Introduction

Alternative splicing (AS) emerges as a substantial contributor to the production of a myriad of transcripts from limited genes, which enriched protein variation and phenotypic diversity in multicellular organisms. Pre-mRNA splicing reactions require a step-wise assembly of spliceosome, a megadalton complex that recognize consensus regulatory sequences, including 5’ and 3’ splice sites marking exon-intron boundary, and the branch point site (BPS), to accomplish exon definition (Shi, 2017) (Wilkinson, Charenton et al., 2020). While a splice site is suboptimal, AS emerges under control of inherent sequence acting in cis, namely exonic or intronic splicing enhancer or silence, and the cognate trans-acting splicing factors, which can strengthen or weaken the recognition of splice sites by spliceosome.

It is well documented that AS dysregulation leads to multiple human diseases, including cancer (Montes, Sanford et al., 2019) (Singh & Cooper, 2012). Tumorigenicity has been linked with cancer-specific splicing switch arising from either cis-elements mutations or deregulated splicing regulatory factors. In some cases, splicing factors, particularly spliceosomal components SF3B1 and U2AF1, are frequently mutated across diverse cancer types (Baralle, 2005) (Harbour, Roberson et al., 2013) (Smith, Choudhary et al., 2019) (Cheruiyot, Li et al., 2021). For instance, recurrent SF3B1 mutations led to retention of intronic region in BRD9 pre-mRNA, conferring loss of core subunit function to the ncBAF chromatin remodeling complex and metastatic advantage to melanoma cells (Inoue, Chew et al., 2019). Another layer of regulation involves the abnormal expression of splicing factors derived from altered regulatory pathway. As an archetypal example, a set of members in SR protein family such as SRSF1 and SRSF3 are upregulated in breast tumors, lung tumors, glioma and exhibit oncogenic ability via altering AS products (for example, Bcl-x, MYO1B, NUMB and p53) (Anczuków, Akerman et al., 2015) (Zhou, Wang et al., 2019) (Tang, Horikawa et al., 2012) (Zhou, Gong et al., 2020). Otherwise, many splicing factors such as RBM proteins can either functions as oncoproteins or tumor suppressors in cancer-type dependent manner (Li, Guo et al., 2021), which may be attributed to tissue-specific AS patterns.

Breast cancer is the most common diagnosed malignancy and remain the second leading cause of cancer mortality (over 41,000 deaths annually in the USA) among females worldwide (Siegel, Miller et al., 2019). Metastasis to vital organs accounts for approximately 90% cancer-related death. The 5-year overall survival rate of patients with metastatic breast cancer is drastically decreased to about 25%, as compared to the primary cancer patients’ prognosis with a survival rate greater than 80% (Rabbani & Mazar, 2007) (Valastyan & Weinberg, 2011). Despite advances in early diagnosis and comprehensive interventions for breast cancer treatment, limited effective strategy against metastasis has been an outstanding challenge for improving prognostication of patients. This highlights the urgent need for deepened understanding of mechanisms underlying development of metastasis. Tumor metastasis is multi-step process involving a sequential cascade of events, such as epithelial-mesenchymal transition (EMT), local invasion from primary tumor site, lymphogenic or hematogenic dissemination, cancer cells seeding and metastatic lesions formation (Liang, Zhang et al., 2020). AS dysregulation has been implicated in multistage process of metastasis. For example, the switch of CD44 or FGFR2 splicing isoforms can lead to EMT phenotype (Chen, Zhao et al., 2018) (Warzecha, Sato et al., 2009). VEGF-A or XBP1 splicing transition is tightly linked to angiogenesis in cancer (Bates, 2008) (Zeng, Xiao et al., 2013). On account of the phenotypic plasticity in metastatic cells conferred by alternative splicing, further investigation in potential role of critical splicing regulatory factors and the associated splicing events in breast cancer metastasis is warranted.

RNA-binding proteins (RBPs) are a large class of conserved proteins that participate in cancer development through precisely control RNA metabolism, including RNA splicing. RBM7, a member of RBP family, is commonly known to exert RNA surveillance activity as a component of the human nuclear exosome target (NEXT) complex (Meola, Domanski et al., 2016). Aberrant expression of RBM7 is not only associated with lung fibrosis onset and motor neuronopathy, but also impacts cancer cell proliferation (Fukushima, Satoh et al., 2020) (Xi, Zhang et al., 2020). Previous findings have established a direct connection of RBM7 with spliceosome factors SF3B2 and SF3B4 (Falk, Finogenova et al., 2016), yet whether RBM7 is involved in the regulation of AS remain elusive. In this study, we report that RBM7 is frequently reduced in a subset of breast cancer metastases from lymph nodes lesions and distant organ as compared to primary tumors and its low expression predicts dismal survival outcomes of patients. Breast cancer cells with loss of RBM7 gain enhanced aggressive capability and metastatic potential. Mechanistically, RBM7 ablation enhanced the splicing switch of MFG-E8 pre-mRNA from canonical isoform MFGE8-L to a truncated isoform MFGE8-S. Ectopic expression of MFGE8-L caused marked suppression of migratory and invasive ability of breast cancer cells, whereas MFGE8-S showed opposite function/ disabled such function. At another layer, RBM7 discordantly regulated the phosphorylation of p65. In concert, NF-κB inhibitor abolished pro-angiogenesis effect of RBM7 ablation in breast cancer. Above all, our findings uncovered a previously unidentified role of RBM7 in breast cancer metastasis via coupling MFGE8 splicing switch and NF-κB signaling activation, suggesting a new therapeutic target and prognostic predictor for breast cancer.

Results

RBM7 is reduced in metastatic lesions of breast cancer and positively correlated with patients’ prognosis

To determine the clinical relevance of RBM7 in breast cancer, we assessed the correlation between RBM7 and important clinical outcomes. Analysis of TCGA, GEO and the METABRIC breast cancer datasets showed that RBM7 expression positively correlates with overall survival of breast cancer patients (Figure 1A and Supplementary Figure 1A-1B). RBM7 low expression predicts the reduced disease-free survival in patients with breast cancer (Figure 1B and Supplementary Figure 1C). Congruently, we found RBM7 is decreased in breast cancer compared to normal tissues (Figure 1C and Supplementary Figure 1D). Further dissection in a cohort of breast cancer patients from the cBioPortal unveiled cases with lymph node metastasis have lower expression of RBM7 as compared to cases without metastasis (Figure 1D). We next performed immunohistochemistry (IHC) detection of RBM7 in breast cancer tissue microarrays. The results showed that patients bearing clinical stage III tumors have significantly decreased RBM7 expression than patients bearing I and II clinical stage tumors (Figure 1E). In tissue chip containing triple negative breast cancer, the more aggressive and mesenchymal breast cancer subtype, RBM7 exhibits reduced protein levels in tumors as compared to paracancer tissues (Figure 1F). In addition, we collect specimen consisted of tumor tissues from 21 breast primary cancer, in 9 cases, from metastatic loci and took advantage of IHC analysis to determine the RBM7 expression. Most (25% strong positive;67% moderate positive) of primary tumor samples displayed a positive IHC staining for RBM7, whereas most of lung/liver metastatic lesions showed weak or negative RBM7 expression (Figure 1G). We also obtained paired primary breast cancer and lymph node metastatic loci and found, when compared to the corresponding primary tumor, the expression of RBM7 is significantly decreased in lymph node metastases (Figure 1H-1I). These results associated RBM7 with breast cancer metastasis, prompting us to further investigate the functional role of RBM7 in metastatic processes of breast cancer.

Figure 1.

RBM7 silencing provokes metastatic potential of breast cancer

In light with the above findings, we generated RBM7-silenced breast cancer cells using lentiviral expression of short hairpin RNA (shRNA) followed by profiling high-throughput transcriptome by RNA sequencing. Strikingly, Gene Ontology and pathway analysis displayed a wide range of RBM7-regulated genes were enriched in extracellular matrix organization, cell migration and angiogenesis (Figure 2A-2B), which have been characterized as crucial processes in tumor metastasis. A heatmap generated from qRT-PCR analysis of differentially expressed genes in RBM7-depleted breast cancer cells verified that a series of metastasis-related genes were upregulated upon RBM7 knockdown (Figure 2C). The lung metastasis is one of the most prevalent distant metastases that breast cancer is prone to and associated with poor prognosis of patients (Medeiros & Allan, 2019). Thus, we performed tail-vein transfer model to evaluate the potential impact of RBM7 on pulmonary metastasis of breast cancer cells. 5×10⁴ 4T1 cells, the mouse breast cancer cell line simulating stage IV human breast tumor, with stable RBM7 silence or control vector were implanted into BALB/c mice via tail vein. After 3 weeks, RBM7-silenced 4T1 cells were able to develop more spontaneous lung metastases than control cells as judged by pulmonary nodules number and haematoxylin-eosin (HE) staining of lung tissues (Figure 2D-2F and Supplementary 2A). Cancer metastasis is comprised of a complex cascade of events, such as migration, invasion and angiogenesis, thus we next to determine which process is mediated by RBM7-loss in facilitating breast cancer metastasis. We utilized transwell chamber with or without matrix coat and would healing assays to examine the potential function of RBM7 in migration and invasion capabilities and found the depletion of RBM7 led to a visible increase in mobility and invasiveness of both the triple-negative breast cancer (TNBC) cell line MDA-MB-231, estrogen receptor positive breast cancer cell line MCF7 and 4T1 cells (Figure 2G-2H and Supplementary 2B-2E). Conversely, RBM7 overexpression notably repressed MDA-MB-231 and MCF7 cells to migrate or invade across transwell inserts (Figure 2I-2J). Similar results were obtained in two additional TNBC cell lines BT549 and HCC1937 (Supplementary 2F-2G), which proved that RBM7 possessed the capability to suppress the migration and invasion of breast cancer cells. To examine the impact of RBM7 on angiogenesis in breast cancer, we cultured human umbilical vein endothelia cells (HUVEC) with condition medium obtained from culture supernatants of RBM7-depleted or scramble shRNA transduced cells for tube formation analysis. Compared to the scramble control, HUVEC incubated with culture medium from RBM7-knockdown breast cancer cells exhibited accelerated tube formation length and branch junction (Figure 2K), suggesting the inhibitory effect of RBM7 on breast cancer angiogenesis. Taken together, the above data demonstrated RBM7 as tumor suppressor by regulating breast cancer migration, invasion and angiogenesis.

Figure 2.

RBM7 exhibits splicing regulatory function in breast cancer

In previous studies, RBM7 is commonly characterized as an RNA-binding constituent of trimeric NEXT complex, which directs the RNA exosome to non-coding transcriptome for degradation (Lubas, Andersen et al., 2015). However, the mechanistical role of RBM7 in cancer remain largely unknown. Intriguingly, we observed that, attributed to the ablation of RBM7, almost 551 splicing events occurred significant decrease in PSI (percent splicing inclusion), whereas 289 splicing events endured increased PSI (Figure 3A). The changed events were divided among various types of AS (alternative splicing), including SE (skipping exon), IR (intron retention), A5SS (alternative 5’ splice site), A3SS (alternative 3’ splice site), and MXE (mutually exclusive exons), of note, SE events were the most prominently regulated (Figure 3B). A majority of AS events in particular of SE and A5SS category were in decrease of PSI upon RBM7 depletion (Figure 3C-3D), indicating that the presence of RBM7 may promote the usage of alternative splice sites. Subsequently, we extracted the candidate binding motifs which reside within cassette exons or A5SS and found UUUCUU motfs, simiar to RBM7 binding site (polyU) were enriched in RBM7 positively regulated AS events to a greater extent that control exons unimpacted by RBM7 (Figure 3E), raising the possibility that RBM7 may serve as a splicing activator through directly tethering cis-element in pre-mRNA. This is coherent with the nuclear localization of RBM7 in breast cancer cells (Supplementary 3A). In order to explore the splicing regulation activity of RBM7 in distinct pre-mRNA context, we validated a series of RBM7-significantly-changed splicing events across different AS types according to the RNA-seq results. The RT-PCR analysis demonstrated that the PSI of selected gene related to cancer progression was either increased or decreased endogenously upon RBM7 depletion (Figure 3F and Supplementary 3B). Considering that the AS is a sophisticated multi-step reaction that is regulated by both pre-mRNA-interacting splicing factor and other assistant factors in a context-dependent manner (Wright, Smith et al., 2022), we doubted the divergent effect of RBM7 on distinct splicing targets was due to a more complicated mechanism. Gene ontology (GO) enrichment analysis identified a majority of genes undergoing AS alterations were involved in cytoskeleton assembly and organization (Figure 3G). While globally analyzing the network enrichment of RBM7-regulated downstream splicing events, we observed its targets are functionally connected to well-linked network containing gene subgroups (Figure 3H), which are associated with biological processes in tumor cell metastasis, again suggesting the cancer-related function of RBM7.

Figure 3.

RBM7 depletion switches splicing outcomes of MFG-E8 by decreasing MFG-E8-L isoform

To ascertain the functional targets that explain the inhibitory role of RBM7 in breast cancer metastasis, we examined RNA immunoprecipitation -seq for RBM7 and RNA-seq data for breast cancer cells expressing two shRNAs targeting RBM7. Venn diagram illustrating overlapping genes pointed forward a hotspot gene, namely MFGE8 (Figure 4A). We focused on the AS control of MFGE8 by RBM7 resulting in a significant alteration in ratio of MFGE8 two isoforms, of which the canonical isoform (MFGE8-L) has been demonstrated to be correlated with tumor formation and aggressiveness of breast cancer (Yang, Hayashida et al., 2011). Importantly, the exon 7 located in F5/8 type C 2 domain, which is obligated to bind to phosphatidylserine, was lacked in the short isoform of MFGE8 (Figure 4B). Further RT-PCR identification verified RBM7 ablation indeed potentiated the production of MFGE8 short splice variant (MFGE8-S) in various TNBC cell lines (Figure 4C). Similar results were obtained from MCF7 cells, suggesting that RBM7 regulated the AS of MFGE8 independently of the ER status. In separate experiments, we transfected breast cancer cells with an exogenous MFGE8 splicing reporter containing exon 6-8 and 600bp of flanking intron sequences to determine the splicing regulatory function of RBM7. Accordingly, the depletion of RBM7 still induced the splicing shift of the minigene reporter by elevating MFGE8-S variant (Figure 4D). To test whether such splicing regulatory effect was directly or indirectly, we analyzed the sequence of alternative exon 7 and the motifs nearby its 5’ or 3’ splice sites, and found two poly(U) motifs are positioned in proximal to the pseudo 3’ splice site. Subsequent RT-PCR assay for the precipitation in RIP assays confirmed RBM7 could bind to the upstream sequence containing UUUCUU motifs adjacent to 3’ splice site of MFGE8 cassette exon, but not another region nearby it (Figure 4E). To pinpoint the location for the potential cis-element for AS regulation by RBM7, we designed antisense oligonucleotides (ASOs), which have been chemically modified to avoid endogenous nuclease degradation without recruiting RNase H, to block the UUUCUU residues in proximity to intron6/exon7 junction. As shown in Figure 4F, when compared to control ASO, treatment with ASOs targeting the potential binding site (UUUCUU) of RBM7, contributed to the exclusion of exon7, further suggesting that RBM7 regulated the exon7 skipping of MFGE8 by directly binding to its pre-mRNA.

Figure 4.

The short variant of MFGE8 abolishes the inhibitory function of its classical isoform on breast cancer cell migration and invasion

MFGE8 encodes a secreted protein, Milk fat globule-EGF factor 8 (MFG-E8), which participates in a variety of cellular processes, including mammary gland branching morphogenesis, angiogenesis and mucosal healing, etc (Raymond, Ensslin et al., 2009). We applied MFGE8-L stably overexpressing-or depleted-breast cancer cells to transwell assays and observed the cellular migratory and invasive capabilities was visibly suppressed by MFGE8-L overexpression (Supplementary 4A-4B). On the contrary, the knockdown of MFGE8-L provoked cell migration and invasion (Supplementary 4C). Strikingly, while MDA-MB-231 or HCC1937 expressing MFGE8-L showed a remarkable decrease in migratory and invasive ability compared to control vector, MFG-E8-S not only totally abolished the inhibitory function of MFGE8-L but elicited modest pro-migration/ invasion actions on breast cancer cells (Figure 5A and Supplementary 4D-4E). In comparison with MFG-E8 containing N-terminal EGF-like domain and C-terminal twice-repeated F5/8 type C domains, the protein encoded by MFGE8-S loses the second F5/8 type C domain, implying a shortage of the function corresponding to membrane-binding. Immunofluorescence assays revealed the MFGE8-L forms foci in cytoplasm, particularly enriched near cell-cell contact sites, whereas MFGE8-S is more evenly distributed in the nucleus and cytoplasm (Figure 5B). Furthermore, we compared transcriptome profiles between MFGE8-L and MFGE8-S ectopically expressing breast cells and found a set of functional genes were down-regulated by MFGE8-L, whereas this negative regulation was abrogated in response to MFGE8-S (Figure 5C). The differentially expressed genes between two splice variants were predominantly enriched in cell adhesion molecules and JAK-STAT cascade as revealed by KEGG pathway and Gene ontology analysis (Figure 5D). Subsequent experimental validation demonstrated that the activation of STAT1 was hindered by MFGE8-L, but not affected by MFGE8-S (Figure 4E). Considering that STAT1 signaling activation plays an important role in tumor cell metastasis via cell adhesion molecules induction (Marimuthu, Lakshmanan et al., 2022) (Chen, Hao et al., 2019), it is highly plausible that MFGE8-L, but not MFGE8-S, impeded breast cancer cell motility and invasive capability in a STAT1-dependent manner.

Figure 5.

RBM7 depletion-stimulated aggressiveness of breast cancer is dependent on MFGE8 splicing switch and NF-kB pathway activation

Next, we sought to assess if MFGE8 splicing switch is responsible for acquisition of aggressive feature in RBM7-silenced breast cancer cells. Flag-tagged MFG-E8-L or an empty vector was co-expressed in MDA-MB-231 cells together with RBM7 shRNA (Supplementary Figure 5A). Following seeding into transwell chamber, cells transfected with RBM7 shRNA exerted a higher migratory and invasive capability than scramble control, consistent with above data. Importantly, the restoration of MFGE8-L expression diminished the acquired aggressive properties of MDA-MB-231 cells resulted from RBM7 depletion (Figure 6A). Similar results were obtained using RBM4-depleted MCF7 cells with or without MFGE8-L complement (Figure 6B). However, a series of shRBM7-upregulated secreted factors that have been reported to drive tumor angiogenesis (Gopinathan, Milagre et al., 2015) (Fousek, Horn et al., 2021) (Feng, Ke et al., 2019), were not impacted by MFGE8-L/S (Supplementary Figure 5B and Figure 6C). Condition medium derived from breast cancer cells overexpressing MFGE8-L/S showed negligible effect on HUVEC tube formation (Supplementary Figure 5C), implying that there exists additional mechanism underlying breast cancer metastasis attributed to RBM7 reduction.

Figure 6.

Given that consecutive gene set enrichment analysis (GESA) of highest-ranking gene sets based on transcriptomic profile of mRNA expression showed a signature for NF-κB pathway that was augmented upon depletion of RBM7 (Figure 6D), we suspected that the transcriptional activation of downstream genes in NF-κB signaling known to facilitate tumor angiogenesis might be involved in shRBM7-exerted pro-metastatic function. As expected, the phosphorylation of p65, a core transcription factor well-known to switch on NF-κB pathway was notably enhanced by RBM7 knockdown, meanwhile, IκBα phosphorylation that can blockade p65 activation was curbed upon RBM7 knockdown (Figure 6E). Conversely, RBM7 overexpression impeded the activation of p65 (Supplementary Figure 5D). Yet MFGE8-L/S had no effect on phosphorylation of p65 or IκBα (Supplementary Figure 5E). We further carried out Matrigel tube formation assays and found NF-κB inhibitor PDTC drastically reversed the promotion of HUVEC angiogenesis by conditional medium harvested from RBM7-depleted cells (Figure 6F-6G), implying that RBM7-p65 axis may participate in HUVEC angiogenesis through reprogramming the secretome of breast cancer cells. Similar effect was observed with respect to cell motility and invasion (Supplementary Figure 5F). Taken together, these data above indicated that both MFGE8 splice switch and NF-κB pathway activation are required for mediating the pro-metastasis activity of RBM7 reduction in breast cancer.

The PSI of MFGE-8 exon 7 is decreased in breast cancer patients and positively correlated with RBM7 expression

To study the clinical relevance of MFGE-8 isoform switch, we analyzed the PSI changes of cassette exon7 in TCGA breast cancer cohort and found exon7 skipping occurs more frequently in tumor than in normal tissues (Figure 7A). Consistently, in comparison with normal counterparts, the splicing of MFGE8 is shifted towards a reduced proportion of MFGE8-L across multiple cancer types, including LUAD, HNSC, COAD, etc (Supplementary 6A). RT-PCR assay was performed in surgical tumor specimens in breast cancer patients and showed MFGE8 exon7 is prone to skipping in tumor (Figure 7B). On the other hand, the splicing switch in lymph node metastases favored MFGE exon7 exclusion when compared to paired primary tumor tissues from 3 breast cancer patients (Figure 7C). This is coherent with decreased RBM7 expression levels found in breast cancer with lymph node metastasis. In addition, patients with high expression of MFGE8 exon7 had a visibly better overall survival than patients with low expression (Figure 7D). We performed Pearson correlation test and revealed a noticeably positive correlation between the expression of RBM7 and exon7 inclusion of MFGE8 in breast cancer (Figure 7E), further supporting MFGE8 isoform switch regulated by RBM7.

Figure 7.

Discussion

Previous reports have recognized the RNA-binding protein RBM7 as a component of the NEXT complex that could facilitate RNA exosome targeting to a subset of non-coding RNAs and aberrant transcripts (Meola et al., 2016) (Pascal Preker, 2008), including short-lived exosome substrates and pre-mRNAs, for surveillance and turnover. In addition to this specific characteristic, RBM7 was shown to interact with subunits of 7SK snRNP complex and thus activate P-TEF downstream genes in response to DNA damage (Bugai, Quaresma et al., 2019). Herein, we discover a novel role of RBM7 in alternative splicing control through directly binding to its cognate cis-element in pre-mRNA. Using whole-transcriptome sequencing, we found the depletion of RBM7 contributed to exon exclusion in approximately 70% SE/A5SS events, implying RBM7 mainly functions as a splicing activator in breast cancer. However, the increase of PSI also occurred due to RBM7 knockdown in a smaller faction of splicing events, such as ZMYND8. We speculated this converse impact may be associated with the intrinsically sophisticated splicing processes driven by distinct core splicing factors, cognate cis-acting element and co-factors in a context-dependent manner (e.g., pre-mRNA structure, base modification, antagonistic or cooperative effect of multiple splicing factors). Similar to many splicing factors, like SR proteins and hnRNPs, RBM7 has an RRM-domain required for target RNA recognition in the N-terminus. Through ectopic expression of a series of RBM7 truncated mutant, we unveiled the disordered region (162-266) rich in polar residues serves as the functional modular to execute alternative splicing program.

Currently, the pathobiological function of RBM7 in breast cancer is underexplored except one study that claimed an oncogenic role of RBM7 since its knockdown inhibited breast cancer cell proliferation (Xi et al., 2020). Nevertheless, in our study, RBM7 knockdown in breast cancer displayed visibly pro-metastasis function with activated NF-κB signaling. More importantly, according to TCGA database (total cases > 1000 used for analysis), RBM7 shows decreased expression in breast cancer, particularly in the metastasis loci and has positive correlation with patients’ survival outcomes. It is therefore highly doubtful that RBM7 could serve as an oncogene in breast cancer as previously reported. Moreover, we can’t exclude additional biological function of RBM7 in breast cancer. RBM7-targeted silencing for breast cancer would likely be detrimental for neighbouring heathy tissue and immune environment as a result of NF-κB-dependent pro-inflammatory mediators and cytokines secretion. Other splicing events switch triggered by RBM7 might also be linked to tumor development. For example, we found RBM7 regulated alternative splicing of AKT2, which is involved in glucose metabolism (Hay, 2011). Further investigation is warranted to elaborate more splicing mechanisms of RBM7 in breast cancer.

In conclusion, we discovered that down-regulated RBM7 and its cognate elevated exon7-exclusion MFGE8 were tightly related to enhanced breast cancer metastatic potential and poor prognosis of patients. MFGE8-L, the canonical variant, possessed the capability to trigger STAT1 phosphorylation and inhibit the motility and invasion of breast cancer cells, whereas splice switch to MFGE8-S was found loss-of-function for this tumor-suppressive role. At another layer, RBM7 reduction led to NF-κB pathway activation, resulting in enhancement of breast tumor aggressiveness. Overall, our findngs implicated that activated NF-κB pathway worked in concert with MFGE8 splicing transition to spur metastasis in RBM7-depleted breast cancer. Additionally, although the detailed clinical significance underlying their cooperation need further in-depth investigation, we propose that RBM7 and its regulated splicing variants might serve as potential therapeutic biomarker and prognostic indicator in breast cancer patients.

Materials and Methods

Cell culture

Human breast cancer cell lines (MDA-MB-231, MCF7, HCC1937 and BT-549) and human embryonic kidney cell HEK293T in this article were obtained from the American Type Culture Collection (ATCC). MDA-MB-231 cells were cultured without carbon dioxide at 37℃ in Leibovitz’s L-15 medium (Gibco) contained 10% FBS (SINSAGE TECH). MCF7 and HEK293T cells were cultured with DMEM mediun (Gibco) supplemented with 10% FBS. HCC1937 cells were cultured with RPMI-1640 media (Gibco) supplemented with 10% FBS, while BT-549 was maintained with RPMI-1640 media (Gibco) supplemented with 10% FBS and 0.023 U/ml insulin. HUVECs were cultured with Endothelial Cell Medium (ScienCell) supplemented with ECGS (ScienCell), P/S solution (ScienCell) and 10% FBS (ScienCell). Except for MDA-MB-231, all the cells above were cultured in a 37 °C humidified incubator.

Constructions of stable cell lines

To generate RBM7 overexpression vector, the coding regions of this gene (NM_016090.4) were cloned into pCDH-CMV-MCS-EF1-Puro with N-terminal Flag tag. For RBM7 knockdown, target-specific shRNA (RBM7-sh1: 5’-CGAAAGGACTATGGATAACAT-3’;RBM7-sh2:5’-CTACCTAGCAGATAGACATTA-3’) was cloned into pLKO.1 vector. And HEK293T cells were transfected these over-expression (or low-expression) lentivirus plasmis together with lentivirus packaging plasmids pPAX2 and pMD2 according to the manufacturer’s instructions. Viruses were collected 48 hours after transfection and then to infect breast cancer cell line with 8 ug/ml polybrene (beyotime biotechnology) for 24 hours. Positive cells were selected with 3 ug/ml puromycin.

Western Blot assay

Cells were lysed with RIPA buffer (10 mM Tris-HCl pH 7.5, 150 mM NaCl, 0.1% SDS, 1% Triton X-100, and 1% sodium deoxycholate) supplemented with protein tyrosine phosphatase inhibitor Na₃VO₄ (1mM), serine protease inhibitor PMSF (1mM) and protease inhibitor Cocktail. Immunoblotting assay was performed using standard methods. The following antibodies were used: anti-RBM7 (Sigma-Aldrich, HPA013993), anti-Flag (Sigma-Aldrich, F1804), anti-GAPDH (Proteintech, 10494-1-AP), anti-Phospho-Akt (Ser473) (Cell Signaling Technology, #4060), anti-Akt (pan) (Cell Signaling Technology, #4691), anti-Phospho-NF-κB p65 (Cell Signaling Technology, #3033), anti-NF-κB p65(Cell Signaling Technology, #8242), anti-IκBα (Cell Signaling Technology, #4814), anti-Phospho-IκBα (Ser32) (Cell Signaling Technology, #2859), anti-Phospho-stat1 (Tyr701) (Cell Signaling Technology, #8062),and anti-MFGE8(R&D systems, MAB2767). Used secondary antibodies were conjugated to horseradish peroxidas. The protein bands were detected with an ECL Enhanced chemiluminescence regent Kit (NCM, Biotech), visualized using the MiniChemi™ Cheminescent image system (Beijing Sage Creation, China).

RNA isolation, RNA sequencing, and PCR

Total RNA from cells and tumor tissues was extracted using Trizol (Invitrogen) according to the manufacturer’s instructions. Total mRNA was cleaned up using RNeasy Lipid Tissue Mini Kit (Qiagen, USA). And RNA-seq was carried out on platform of Novogene Illumina sequencing. cDNAs were obtained using Evo M-MLV Plus 1st Strand cDNA Synthesis Kit (Accurate Biotechnology, AG11615) with random primers. qRT-PCR was carried out using MonAmp^TM Universal ChemoHS Specificity Plus qPCR Mix. GAPDH mRNA was examined as control for normalization. Gene expression relative to GAPDH was calculated using the formula 2^−ΔΔCT. For identified alternative splicing events, primers of the upstream and downstream of variable exons or introns were designed and PCR assay was performed. PCR products were subject to DNA agarose gel electrophoresis. AS events change of percent-spliced-in (PSI) values were the ratio of retained splices to total transcripts. The amount of each splicing isoform was measured with ImageJ. All the primers are showed in Table1.

Migration and invasion assays

For transwell migration assay, 10 × 10⁴ MDA-MB-231 or other breast cancer cells (MCF7, BT-549 and HCC1937) were plated in the upper chamber of 6.5 mm insert (Costar, USA) with 8.0 μm polycarbonate membrane with 100 μL serum-free medium. For invasion assay, 10 × 10⁴ breast cancer cells were seeded in the top chamber of each Matrigel-coated insert (BD Biosciences, USA). After 24 hours or 48 hours of incubation at 37 °C, migrated or invaded cells were fixed with 4% PFA and stained in 0.1% crystal violet. The cell number was counted by ImageJ software in 3 views from 3 independent experiments.

Wound healing

Knockdown RBM7 in MDA-MB-231 and 4T1 cells or control cells were implanted into 6-well culture dishes. When the cells grew about 90% confluence, lines of the same width were drawn on the bottom of the culture dish using autoclaved sterile tips. Images were captured at 0 and 36 hours after the wounding. Data shown were representative of three independent repeats.

Tube formation of HUVECs assay

Capillary tube formation assay was used to observe angiogenesis of HUVECs. HUVECs were seeded into 48-well plate coated with growth factor reduced Matrigel (BD Biosciences). HUVECs cells were treated with conditioned medium for 4-5 hours and then formation of capillary-like tubes was observed and stained with calcein for 30 minutes. Images of blood vessels were taken using fluorescence microscopy. And the tube length was analyzed by ImageJ software.

Immunofluorescence assay

After seeded on coverslips for 36 hours, the MDA-MB-231 cells were immersion fixed with 4% PFA for 20 minutes at room temperature, washed 3 times in PBS. Then the cells were treated with 0.2 % Triton-X100 in PBS for 10 minutes and blocked with 3% BSA for 10 minutes. Next the cells were immersed in primary antibody at 4℃ overnight, washed in PBS 3 times for 5 minutes each, and incubated with secondary antibody labelled with Alexa fluorophore 488/592 for 1 hour at RT. After washing 3 times in PBS, the cells were dyed with DAPI. In addition, antibodies targeting RBM7 (Sigma, HPA013993), and Flag (Sigma, 3165) were used in the assay. The pictures are captured with fluorescent microscope (Leica, USA).

RNA immunoprecipitation assay

RNA immunoprecipitation was performed as previously described. In brief, HEK293T cells were transfected Flag-RBM7 or control plasmid. After 48 hours of transfection, Flag-RBM7 or control cells were collected and cross-linked by 1% formaldehyde, homogenized in lysis buffer (50 mM Tris-HCl Ph 7.5, 0.4 M NaCl, 1 mM EDTA, 1mM DTT, 0.5% TritonX-100, 10% Glycerol supplemented with RNase inhibitor and broad-spectrum protease inhibitors Cocktail), and broke with an ultrasonic crusher. And solubilized cell lysate was obtained and precleared by ProteinG-agarose beads together with nonspecific tRNA to remove nonspecific binding. Then binding RNAs with RBM7 protein were precipitated with Anti-Flag M2 Affinity beads. Co-precipitated RNAs were subject to RT-PCR. Input controls (total RNAs) and Flag-controls were assayed simultaneously to prove that the detected signals were the result of RNAs specifically binding to RBM7. The gene-specific primers were designed MFGE8-RIP1-F:5’-tatttatgccccctcaccgc-3’; MFGE8-RIP1-R:5’-atggtggctgcctgtaat-3’;MFGE8-RIP2-F:5’-ctgaggctcacccaagtag-3’;MFGE8-RIP2-R:5’-GTAGGATGCCACAAACTG-3’). GAPDH mRNA was as negative control (GAPDH-F 5’-GGGGAAGGTGAAGGTCG-G-3’; GAPDH-R 5’-TTGAGGTCAATGAAGGGG-3’).

Antisense oligonucleotide (ASO)-mediated splicing blockade

MFGE8-ASOs were synthesized with a full-length phosphorothioate backbone and 2’-O-methyl modified ribose molecules (GenScriptBiotech Corp). ASO was transfected at final concentration of 100 nM using Lipofectamine3000 (Thermo Fisher Scientific) according to the manufacturer’s instructions. The MCF7 cells were harvested for RNA and splicing assays 48 h after transfection.

Animal experiments

All procedures and experimental protocols involving mice were approved by the Institutional Animal Care and Use Committee (IACUC) of Dalian Medical University. Male NYG mice (6 weeks old, 4-5 mice per group as indicated in figures and/or figure legends) were injected into tail vein with murine 4T1 breast cancer cells (6 × 10⁵ cells) to develop mice model of pulmonary metastasis. After about 2 weeks, all mice were euthanized, and lungs were harvested, fixed and paraffin embedded. After lungs were H&E stained, the number of metastatic lesions were enumerated by pathologist using brightfield microscopy.

Clinical tissues samples collection

All human breast cancer tissues or normal tissue were obtained with written informed consent from patients or their guardians prior to participation in the study. The Institutional Review Board of the Dalian Medical University approved use of the tumor specimens in this study.

Statistical analyses

Statistical analyses and statistical graphs were conducted by GraphPad Prism 8 software. To compare the differences between the two variables, Student’s T-test, one-way ANOVA with Dunnett multiple comparisons, and one-way ANOVA with Tukey’s multiple comparisons according to the specific situation were applied. *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001 indicated the significant difference.

Data availability

The authors declare that all the data supporting the findings of this study are available within the article and its supplementary information files. RNA-seq data in this study have been deposited in Gene Expression Omnibus of NCBI with the accession code GSE248938.

Author contributions

YW and YQ conceived the project and designed the experiments. FH, JY, ZD, KW, and DC performed most of the experiments, whereas CC, JZhang, JZhao and WZ performed data analysis. YW and YQ wrote the manuscript.

Acknowledgements

This work was supported by the National Key R&D Program of China (2022YFA1104002 to YW); the National Natural Science Foundation of China (82225034, 81830088 to YW; 82103148 to YQ; 81872247 to WZ); the Science and Technology Innovation Foundation of Dalian (2022JJ11CG009 to YW), Liaoning Revitalization Talents Program XLYC1802067 to YW; the Department of Science and Technology of Liaoning Province (2021JH6/10500160 to YW); the Basic Scientific Research Project of Education Department of Liaoning Province (LJKQZ2021104 to YQ); the Science and Technology Innovation Talent Support Program of Dalian (2022RQ056 YQ); Dalian High Level Talents Renovation Supporting Program (2019RQ097 to WZ).

Study Approval

The Institutional Animal Care and Use Committee of the Dalian Medical University approved use of animal models in this study. All human tumor tissues were obtained with written informed consent from patients or their guardians prior to participation in the study. The Institutional Review Board of the Dalian Medical University approved use of the tumor specimens in this study