Overexpression of CDCP1 is Associated with Poor Prognosis and Enhanced Immune Checkpoints Expressions in Breast Cancer

CUB-domain containing protein 1 (CDCP1) is a transmembrane protein acting as an effector of SRC family kinases, which play an oncogenic role in multiple human cancers. However, its clinical and immune correlations in breast cancer (BrCa) have not been explored. To define the expression, prognostic value, and potential molecular role of CDCP1 in BrCa, multiple public datasets, and an in-house cohort were used. Compared with paratumor tissue, CDCP1 was remarkably upregulated in the tumor tissues at both mRNA and protein levels. In the in-house cohort, CDCP1 protein expression was related to several clinicopathological parameters, including age, ER status, PR status, molecular type, and survival status. Kaplan–Meier analysis and Cox regression analysis exhibited that CDCP1 was an important prognostic biomarker in BrCa. In addition, enrichment analysis uncovered that CDCP1 was not only involved in multiple oncogenic pathways, but correlated with overexpression of immune checkpoints. Overall, we reported that increased expression of CDCP1 is a favorable prognostic factor in patients with BrCa. In addition, the correlations between CDCP1 and immune checkpoints provide a novel insight into the adjuvant treatment for immune checkpoint blockade via targeting CDCP1.


Introduction
Breast cancer (BrCa) is a common malignancy with the highest morbidity and terrible mortality among all cancers worldwide [1]. According to the latest statistical data, there will be 290,560 estimated new cases and more than 43,000 estimated deaths in 2022 in the USA [1]. In addition, the morbidity of BrCa has been slowly increasing by approximately 0.5% per year since the mid-2000 s partly due to continued decreases in fertility and increase in excess body weight [2]. Although the prognosis of BrCa has been persistently improved with the rapid development of comprehensive and personalized therapeutic strategies, including chemotherapy, radiotherapy, targeted therapy, and immunotherapy, not all patients could benefit from the established treatment options [3]. us, reliable biomarkers are important for the prediction of drug-specific responses and prognosis in BrCa patients.
CUB-domain containing protein 1 (CDCP1) encodes a transmembrane protein that contains three extracellular CUB domains and functions as an effector of SRC family kinases [4]. Previous studies have revealed that CDCP1 is oncogenic in several human cancers via regulating tyrosine phosphorylation-dependent cellular functions, and then promotes tumor invasion and metastasis [5,6]. A growing number of studies uncover the multiple roles of CDCP1 in cancers. CDCP1 is highly expressed in mesenchymal glioma subtypes, which may promote proneural-mesenchymal transformation [7]. Given CDCP1 is highly expressed in RAS-driven cancers, targeting a proteolytic neoepitope on CDCP1 is a pan-cancer approach to control RAS-driven cancers [8]. In addition, CDCP1 is a prognostic biomarker in early non-small-cell lung cancer, and its high expression predicts a poor prognosis [9]. Although several studies have preliminarily investigated the oncogenic role of CDCP1 in BrCa [10,11], systematic analysis based on transcriptomics and its prognostic value in BrCa has not been defined yet.
In the current research, we aimed to investigate the expression, prognostic value, and potential molecular role of CDCP1 in BrCa using multiple public datasets and an inhouse cohort. We reported that CDCP1 was remarkably upregulated in BrCa tissues and enriched in the HER2positive and the triple-negative subtypes. In addition, high expression of CDCP1 predicted poor prognosis in BrCa. Moreover, we also performed a systematic analysis of CDCP1 using the transcriptomic data and found that CDCP1 was not only involved in multiple oncogenic pathways but correlated with overexpression of immune checkpoints. Overall, we systematically analyzed the role of CDCP1 and emphasized the remarkable correlation between CDCP1 and immune checkpoints in BrCa.

UALCAN Database
Analysis. UALCAN (https://ualcan. path.uab.edu/) is an online open-access platform using omics data and clinical information from e Cancer Genome Atlas (TCGA) and the Clinical Proteomic Tumor Analysis Consortium (CPTAC) databases [12]. It could be utilized to analyze transcriptional and protein levels of potential genes of interest between tumor and paratumor tissues and their association with clinicopathologic features. In the current study, the UALCAN tool was utilized to analyze the transcriptional and protein levels of CDCP1 in BrCa and paratumor tissues and its association with clinical stages and molecular subclasses. All the BrCa cases available in the TCGA and the CPTAC subdatabases were included in our study.

2.2.
Kaplan-Meier Plotter Database Analysis. Kaplan-Meier plotter (https://kmplot.com/analysis/) is a web-based tool integrating gene expression cohorts, clinical information, and survival data [13]. All cancer samples accessible on the Kaplan-Meier plotter were utilized to assess the prognostic values of CDCP1 in BrCa. e mean expression of probe sets (1554110_at and 218451_at) was used to estimate the CDCP1 expression. BrCa patients were divided into the low-and high-CDCP1 expression groups according to the median level of CDCP1, with the rest of the settings set to default. Kaplan-Meier survival plots were derived to display all of the cohorts. e log-rankP value, 95 percent confidence interval (95%CI), and hazard ratio (HR) were computed and shown online.

Correlation Genes Screen and Enrichment Analysis.
Linked Omics (https://www.linkedomics.org/login.php) is a web-based tool used to handle the TCGA data [14]. In this research, the Linked Omics was used to screen genes that correlated with CDCP1 in BrCa. Genes with a correlation coefficient ≥ 0.2 or ≤ −0.2 were deemed to be candidates. For all parameters, the default choices were utilized. To identify the CDCP1-related biological functions and pathways, all correlated genes were used for enrichment analysis. We downloaded the h.all.v7.4.symbols.gmt and c2.cp.wikipathways.v7.4.symbols.gmt subclasses from the molecular signatures database [15], which were used as the background. e enrichment analysis was conducted using the R package "clusterProfiler." To obtain the results of gene set enrichment, the minimum gene set was set to 5 and the maximum gene was set to 5000. e top 5 terms were exhibited in this research.

Estimation of the Immunological Characteristics of the TME.
e RNA-sequencing (RNA-seq) data of BrCa in the TCGA database was obtained from the UCSC Xena (https:// xenabrowser.net/datapages/). e public data was utilized to investigate the immunological features. First, the ESTI-MATE algorithm was conducted to estimate tumor purity, ESTIMATE score, immune score, and stromal score [16], and their correlations with CDCP1 expression were next assessed. Next, several gene markers related to the tumor microenvironment (TME) as well as immune checkpoints were obtained from a previous publication [17] and their correlations with CDCP1 expression were evaluated. Furthermore, the correlations between CDCP1 expression and 150 immune-related genes, including chemokines, receptors, MHC molecules, immunoinhibitors, and immunostimulators, were assessed. In addition, the CIBERSOR method [18] was used to estimate the abundance of tumorinfiltrating immune cells (TIICs) based on gene expression profiles using the R package IOBR, and the correlations between CDCP1 expression and TIICs abundance were also evaluated.

Collection of BrCa Specimens.
e BrCa (Cat. HBre-Duc159Sur-01) tumor tissue microarray (TMA) was purchased from Outdo BioTech (Shanghai, China). A total of 119 tumor samples and 40 paired paratumor samples were contained in this research. Detailed clinic-pathological and follow-up data were provided by Outdo BioTech. Ethical approval was granted by the Clinical Research Ethics Committee in Outdo Biotech (Shanghai, China).

IHC Staining and Semiquantitative Assessment.
Immunohistochemistry (IHC) staining was conducted on the above sections according to the standardized procedures.

Acquisition of GSE173839 Dataset.
e GSE173839 dataset included RNA-seq data of BrCa from 71 patients on the durvalumab/olaparib arm, which were downloaded from the Gene Expression Omnibus (https://www.ncbi.nlm.nih. gov/geo/) [20]. We extracted the expression data of CDCP1 and PD-L1, explored the predictive value of CDCP1 for immunotherapy, and compared its predictive value with PD-L1.

Statistical Analysis.
All statistical analyses were conducted using SPSS 26.0 and R 4.0.2. All data are presented as means ± SDs. e difference between the two groups was analyzed by Student's t-test or Mann-Whitney test. Survival analysis was performed by log-rank test and Cox regression analysis. Associations between CDCP1 expression and clinic-pathological features were assessed using the chisquare test or corrected chi-square test. Correlation analysis between two variables was analyzed by the Pearson test. All statistical tests were two-sided, and P value ≤ 0.05 was considered statistically significant.

CDCP1 was Upregulated in BrCa
Tissues. First, we compared the expression levels of CDCP1 in tumor and paratumor samples using the TCGA, the CPTAC, and the in-house cohorts. In the TCGA cohort, the transcriptional level of CDCP1 was notably upregulated in BrCa tissues ( Figure 1(a)). In addition, CDCP1 protein was also overexpressed in tumor samples in the CPTAC cohort ( Figure 1(b)). Moreover, we utilized the IHC staining to detect CDCP1 expression BrCa and paratumor tissues, and the results showed that CDCP1 protein was significantly enhanced in tumor samples (Figure 1(c)-1(d)). Overall, CDCP1 was highly expressed in BrCa tissues, which could participate in the oncogenesis of BrCa.

CDCP1 Was Related to the Molecular Type of BrCa.
Next, the associations between CDCP1 protein expression and clinicopathological features in BrCa were evaluated in the in-house cohort. As shown in Table 1, the expression of CDCP1 was not related to tumor differentiation, T stage, AJCC stage, and HER2 status. However, CDCP1 was significantly associated with age, N stage, ER status, PR status, molecular type, and survival status. We also compared the expression levels of CDCP1 in different TNM stages and molecular subtypes in the TCGA, the CPTAC, and the inhouse cohorts. e results exhibited that CDCP1 was not varied in tumor tissues with different TNM stages ( Figures 2(a), 2(c), 2(e)), but upregulated in HER-positive and triple-negative subtypes (Figures 2(b), 2(d), 2(f )). Taken together, the expression of CDCP1 was associated with the molecular type of BrCa.

Overexpression of CDCP1 Predicted Poor Prognosis of
BrCa. Given the notable association between CDCP1 expression and survival status, we subsequently investigated the prognostic value of CDCP1 in BrCa. In the Kaplan-Meier plotter database, high transcriptional expression of CDCP1 was remarkably associated with poor relapse-free survival (RFS), overall survival (OS), and distant-metastasis-free survival (DMFS) (Figures 3(a)-3(c)). In addition, in the in-house cohort, CDCP1 was upregulated in the tumor tissues of patients who died during the follow-up processes (Figure 3(d)). Similarly, high expression of CDCP1 protein expression predicted poor OS in the inhouse cohort (Figure 3(e)). Furthermore, both univariate and multivariate Cox regression analyses revealed that high expression of CDCP1 was an independent prognostic factor in BrCa patients (Table 2). Collectively, CDCP1 was a significant prognostic biomarker in BrCa.

Analysis of CDCP1-Related Potential Functions in BrCa.
Subsequently, we tried to investigate CDCP1-related functions in BrCa. Genes correlated with CDCP1 in BrCa with correlation coefficient ≥0.2 or ≤ −0.2 were deemed to be candidates (Figures S1A-S1C). en, hallmark and Wikipathways gene set analyses of positively correlated genes (PCGs) and negatively correlated genes (NCGs) were conducted, respectively. PCGs mainly participated in an inflammatory response, TNF-α signaling, hypoxia, epithelial-mesenchymal transition (EMT), and interferon-c response (Figure 4(a)), and was involved in focal adhesion, primary focal segmental glomerulosclerosis, PI3K-AKT signaling pathway, and AGE-RAGE pathway (Figure 4(b)). e enrichment results of Wikipathways were visualized in Figure 4(c). Given that EGFR was as a significant gene that positively correlated with CDCP1, we validated the correlation between these genes in the in-house cohort, and the result exhibited that CDCP1 was significantly correlated with EGFR (Figures 4(d)-4(e)). In addition, the enrichment results of NCGs were scattered, which were exhibited in Figure S2. To sum up, CDCP1 may be related to inflammatory and immune responses via regulating multiple pathways in BrCa.

CDCP1 Was Correlated with Immune Checkpoints Expressions in BrCa.
Considering the potential relationship between CDCP1 and inflammatory and immune response in BrCa, we next explored the correlations between CDCP1 and gene markers of immune-related events. CDCP1 showed no significant correlation with the stromal score, immune score, and ESTIMATE score ( Figure 5(a)). In addition, CDCP1 was also not correlated with MHC molecules, gene markers of multiple immune cells, but positively related to immune checkpoint expressions, including CD274        Journal of Oncology   IL2RA   IL6R   RAET1E   NT5E   ICOS   TNFSF15   TNFSF14   CD276   TNFSF13B   CD48   C10orf54   TMEM173   ICOSLG   CD28   CD27   TNFRSF4   IL6   TNFRSF14   TMIGD2   TNFRSF18   TNFSF13   TNFSF18   TNFSF4   CD80   CD86   TNFSF9   PVR   CD40   TNFRSF13C   LTA   ENTPD1   TNFRSF25   CD70   KLRC1   CD40LG   ULBP1   TNFRSF9   KLRK1   MICB   HHLA2   BTNL2   TNFRSF13B   TNFRSF17   TNFRSF8 LGALS9 (PD-L1), CD276 (B7-H3), and VTCN1 (B7-H4) ( Figure 5(b)). In addition, a larger throughput analysis showed that CDCP1 was not significantly associated with immune-related genes and TIICs abundance (Figures 5(c)-5(d)). Since CDCP1 was positively correlated with PD-L1, we also examined whether CDCP1 could be a biomarker for immunotherapy in BrCa. e results showed that CDCP1 and PD-L1 were highly expressed in BrCa tissues with a good response ( Figure 6(a)), and the predictive value of CDCP1 was even higher than PD-L1 in the GSE173839 dataset ( Figure 6(b)). Overall, CDCP1 was related to enhanced immune checkpoint expressions and could predict the response to immunotherapy in BrCa.

Discussion
CDCP1 has been revealed to be significantly dysregulated in tumor tissues and accelerates progression in several malignancies [21]. CDCP1 is eminently located on the cytomembrane, which lies at the nexus of critical tumorigenic signaling cascades, containing the SRC-PKCδ, PI3K-AKT, WNT, and RAS-ERK axes, the oxidative pentose phosphate pathway, and fatty acid oxidation, making significantly functional contributions to tumor progression and development [21]. In addition, CDCP1 has a notable prognostic role in cancer. Ikeda et al. performed a multivariate Cox regression analysis of 200 lung adenocarcinoma patients and revealed that high-CDCP1 expression was an independent prognostic factor for OS in lung adenocarcinoma [22]. Dagnino et al. suggested that the circulating serum level of CDCP1 was related to the risk of developing lung cancer, especially in patients with tobacco exposure [23]. However, a systematic analysis of CDCP1 in BrCa has not been performed yet.
In this research, we reported that CDCP1 was significantly overexpressed in BrCa tissues and highly expressed in the HER2-positive and triple-negative subtypes. Previous research has revealed that CDCP1 is a novel marker of triple-negative breast cancer [24] and promotes tumor progression via reduction of lipid-droplet abundance and stimulation of fatty acid oxidation [25]. In addition, CDCP1 could interact with HER2 and enhance HER2driven tumorigenesis in BrCa [26]. us, the enrichment of CDCP1 might be crucial for the aggressiveness of the HER2-positive subtype. Furthermore, high expression of CDCP1 predicted poor prognosis in BrCa, which could be a novel biomarker for prognostic assessment in BrCa. Moreover, we also performed a systematic analysis of CDCP1 using the transcriptomic data and found that CDCP1 was not only involved in multiple oncogenic pathways, but correlated with overexpression of immune checkpoints.
With the rapid development of bioinformatics-assisted tumor immunity studies, immuno-correlations analysis has been emerging as a hotspot in the field of cancer research. A growing number of novel immune biomarkers has been identified [27][28][29]. Most immune biomarkers in the tumor were correlated with the inflamed immune microenvironment, such as enhanced chemokines, MHC molecules, and effective TIICs, and also correlated with immune checkpoint expressions [30,31]. In the current research, we found that CDCP1 was not related to the inflamed immune microenvironment, but positively correlated with immune checkpoint expressions, including CD274 (PD-L1), CD276 (B7-H3), and VTCN1 (B7-H4). us, CDCP1 might be a crucial regulator that contributed to immune evasion via promoting immune checkpoint expressions.
It has been reported that CDCP1 is crucial for the activation of RAS in cancer [8], and participates in multiple oncogenic pathways, such as EGF signaling [32] and HGF signaling [33]. In addition, we predicted that CDCP1 was involved in TNF-α signaling, hypoxia, EMT, interferon-c response, PI3K-AKT signaling, and AGE-RAGE signaling. Most of these pathways are associated with the regulation of immune checkpoints in cancer. For example, PD-L1 could be upregulated in ZEB1 and miR-200 dependent manners EMT-activated human breast cancer cells [34]. In addition, immune checkpoint molecules PD-L1 and B7-H3 were notably upregulated during TGF-β1-induced EMT [35]. Although our current study suggested potential relationships of CDCP1 to these pathways, the lack of confirmation from the molecular biology level remained an unavoidable shortcoming of this study.

Conclusion
In conclusion, we revealed that CDCP1 was highly expressed in BrCa tissues and enriched in the HER2positive and triple-negative subtypes, which also functioned as a novel prognostic biomarker in BrCa. In addition, CDCP1 was positively correlated with immune checkpoint expressions in BrCa, and several possibly related pathways were also suggested. Overall, we systematically investigated the role of CDCP1 in BrCa and provided a possible insight into the CDCP1-mediated overexpression of immune checkpoints.

Data Availability
All data supporting the results of this study are shown in this published article and supplementary documents. In addition, original omics data for bioinformatics analysis could be obtained from corresponding platforms.

Conflicts of Interest
e authors declare that have no conflicts of interest. Figure S1. e genes co-expressed with CDCP1 in BrCa.