NCBP2 predicts the prognosis and the immunotherapy response of cancers: a pan-cancer analysis

Data source and processing

Transcriptional data and clinical information from the TCGA Pan-cancer cohort, as well as normal human tissue data from the Genotype-tissue expression (GTEx) database, were acquired through university of Cingifornia Sisha Cruz (UCSC) Xena (Goldman et al., 2020) (https://xenabrowser.net/). The expression of NCBP2 in TCGA and GTEx datasets was analyzed with transcripts per million (TPM) using the same sequencing platform and library preparation to reduce potential batch effects. The abbreviations for different types of cancers can be found in Table S1.

Two immunotherapy studies were included in this study. The sequencing and clinical data of 298 patients with urological cancer who received atezolizumab treatment (IMvigor210 cohort) were obtained from the website research-pub.gene.com/IMvigor210CoreBiologies/packageVersions/. The cohort of melanoma patients treated with nivolumab (anti-PD1) (GSE91061) was acquired from the Gene Expression Omnibus database (GEO).

The TIMER (Li et al., 2016) 2.0 database (http://timer.cistrome.org/) was utilized to acquire NCBP2-associated immune cell infiltration correlations for the TCGA Pan-Cancer study. We summarized the correlations between NCBP2 mRNA expression and 21 immune cell subsets, including CD4+ T cells, CD8+ T cells, γδT cells, T cell regulatory (Tregs), T cell follicular helper (Tfh), cancer-associated fibroblast (CAF), endothelial cells (Endo), eosinophils (Eos), mast cells, progenitors of lymphoid/myeloid/monocyte, hematopoietic stem cell (HSC), myeloid-derived suppressor cell (MDSC), B cells, macrophages, dendritic cells, NK cells, nature killer T cell (NKT), monocytes and neutrophils, which were summarized across various cancers using the R package “ggplot2”.

Genomic alterations, such as mutations, structural variations, amplifications, deep deletions, and multiple alterations, were obtained from the cBioPortal database (Gao et al., 2013; https://www.cbioportal.org/). Furthermore, the cBioPortal web tool was employed to visualize the rate of genomic alterations.

NCBP2 protein localization and interaction

The Human Protein Atlas (Uhlen et al., 2010) provides an interactive presentation showcasing the protein expression profiles of various human tissues. In order to illustrate the subcellular localization of NCBP2, immunofluorescence staining images of human cancer cell lines (U-251, A-431, U2-OS, MCF7, and PC-3) were utilized (Human Protein Atlas).

Clinical cancer samples

Several cancers samples, including lung cancer, kidney cancer and pancreatic cancer were collected from patients hospitalized in General Hospital of the Xinjiang Military Command from 2018 to 2022. The Medical Ethics Committee of the General Hospital of the Xinjiang Military Command (Urumqi, China) granted approval for the utilization of clinical excisions (2024RR0313), with written informed consent obtained from patients. The collection and utilization of clinical samples adhered strictly to established guidelines, as sanctioned by the Medical Ethics Committee of the General Hospital of the Xinjiang Military Command.

Immunohistochemistry

Formalin-fixed paraffin-embedded (FFPE) human cancer tissues, including lung squamous cell carcinoma (LUSC), KIRP, PAAD, and KIRC, were sliced into 3 µm sections and subsequently deparaffinized, rehydrated, and subjected to antigen retrieval. The sections were then incubated overnight at 4 °C with a primary NCBP2 rabbit polyclonal antibody (A7239; ABclonal, Woburn, MA, USA) and subsequently detected using a secondary antibody (goat anti-rabbit). Color development was achieved by applying DAB (K3486, Dako, Santa Clara, CA, USA), and hematoxylin was used as a counterstain. The immunostained sections were scanned using Motic EasyScan.

Expression and prognosis analysis

The expression levels of NCBP2 between tumor and normal groups in pan-cancer were compared using the “limma” package of R studio software. Additionally, the prognostic value of NCBP2 in pan-cancer was assessed through the utilization of univariate Cox regression analysis. To investigate the relationship between NCBP2 and prognosis, the OS, progression-free interval (PFI), and DSS were chosen as outcome measures. Kaplan–Meier analysis and univariate Cox regression were employed to evaluate the prognostic relevance of NCBP2 for different cancer prognosis types. The NCBP2 FPKM expression data was utilized for the univariate Cox regression analysis, while the bivariate Kaplan–Meier curve analysis was performed using median cutoffs. Subsequently, the log-rank P value and hazard ratio (HR) were determined using the Kaplan–Meier method, and the results were visually depicted on the Kaplan–Meier curve.

Differentially expressed genes analysis

According to the expression of NCBP2 mRNA, cancer patients were categorized into high-NCBP2 and low-NCBP2 groups, with the top 30% classified as high-NCBP2 patients and the bottom 30% as low-NCBP2 patients. The transcriptional profile in pan-cancer between these subgroups was analyzed using the “limma” package of R studio software. Genes with an adjusted P-value <0.05 were considered as DEGs.

Gene set enrichment analyses

The hallmark gene set (h.all.v7.5.1. entrez.gmt) was obtained from MSigDB Database (Subramanian et al., 2005) (https://www.gsea-msigdb.org/gsea/msigdb/) and employed to compute the normalized enrichment score (NES) and P value for each biological process associated with each cancer type (Subramanian et al., 2005). Following this, GSEA was performed using the R packages “clusterProfiler” and “ggplot2”.

Immunotherapy prediction analysis

A Spearman correlation analysis was employed to investigate the associations between NCBP2 and tumor mutational burden (TMB), microsatellite instability (MSI). In order to explore the connection between NCBP2 and the response to immune checkpoint blockade (ICB) therapy, data from two sources: IMvigor210, which consisted of 298 bladder cancer patients treated with atezolizumab (anti-PD-L1), and GSE91061, which included 26 melanoma patients treated with nivolumab (anti-PD-1), were utilized. Patients were categorized into low- and high-NCBP2 groups based on the cut-off value determined by the “ROC” R package. The proportion difference between low- and high-NCBP2 cancer groups was evaluated through Chi-square test.

Stemness index calculation and analysis

In order to assess tumor stemness, six prediction models were acquired from prior studies, which were based on mRNA expression data or DNA methylation data (Malta et al., 2018). Subsequently, a model utilizing the aforementioned equation was employed to compute stem cell indices for TCGA samples. The measurement of stemness involves comparing tumor cells and stem cells through an index that ranges from 0 to 1. To investigate the association between NCBP2 and stemness indices, Spearman correlation analysis was conducted and the outcomes were visualized using the R package “ggplot2”.

Drug sensitivity and NCBP2 expression analysis

The expression of NCBP2 in various cell lines was obtained from the Cancer Cell Line Encyclopedia (Ghandi et al., 2019) (CCLE; http://sites.broadinstitute.org/ccle/) database. The AUC value, which indicates cellular sensitivity to drugs of paired cell lines, was acquired from the Genomics of Drug Sensitivity in Cancer (GDSC) (Yang et al., 2013) database (https://www.cancerrxgene.org/celllines). Subsequently, Spearman correlation was used to estimate the relationship between drug sensitivity and NCBP2. A positive correlation suggests that cell lines exhibiting high NCBP2 expression were resistant to certain drugs, while a negative correlation suggests that cell lines with high NCBP2 expression were sensitive to specific drugs.

Single cell sequencing analysis

A single-cell assay was conducted on the CancerSEA (Yuan et al., 2019) platform (http://biocc.hrbmu.edu.cn/CancerSEA/) to examine the association between NCBP2 and different functional states of various cancers. A correlation threshold of correlation ≥0.3 and a p-value<0.05 were established to determine the significance of the relationship between NCBP2 and cancer functional status. Single-cell expression profiles of NCBP2 were visualized using t-SNE diagrams.

Statistical analyses

To ascertain statistical significance, the Wilcoxon rank-sum test was employed to compare the expression levels of NCBP2 in tumor and normal tissues. The prognostic significance of NCBP2 expression in each cancer type was evaluated through univariate Cox regression and Kaplan–Meier techniques, specifically log-rank tests. Furthermore, a Spearman correlation analysis was performed to examine the statistical relationship between NCBP2 and multiple factors, including levels of immune cell infiltration, genes involved in immune responses, TMB and TMI. To determine the statistical significance of the proportion of ICI-therapy responders and non-responders between low- and high-NCBP2 subgroups, a chi-square test was employed. A P value less than 0.05 was deemed to be statistically significant.

The genetic alteration of NCBP2 in pan-cancer

To investigate the role of nuclear cap binding proteins (NCBPs) in cancers, we analyzed the genomic alteration profiles of NCBPs including NCBP1, NCBP2, NCBP3, NCBP2AS2 and NCBP2-AS1 and NCBP2L in pan-cancer (TCGA, Firehose Legacy) using cBioPortal database (https://www.cbioportal.org/). As shown in Fig. 1A, NCBP2 was found to be the most frequently altered of the six genes, with nearly 7% of alterations occurring in pan-cancer, while the remaining genes showed relatively low mutation rates. In addition, Kaplan–Meier survival analysis revealed that patients with NCBP2 alteration exhibited a shorter OS and disease-free survival (DFS) compared with those without NCBP2 alterations (Figs. 1B & 1C). Then, we conducted a cancer type-based genetic alteration analysis of NCBP2 in the TCGA pan-cancer tumor samples. As shown in Fig. 1D, the general alteration frequency of NCBP2 was relatively high in the top five cancers, which had an alteration frequency of >20% with “Amplification” as the primary type, while the most frequent NCBP2 alteration was identified among LUSC patients (>40%). As a consequence, we further analyzed the relationship between the expression of NCBP2 and copy numbers in the top five cancers. According to our findings, compared with diploid samples, samples with NCBP2 amplification/gain/deletion exhibited highest/higher/lower mRNA expression level of NCBP2, respectively (Figs. S1A–S1E), indicating that copy numbers of NCBP2 appeared to be associated with its mRNA expression.

The expression of NCBP2 in pan-cancer

We further assessed the expression levels of NCBP2 across cancers using integrated data from TCGA and GTEx databases. NCBP2 was differentially expressed in 27 of 32 kinds of cancers. The expression of NCBP2 was higher in LUSC, esophageal carcinoma (ESCA), ovarian serious cystadenocardinoma (OV), cervical squamous cell carcinoma and endocervical adenocarcinoma (CESC), head and neck squamous cell carcinoma (HNSC), uterine carcinosarcoma (UCS), uterine corpus endometrial carcinoma (UCEC), lymphoid neoplasm diffuse large B-cell lymphoma (DLBC), stomach adenocarcinoma (STAD), bladder urothelial carcinoma (BLCA), lung adenocarcinoma (LUAD), prostate adenocarcinoma (PRAD), breast invasive carcinoma (BRCA), cholangio carcinoma (CHOL), brain lower grade glioma (LGG), pancreatic adenocarcinoma (PAAD), liver hepatocellular carcinoma (LIHC), skin cutaneous melanoma (SKCM), glioblastoma multiforme (GBM), kidney renal papillary cell carcinoma (KIRP), testicular germ cell tumors (TGCT), colorectal carcinoma (COADREAD), thyroid carcinoma (THCA) and thymoma (THYM), compared with corresponding normal tissues. In contrast, the expression of NCBP2 was lower in kidney renal clear cell carcinoma (KIRC), adrenocortical carcinoma (ACC), acute myeloid leukemia (LAML) (expression data for mesothelioma (MESO), pheochromocytoma and paraganglioma (PCPG), sarcoma (SARC) and uveal melanoma (UVM) were not shown for the absence of enough control normal tissue data) (Fig. 2A). As observed in the genetic analysis, most cancers expressed altered levels of NCBP2.

To investigate the location of NCBP2 in cells, the immunofluorescence (IF) staining images from Human Protein Atlas (HPA) database were acquired which indicated that NCBP2 protein was predominantly located in the nucleus and partially in cytoplasm of U-251, A-431, U2-OS, PC3 and MCF7 tumor cell lines (Fig. 2B). Moreover, the expression of NCBP2 protein levels were validated in clinical cancers and paired normal tissues by immunohistochemical staining, which showed that NCBP2 protein were up-regulated in LUSC, KIRP and PAAD compared with normal tissues, since lower expression of NCBP2 was observed in KIRC consistent with previous analysis (Figs. 2C–2F, Figs. S2A–S2D).

Clinical prognostic significance of NCBP2 in pan-cancer

To further investigate the prognostic potential of NCBP2 in cancers, univariate cox regression and Kaplan–Meier method (log-rank test) were performed for each cancer. A positive relationship between poor OS and NCBP2 was observed in UCEC, PAAD, LIHC, KIRP, KICH and KIRC as shown in the forest plot (Fig. 3A) and patients with high expression of NCBP2 shown worse OS compared to patients with low expression of NCBP2 in all of the 6 cancers (Figs. 3B–3G). Regarding the expression of NCBP2 and DSS, a significant positive association was observed in LIHC, PAAD, LUAD, and UCEC, and a negative association was observed in BLCA (Fig. S3A), and patients (LIHC, PAAD, UCEC) with higher expression of NCBP2 shown worse DSS compared to those with lower expression of NCBP2 (Figs. S3B–S3G). In view of PFI, higher expression of NCBP2 had a favorable influence in BLCA, however, it deemed to be a hazard factor in PAAD and UCEC (Figs. S3H–S3N). It is worth noting that NCBP2 was a risk factor for OS, DSS and PFI of PAAD and UCEC, which were significantly correlated with poor outcomes. Considering that highly expression of NCBP2 was associated with poorer OS for UCEC, PAAD, LIHC, KIRC, KICH and KIRP, multivariate cox regression analysis was performed to further explore the relationship between NCBP2 and OS. As shown in Figs. 3H–3M, NCBP2 was an independent risk factor of OS in PAAD, LIHC and KIRP. Overall, these data suggested that NCBP2 plays a prognostic role in several cancers, but the roles were complicated and multifaceted across cancers. Further investigation should focus on the function of the NCBP2 protein in cancer cells.

Gene set enrichment analysis of NCBP2 revealed its association with the cancer immune response

To elucidate the biological processes associated with NCBP2 in cancers, we performed DEGs analysis between the top (71–100%) and bottom (0–30%) NCBP2 expression subgroups in each cancer type. Subsequently, based on the DEGs, GSEA was executed across 33 cancer types to investigate the NCBP2-associated cancer processes. The results showed that immune-related pathways: IFN-alpha response, IFN-gamma response, IL-6/JAK/STAT3 signaling, IL-2/STAT5 signaling, allograft-rejection pathways, inflammatory response, and TNF alpha signaling-via NFKB, were remarkably enriched in a variety of tumors, especially in COAD, HNSC, LUSC and READ (Fig. 4), which indicated that NCBP2 may be involved in tumor immune microenvironment and ligand–receptor interactions between malignant tumor cells and immune cells. Moreover, we also found that NCBP2 was associated with partial proliferative pathways, including G2M checkpoint and E2F targets of many kinds of tumors which possessed significant positive correlation with BRCA, KIRP, LUAD, PAAD, PRAD and STAD, suggesting that NCBP2 could play a significant role in tumor proliferation. In summary, these results provide evidences that the abnormal expression of NCBP2 may be involved in the immune response and progress of cancers.

Association between immune-related factors and NCBP2

Since NCBP2 was correlated with several immune-related pathways, such as IFN-alpha response and TNF alpha signaling-via NFKB (Fig. 4), we further explored the correlations between NCBP2 and tumor microenvironment through Estimation of STromal and Immune cells in MAlignant Tumor tissues using Expression data (ESTIMATE) algorithm, which was performed to calculate the level of tumor stromal cells, and the infiltration immune cells in tumor tissues based on RNA sequencing data. The correlation between NCBP2 and stromal score, immune score, and ESTIMATE score were shown in Fig. 5A. Notably, NCBP2 expression were negatively associated with stromal score, immune score and ESTIMATE score in almost all cancer types except for ACC, CHOL, DLBC, KICH, LIHC and PAAD.

We further employed the TIMER2.0 database to assess the association between NCBP2 with various tumor infiltrating immune cells, which were determined by a variety of quantitative immune infiltration algorithms based on expression data, to show the infiltration levels of CD4+ T cells, CD8+ T cells, γδT cells, T cell regulatory (Tregs), T cell follicular helper (Tfh), cancer-associated fibroblast (CAF), endothelial cells (Endo), eosinophils (Eos), mast cells, progenitors of lymphoid/myeloid/monocyte, hematopoietic stem cell (HSC), myeloid-derived suppressor cell (MDSC), B cells, macrophages, dendritic cells, NK cells, nature killer T cell (NKT), monocytes and neutrophils in pan-cancer (Fig. 5B). In general, a positive correlation was observed between NCBP2 and various immune cells, such as MDSC, progenitors of lymphoid, CD4+ T cell, CAF, neutrophils and monocytes, however a negative correlation was identified with NKT cells in various cancers (Fig. 5B). Likewise, we found that NCBP2 expression was correlated with various immune infiltrating cells in each cancer, and the landscape of this correlation was different, which may be due to the other immune infiltration rates of certain tumors. Interestingly, we found that NCBP2 expression was significantly negatively correlated with CD8+ T cells, dendritic cells and neutrophils in various cancers, including COAD, HNSC and LUSC (Fig. 5B), which exhibited negatively enriched immune-related pathways (Fig. 4).

Due to the essential role of microenvironment in immunotherapy and the potential role of NCBP2 in microenvironment regulation, we aimed to investigate the correlation between NCBP2 and twenty-six immunomodulatory genes using Spearman correlation analysis. We found that NCBP2 was positively correlated with most of the immunomodulatory genes in LIHC, PAAD, PRAD, and UVM, while negatively correlated with most of immunomodulatory genes in CESC, COAD and LUSC (Fig. 6A). To dissect the role of NCBP2 in predicting the efficacy of ICB therapy, we analyzed the correlation between the expression of NCBP2 and TMB/MSI, which showed that NCBP2 was positively correlated with TMB in LUAD and THYM, and a negative association was found in COAD and UCEC (Fig. 6B). For MSI, a positive association in HNSC, KIRC, LUSC, PRAD and THYM, as well as a negative association in COAD and TGCT, was identified (Fig. 6C). Then, we analyzed the predictive role of NCBP2 in patients who received ICB therapy. In the urinary system tumors of the IMvigor210 cohort (bladder cancer), relationship between NCBP2 and the response to anti-PD-L1 treatment was analyzed. The patients exhibited four degrees of response, ranging from progressive disease (PD), stable disease (SD), partial response (PR), to complete response (CR). Results showed that patients with high NCBP2 expression had an anti-PD-L1 response rate of 30.48% (32/105), compared with that of 18.65% (36/193) in patients with low NCBP2 expression (Fig. 6D). Kaplan–Meier survival analysis revealed that patients with high NCBP2 expression had longer OS than those with low NCBP2 expression, and the hazard ratio for univariate cox was 0.62 (0.46–0.85) (Fig. 6E). In addition, the opposite results were found in patients with melanoma who received anti-PD-1 therapy in GSE91061 cohort (Figs. 6F & 6G).

Tumor stemness and chemotherapeutic value of NCBP2

Cancer stemness can be associated with immunotherapy resistance, but direct clinical evidence is lacking (Bayik & Lathia, 2021). We next employed several established algorithms to assess stemness of TCGA samples based on transcriptomic and epigenetic data. These stemness-indexes (mRNAsi, EREG-mRNAsi, mDNAsi, EREG-mDNAsi, ENHsi, DMPsi) derived from the above algorithms represent the degree of resemblance between tumor cells and stem cells. In general, NCBP2 was positively correlated with stemness-indexes in most cancers, except for ACC, CHOL, LAML, PCPG, and SARC (Fig. 7A). Notably, an apparent correlation between NCBP2 expression and all six stemness-indexes was found in LUSC and TCGT (Fig. 7A). These data suggested that NCBP2 might promote cancer progression by regulating cancer stemness in multiple cancers. Subsequently, drug sensitivity analysis was performed to appraised the correlation between NCBP2 and response to drugs in multiple cancer cell lines using data downloaded from the GDSC database, which indicated that NCBP2 was positively associated with sensitivity of 15 drugs (drug resistant), while negatively correlated with sensitivity of 54 drugs (drug sensitive) (Fig. 7B upper part). Furthermore, targets of the above drugs were analyzed, which showed that these NCBP2-associated drugs mainly targeted the DNA replication, chromatin histone methylation, ABL signaling, cell cycle and PI3K signaling (Fig. 7B lower part).

Functional analysis of NCBP2 at single cell level

As a means of deciphering the role of NCBP2 in cancers at the single-cell levels, CancerSEA was used to investigate its functional states, which showed that NCBP2 was positively associated with cell cycle, DNA damage, DNA repair, invasion and stemness for majority cancers excluding RB, UM, while negatively associated with apoptosis, inflammation, and hypoxia in majority cancers (Fig. 8A). Subsequently, we further examined the association between NCBP2 and several biological process of specific cancer types. Results indicated that NCBP2 was positive correlated with differentiation, metastasis, proliferation, inflammation and quiescence in AML; negative correlation with DNA repair, DNA damage, apoptosis, metastasis, invasion and quiescence in UM, noting that NCBP2 was positive correlation with angiogenesis, differentiation and inflammation, while negative correlation with DNA repair and cell cycle in RB (Figs. 8B–8D). Moreover, T-SNE diagrams were used to visualize NCBP2 expression profiles at the single-cell level in AML, UM and RB (Figs. 8E–8G).