Publicly available transcriptome data of TCGA pan-cancer and the related clinical features were obtained from the UCSC XENA (https://xena.ucsc.edu/). The expression matrices of 31 human normal tissues were downloaded from GTEX web portal (https://www.gtexportal.org/). The strawberry Perl script was developed (version 5.30.0.1, http://strawberryperl.com/) to extract the SRI expression data in 33 TCGA tumor types and GTEX normal tissues. The mRNA level of SRI in the healthy men and women tissues was visualized with “gganatogram” R package. The expression data were log2(TPM) transformed excluding missing data and duplicated values. The differences in the SRI expression between the tumor and normal tissues were examined by the Wilcoxon rank-sum test. The differential expression of SRI mRNA was evaluated in various tumor types using the Oncomine database (www.oncomine.org) (24). All the analyses were conducted using the R version 4.0.2 software (https://www.Rproject.org/). The overall workflow of our study is presented in Figure 1A.

The survival and clinical characteristics data were obtained from the UCSC XENA repository. The association between the SRI expression level and survival outcomes, such as overall survival (OS), progression-free interval (PFS), disease-specific survival (DSS), and disease-free survival (DFS), was evaluated using Kaplan–Meier method and Cox proportional hazards model. The “survival,” “survminer,” and “forestplot” packages were employed to draw the Kaplan–Meier and forest plots. The clinical records, such as patents age, tumor stage, and tumor status, were applied to investigate their relationship with the SRI expression. The data representations were performed using “limma” and “ggpubr” R-packages.

The SRI gene alternations in TCGA pan-cancer datasets across 10,953 patients were analyzed using cBioPortal database (http://www.cbioportal.org/) (25). The “Oncoprint” and “Cancer Type Summary” modules were used to investigate the genetic alterations of SRI. The “Mutations” module was applied to obtain the mutated site information of SRI. The effect of SRI alternations on the OS and gene mutation co-occurrence analysis were obtained from the “Comparison/Suvival” module. Furthermore, the copy number alterations data, mRNA Expression Z-scores data, and protein level Z-scores data of SRI in ovarian cancer were downloaded from the TCGA PanCancer Atlas dataset. To investigate the co-expression genes of SRI in ovarian cancer, the TCGA-OV RNA-seq data were accessed using the Linkedomics online tool (http://www.linkedomics.org/login.php/). The top 50 positive and negative correlated genes were selected to construct a co-expression network. The network was visualized in the STRING database (https://string-db.org/).

Cancer stemness was reported to be capable of evaluating by RNA stemness score (RNAss) based on mRNA expression (26). Correlation analysis between SRI expression and RNAss was examined using Spearman rank-based testing. Data were visualized with the R-package “corrplot”. We further determined the correlation between the level of SRI and cancer stem cell marker genes, epithelial-mesenchymal-transition (EMT)-related genes. The heatmaps were generated with the “reshape2” and “RColorBrewer” R-packages.

Recent studies have revealed that TMB and MSI could become independent biomarkers for immune checkpoint inhibitors response (27, 28). The TCGA pan-cancer mutation data were applied to calculate the TMB scores of each sample. The MSI scores of the TCGA pan-cancer samples were obtained according to a previous study (29). The analyses regarding the correlations between the SRI expression and TMB, MSI were calculated using Spearman's coefficient. The results were displayed as radar plots using the R-package “fmsb.” The tumors with high microsatellite instability are often characterized by a defective DNA mismatch repair system (MMR) system (30). The association between the SRI expression and MMR-related genes was further explored in cancers. The results were visualized as the heatmaps using the “reshape2” and “RColorBrewer” R-packages.

The Estimation of Stromal and Immune cells in Malignant Tumors using Expression data (ESTIMATE) algorithm was performed to calculate the ImmuneScore, StromalScore, and ESTIMATEScore using the R package “ESTIMATE” (31). The results were displayed using the R-package “corrplot” using Spearman's rank-based testing. Moreover, we analyzed the correlation of SRI expression with the abundance of various immune cell infiltrates in pan-cancer using the TIMER database (32) (https://cistrome.shinyapps.io/timer/). In addition, a correlation analysis of SRI and immune checkpoint genes was performed. The results were processed using the “reshape2” and “RColorBrewer” R-packages.

The NCI-60 compound activity data and the RNA-seq expression profiles were obtained from CellMiner (https://discover.nci.nih.gov/cellminer/home.do) (33). The drugs that were considered FDA approved or in the clinical trials were selected for further analysis. Then, the effect of SRI on the drug sensitivity was analyzed using the “impute,” “limma,” “ggplot2,” and “ggpubr” R package. The GSEA analysis was performed to explore the biological functions of SRI in cancer. The gene sets of Kyoto Encyclopedia of Genes and Genomes (KEGG) and Gene Ontology (GO) signature were obtained from GSEA online (https://www.gsea-msigdb.org/gsea/downloads.jsp). The GSEA analysis was processed with the R-packages “enrichplot,” “org.Hs.eg.db,” “clusterProfiler,” “DOSE,” and “limma”.

The cell culture and sphere formation assays were performed according to the previously described method (34). The sorcin expression in the ovarian cancer sphere cells was evaluated using a western blot. Anti-SRI antibody (ab71983) was purchased from Abcam (Cambridge, UK).

A log-rank test was applied in the Kaplan–Meier survival curves. Hazard ratio (HR) was determined by a Cox proportional hazard regression model. The correlation analysis was executed by Spearman's rank test. The statistical analyses were performed using the R version 4.0.2 software (R Foundation for Statistical Computing, Vienna, Austria) or GraphPad Prism 8 (San Diego, CA, USA). Results with P < 0.05 were considered statistically significant (^*P < 0.05, ^**P < 0.01, ^***P < 0.001, and ^****P < 0.0001).

To gain insights into the expression pattern of SRI in the human normal tissues, the SRI expression in tissue physiological state was investigated according to GTEX dataset. SRI was highly expressed in the small intestine, bone marrow, brain, and breast tissues, while the skeletal muscle and liver tissues expressed low levels of SRI (Figure 1B). The SRI expression abundances of various tissues in men and women are displayed in Supplementary Figures 1A,B. Overall, no gender difference was observed in the mRNA expression levels of SRI (Supplementary Figure 1C).

To further explore the SRI expression in human cancers, the SRI expression in various types of cancers was analyzed using the RNA-seq data of TCGA database (Figure 1C). The aberrant expression of SRI was detected in 28 types of cancer except for those cancers where no normal tissue data were available. The SRI expression was significantly higher in bladder urothelial carcinoma (BLCA), breast invasive carcinoma (BRCA), cholangiocarcinoma (CHOL), colon adenocarcinoma (COAD), lymphoid neoplasm diffuse large B-cell lymphoma (DLBC), esophageal carcinoma (ESCA), glioblastoma multiforme (GBM), head and neck squamous cell carcinoma (HNSC), kidney chromophobe (KICH), kidney renal clear cell carcinoma (KIRC), kidney renal papillary cell carcinoma (KIRP), brain lower grade glioma (LGG), liver hepatocellular carcinoma (LIHC), lung adenocarcinoma (LUAD), lung squamous cell carcinoma (LUSC), ovarian serous cystadenocarcinoma (OV), pancreatic adenocarcinoma (PAAD), prostate adenocarcinoma (PRAD), rectum adenocarcinoma (READ), skin cutaneous melanoma (SKCM), stomach adenocarcinoma (STAD), testicular germ cell tumors (TGCT), thymoma (THYM), uterine corpus endometrial carcinoma (UCEC), and uterine carcinosarcoma (UCS). In contrast, the SRI levels were significantly downregulated in adrenocortical carcinoma (ACC), acute myeloid leukemia (LAML), and thyroid carcinoma (THCA). Aside from this, the higher level of SRI was reconfirmed in the brain, bladder, esophageal, gastric, head and neck, kidney, liver, melanoma, myeloma, pancreatic, and prostate cancer compared with the normal tissues using ONCOMINE database (Supplementary Figure 1D).

To further explore the prognostic value of SRI in pan-cancer, the association between the SRI expression level and survival of patients was evaluated using the Cox proportional hazards model and the Kaplan-Meier analysis. The results from a Cox proportional hazards regression model revealed that the SRI expression levels were correlated with OS in BLCA (P = 0.030), HNSC (P = 0.023), KIRP (P = 0.009), LIHC (P < 0.001), LGG (P = 0.001), PAAD (P = 0.033), SKCM (P = 0.006), STAD (P = 0.043), THYM (P = 0.05), and uveal melanoma (UVM) (P = 0.048) (Figure 2A). SRI served as a high-risk factor for HNSC, KIRP, LIHC, PAAD, STAD, and UVM, while it acted as a low-risk gene for BLCA, LGG, SKCM, and THYM. Furthermore, the Kaplan–Meier survival analysis demonstrated that the high level of SRI predicted poor OS in cervical squamous cell carcinoma and endocervical adenocarcinoma (CESC) (Figure 2B, P = 0.020), CHOL (Figure 2C, P = 0.044), KIRP (Figure 2D, P = 0.050), LIHC (Figure 2E, P = 0.023), and PAAD (Figure 2F, P = 0.026). While low expression of SRI was correlated with the shortened OS in SKCM (Figure 2G, P = 0.003). Regarding the PFS, the high SRI level represented an adverse factor in CESC (P = 0.012), HNSC (P = 0.016), LIHC (P = 0.049), PAAD (P = 0.023), PRAD (P = 0.009), STAD (P = 0.008), and UVM (P = 0.008), while the high SRI level was considered as a favorable factor in BLCA (P = 0.002), BRCA (P = 0.004), and LGG (P = 0.017) (Figure 3A). The Kaplan–Meier curves for PFS revealed a correlation between the high SRI expression level and poor survival time in the patients with CESC (Figure 3C, P = 0.038), HNSC (Figure 3D, P = 0.040), KIRC (Figure 3E, P = 0.009), and UCEC (Figure 3G, P = 0.016). The patients with high SRI expression had significantly longer PFS than the patients with low expression in BLCA (Figure 3B, P = 0.012) and THCA (Figure 3F, P = 0.028). Regarding the associations between the SRI expression and DFS, the forest plots showed high SRI expression predicted poor DFS in CESC (P = 0.002), OV (P = 0.020), and STAD (P = 0.040) (Figure 4A). Significant relationships between the SRI expression and DFS were observed in CESC (Figure 4B, P = 0.013) and THCA (Figure 4C, P = 0.005) by the Kaplan–Meier survival analysis. Furthermore, SRI exhibited a significant prognostic value in BLCA (P = 0.004), BRCA (P = 0.006), LIHC (P = 0.008), PAAD (P = 0.034), STAD (P = 0.021), UCEC (P = 0.014), and UVM (P = 0.05) in a Cox proportional hazards regression model for DSS (Figure 4D). The Kaplan–Meier survival analysis found that the PAAD (Figure 4E, P = 0.025) and UCEC (Figure 4F, P = 0.008) patients with high expression of SRI had shortened DSS. Additionally, the survival analysis of SRI in some types of cancer was validated in the GEO database (Supplementary Figure 2).

Next, the correlations between the SRI expression and the clinicopathological features of patients were investigated in pan-cancer. Patients with age ≥65 years had higher expression of SRI in KIRC (Supplementary Figure 3A; P = 0.011) and PRAD (Supplementary Figure 3B; P = 0.012). No significant difference was observed between the SRI expression and age in other cancers. We further compared the differential mRNA level of SRI in different tumor stages. SRI expression tended to decrease from stage I to stage IV in BLCA (Figure 5A). In contrast, the high SRI expression tended to associate with the advanced tumor stages in LIHC (Figure 5B). The difference between stage I and IV tumors was not statistically significant in LIHC, one reason that might be responsible is the small patient numbers in the advanced stage. Tumor status after treatment was closely associated with disease recurrence. We found that the high level of SRI was significantly correlated with-tumor status in ESCA (Figure 5D), GBM (Figure 5E), KIRC (Figure 5F), LIHC (Figure 5G), PRAD (Figure 5H), and UVM (Figure 5I). While a high expression of SRI was significantly related to the tumor-free status in BLCA (Figure 5C).

Next, the cBioPortal database was applied to investigate the genetic alterations of SRI in the TCGA pan-cancer datasets. As shown in Figure 6A, the SRI gene was altered in 168 patients with cancer which accounted for only 1.5% across 10,953 samples. Regarding the SRI alterations in different cancer types (Figure 6B), the most frequent alterations of SRI gene were amplification in ESCA, DLBC, STAD, LUSC, HNSC, CHOL, OV, PAAD, GBM, UCS, BLCA, PRAD, CESC, LIHC, TGCT, ACC, LUAD, BRCA, KIRC, LGG, and KIRP. Patients with UCEC, SKCM harbored High frequency of SRI mutations was observed in UCEC, SKCM, while patients with LAML and THCA harbored high frequency of SRI deep deletion of SRI gene was the most frequent mutation type in LAML and THCA. The mutation types, number, and sites of the SRI genetic alterations are displayed in Figure 6C. The results showed that the missense mutations were the major mutation type of SRI. Intriguingly, co-occurrence of STEAP4, ADAM22, ZNF804B, DBF4, CFAP69, ABCB1, STEAP2, ABCB4, RUNDC3B, and TEX47 alterations was observed with the SRI alterations (Figure 6D). While the patients with SRI alterations represented only a small part, the patients in the SRI genetic altered group had poorer OS than those in the SRI unaltered group (Figure 6E). To explore whether the SRI amplification had an influence on its mRNA and protein level, the copy number alterations data and expression data of SRI were acquired in TCGA ovarian cancer. The results indicated that SRI was amplified along with the significantly high mRNA and protein level in TCGA-OV cohort (Figures 6F,G). Regarding the co-expression genes of SRI, the co-expression analysis was conducted using the Linkedomics database in the TCGA ovarian cancer dataset. The results are presented in Supplementary Table 1. The top 50 positive and negative co-expression genes of SRI were visualized in the STRING database (Supplementary Figure 4).

Our previous work revealed that SRI promoted the stemness and EMT process in ovarian cancer (35), we, therefore, wanted to investigate the association of SRI expression with stemness score and EMT-related genes in cancers. First, the sorcin expression was upregulated in the OVCAR-3 and SKOV-3 spheres cells compared with the respective adherent cells (Supplementary Figure 5). Correlation analysis indicated that the SRI expression was positively associated with the RNAss in GBM, KIRC, LAML, LGG, OV, PADD, pheochromocytoma and paraganglioma (PCPG), UCEC, and UCS. While a negative relationship between the SRI level and RNAss was observed in ACC (Figure 7A). Figure 7B depicts the correlations between the SRI expression and 18 tumor stem cell markers. More than 10 cancer stem cell markers expression were significantly positively correlated with the SRI levels in GBM, KIRC, LGG, LIHC, SKCM, THCA, and UCEC. Furthermore, a heatmap showed a positive correlation between the SRI expression and 25 EMT-related genes (Figure 7C). The expression of over 15 EMT-related genes was positively associated with the SRI expression in COAD, GBM, KIRC, LIHC, and OV. The correlations among the TMB, MSI, and SRI expression were further analyzed in cancers. The results demonstrated that the SRI expression was significantly associated with increased MSI in BRCA, DLBC, ESCA, HNSC, KIRC, PAAD, THCA, and UCEC, while the SRI expression was negatively associated with MSI in COAD, GBM, LUAD, and LUSC (Figure 8A). In addition, SRI was positively correlated with TMB in ESCA, HNSC, LGG, PAAD, SKCM, and STAD, while a negative correlation between the SRI expression and TMB was found in ACC, BRCA, CESC, COAD, OV, and THCA (Figure 8B). As shown in Figure 8C, the SRI expression was significantly positively correlated with the MMR-related genes level in most of the tumors, especially in KIRP, LIHC, SKCM, UCEC, and UVM.

Presently, the predictive role of SRI in tumor immune microenvironment has received little attention. Here, the association of SRI expression with the tumor immune microenvironment was evaluated according to the ESTIMATE algorithm and TIMER database. Our findings showed that the SRI expression had a positive correlation with the immune scores and estimate scores in ACC, ESCA, CHOL, LIHC, PRAD, and SARC (sarcoma). A strong negative correlation between the SRI and immune scores and estimate scores was found in LGG, MESO, and PAAD (Figures 9A,B). In addition, the correlation analyses revealed that the stromal scores were negatively correlated with the SRI expression in GBM, LGG, MESO, PADD, and UCS (Figure 9C). Then, the relationship between the SRI expression and immune cells infiltration was investigated in pan-cancer. The results indicated that the SRI expression was significantly associated with tumor purity in five cancer types. Furthermore, the SRI expression was significantly correlated with the infiltration levels of B cells in 14 cancer types, CD8+T cells in 14 cancer types, CD4+T cells in 13 cancer types, the macrophages in 21 cancer types, the neutrophils in 16 cancer types, and the dendritic cells in 12 cancer types (Figures 9D–H; Supplementary Figures 6, 7). Combined results from the correlation analysis and TIMER database, the SRI expression was strongly correlated with the immune infiltrating level in ACC, ESCA, LIHC, LUSC, and PRAD. The correlation between the immune checkpoint genes expression and SRI expression was explored. The results demonstrated that the most immune checkpoint genes were positively correlated with the SRI expression, especially in LIHC, PRAD, UVM, and ACC (Supplementary Figure 8), which suggested that the high level of SRI might mediate immune escape.

Since the drug resistance role of SRI in cancer has been gradually valued, we further investigated the potential correlation analysis between the drug sensitivity and SRI expression using the CellMiner™ database. Our results indicated that the SRI expression was positively related to JNJ-38877605, Simvastatin, BMS-777607, PF-04217903, and XAV-939 sensitivity (Figures 10A,B,G,J,O). Notably, the SRI expression was negatively correlated with the drug sensitivity of fluorouracil, dolastatin 10, paclitaxel, actinomycin D, tegafur, docetaxel, isotretinoin, mithramycin, elesclomol, EMD-534085, and dinaciclib (Figures 10C–F,H,I,K–N,P).

The GSEA was then performed to explore the main biological process affected by SRI in cancer. In the KEGG pathway gene set analysis, our results suggested that SRI positively regulated the immune-related pathways in CHOL, COAD, PRAD, such as natural cell-mediated cytotoxicity, antigen processing, and presentation. SRI was positively enriched in the metabolism-related pathways of OV and PRAD, such as oxidative phosphorylation, ascorbate and aldarate metabolism, and chlorophyll metabolism. In contrast, SRI was negatively enriched in the glyoxylate and dicarboxylate metabolism and drug metabolism cytochrome P450 in READ and SKCM. In addition, SRI was identified as a negative regulator for olfactory transduction, autophagy, WNT signaling pathway in LGG, READ, OV, UCEC, and STAD (Figure 11). The GO results of GSEA analysis of SRI in pan-cancer are displayed in Supplementary Figures 9, 10. In ACC, CESC, HNSC, LIHC, LAML, and SKCM, the SRI expression showed positive enrichment in the gene regulatory mechanisms, such as gene silencing, alternative mRNA splicing, transcription activator activity, and methylation CPG binding. Several immune response pathways, such as immune response regulating cell surface receptor signaling, regulation of immune effector process, and response to interleukin 12 were positively correlated to the SRI expression in STAD, CHOL, and SARC. Furthermore, a negative enrichment among the cell cycle G1/S transition, regulation of epidermal cell growth, and SRI expression was observed in OV, LUSC, PRAD, UCS, PCPG, READ, and ESCA.