arginine dimethylation of histones and subsequent DNA methylation, along with RNA pol II pausing at the TCF3-ICR region, led to the exclusion of the upstream 18A exon. Previously, it was shown that slowing of RNA Pol II could recruit a negative splicing factor, causing the exclusion of the upstream exon [59]. Therefore, we investigated whether RNA Pol II pausing at the TCF3-ICR region could possibly recruit a negative splicing factor, which could lead to the exclusion of exon 18A under hypoxia. To check this, we first performed a motif analysis of the TCF3-ICR region and found PTBP1 to have the highest binding score (S5B Fig). Therefore, we questioned whether PTBP1 could be the negative splicing factor responsible for exon 18A exclusion under hypoxia. To verify the involvement of PTBP1 in governing TCF3 splicing, we performed shRNA-mediated knockdown of PTBP1 under hypoxia in both MCF7 and MDA-MB-231 cell lines (S5C Fig) and checked if it affected TCF3 splicing. Our qRT-PCR analysis revealed a significant increase in the exon 18A/18B ratio upon PTBP1 knockdown under hypoxia (Fig 6D), indicating that the loss of PTBP1 resulted in increased exon 18A inclusion. This suggested that PTBP1 is a critical regulator of TCF3 splicing which mediates exon 18A exclusion under hypoxia. To establish that PTBP1 was indeed recruited to the TCF3-ICR region at the RNA, we performed PAR-CLIP-qPCR analysis under hypoxia with or without GSK591 treatment in MCF7 cell line. We observed a significant enrichment of PTBP1 binding on TCF3 RNA at the TCF3-ICR region, which was decreased upon GSK591 treatment (Figs 6E and S5D), suggesting that PTBP1 recruitment was also dependent on PRMT5-mediated histone arginine symmetric dimethylation. Similarly, our dCAS9-PRMT5 targeted to the ICR region in PRMT5-depleted MCF7 cells under hypoxia rescued PTBP1 binding at the ICR region (Fig 6F) further establishing that PRMT5-mediated histone methylation is crucial for regulation of TCF3 splicing under hypoxia. Having hypothesized that RNA Pol II pausing allows PTBP1 recruitment to TCF3 premRNA, our next objective was to establish the significance of RNA Pol II pausing in PTBP1 recruitment. Since we had established that MeCP2 binding at the TCF3-ICR region resulted in increased RNA Pol II occupancy, we performed shRNA-mediated knockdown of MECP2 in MCF7 cell line followed by PTBP1 PAR-CLIP-qPCR analysis to observe its subsequent effect on PTBP1 recruitment. Our PAR-CLIP-qPCR analysis revealed a significant decrease in PTBP1 enrichment on TCF3 RNA at the TCF3-ICR region in shMECP2 MCF7 cells under hypoxia (Fig 6G) indicating that RNA Pol II pausing indeed regulates PTBP1 recruitment to the TCF3-ICR splicing locus under hypoxia. Put together, these observations demonstrated that MeCP2-mediated RNA pol II pausing was a critical step which facilitated PTBP1 recruitment to the TCF3 splicing locus thereafter affecting TCF3 splicing event under hypoxia. In addition to the role of PTBP1 in regulating TCF3 splicing, a previous report from [55] underscored the importance of change in hnRNPH1 levels also being a determining factor regulating exon 18A exclusion. Therefore, we analyzed the expression status of hnRNPH1 under hypoxia to understand how it affected TCF3 exon 18A exclusion. Our immunoblotting analysis revealed a reduction in the protein levels of hnRNPH1 (Fig 6H). We speculate that reduced hnRNPH1 levels assist in exclusion of Exon18A under hypoxia. As mentioned in [55], reduced hnRNPH1 levels promote Exon18B inclusion which is consistent with our observations. Under hypoxia, PRMT5-mediated histone methylation subsequently recruits PTBP1 to the ICR region that eventually represses 18A inclusion under hypoxia. In normoxic condition or in the absence of PRMT5, PTBP1 fails to bind to ICR which allows hnRNPH1 to suppress 18B inclusion thereby promoting 18A inclusion.

Our study so far elucidated that the PRMT5-mediated symmetric arginine dimethylation of histones at the ICR, regulates TCF3 alternative splicing under hypoxia via inducing a cascade of events. Therefore, we further wanted to establish the role of this regulatory region of ICR which is important for PRMT5-mediated DNA methylation and subsequent MeCP2 binding. To achieve this, we intended to delete this ICR regulatory region endogenously to check its effect on the alternative splicing pattern of TCF3 under hypoxia. We made use of the CRISPR/Cas9 approach to excise the ICR regulatory region containing DNA methylation and MeCP2-binding sites (Fig 6I). We designed two sgRNAs flanking the ICR regulatory region (Fig 6I) which resulted in a successful deletion of our region of interest (S5E and S5F Fig). We then treated these ICR mutant cells under hypoxia and performed qRT-PCR analysis to check the effect of the deletion of the regulatory ICR region. We observed that the exon 18A/18B ratio was significantly upregulated in ICR mutant cells under hypoxia compared to wild type cells (Fig 6J) indicating that the regulatory region of ICR is important for PRMT5 to exert its effects to regulate of TCF3 alternative splicing under hypoxia. Together, these results highlighted that the regulatory region of the TCF3-ICR is indeed important for regulation of TCF3 alternative splicing under hypoxia.

Alternative splicing of TCF3 promotes EMT and invasion of breast cancer cells

Having established that PRMT5 regulates TCF3 alternative splicing under hypoxia, we next wanted to study the importance of the TCF3 alternative splicing event and its role in regulation of EMT and invasion under hypoxia. Under hypoxia, we observed that PRMT5 promoted exclusion of exon 18A and thereby leading to the production of the pro-invasive TCF3-18B (E47) isoform. TCF3-18B (E47) isoform is known to inhibit E-CAD expression promoting cell invasion [51,54,60,61]. However, there have been no reports of TCF3-mediated regulation of invasion under hypoxia. To address this, we first cloned both and TCF3-18A and TCF3-18B isoforms into pCMV-3tag-1A FLAG-tag vector (S5D Fig). Since normoxia mimics low TCF3-18B (E47) isoform condition, we overexpressed TCF3-18A (E12) and TCF3-18B (E47) isoforms independently in MCF7 cell line. We then performed immunoblotting analysis to observe the change in E-CAD protein expression. Expectantly, our immunoblotting analysis revealed a decrease in E-Cad protein levels upon TCF3-18B (E47) isoform overexpression (S6A Fig) indicating that TCF3-18B (E47) isoform induces EMT. Further to check the effect of E47 isoform on the invasion of breast cancer cells, we overexpressed FLAG-tagged TCF3-18A (E12) and TCF3-18B (E47) isoforms (Fig 6B and 6C) and performed Matrigel invasion assay in both MCF7 and MDA-MB231 cell lines. We observed that overexpression of TCF3-18B (E47) isoform led to a significant increase in invasive potential of both MCF7 and MDA-MB-231 cells indicating that TCF3-18B (E47) isoform of TCF3 gene promotes tumor cell invasion (Fig 6D). Together, these results suggested that PRMT5-mediated regulation of TCF3 alternative splicing under hypoxia, led to the production of E47 isoform which essentially promoted EMT and invasion under hypoxia. Since we had established that PRMT5-mediated H4R3me2s mark is sufficient to regulate TCF3 splicing which subsequently regulates tumor cell invasion, we asked if targeting dCAS9-PRMT5 to the TCF3 splicing locus will be sufficient to enhance invasion by altering TCF3 splicing event. We performed Matrigel invasion assay, and we observed a significant increase in cell invasion upon transfection with dCAS9-PRMT5 (S6E Fig) under hypoxia indicating that PRMT5-mediated histone modification at the TCF3-ICR splicing locus is sufficient to induce cell invasion. This observation was indicative of the importance of histone methylation in regulating TCF3 alternative splicing under hypoxia.

Ethics statement

All the breast cancer patient data and tissue sections collected for this study was approved by the Institute ethics committee of Indian Institute of Science Education and Research-Bhopal, India (IISERB/IEC/Certificate/2021-I/09) dated December 10th, 2021. The written informed consents were taken from all the patients before collecting tissue samples.

Cell culture

Human breast cancer cell lines MCF7 and MDA-MB-231 were obtained from American Type Culture Collection (ATCC). BC8322 primary cell line derived from breast cancer patient as described in [5]) was used in this study. MCF7, MDA-MB-231, BC8322 and HEK293T were cultured at 37°C, 5% CO₂ in DMEM (Invitrogen, 12,800,017, lot no. 2248833) supplemented with 10% fetal bovine serum (FBS; Sigma, F7524, lot no. BCBX8466), 100 units/ml of penicillin and streptomycin (Invitrogen, 15,140,122, lot no. 2321120), and 2 mM/l L-glutamine (Invitrogen, 25,030,081, lot no. 1917006). The cell lines used in the study were routinely tested for mycoplasma contamination using PCR based method. For GSK591 (Cayman Chemical, 18,354) treatment, 5 µM GSK591dissolved in DMSO was added to media used for culturing cells. Cells were incubated with GSK591 for a total of 24 hours. Cells were maintained at 1% O₂ for hypoxia treatment in a Ruskinn INVIVO2 hypoxia chamber.

Plasmids

The PRMT5 overexpression construct was generated by cloning the insert between BamHI and EcoRI sites of the plasmid pCMV-3tag-1a (Agilent, 240195). TCF3-18A and TCF3-18B overexpression constructs were cloned into pCMV-3tag-1a using BamHI and HindIII. For creating the dCAS9-PRMT5 construct, firstly, the DNMT3A CD was digested out of the dCAS9-DNMT3A-puroV2 plasmid (Addgene, 74407) using BamHI and FseI. Following this, MCS taken from pCMV-3tag1A was annealed and cloned in place of DNMT3A CD to create a base parent vector named as dCAS9-MCS. Subsequently, PRMT5 full-length construct was cloned into the dCAS9-MCS using BamHI and Sal1 to generate the dCAS9-PRMT5 construct. All of the constructs generated have been sequence verified using Sanger sequencing. Primers used in construction of these vectors have been listed in the supplementary S1 Table.

dCAS9-mediated epigenome editing

Guide RNAs (sgRNA) were annealed and cloned at the BbsI restriction site downstream of the U6 promoter in dCAS9-DNMT3A-PuroR_v2 (Addgene, 74,407) [4,81] or dCAS9-PRMT5 vector (This paper). Nontargeting sgRNA containing dCAS9-DNMT3A vector (Addgene, 71830) was used along with dCAS9-DNMT3A system as control. Transfections were performed in 60 or 100 mm dishes using TurboFect (Thermo Scientific, R0531). For dCAS9-DNMT3A-mediated epigenetic editing studies, cells were subjected to hypoxia, 12 h post transfection. Following hypoxia treatment, cells were harvested for either immunoblot assay, MeDIP or ChIP analysis. For dCAS9-PRMT5-mediated epigenetic editing, cells were harvested 24 h post transfection for either ChIP qPCR analysis or RNA was collected for q-RT PCR analysis.

Quantitative RT-PCR

Total RNA was extracted using TRIZol reagent (Ambion, 15,596,018, lot no. 260712) as per the manufacturer’s instructions. cDNA was synthesized using PrimeScript 1st strand cDNA Synthesis Kit (Takara, 6110A). qRT-PCR reactions were setup using KAPA SYBR FAST QPCR master mix (Sigma, KK4618) in a qTOWER³-G (Analytik Jena) qPCR machine according to the manufacturer’s protocol. The primers were designed using the IDT Primer Quest tool (https://www.idtdna.com) and are enlisted in S1 Table. Housekeeping control gene RPS16 was used to normalize using the formula 2^{(Ct Control − Ct Target)} [82] Student t test or one-way ANOVA was used to compare gene/exon expression between two and three different groups, respectively. p < 0.05 was considered statistically significant.

Immunoblotting

Urea lysis buffer (8 M urea, 2 M thiourea, 2% CHAPS, 1% DTT) supplemented with 1 × protease inhibitor cocktail (PIC; leupeptin 10–100 μM, pepstatin 1 μM, EDTA 1–10 mM, AEBSF < 1 mM) and 10 mM PMSF was used to lyse the cells. Equal concentration of protein samples was loaded for every experiment. Recommended dilutions of primary antibodies were used to incubate blots overnight at 4°C, followed by 1 hour incubation with secondary antibody. The blots were scanned using Odyssey membrane Scanning system. Quantification of the bands was performed using ImageJ. The uncropped images of all the blots are available in S1 Raw Images file. Antibodies used for immunoblotting have been listed in S4 Table.

PAR-CLIP-qPCR

PAR-CLIP was performed as described earlier [83] with slight modifications. 100 μM 4SU (4-thiouridine, Cayman Chemical, T4509) was added in the culture medium 14 h. prior to crosslinking. Cells were crosslinked at 0.15 J cm⁻² total energy of 365-nm UV light using a UV-crosslinker. Cells were scraped in 3 mL phosphate-buffered saline (PBS) and were pelleted by centrifugation at 500g for 5 min at 4°C. Following this, cells were resuspended in NP40 lysis buffer (50 mM HEPES-KOH, pH 7.5, 150 mM KCl, 2 mM EDTA-NaOH, pH 8, 1 mM NaF, 0.5% v/v NP40) supplemented with 1× PIC, 10 mM PMSF, 0.5 mM DTT, and 40 U/mL RNaseOUT (Invitrogen, 10777019). Cells were allowed to lyse in ice for 10 min and cleared at 13,000g for 15 min at 4°C. Lysate was incubated with PTBP1, or rabbit IgG antibodies conjugated to Dynabeads Protein G (Invitrogen, Cat No. 10004D) overnight at 4°C. Following antibody incubation, immunoprecipitate complex was washed with IP wash buffer (50 mM HEPES-KOH, pH 7.5, 300 mM KCl, 0.05% v/v NP40, 0.5 mM DTT (fresh), 1× PIC, 10 mM PMSF, 40 U/mL RNaseOUT) and high salt wash buffer (50 mM HEPES-KOH, pH 7.5, 500 mM KCl, 0.05% v/v NP40, 0.5 mM DTT (fresh)), 1× PIC, 10 mM PMSF, 40 U/mL RNaseOUT). 20 μL aliquots of the immunoprecipitate complex were taken for immunoblot analysis. The remaining volume was subjected to decrosslinking in Proteinase K buffer (100 mM Tris–HCl, pH 7.5, 12.5 mM EDTA-NaOH, pH 8, 150 mM NaCl, 2% v/v SDS, 1.2 mg mL⁻¹ Proteinase K) at 55°C for 30 min. TRIZol was added to the mixture and RNA was isolated as per the manufacturer’s instructions including 1 µL of Glycoblue Coprecipitant (Invitrogen, AM9515). RNA obtained was converted to cDNA with random hexamers using PrimeScript 1st strand cDNA Synthesis Kit (Takara, 6110A). qRT-PCR was performed using KAPA SYBR FAST QPCR master mix (Roche, KK4618). The enrichment of IP was normalized to its respective input using (2^{(Ct IP − Ct Input)}). Fold change was calculated relative to IgG enrichment for representation.

Chromatin immunoprecipitation (ChIP) assay

ChIP assay was performed as previously described [5]. Briefly, about 10 million cells were crosslinked using 1% formaldehyde and scraped in PBS, followed by lysis and sonication. About 10–25 μg of sheared chromatin was incubated with an antibody of interest overnight at 4°C. Immunoprecipitation was performed using Dynabeads Protein G (Invitrogen, Cat No. 10004D). The immunoprecipitated protein–DNA complexes and 5% input were analyzed by qRT-PCR using KAPA SYBR FAST master mix (Sigma, KK4618). Normalization was performed using the formula [2^{(Ct input − Ct immunoprecipitation)}]. The obtained values were further normalized to the relative rabbit IgG and control IP values. Significance between the two groups was calculated using Student t test or one-way ANOVA with a p value of <0.05 considered statistically significant. Primers used for ChIP-qPCR analysis are listed in S1 Table. Antibodies used for ChIP assay have been listed in S4 Table.

Immunohistochemistry-fluorescence (IHC-F)

This study was approved by the Institute Ethics Committee of Indian Institute of Science Education and Research-Bhopal, India (IISERB/IEC/Certificate/2021-I/09) dated December 10th, 2021. The tissue sections were obtained following the written consent from the patient. Breast tumor tissues of 5 microns size were obtained from Bansal Hospital, Bhopal, India, for immunohistochemical analysis. Briefly, the sections were deparaffinized at 65°C followed by xylene treatment and were rehydrated using ethanol gradients. Heat-induced antigen retrieval was performed in citrate buffer (pH 6.0) for 10 min. Blocking was done using 5% BSA for 1 h at RT. Primary antibodies against PRMT5 and CA9 were incubated overnight at 4°C. Subsequently, tumor sections were incubated with Alexa-Flour 555 anti-rabbit IgG secondary antibody for 1 h following which nuclear counter stain was performed using DAPI (Invitrogen, D1306, lot no. 1673432). Imaging was performed using Olympus FV3000 confocal laser scanning microscope with a 40× objective. Image analysis was performed using ImageJ software. The details of patient samples are listed in S3 Table.

Luciferase reporter assay

PRMT5 promoter fragments 500, 1,000, 1,500, and 2,000 bp upstream of TSS were amplified individually. Genomic DNA (gDNA) obtained from MCF7 served as a template. Amplified fragments were then cloned between KpnI and HindIII restriction sites of the pGL3 basic expression vector (Promega, E1751). MCF7 or MDAMB231 cells were seeded in 24-well plates and were co-transfected with different PRMT5 promoter luciferase constructs along with pRL-TK renilla luciferase plasmid (Promega, E2231) using PEI (Polyethyleneimine, 23966-100, Polysciences). Sixteen hours posttransfection, the cells were subjected to appropriate hypoxia treatment. Subsequently, the cells were lysed, and the luciferase activity was determined using GloMax- Multi Detection System (Promega) by normalizing to the Renilla luciferase activity. Primers used for amplification of different fragments are listed in S1 Table.

Site-directed mutagenesis

To generate CTCF-binding site mutant in the PRMT5 promoter luciferase construct, SDM primers containing mutations as mentioned in S1 Table were used. PRMT5-1500 promoter construct was used as a template. Following DpnI digestion and transformation, positive colonies were screened and the presence of intended mutations in the CTCF-binding site were confirmed by Sanger sequencing.

Matrigel invasion assay

Transwell with 8.0 µm pore Polycarbonate membrane insert was first coated with Matrigel (Corning, 356,230, lot no. 2010001). MCF7 or MDA-MB-231 cells (3 × 10⁴) were resuspended in serum-free media and were added onto the Matrigel layer of the insert. The insert was placed in a well containing media supplemented with 10% FBS. The whole setup was placed in a hypoxia incubator for 24–72 h. Following hypoxia incubation, cells were fixed with 4% formaldehyde and stained with crystal violet (0.05% in 10% methanol in 1× PBS). Images were taken at 20× in an inverted microscope (Olympus CKX41) and image analysis was performed using Image J. Numerical data of all replicates including the type of statistical analysis used and p values of all invasion assay analyses has been included in S1 Data.

RNA interference

The lentivirus containing small hairpin RNA (shRNA) (Sigma, Mission Human Genome shRNA Library) against target genes along with the packaging plasmids were transfected into HEK293T cells using PEI. Lentivirus was collected 24 and 48 hours posttransfection. MCF7 and MDA-MB-23 cells were seeded at a density of 3 × 10⁵ cells per well of a six-well culture plate. Lentivirus-containing media obtained from HEK293T cells was used to inoculate target cells along with 8 μg/ml polybrene (Sigma, H9268, lot no. SLBH5907V). Twelve hours postviral inoculation, the media was changed, and the cells were selected with 1 μg/ml puromycin (Sigma, P9620, lot no. 034M4008V) for 72 h following which cells were used for different assays as desired. The list of shRNA sequences used in this study are provided in S2 Table.

Methylation-Dependent Immunoprecipitation (MeDIP)

gDNA was isolated using gDNA isolation kit (Sigma, G1N70) as per manufacturer’s protocol. MeDIP was performed as described earlier [31]. About 3 μg of gDNA was sonicated using the Bioruptor sonicator (Diagenode) to obtain a fragment size approximately ranging from 100 to 400 bp. Subsequently, sonicated DNA was denatured at 95°C for 10 min and then incubated with 5methyl cytosine antibody and rabbit IgG antibody overnight at 4°C. Five percent of denatured DNA was taken as input. Immunoprecipitation was performed using Dynabeads Protein G (Invitrogen, 10004D). Decrosslinking of the immunoprecipitated complexes was performed in TE buffer with 1% SDS and Proteinase K (Invitrogen, 25,530,049) overnight at 65°C. De-crosslinked DNA and input was purified using a PCR purification kit (QIAGEN, 28,106). The immunoprecipitated samples along with input were analyzed by q-RT PCR using KAPA SYBR FAST (Sigma, KK4618). The enrichment was calculated using input as normalization factor and the following formula: 2^{(Ct input − Ct Immunoprecipitation)}. Additionally, the resultant values were normalized by the enrichment values of normal rabbit IgG. The final values are represented as mean ± SD of triplicates. The details of primers used for MeDIP qPCR are provided in S1 Table.

CRISPR/Cas9-mediated ICR regulatory region editing

To delete the ICR regulatory region necessary for PRMT5-mediated regulation of TCF3 alternative splicing, we made use of the CRISPR/Cas9 approach as reported in [84] with slight modifications. sgRNA were annealed and cloned at the BsmbI restriction site downstream of the U6 promoter in pLentiCRISPR-E (Addgene, 78852) plasmid. Two sgRNAs, one upstream of the ICR regulatory site (sg1) and one downstream of the regulatory site (sg2) were picked. HEK293T cells were transfected with equal concentration of plasmids containing sgRNA1 and sgRNA2 along with the packaging plasmids using PEI. Supernatant, containing viral particles was collected 48 h post transfection. Lentivirus-containing media obtained from HEK293T cells was used to inoculate MCF7 cells along with 8 μg/ml polybrene (Sigma, H9268, lot no. SLBH5907V). Twenty-four hours post viral inoculation, the media was changed, and the cells were selected with 1 μg/ml puromycin (Sigma, P9620, lot no. 034M4008V) for 72 h. After 72 h, gDNA was isolated and amplified using flanking primers. Amplified product was sent for sanger sequencing for Indel analysis. Indel analysis was performed using DECODR. Cells showing positive edits were subjected to hypoxia treatment and were further used for TCF3 splicing analysis. Sequences of sgRNAs and the flanking primers used for amplifying the edit region are provided in S1 Table.

RNA-seq and bioinformatic analysis

Total RNA from hypoxia-treated shPRMT5 and shCTRL cells was isolated using PureLink RNA Mini Kit (Invitrogen, 12,183,025). One microgram of total RNA was used to make cDNA libraries. Prior to library preparation, mRNA enrichment was performed using NEBNext Poly(A) mRNA Magnetic Isolation Module (E7490S). The libraries were constructed using NEBNext Ultra II RNA Library Prep Kit for Illumina (E7770S).

Paired-end (2 × 150 bp) sequencing was performed on Illumina NovaSeqX platform (MedGenome Labs Ltd, Bengaluru, India). Raw sequencing reads were aligned to the human reference genome (hg38 assembly) with STAR Aligner (version 2.7.1a) [85]. For alternative splicing analysis, aligned reads were analyzed using rMATS 4.0.2 [86]. Differential splicing analysis was performed by calculating PSI (|ΔPSI| = |PSI (shCTRL × 100) − PSI (shPRMT5 × 100) | > 3%, p < 0.05) as described previously [W. juan Li and colleagues, 2023]. Genes undergoing mutually exclusive exon events upon PRMT5 knockdown under hypoxia have been listed in S1 File. Gene ontology was performed using ShinyGo 0.80 [87]. The list of genes in highlighted categories in listed in S1 File. List of EMT transcription factors was obtained from [50]. The list of EMT transcription factors has been provided in S1 File. Spearman’s correlation analysis of PRMT5 mRNA expression with hypoxia marker genes was performed using Gene Expression Profiling Interactive Analysis 2 (GEPIA 2) using TCGA BRCA tumor dataset [88] (http://gepia2.cancer-pku.cn/#correlation). Microarray data from GSE76250 was analyzed as described earlier [7,89,90]. The breast cancer patient microarray data from GSE76250 was downloaded from https://xenabrowser.net/. These samples were then stratified as either hypoxia high or hypoxia low based on signature gene expression obtained from Hypoxia Hallmark signature from MSigDB. Samples with higher scores (Top 20%) were stratified as hypoxia high whereas samples with lower scores (Bottom 20%) were stratified as hypoxia low following which gene expression levels for PRMTs was checked. Transcription factor binding sites with stringency set at 500 of breast cells were obtained from ChIP Atlas [91] (https://chip-atlas.org). UCSC genome browser hg38 tracks were used for CpG island and conservation analysis.

ChIP-seq analysis

MCF7 cells CTCF ChIP-seq in normoxic versus hypoxic conditions was obtained from GSE216843 [4]. Raw fastq files were trimmed by Trimmoatic (v0.39) [92] with default parameters. Reads were uniquely aligned to GRCh38 using STAR (v2.7.3a) aligner [85]. The biological replicates were merged by Samtools v1.9 [93]. Peak calling was performed using MACS v2.1.2 with default parameters [94]. Significant CTCF binding peaks were scanned at PRMT5 promoter. Integrative Genomics Viewer (IGV) was used to visualize CTCF-binding peaks on PRMT5 promoter.

Statistical analysis

Data are presented as mean ± SEM unless otherwise stated. At least three independent biological replicates have been performed for each experiment. All statistical analyses were conducted using GraphPad Prism 10 software. Statistical tests used and specific p values are indicated in the figure legends.