Discovery of C1orf50 as a prognostic marker for Luminal A breast cancer
We investigated whether C1orf50 mRNA expression plays a prognostic role using the TCGA-BRCA dataset, in which RNAseq data is accompanied by survival information. For the analysis, we included 747 cases of primary tumor and infiltrating ductal carcinoma (IDC) in the TCGA-BRCA dataset. When comparing the expression levels of C1orf50 by IDC subtype, it was found that Luminal A breast cancer showed significantly higher expression levels of C1orf50 than non-Luminal A breast cancer (Fig. 1A). When examining the primary IDC breast cancer patient population registered in TCGA, it was found that there were many patients with stage II breast cancer (Fig. 1B). Since the prognosis for stage I breast cancer is widely known, this analysis focused primarily on patients with stage II breast cancer. Using the median of C1orf50 mRNA expression values, we divided stage II breast cancer patients into C1orf50-high and C1orf50-low groups and performed survival analysis of stage II breast cancer patients, finding a significant (p = 0.045) difference in 10-year survival (Fig. 1B). In particular, a trend towards a greater divergence in the survival curve was observed from 5 years. In the analysis of stage II breast cancer patients by histological subtype, a significant difference in 10-year survival was observed in patients with Luminal-type breast cancer (p = 0.044), which is characterized by estrogen receptor expression (Fig. 1C). This difference was particularly significant (p = 0.01) in the group of patients with Luminal A breast cancer (Fig. 1D). There was no significant difference in 10-year survival rates among patients with triple negative, HER2, and Luminal B stage II breast cancer (Fig. 1E and Fig. S1). This data suggest that the expression level of C1orf50 mRNA is a prognostic marker, especially in patients with stage II luminal A breast cancer. In addition, we investigated whether C1orf50 has different effects in premenopausal and postmenopausal breast cancer. In this study, we re-categorized women aged 50 years or older as postmenopausal, and found that C1orf50 may be a factor involved in survival in stage II breast cancer in postmenopausal patients (Fig. S1C-F).
Next, we determined whether C1orf50 is expressed at the protein level in breast cancer tissues. Using a cohort of breast cancer patients from the Clinical Proteomic Tumor Analysis Consortium (CPTAC), we analyzed whether C1orf50 mRNA and C1orf50 protein expression were correlated (r = 0.49, p < 0.001), and found a positive correlation (Fig. 2A). Furthermore, immunostaining with anti-C1orf50 antibody in tissue arrays composed of normal mammary tissues and breast cancer tissues showed that C1orf50 protein expression was low in normal mammary tissue (Fig. 2B), whereas the expression of C1orf50 protein was high in breast cancer tissues of all subtypes (Figs. 2C, 2D upper panels). Importantly, C1orf50 expression was found to be maintained at high levels not only in primary lesions but also in metastatic lesions of the lymph nodes (Figs. 2C, 2D lower panels). This data suggests that C1orf50 expression is upregulated in cancer cells compared to normal cells and that its expression is independent of the environment in which the cancer cells are located. In addition, the positive correlation between C1orf50 mRNA and C1orf50 protein expression suggests that it is reasonable to evaluate C1orf50 protein as a prognostic marker in pathological specimens by immunostaining or other methods.
Pathway analyses of C1orf50 in Luminal A breast cancer
The data above shows that stage II breast cancer patients with high C1orf50 expression have a significantly worse prognosis (Fig. 1). The function of C1orf50 has not been previously reported, and its physiological and pathological roles and association with cancer biology remain completely unknown. To elucidate the molecular mechanisms of how C1orf50 promotes cancer progression, we performed pathway analysis focusing on C1orf50 mRNA expression levels using the TCGA-BRCA dataset. First, we divided the TCGA data of stage II Luminal A breast cancer patients into C1orf50-high and C1orf50-low groups and performed Gene Set Enrichment Analysis (GSEA) using the Molecular Signatures Database (MSigDB) hallmark gene sets, and found a significant increase in the MITOTIC_SPINDLE gene set, which is related to the cell cycle. We also found a decrease in the OXIDATIVE_PHOSPHORYLATION gene set, which is related to mitochondrial function, as well as a decrease in INTERFERON_ALPHA_RESPONSE, INTERFERON_GAMMA_RESPONSE, and ALLOGRAFT_REJECTION, which are related to immune response (Fig. 3A). Interestingly, we observed a decrease in the ESTROGEN_RESPONSE_LATE gene set: it has been reported that the value of this pathway correlates with estrogen reactivity [19], suggesting that a decrease in estrogen reactivity occurs in the C1orf50-high patient group. Furthermore, the Gene Set Variation Analysis (GSVA) confirmed a similar trend to the GSEA results, as well as stronger association with transforming growth factor (TGF) beta signaling in the C1orf50-high group (Fig. 3B). This data suggests that in stage II Luminal A breast cancer patients with higher levels of C1orf50 expression are associated with an increased cell cycle activity, while lower levels of C1orf50 expression are associated with decreased expression of immunoreactive and estrogen responsive gene groups.
Next, we performed the same analysis on the MsigDB C6 (Oncogenic signature) gene sets and found that the MEL18 gene signature and the BMI1 gene signature were decreased in the C1orf50-high group (Fig. S1A). MEL18 and BMI1 are Polycomb proteins involved in gene silencing [20]. In addition, since it has been reported that MEL18 deficiency leads to the reduction of estrogen receptors, which results in hormone-sensitive breast cancer cells acquiring the ability to grow in a hormone-independent manner [21], C1orf50 may have a strong role in hormone insensitivity in Luminal breast cancer patients. In the C6 gene set, several KRAS-related pathways have been shown to be upregulated; previous studies have shown that increased expression of KRAS in the TCGA-BRCA dataset is associated with an increase in PD-L1 [22], suggesting that C1orf50 may indirectly contribute to immune reactivity or immune evasion. The Gene Ontology Biological Process (GOBP) gene sets showed a trend toward decreased pathways related to mitochondrial function, as well as decreased pathways related to immune response (Fig. S1B). Further analysis using the Kyoto Encyclopedia of Genes and Genomes (KEGG) gene sets, revealed decreased oxidative phosphorylation and decreased antigen processing and presentation (Fig. S1C). This data suggests that C1orf50 is widely implicated in cancer signaling, regulation of mitochondrial function, and immune evasion.
Expression levels of C1orf50 determine cell cycle and response to CDK4/6 inhibitors of Luminal breast cancer
Since the C1orf50-high group had enhanced cell cycle gene sets in both GSEA and GSVA, we analyzed the correlation between the expression level of C1orf50 and that of cell cycle-related factors. We found a positive correlation with components of the cyclin D:CDK4/6 complex and the cyclin E:CDK2 complex (Fig. 4A). The cyclin D:CDK4/6 complex is a factor that promotes breast cancer progression [23, 24], and CDK4/6 inhibitors have recently been administered to patients with unresectable or recurrent Luminal breast cancer [25, 26]. The cyclin E:CDK2 complex has been reported to be involved in breast cancer progression [27, 28]. Interestingly, BT474 cells exogenously transfected with C1orf50 showed increased sensitivity to abemaciclib, a CDK4/6 inhibitor drug (Fig. 4B). We compared the relationship between the expression levels of C1orf50 and eight CDKs in the dataset and found that the expression of CDK2, CDK4, CDK6, CDK7, and CDK8 was significantly higher in the C1orf50-high group (Fig. 4C). CDK7 forms a complex with cyclin H and MAT1 and acts as a CDK-activating kinase that phosphorylates and activates CDK2 and CDK4/6 [29]. It has been reported that when CDK8 expression is high in various carcinomas, especially breast cancer, a poor prognosis is found [30]. This data suggests that high levels of C1orf50 expression contribute to cell cycle acceleration.
To confirm that C1orf50 is indeed associated with cell cycle progression, we transfected siRNAs against C1orf50 mRNA into Luminal-type BT474 cells and tested whether the proliferative capacity of the cells would be affected. Two unique siRNAs against C1orf50 mRNA each attenuated the protein level of C1orf50 in BT474 cells (Fig. 4D). This indicates that the anti-C1orf50 antibodies used in this study accurately recognize the C1orf50 protein. After siRNA transfection at 24, 48, 72, and 96 hours, cell numbers were assessed using the MTS assay and showed that the proliferation of siC1orf50-transfected cells was significantly attenuated compared to siControl-transfected cells (Fig. 4E). This data indicates that C1orf50 protein is indeed expressed in breast cancer cells and is imperative to cell cycle progression.
C1orf50 promotes Luminal breast cancer stemness properties
Cancer stem cell populations have been implicated in the poor prognosis of various cancers [31–33]. We confirmed that C1orf50 expression levels correlate with cancer stem cell-related signatures. Pathway analysis of the association between C1orf50 expression levels and cancer stemness showed that the stemness-related REACTOME_YAP1_AND_WWTR1_TAZ_STIMULATED_GENE_EXPRESS, and RAMALHO_STEMNESS_UP scores were positively correlated with C1orf50 expression, and RAMALHO_STEMNESS_DN scores were negatively correlated with C1orf50 expression (Fig. 5A). The Hippo signal transducers, YAP/TAZ, are one of the most important factors in the molecular mechanisms that promote cancer stem cells [16, 17, 34]. Many reports have shown that the expression level of YAP/TAZ defines cancer stemness [35]. Our data suggests that C1orf50 progresses breast cancer stemness through YAP/TAZ signaling.
Immunostaining with anti-YAP/TAZ, anti-C1orf50, and anti-NANOG antibodies in tissue arrays showed that C1orf50-high breast cancer cells express YAP/TAZ and NANOG at high levels in Luminal breast cancer cells (Fig. 5B). Having examined 48 human breast cancer samples, the C1orf50 mean fluorescence intensity (MFI) strongly correlates with both YAP/TAZ (r = 0.80, p < 0.001), and NANOG (r = 0.77, p < 0.001) MFI scores (Fig. 5C). We investigated the effect of C1orf50 on breast cancer cell stemness in vitro. First, we infected Luminal-type MCF7 and BT474 cell lines with lentiviruses expressing shRNA against C1orf50 mRNA and performed Western blotting of cell extracts. We observed that C1orf50 protein deficiency results in decreased YAP/TAZ proteins. We confirmed that the expression levels of AXL and CYR61, target proteins of YAP/TAZ signaling, as well as the expression levels of c-MYC and KLF4, factors representing the cancer cell undifferentiated state, were similarly decreased (Fig. 5D). Since stemness is generally assessed by self-renewal capacity, we evaluated C1orf50-depleted breast cancer cells and confirmed C1orf50 expression is imperative to the self-renewal capacity in breast cancer cells (Fig. 5E). This was not restricted to Luminal breast cancer cell lines, but also in other breast cancer molecular subtypes (Fig. S3). This suggests that C1orf50 is essential for maintenance of breast cancer stemness.
C1orf50 associates with immune evasion signatures in Luminal breast cancer
We have shown that C1orf50 expression levels are particularly detrimental to prognosis in a subset of patients with Luminal A stage II breast cancer. The results of the Hallmark pathway analysis suggest that the patients with high C1orf50 expression may have suppressed immunity, but it remains unclear whether there is a patient population in Luminal breast cancer for whom immune checkpoint inhibitors, currently widely used in triple-negative breast cancer, are effective. Therefore, we performed in silico analysis to examine whether the use of immune checkpoint inhibitors may be applicable in patients with high C1orf50 expression.
The Hallmark pathway analyses showed that immune response-related pathways were downregulated in the C1orf50-high group (Fig. 3A-B, Fig S2). We then found that C1orf50 expression negatively correlated with T-cell mediated cytotoxicity (Fig. 6A). To further investigate the mechanisms behind these findings, we examined the mRNA expression levels of immunosuppressive molecules. As shown in Fig. 6A-C, the expression levels of PD-L1 (CD274) and PD-L2 (PDCD1LG2) were positively correlated with the expression level of C1orf50, and the expression levels of CMTM4 and CMTM6, regulators of PD-L1 [36], were also positively correlated with that of C1orf50. This data suggests that the expression level of C1orf50 may have a suppressive effect on immune checkpoint mechanisms regulated by PD-1/PD-L1. Therefore, the expression level of C1orf50 may be a useful marker when considering the application of PD-L1 inhibitors in the Luminal breast cancer patient population.