Expression patterns of NUDCD1 in normal tissues and cells
To detect the mRNA and protein expression profiles of NUDCD1 in human normal and cancer tissues, the NUDCD1 expression data for normal tissues were evaluated using the HPA database. An overview of NUDCD1 mRNA and protein expression data indicated that NUDCD1 distribution possessed only a low tissue specificity (Fig. S1a). NUDCD1 exhibited high mRNA expression levels in placenta, testis and urinary bladder (Fig. 1a). The NUDCD1 mRNA expression in GTEx (Genotype-tissue expression), FANTOM5 (Function annotation of the mammalian genome 5) and consensus dataset (dataset created by combining the HPA and GTEx) was shown in Fig. S1b-d. NUDCD1 protein expression was widely expressed with medium-low levels in most normal tissues (Fig. 1b) and was primarily located in the nucleoplasm and cytosol (Fig. 1c). Moreover, NUDCD1 was expressed in several cancers including testicular and colorectal cancer (Fig. 1c). The bulk tissue gene expression from GTEx analysis revealed that NUDCD1 was highly expressed in cultured fibroblasts and EBV-transformed lymphocytes but expressed at only low levels in whole blood (Fig. 1d). Single cell snRNA-seq single tissue expression for NUDCD1 was shown in Fig. 1e. NUDCD1 possessed 4 isoforms (ENST00000521439.1, ENST00000427660.6, ENST00000519607.5 and ENST00000239690.8) differentially expressed in tissues (Fig. S1e) and the junctions and exon expression levels were defined for these isoforms (Fig. S2a, b). NUDCD1 immune cell expression indicated a low cell type-specificity (Fig. S2c-e) and NUDCD1 expression was elevated in extracellular vesicles found in urine and lowest in the glomerular basement membrane (Fig. S2f).
Expression patterns of NUDCD1 in cancer tissues and cells.
NUDCD1 has been previously defined as a cancer antigen and is abnormally upregulated in multiple tumors and acts as an oncogene [10]. NUDCD1 expression level was elevated in tumor tissues compared with the corresponding normal counterparts for BLCA (Bladder urothelial carcinoma), BRCA (Breast invasive carcinoma), CESC (Cervical squamous cell carcinoma and endocervical adenocarcinoma), CHOL (Cholangio carcinoma), COAD (Colon adenocarcinoma), ESCA (Esophageal carcinoma), GBM (Glioblastoma multiforme), HNSC (Head and Neck squamous cell carcinoma), LIHC (Liver hepatocellular carcinoma), LUAD (Lung adenocarcinoma), LUSC (Lung squamous cell carcinoma), READ (Rectum adenocarcinoma) and STAD (Stomach adenocarcinoma) (Fig. 2a). NUDCD1 expression was abnormally expressed in DLBC (Lymphoid neoplasm diffuse large B-cell lymphoma), LAML (Acute myeloid leukemia), THYM (Thymoma) compared with normal tissue (Fig. 2b) and was also present in BRCA subtypes (Fig. 2c). NUDCD1 total protein was elevated in ovarian, colon, clear cell RCC (Renal cell carcinoma), UCEC (Uterine corpus endometrial carcinoma) and LUAD compared with normal tissues (Fig. 2d). The enrichment score rankings (Fig. 2e) were then associated with pathways effected by NUDCD1 (Fig. 2f).
Tissue-based RNA expression can be affected by tumor heterogeneity or individual patient differences so we further analyzed NUDCD1 RNA expression in different cell lines. NUDCD1 was highly expressed in the cells from meningioma and CML (Chronic myelogenous leukemia), and low in the cells from chondrosarcomas and giant cell tumors (Fig. S3a). Immunofluorescence microscopy confirmed that the NUDCD1 protein was primarily localized to the nucleoplasm (Fig. S3b). NUDCD1 mRNA expression and copy numbers were linearly correlated (Spearman 0.48, p < 0.01, Fig. S3c) while there was no statistically significant correlation between NUDCD1 mRNA and DNA methylation (RRBS) (p = 0.847, Fig. S3d). The overall expression of NUDCD1 in cell lines indicated that mesenchymal cells expressed the highest levels of NUDCD1 mRNA (Fig. S3e). We examined the effect of NUDCD1 expression on the pathological stages of cancers from the “Pathological Stage Plot” module of GEPIA2, TISIDB (web portal for tumor and immune system interaction), and GSCA (Fig. S4a-c), that could also correlate NUDCD1 expression to tumor pathologic stage for KICH (Kidney chromophobe), KIRP (Kidney renal papillary cell carcinoma) and LUAD (Fig. S4d). NUDCD1 mRNA expression in tumor tissues was significantly higher than that in normal tissues for in BRCA, BLCA, HNSC, LUSC, ESCA, LUAD, LIHC (Liver hepatocellular carcinoma), STAD and COAD but not THCA (Thyroid carcinoma) (Fig. S5a). Subtype analysis indicated associations between NUDCD1 expression and immune or molecular subtypes for BRCA, LUAD and UCEC (Fig. S5b, c) and subtypes could be grouped between high and low NUDCD1 expression (Fig. S5d). Moreover, NUDCD1 expression was also positively correlated to tumor grades in CESC, LGG (Brain lower grade glioma), LIHC and UCEC and negatively correlated to STAD tumor grade (Fig. S5e).
Prognostic Significances Of Nudcd1 In Different Cancers
NUDCD1 is aberrantly expressed in numerous tumor tissues and cells so we also explored the prognostic relevance of NUDCD1 in cancers. First, analyses based on TISIDB revealed that high expression of NUDCD1 indicated a shorter OS (overall survival) in 8 types of cancers (Fig. 3a). Data from GEPIA2 suggested that NUDCD1 expression was correlated with OS in 7 types of cancers (Fig. 3b). Moreover, Kaplan-Meier Plotter (KM Plotter) was used to draw overall survival curves of cancer patients (Fig. 3c) and the OS from GSCA indicated distinct survival differences between high and low NUDCD1 expression groups in multiple cancers (Fig. 3d). The combined results of above 4 databases indicated that in BRCA, LUAD and SARC (Sarcoma), increased expression of NUDCD1 predicted a poor overall survival (Fig. 3e).
Immunohistochemistry data from HPA indicated that the NUDCD1 protein was more highly expressed in tumor samples compared to normal tissues in BRCA, LUAD and SARC (Fig. S6a). Regrettably, prognostic analysis of the NUDCD1 signatures in these 3 cancers revealed that its expression had weak to moderate implications for 1, 3, 5-year survival (Fig. S6b-d). We also calculated disease-free survival (DFS) (Fig. S7a) and relapse-free survival (RFS) (Fig. S7b). We found a significant effect of NUDCD1 expression on DSS (disease-specific survival), DFI (disease-free interval) and PFS (progression-free survival) (Fig. S7c). Increased expression of NUDCD1 was a predictor of a poor DFS especially in SARC.
The Genetic And Epigenetic Features Of Nudcd1 In Cancers
Genetic alterations of NUDCD1 in different tumor samples indicated that the highest alteration frequency (> 15%) appeared for patients with ovarian serous cystadenocarcinoma. In contrast, low NUDCD1 alteration frequencies were detected with adrenocortical cancer, cholangiocarcinoma, kidney renal papillary cell carcinoma and thymoma. The primary type for these mutations was gene amplification (Fig. 4a) and the general mutation type, structural variants and copy numbers in TCGA cancer types also varied (Fig. 4b). Survival analysis revealed that cancer patients with a mutation in NUDCD1 had a significantly worse DFS and PFS than the patients without NUDCD1 mutations (Fig. 4c) and there was no significant difference between DSS and OS (Fig. S8a).
A total of 103 mutation sites for the gene was identified, including 80 missense, 16 truncations, 2 splice and 5 fusion mutations between codons 1 and 583 (Fig. 4d). In addition, “Single Nucleotide Variation” (SNV) (Fig. S8b) and “Copy Number Variation” (CNV) (Fig. S8c, d) were found that included 32 cancer types. There were also significant correlations of NUDCD1 CNV with its mRNA expression in most types of cancers with the exception of THCA, LAML and DLBC (Fig. S8e). The SNV percentage heatmap and CNV percentage in each cancers were displayed in Fig. 4e and Fig. 4g, respectively. The highest SNV mutation frequency of NUDCD1 (21%) appeared in patients with UCEC while there were none found for patients with CESC and DLBC (Fig. 4e). Survival differences between mutant and WT for SNV indicated a significantly worse DFI profile for LUAD, DSS in BRCA and PFS in UCEC (Fig. 4f). Moreover, in KIRP and UCEC, NUDCD1 CNV was significantly worse for DFI, DSS, OS and PFS (Fig. 4h).
Methylation of NUDCD1 in cancers.
DNA methylation of genes also plays a key role in the regulation of cancer progress and we used the GSCA database to explore the DNA methylation of NUDCD1 in pan-cancers. Methylation differences between tumor and normal samples occurred for BLCA, BRCA, COAD, HNSC, LIHC and PAAD (Pancreatic adenocarcinoma) (Fig. 5a). DNA methylation of NUDCD1 was significantly and negatively correlated with its mRNA expression in most cancers with the exception of GBM, CHOL, LAML, THYM and THCA (Fig. 5b). UVM (Uveal melanoma) patients with NUDCD1 methylation had better DSS, OS and PFS than the patients without methylation. Similarly, SARC patients with NUDCD1 methylation had a better prognosis for DSS and OS while KICH patients with NUDCD1 methylation possessed a greater PFS (Fig. 5c). Clinical data and expression, copy number and DNA methylation data for UVM, SARC and KICH are depicted in Fig. S9a-c.
Interactions Between Tumor-immune System And Nudcd1
The interaction between tumors and the immune system plays a crucial role in cancer initiation, progression, and treatment. Therefore, elucidation of tumor and immune cell interplay would assist both the prediction of immunotherapy responses and the development of novel immunotherapy targets. TISIDB is a web portal for tumor and immune system interaction and integrates multiple heterogeneous data types. This portal was used to identify potential correlations between NUDCD1 expression and TILs, immunoinhibitors, immunostimulators, MHC, chemokines and receptors (Fig. 6a-f). It was found that NUDCD1 expression was not significantly different between responders and non-responders (Fig. 6g). However, NUDCD1 did possess significant mutation differences between responders and non-responders for anti-CTLA-4 (ipilimumab) therapy for melanoma (Fig. 6h).
The mismatch repair (MMR) pathway plays a critical role in identifying and repairing mismatched bases during DNA replication and genetic recombination [31]. It was found that NUDCD1 expression was positively correlated with MSI (microsatellite instability) in STAD while negatively correlated with MSI in DLBC (Fig. 6i). NUDCD1 expression was positively correlated with TMB (tumor mutation burden) in STAD, LUAD and BRCA while negatively correlated with TMB in UVM and THCA (Fig. 6j).
Interactions Between Nudcd1 And Immune Infiltration In Cancer
To explore whether NUDCD1 relates to the process of immune infiltration in cancers, TIMER2 was employed to examine NUDCD1 expression in tumor-immune infiltrates. Overall, its gene expression was positively correlated with immune infiltrating levels of CD4 + and CD8 + T cells, macrophages, neutrophils, CAP, common myeloid progenitor X cells and MDSC (Fig. 7). The negative and positive Spearman correlations between NUDCD1 expression and immune infiltrates are presented respectively in Fig. S10a, b. We also explored relationships between NUDCD1 methylation and immune infiltrates (Fig. S10c, d) and identified differences of immune infiltration between mutant and WT NUDCD1 in specific cancers (Fig. S10e) and correlations between NUDCD1 CNV and immune infiltrates (Fig. S10f, g).
Correlations Between Nudcd1 Expression And Drug Sensitivity In Cancers
The Cancer Therapeutics Response Portal (CTRP) was used to link genetic, lineage and other cellular features of cancer cell lines to small-molecule sensitivity and Genomics of Drug Sensitivity in Cancer (GDSC) portal was used to link drug sensitivity in cancer cells and molecular markers of drug response. We found that NUDCD1 expression was significantly and negatively correlated with the CTRP drug sensitivity (IC50) for BRD-K01737880, BRD-K33514849 and GSK-J4 (Fig. 8a, Table S1). However, NUDCD1 expression was significantly and positively related to GDSC IC50 for PLX4720, SB590885 and selumetinib while significantly and negatively correlated with NPK76-II-72-1, AICAR and BX-795 (Fig. 8b, Table S2). The CTD was also used to establish an interaction net-work between NUDCD1-chemicals-cancers (Fig. 8c).
Enrichment analysis of NUDCD1-related partners
To further explore the molecular mechanism of NUDCD1 in human cancers, we established a protein-protein interaction (PPI) network including 50 NUDCD1-interacted proteins derived from STRING (Fig. 9a). The top 100 genes that correlated with NUDCD1 expression indicated 4 common members: FAM91A1, DCAF13, MED30 and DDX21 (Fig. 9b), and their corresponding expression in different tumor types could be established (Fig. 9c). 146 NUDCD1-related -interacting and -correlated genes were also identified (Fig. 9d). Protein-protein interactions were also identified for NUDCD1-related genes (Fig. S11a) and included 6 MCODE algorithm components (Fig. S11b). NUDCD1-related genes were compiled according to gene set enrichment analysis (GSEA) and gene set variable analysis (GSVA) scores to investigate genomic variations and clinical outcomes. The highest GSEA scores were found for COAD, STAD and ESCA (Fig. 9e and Fig. S11c). NUDCD1-related genes were correlated with the cancer-related pathways of apoptosis, cell cycle and DNA damage (Fig. 9f).
The GSVA scores in different tumors and subtypes of cancers are summarized in Fig. 9g and Fig. S12a, respectively. Most cancer types when compared to non-tumor tissues possessed significantly higher GSVA scores in tumors. These scores were also linked to clinical/mutation/CNV/expression in THYM, UVM, HNSC, THCA, SARC, PAAD, MESO (Mesothelioma), LUAD, LIHC, KIRP, KICH, BRCA and ACC (Adrenocortical carcinoma). The high GSVA score groups had a higher hazard ratio for survival (Fig. 9h). THYM, SKCM (Skin Cutaneous Melanoma), LAML, GBM, BLCA patients with mutation of NUDCD1-related gene set suggests shorter survival than those in the WT group (Fig. 9i). There was an association identified between the NUDCD1-related gene set and CNV survival (Fig. 9j). The NUDCD1-related gene set was also negatively correlated with immune infiltration score, CD4-T, NK, Tfh, MAIT, NKT, Gamma delta, CD8-T, cytotoxic and Th2; while positively correlated with nTreg, neutrophil, central memory and iTreg (Fig. S12b).
The association between NUDCD1-related gene set mutation and immune infiltration in different cancer types is summarized in Fig. S12c. NUDCD1-related genes also impacted drug susceptibility. The expression of NUDCD1-related gene set was negatively correlated CTRP and GDSC drug susceptibility (IC50) in most cases. Notably, expression of NUDCD1-related gene set (TXNRDI, STK3, PSMD5, DHX32, COPZ2 and COPB2) was positively correlated with the drug susceptibility (IC50) in both CTRP (Fig. S12d) and GDSC (Fig. S12e).
Validation Of NUDCDI Expression And Function In STAD
To further validate our results, we examined NUDCD1 mRNA and protein expression using STAD and pericarcinous tissues. We found that the mRNA expression of NUDCD1 was significantly higher than that in pericarcinous tissues (Fig. 10a); moderate and strong staining of NUDCD1 was mainly observed in high-grade STAD tissues, while most of the low-grade tumor tissues showed weak staining (Fig. 10b). Among the five human cell lines (CES-1, AGS, SGC7901, MKN28 and HGC-27), NUDCD1 showed highest expression in AGS and HGC-27 (Fig. 10c). Next, stably NUDCD1-knockdown or NC AGS and HGC-27 cells were constructed and validated (Fig. 10d, e). Base on above, NUDCD1 was correlated with the cancer-related pathways of apoptosis and cell cycle in STAD (Fig. 9f), so we detected the apoptosis and cycle in AGS and HGC-27 cells. Results showed that NUDCD1-knockdown increased the percentage of apoptotic STAD cells (Fig. 10f); moreover, elevated the proportion of cells in G0/G1 phase and decreased proportion of cells in S phase were also observed in the comparison to negative control (Fig. 10g). As expected, colony formation assays revealed that NUDCD1 knockdown significantly decreased the formation capacity in vitro (Fig. 10h). Consistent with that, to knockdown NUDCD1 in vivo could significantly suppress the carcinogenesis of STAD cells in nude mice (Fig. 10i).
