PSPC1 is a potential prognostic marker for hormone-dependent breast cancer patients and modulates RNA processing of ESR1 and SCFD2

Breast cancer is the most common cancer type among women worldwide. The majority of breast cancer expresses estrogen receptor (ER) and endocrine therapy is a standard treatment of ER-positive breast cancer. However, development of the therapy resistance is still a major challenge and thus new therapeutic approaches are needed. Here we show that an RNA-binding protein, PSPC1, play a crucial role in ER-positive breast cancer growth through post-transcriptional gene regulation. We showed that siRNA-mediated PSPC1 silencing suppressed the proliferation of ER-positive breast cancer cells. Strong immunoreactivity (IR) of PSPC1 was correlated with poor prognosis for ER-positive breast cancer patients. Using immunoprecipitation, RNA-immunoprecipitation (RIP) and quantitative PCR (qPCR) experiments, we showed that PSPC1 interacted with PSF and was involved in post-transcriptional regulation of PSF target genes, ESR1 and SCFD2. Strong SCFD2 IR was correlated with poor prognosis for ER-positive breast cancer patients and combinations of PSPC1, PSF, and SCFD2 IRs were potent prognostic factors. Moreover, we identified DDIAS and MYBL1 as SCFD2 downstream target genes using microarray analysis, and finally showed that SCFD2 silencing suppressed tamoxifen-resistant breast tumor growth in vivo. These results indicated that PSPC1 and SCFD2 axis could be a promising target in the clinical management of the disease.


-Hydroxytamoxifen
Breast cancer is the most common cancer type with the highest incidence and mortality rates among women worldwide 1,2 . The majority (70-80%) of breast cancers are hormone-dependent with expression of estrogen receptor (ER) and its target progesterone receptor (PR) 3 . Although they initially respond to endocrine therapy such as ER antagonist tamoxifen 3 , acquired endocrine therapy resistance is often occurred in hormone-dependent tumors as a risk of distant recurrence of ~ 10% during years 5-20 even among patients with small, node-negative, low-grade tumor 4 . Novel therapeutic targets are therefore required to overcome the endocrine therapy resistance. RNA-binding proteins (RBPs) are key players in gene expression, particularly at post-transcriptional levels 5 . RBPs regulate the cellular dynamics of RNAs, such as RNA processing, localization, and decay. Recent studies demonstrated that RBPs play an important role in the development and progression of various cancers, thus suggesting that RBPs may serve as new therapeutic targets for different cancers [6][7][8][9] . We previously demonstrated that PSF, an RBP belonging to Drosophila behavior human splicing (DBHS) family 10 , exerts oncogenic roles in breast and prostate cancers through post-transcriptional gene regulation 11,12 . PSF contributes to RNA processing of ESR1 and SCFD2 in ER-positive breast cancer, which promotes the cancer progression 12 . We also showed that PSF upregulates spliceosome gene expression, which plays a role in the splicing of androgen receptor (AR) in hormone-refractory prostate cancer 11 . Recent studies reported that another DBHS family RBP PSPC1 acts as a transcriptional regulator in cancers, facilitating cancer stemness, epithelial-to-mesenchymal transition (EMT), and metastasis in breast and lung cancers and hepatocellular carcinoma (HCC) 13,14 . PSPC1-mediated posttranscriptional gene regulation in cancer, however, remains to be elucidated.
Here, we demonstrate that PSPC1 is associated with poor prognosis in ER-positive breast cancer patients and involved in the post-transcriptional regulation of ESR1 and SCFD2, both were previously defined as PSF target genes by our previous study 12 . We here identified anti-apoptotic genes DDIAS and MYBL1 as downstream target genes of SCFD2 in ER-positive breast cancer cells. We further defined that combinations of strong intensities for PSPC1, SCFD2, and PSF immunostaining could be potent prognostic factors for ER-positive breast cancer patients. PSPC1 is associated with poor prognosis for ER-positive breast cancer patients. Next, to examine the clinical significance of PSPC1 in breast cancer, we performed immunohistochemical analysis of PSPC1 in 114 ER-positive breast cancer samples. PSPC1 signals were strong in 30 tumor samples, whereas weak in 84 tumor samples and normal mammary tissues (Fig. 1I-K). In correlation analysis between PSPC1 status and clinicopathological parameters, P values in correlation were 0.080 and 0.089 for stage and PSF IR status that we previously determined 12 , respectively (Table 1). In Kaplan-Meier survival analysis, PSPC1 strong IR was significantly correlated with shorter disease-free survival (Fig. 1L, P = 0.0002) and overall survival (Fig. 1M, P = 0.0023) of the patients. Intriguingly, univariate and multivariate analyses indicated that PSPC1 IR status is an independent prognostic factor of ER-positive breast cancer (Table 2). Notably, we observed that PSPC1 expression was elevated in tamoxifen-resistant OHTR cells compared with parental MCF-7 cells (Fig. 1N).   16 . Consistent to these findings, we observed the interaction of these proteins in MCF-7 and OHTR breast cancer cells by immunoprecipitation assay ( Fig. 2A-D). siRNA-mediated silencing experiments in these cells showed that PSF expression was not affected by PSPC1 silencing, while PSPC1 expression was elevated by PSF knockdown (Fig. 2E,F). To examine if PSPC1 cooperatively functions with PSF in ER-positive breast cancer cells, we next analyzed whether PSPC1 affects the post-transcriptional regulation of ESR1 and SCFD2, which we previously defined as PSF downstream targets 12 . RNA-immunoprecipitation (RIP) assay indicated that ESR1 and SCFD2 mRNAs were specifically precipitated by immunoprecipitation of PSPC1, indicating that PSPC1 protein functionally associates with these mRNAs (Fig. 2G,H). The qRT-PCR analysis with primer sets for ESR1 and SCFD2 mRNAs demonstrated that PSPC1-specific siRNAs substantially repressed ESR1 and SCFD2 expression compared with control siRNA (Fig. 2I,K,M,O). However, qRT-PCR with another primer set for intron reveals that PSPC1-specific siRNAs increased or had no effect on intron-containing pre-mRNAs of ESR1 and SCFD2, implicating a discordant effect on pre-mRNA maturation or splicing (Fig. 2J,L,N,P). We further performed siRNA-mediated double knockdown experiments for PSPC1/PSF. While PSPC1 single knockdown significantly impaired MCF-7 and OHTR cell proliferation, double knockdown of PSPC1/PSF markedly suppressed cell proliferation ( Supplementary Fig. S1), suggesting possible cooperative functions of PSPC1 with PSF.
SCFD2 is associated with poor prognosis for ER-positive breast cancer patients. We demonstrated that SCFD2 mRNA levels were associated with poor prognosis for ER-positive breast cancer patients using Kaplan-Meier plotter database in the previous study 12 . Immunohistochemical analysis was performed to reveal the role of SCFD2 protein in clinical samples, and SCFD2 IR was strong in 45 cases whereas weak in 69 cases among the present 114 cases (Fig. 3A,B). Normal mammary tissues showed weak or negligible SCFD2 IR (Fig. 3C). SCFD2 status was positively correlated with histological grade, PSF status, and HER2 status, and had www.nature.com/scientificreports/ a tendency of correlation (P < 0.1) with LVI and PSPC1 IR status (Table 3). In Kaplan-Meier survival analysis, strong SCFD2 IR was significantly associated with shorter disease-free survival (P = 0.0253) and overall survival (P = 0.008) of ER-positive breast cancer patients (Fig. 3D,E) Based on univariate and multivariate analyses, SCFD2 IR had a tendency to be an independent prognostic factor for disease-free and overall survivals in these patients ( Table 2).
Combinations of PSPC1, PSF, and SCFD2 IRs are potent prognostic factors for ER-positive breast cancer. As PSPC1, PSF and SCFD2 IRs were shown to be potential prognostic factors for ER-positive breast cancer patients 12 , we further examined whether combinations of IRs of these proteins could efficiently predict the prognosis of ER-positive breast cancer patients. Among the 114 patients, those with strong IR of any two proteins tended to have higher rates of disease relapse and death compared with those with strong IR of any single protein alone (Fig. 3F,G). In particular, the disease relapse event was not observed among patients with strong PSF or SCFD2 IR alone, while it occurred in 15% of those with strong IR of PSF/SCFD2 (Fig. 3F). While the rate of disease relapse was 20% in patients with strong PSPC1 IR alone, it was much higher as 50% in those with strong IR of PSPC1/SCFD2 (Fig. 3F). It is notable that no death was observed in patients with strong IR of any single protein alone (Fig. 3G). The death rate was increased in patients with strong IR of PSPC1/PSF, PSPC1/SCFD2, and PSF/SCFD2 (22%, 17%, and 15%, respectively), and was 30% in those with strong IR of all three proteins (Fig. 3G). In Kaplan-Meier survival analysis, patients with tumors exhibiting score 2 IR of any two proteins were associated with shorter overall survival compared to those with tumors exhibiting score 0 IR of all proteins or score 1 IR of any single protein (P = 0.0216 or 0.042, respectively, corrected by Bonferroni method). Moreover, combinations of score 3 IR of all three proteins were significantly correlated with shorter disease-free survival compared to the score 0 or 1 IR (P = 0.0096 or 0.0072, respectively, corrected by Bonferroni method), and with overall survival compared to the score 0 or 1 IR (P = 0.0012 or 0.0054, respectively, corrected by Bonferroni method) (Fig. 3H,I). These results indicate that double or triple strong IRs of PSPC1/PSF/SCFD2 could be potent prognostic factors for ER-positive breast cancer patients.
High expression of SCFD2 downstream target genes DDIAS and MYBL1 is associated with poor prognosis in ER-positive breast cancer patients. To elucidate SCFD2-dependent signaling in ER-positive breast cancer, we further performed transcriptomic analysis in MCF-7 cells with or without SCFD2 knockdown. We picked up 150 genes most significantly repressed by SCFD2 siRNA versus control siRNA and further analyzed enriched pathways among the genes based on gene ontology (GO) terms using AmiGO2 database. "Cell cycle process" was most significantly associated with the genes (Supplementary Table S1 and Supplementary Fig. S2). Among cell cycle process-related genes, we found that DDIAS and MYBL1 were particularly overexpressed in invasive ductal breast carcinoma (IDC) or invasive lobular breast carcinoma (ILC) samples than normal breast tissues in TCGA breast cancer dataset retrieved from the Oncomine™ Platform (Fig. 4A,B). Online Kaplan-Meier plotter (http:// kmplot. com/) showed that high expression of DDIAS and MYBL1 mRNAs www.nature.com/scientificreports/ www.nature.com/scientificreports/ www.nature.com/scientificreports/ in tumors was significantly associated with shorter relapse-free survival in ER-positive breast cancer patients (Fig. 4C,D). We found that both DDIAS and MYBL1 were substantially downregulated in MCF-7 and OHTR cells treated with SCFD2 siRNAs (Fig. 4E-J). DDIAS and MYBL1 expression levels were higher in OHTR cells compared to MCF-7 cells (Fig. 4K,L).

SCFD2 silencing suppresses in vivo tumor growth of tamoxifen-resistant breast cancer cells.
Finally, we examined the role of SCFD2 in in vivo tumor growth of tamoxifen-resistant breast cancer cells. To this end, OHTR cells were xenografted to female athymic mice and an siRNA against SCFD2 was administrated into generated tumors. As a result, siSCFD2 #1 suppressed OHTR tumor growth (Fig. 5A,B). SCFD2 was efficiently downregulated by siSCFD2 #1 (Fig. 5C). Moreover, DDIAS and MYBL1 mRNA expression was downregulated by administration of siSCFD2 #1 (Fig. 5D,E). These results support the findings of our in vitro experiments and indicated the significance of SCFD2 in vivo.

Discussion
In the present study, we showed that PSPC1 post-transcriptionally regulates ESR1 and SCFD2 and potentially contributes to the proliferation of ER-positive breast cancer cells. PSPC1 interacted with PSF and transcripts of ESR1 and SCFD2 genes in MCF-7 and OHTR cells. Moreover, PSPC1 silencing caused a decrease in amount of ESR1 and SCFD2 mRNAs but not in those of intron region, suggesting that PSPC1 positively affects splicing process of intron-containing mRNA precursors. We have previously reported that siRNA-mediated suppression of PSF increases or has no effect on intron levels in ESR1 and SCFD2 genes 12,17 . In addition, the double silencing of PSPC1 and PSF potently suppressed the growth of MCF-7 and OHTR cells than single silencing Table 3. Association between SCFD2 status and clinicopathological parameters in 114 breast cancer. P value < 0.05 and 0.05 ≤ P value < 0.10 were considered significant and borderline significant, and were listed in bold and italic, respectively. a Data are presented as mean ± SD. All other values represent the number of cases. www.nature.com/scientificreports/ of PSPC1. Thus, PSPC1 and PSF will cooperatively increase the expression levels of their target genes through post-transcriptional regulation. We previously reported that SCFD2 is a PSF target and promotes the proliferation and tamoxifen resistance of ER-positive breast cancer cells 12 . In the present study, we revealed that strong SCFD2 IR was a potential prognostic marker for ER-positive breast cancer patients, and the combinations of strong intensities of PSPC1, PSF, and SCFD2 IRs were more potently associated with poor prognosis in these patients.
Notably, we further showed DDIAS and MYBL1 as downstream target genes of SCFD2 in ER-positive breast cancers based on transcriptomic analysis. DDIAS is an anti-apoptotic factor that suppresses cell death caused by various DNA-damaging stresses such as cisplatin treatment and UV irradiation [18][19][20] . Recently, it was demonstrated www.nature.com/scientificreports/ that DDIAS inhibits the death-inducing signaling complex (DISC) formation and promotes destabilization of caspase-8, thus suppressing TRAIL-mediated apoptosis 21 . DDIAS is also involved in homologous recombination DNA repair and IL-6/STAT3 signaling activation, which may contribute to cancer progression 22,23 . Immunohistochemical analyses showed that DDIAS expression was associated with shorter survival of breast cancer and non-small-cell lung cancer patients 24,25 . DDIAS regulates cell cycle progression in a context-dependent manner, as it promotes G1 to S phase transition in HCC cells 20 whereas hardly affects cell cycle progression in osteosarcoma U2OS-SCR cells 22 . Interestingly, there is a report that PSPC1 knockdown increased apoptosis in HeLa cells treated with DNA-damaging agent methyl methanesulfonate 26 . Considering that DDIAS expression is induced by DNA damage response in cancer 19 , PSPC1 and SCFD2 may contribute to the suppression of apoptosis partly dependent on DDIAS. MYBL1 is a MYB family transcription factor and its gene amplification and rearrangements were observed in some cancers [27][28][29][30][31] . MYBL1 high expression was shown to be associated with shorter survival of HCC and those with ER-positive breast cancer 32,33 . In HCC, MYBL1 is shown to promote cancer cell proliferation by activating TWIST1 transcription 32 . It was also suggested that MYBL1 rescues murine B-cell lymphoma cells form the growth arrest and apoptosis induced by anti-immunoglobulin M (IgM) antibody through maintaining c-Myc expression 34 . Further study on the function of DDIAS and MYBL1 will be useful to elucidate the role and mechanism of SCFD2 in ER-positive breast cancer progression.
Finally, we demonstrated that siRNA-mediated SCFD2 silencing suppressed in vivo tumor growth of tamoxifen-resistant breast cancer cells, suggesting that SCFD2 can be a promising therapeutic target of tamoxifenresistant breast cancer.
In conclusion, the present results suggest that PSPC1 would facilitates hormone-dependent breast cancer proliferation by exhibiting RNA processing of ESR1 and SCFD2, the latter further contributes to anti-apoptotic or proliferative responses by modulating the expression of DDIAS and MYBL1 (Fig. 5F).

Methods
Cell culture and siRNA transfection. Human ER-positive breast cancer MCF-7 cells and its 4-hydroxytamoxifen (OHT)-resistant OHTR cells were previously described 12,15 . Cell authentication was confirmed by STR profiling. Cells were transfected with siRNAs using Lipofectamine RNAiMAX reagent (Thermo Fisher Sci-  Cell cycle analysis. Flow cytometric analysis of cell cycle was performed as previously described 12  Immunohistological analysis. Immunohistochemical analyses were performed as previously described 37 . Briefly, 4-μm tissue sections were deparaffined with xylene, hydrated stepwise with ethanol, and then washed with TBS. Subsequently, the sections were heated in 10 mM sodium citrate buffer (pH 6.0) at 121 °C for 5 min for PSPC1 staining or 20 min for SCFD2 staining. Microarray and database analysis. To examine the gene expression of siRNA-treated MCF-7 cells, microarray analysis was performed using GeneChip Human Gene 1.0 ST Array (Thermo Fisher Scientific) according to the manufacturer's protocol. Gene ontology (GO) analysis was performed using AmiGO2 database (http:// amigo. geneo ntolo gy. org/ amigo). The Oncomine™ Platform (https:// www. oncom ine. org) was used to analyze gene expression in normal breast tissues, invasive ductal breast carcinoma (IDC), and invasive lobular breast carcinoma (ILC). For analyses of association between gene expression and relapse-free survival of ERpositive breast cancer patients, Kaplan-Meier Plotter software (https:// www. kmplot. com) was used.
Mice. BALB/cAJcI-nu/nu mice were purchased from CREA Japan (Tokyo, Japan). Animal care and experimental procedures were performed at animal facility of Tokyo Metropolitan Institute of Gerontology, and mice were maintained under a 12-h light-dark cycle (light on at 8:00 a.m.). All animal experiments were approved by the ethics committee of animal experiments at the Tokyo Metropolitan Institute of Gerontology (approval no. 18019) and performed in accordance with the animal experimental guidelines of the Tokyo Metropolitan Institute of Gerontology. This study complied with ARRIVE guidelines (https:// arriv eguid elines. org/).
In vivo tumor formation and siRNA administration. OHTR cells (1 × 10 6 cells) were mixed with the equal volume of Matrigel matrix (Corning, Corning, NY, USA) and injected subcutaneously into the side flank of 10-weeks-old BALB/cAJcI-nu/nu mice (CREA Japan, Tokyo, Japan). When the tumor volume reached 100 mm 3 , mice were divided randomly into two groups. siControl or siSCFD2 #1 (5 μg each) was prepared with GeneSilencer Reagent (Genlantis, San Diego, CA, USA) and injected into the generated tumors twice a week. The tumor volumes were measured once a week as previously described 12 .
Statistical analysis. Statistical analysis was performed using JMP version 9.0.0 (SAS Institute, Cary, NC, USA). Kaplan-Meier curves were generated using JMP, and P values were evaluated by log-rank test. For multiple comparisons, significance is assessed at P < 0.05 following Bonferroni correction. The correlation between PSPC1 or SCFD2 intensity scores and clinicopathological factors was evaluated by the Student's t-test or Pearson's chi-squared (χ 2 ) test. Univariate and multivariate analyses were performed using a Cox proportional hazard model. For statistical analyses in studies using cultured breast cancer cells, P-values were evaluated by Student's t-test or two-way ANOVA. www.nature.com/scientificreports/