Transcriptome data and clinical information for patients of 33 types of cancers were downloaded from the TCGA data portal (http://cancergenome.nih.gov/), and the Genotype-Tissue Expression (GTEx) portal. 33 cancer types including ACC, adrenocortical carcinoma; BLCA, bladder urothelial carcinoma; BRCA, breast invasive carcinoma; CESC, cervical squamous cell carcinoma; CHOL, cholangiocarcinoma; COAD, colon adenocarcinoma; DLBC, lymphoid neoplasm diffuse large B-cell lymphoma; GBM, glioblastoma multiforme; HNSC, head and neck squamous cell carcinoma; KICH, kidney chromophobe; KIRC, kidney renal clear cell carcinoma; KIRP, kidney renal papillary cell carcinoma; LAML, acute myeloid leukemia; LIHC, liver hepatocellular carcinoma; LUAD, lung adenocarcinoma; LUSC, lung squamous cell carcinoma; MESO, mesothelioma; OV, ovarian serous cystadenocarcinoma; PAAD, pancreatic adenocarcinoma; PCPG, pheochromocytoma and paraganglioma; PRAD, prostate adenocarcinoma; READ, rectum adenocarcinoma; SARC, sarcoma; SKCM, skin cutaneous melanoma; TGCT, testicular germ cell tumors; THCA, thyroid carcinoma; THYM, thymoma; UCEC, uterine corpus endometrial carcinoma; UCS, uterine carcinosarcoma; UVM, uveal melanoma.

The RNA sequencing data in TPM (transcripts per million reads) format were downloaded from TCGA and GTEx portal and processed by the Toil (Vivian et al., 2017). SERPINH1 expression data were normalized by log2 conversion. The Wilcoxon rank sum test was for comparing two independent samples, and Wilcoxon signed rank test was for paired sample.

The pan-cancer samples were divided into SERPINH1 high and SERPINH1 low expression groups based on the minimum p-value approach (Kouyama et al., 2019). The Kaplan–Meier survival curves were utilized to exhibit the correlation of SERPINH1 expression with the prognosis of patients’ overall survival (OS), disease-specifific survival (DSS), progression-free interval (PFI).

Interaction network analysis was obtained by employing STRING v11.5 database (http: //string-db.org/), keeping default parameters. The topological properties of the PPI network included average shortest path length, betweenness centrality, closeness centrality, degree, eccentricity, neighborhood connectivity, radiality, stress, and topological coefficient. Gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis were performed to identify significant pathways via the “cluster Profiler” package in R (Yu et al., 2012).

We applied the cBioPortal website (https://www.cbioportal.org/) to query the genetic variation characteristics of SERPINH1 (Cerami et al., 2012; Gao et al., 2013). The correlations of SERPINH1 expression with TMB and MSI in human pan-cancer were based on the TCGA pan-cancer atlas, and analyzed by spearman’s rank correlation method.

Mismatch repair (MMR) is a post-replicative repair mechanism that can repairs errors in DNA replication and attenuate chromosomal rearrangements, and is closely related to tumorigenesis (Iyer et al., 2015; Baretti and Le, 2018). In this study, the expression levels of MMR genes (MLH1, MSH2, MSH6, EPCAM, and PMS2) in human pan-cancer were obtained from the TCGA database. The correlation between MMR genes and SERPINH1 expression level were analyzed by pearson’s correlation method. Another epigenetic factor affecting gene expression is DNA methylation. We also evaluated the correlation of four methyltransferases (DNMT1, DNMT2, DNMT3A, and DNMT3B) with SERPINH1 expression by the pearson’s correlation method.

The relationship between SERPINH1 and immunomodulators (such as immunostimulators, immunoinhibitors, MHC molecules, TILs, receptors, and chemokines) across multiple types of human cancer were evaluated through the TISIDB portal (http://cis.hku.hk/TISIDB/). The correlation analysis was performed to explore the relationship between SERPINH1 and 47 common immune checkpoint genes. To explore the prognostic value of SERPIHN1 on immunotherapy, we analyzed the relationship between SERPINH1 and immunotherapy targets (PD-1, PD-L1, PD-L2, and CTLA4). We utilized the Tumor Immune Estimation Resource (TIMER, https://cistrome.shinyapps.io/timer/) database to explore potential associations between SERPINH1 and the infiltrating immune cells across human cancers.

The TISIDB web portal (http://cis.hku.hk/TISIDB/index.php) was applied to explore the association between SERPINH1 and multiple immune regulatory factors across multiple cancer types (Ru et al., 2019). The TIMER database (https://cistrome.shinyapps.io/timer) was used to analyze the correlation between SERPINH1 expression and immune cell infiltration level across human cancers (Li et al., 2017). Spearman correlation analysis was applied to evaluate the correlation between SERPINH1 expression and the scores of multiple immune cells, including B cells, CD4⁺ T cells, CD8⁺ T cells, macrophages, neutrophils, and myeloid denedritic cells.

All statistical analysis and visualization were conducted by R (version 3.6.3). Kaplan-Meier analysis was carried out using the survival package, correlation heatmap was visualized by ggplot2 package, radar chart was illustrated by ggradar and ggplot2 package. Tumor purity is a major confounding factor in this analysis, since most immune cell types are negatively correlated with tumor purity. Therefore, we use the partial spearman’s correlation to perform purity adjustment association analysis, and explore the relationship between infiltrates estimation value and SERPINH1 expression. p < 0.05 was considered statistically significant.

In this study, we aimed to analysis the expression of SERPINH1 across various cancer types. Based on the TCGA database, we found that SERPINH1 expression is relatively higher in BRCA, CHOL, COAD, ESCA, GBM, HNSC, KIRP, LIHC, LUAD, LUSC, READ, STAD, and THCA (all p < 0.001), BLCA (p < 0.01), and LGG (all p < 0.05) than those in nomal tissues (Figure 1A). Taking into account the lack of normal samples in the TCGA database for some cancer types, we integrated the GTEx database for further analysis and the results showed that the expression level of SERPINH1 in the tumor tissues of ACC, BLCA, BRCA, CHOL, COAD, DLBC, ESCA, GBM, HNSC, KIRC, KIRP, LAML, LGG, LIHC, LUAD, LUSC, OV, PAAD, PRAD, READ, SKCM, STAD, TGCT, THYM, and UCS (all p < 0.001); UCEC (p < 0.01); and SARC, THCA (all p < 0.05) are much higher than the corresponding control tissues (Figure 1B). In paired samples, the expression of SERPINH1 in tumor tissues of BRCA, COAD, HNSC, KIRC, KIRP, LIHC, LUAD, LUSC, STAD, and THCA (all p < 0.001); BLCA, CHOL, and KICH (all p < 0.01); and ESCA, READ, and UCEC (all p < 0.05) are significantly higher than the corresponding control tissues (Figure 1C). The above results showed that SERPINH1 expressed abnormally across multiple cancers. Interestingly, we noticed that in PRAD, compared with control tissues, the expression of SERPINH1 has no significant difference in independent samples, but there is a significant difference in paired samples (p < 0.05).

Furtherly, we evaluated the relationship between the expression of SERPINH1 and the prognosis of pan-cancer patients. According to the Log-rank analysis, we found that in 33 cancers, the high expression of SERPINH1 significantly reduces the OS in 18 types of cancers including ACC, BLCA, CESC, CHOL, COAD, GBM, HNSC, KIRC, KIRP, LGG, LIHC, LUAD, MESO, PRAD, SARC, SKCM, UCEC, and GBMLGG (Figures 2A,B). Since there may be non-tumor death factors during follow-up, we further explored the relationship between SERPINH1 expression and patients’ DSS. Results indicated that high expression of SERPINH1 significantly impacted DSS in 20 types of cancers including ACC, BLCA, BRCA, CESC, COAD, ESCA, GBM, HNSC, KIRC, KIRP, LGG, LIHC, LUSC, MESO, PAAD, SKCM, STAD, UCEC, COADREAD, and GBMLGG (Figures 3A,B). In addition, we also explored the correlation of SERPINH1 with patients’ PFI, which indicated that high expression of SERPINH1 unfavorably impacted PFI in 21 types of cancers including ACC, BLCA, BRCA, CESC, COAD, GBM, HNSC, KIRC, KIRP, LGG, LIHC, LUAD, MESO, PAAD, PRAD, SARC, THCA, UCEC, UVM, COADREAD, and GBMLGG (Figures 4A,B).

The PPI network of SERPINH1 was shown in Figure 5A. The results of GO enrichment analysis showed that SERPINH1 significantly focused on protein digestion and absorption, AGE-RAGE signanling pathway in diabetic complications, ECM-receptor interaction (biological process, BP); unfolded protein binding, extracellular matrix structural constituent conferring tensile strength (cell components, CC); and endoplasmic reticulum lumen, collagen trimer (molecular function, MF); According to the p-value, the top three items from the three categories were selected to plot a bubble (Figure 5B). KEGG enrichment analysis (Figure 6B) indicated that SERPINH1 was significantly enriched with chaperone-mediated protein folding and peptidyl-proline modification etc.

The genetic alteration status of SERPINH1 in different cancer samples were observed via the TCGA cohorts. As shown in Figure 7A, the highest alteration frequency of SERPINH1 is “amplification” type, accounting for 7% of 13 patients with undifferentiated STAD. The “mutation” was the only type in 48 patients with mature B-cell neoplasms, which show an alteration frequency of 4.17%. 63 seminoma cases with genetic alteration (1.59% frequency) had copy number deletion of SERPINH1.

Tumor mutation burden (TMB), which counts the number of somatic mutations per megabase (mut/Mb), is an emerging potential biomarker for immunotherapy (Aggarwal et al., 2020). Microsatellite instability (MSI) is closely association with the development of most cancers (Matsuno et al., 2019). Here, we evaluated the correlation of SERPINH1 expression with TMB and MSI. Results showed that SERPINH1 expression was positively related to TMB in BRCA, KIRC, LAML, LGG, LUAD, and THYM; while negatively associated with TMB in ESCA, HNSC, LIHC, and SKCM (Figure 7B). In terms of MSI, the SERPINH1 expression was positively related to MSI in SARC, and TGCT (Figure 7C).

Correct replication of the genome is a prerequisite for the maintenance of genome stability (Barra and Fachinetti, 2018). MMR serves as an important factor in maintaining genome stability in response to spontaneous DNA damage. MMR-deficient cells usually show microsatellite instability but display low levels of genome instabilities (Jia et al., 2017). To determine the role of SERPINH1 in tumor progression, we evaluated the association of the SERPINH1 expression with five MMR genes mutation levels. As results shown in Figure 6A, SERPINH1 expression was highly related to five MMR genes in seven cancers, including ESCA, KIRP, LGG, LIHC, OV, PRAD, and UCEC.

Global DNA hypomethylation and focal DNA hypermethylation are associated with tumorigenesis. Methylation of DNA is catalyzed by the DNA methyltransferase (DNMT) family of enzymes, with DNMT3A and DNMT3B catalyzing de novo DNA methylation and DNMT1 mediating both de novo and maintenance methylation of DNA (Suppiah et al., 2019). We therefore sought to determine whether SERPINH1 influences methyltransferases -mediated DNA repair, then we evaluated the relationships between SERPINH1 and four DNMTs. As shown in Figure 6B, SERPINH1 expression was highly associated with these four DNMTs in multiple cancers, specifically in ACC, BLCA, BRCA, CESC, COAD, ESCA, GBM, HNSC, KICH, KIRP, KIRC, LAML, LGG, LIHC, MESO, OV, PRAD, SKCM, STAD, TGCT, THCA, and UVM. Results indicate that SERPINH1 may regulate the tumor progress by MMR-mediated DNA repair and DNA methylation.

The relationship between SERPINH1 and immunomodulators were shown in Figures 8A–F). Take COAD for example, we founded that SERPINH1 was significant positive correlation with several immunomodulators, including immunostimulator CD 276 and CD70 expressions (spearman correlation of 0.554 and 0.448 respectively), immunoinhibitor TGFB1 expression (spearman correlation of 0.553), and CCL21 expression (spearman correlation of 0.38) (Figures 8G–I). Based on the results, we can infer that the immunomodulators CD 276, TGFB1, CD70, and CCL21 might be regulated by SERPINH1 in COAD. Thus it can be seen that SERPINH1 interacts with immune regulation and may become a potential biomarker which has an important impact on the development of cancers and prognosis of patients.

Furtherly, we performed a correlation analysis to explore the relationship between SERPINH1 and 47 common immune checkpoint genes. Results showed that SERPINH1 expression correlated with 38 immune checkpoint genes in TGCT, and 35 immune checkpoint genes in LGG (Figure 9A). The relationship between SERPINH1 and immunotherapy targets (PD-1, PD-L1, PD-L2, and CTLA4) was shown in Figure 9B. Results indicated that SERPINH1 was positively correlated with immunotherapy targets in BLCA, BRCA, COAD, LGG, PAAD, PCPG, PRAD, READ, COADREAD, and LUADLUSC; and negatively correlated with immunotherapy targets in SKCM, TGCT, and THYM.

The tumor microenvironment consists of tumor cells, infiltrating immune cells, and stromal cells (Zhang et al., 2020). We utilized the TIMER database to explore potential associations between SERPINH1 and the infiltrating immune cells across human cancers (Figure 10A). This study focuses on showing the correlation between SERPINH1 expression and six immune infiltrating cells, including B cell, CD4⁺ T cell, CD8⁺ T cell, dendritic cell macrophage cell and neutrophil cell in BRCA, LGG, and LIHC (Figure 10B). From the above results, we can infer that SERPINH1 acts significantly in tumor immunity.

The immunohistochemistry data of SERPINH1 was acquired from the Human Protein Atlas database (http://www.proteinatlas.org) for validation. As shown in Figure 11, SERPINH1 staining was higher in tumor samples than in normal tissue, which was consistent with the result of survival analysis, indicating that high expression of SERPINH1 is a risk factor in COAD, HNSC, KIRP, and CESC.