Expression of PPIA in pan-cancer and prognostic value of PPIA in GC
To assess the effects of PPIA on the genesis of human tumor, the Cancer Genome Atlas (TCGA) database was utilized to detect the mRNA levels of PPIA in 33 types of cancer. The findings demonstrated that the levels of PPIA were upregulated in 17 tumor types including BRCA, BLCA, CHOL, COAD, CESC, ESCA, HNSC, GBM, KIRP, KIRC, LIHC, LUAD, LUSC, READ, PRAD, UCEC and STAD compared to the corresponding normal tissues (Fig. 1A). However, there was no obvious difference of PPIA in DLBC, ACC, LGG, LAML, OV, MESO, TGCT, UVM and UCS. We also demonstrated that PPIA was downregulated in KICH. In previous studies, Yu and colleagues demonstrated that PPIA was upregulated in the 17 different types of cancer available through TIMER database, including BRCA, BLCA, COAD, CHOL, HNSC, ESCA, KIRP, KIRC, KICH, LUSC, LUAD, LIHC, READ, PRAD, UCEC, THCA and STAD. TIMER database was also used to validate PPIA expression in multiple tumor types. Notably, the high expression levels of PPIA could be observed in BRCA, BLCA, COAD, CHOL, CESC, HNSC, GBM, ESCA, KIRP, KIRC, LUSC, LUAD, LIHC, PRAD, PCPG, STAD, UCEC and READ (Supplementary Fig. 1). Taken together, PPIA was increased in BRCA, BLCA, COAD, CHOL, HNSC, ESCA, KIRP, KIRC, LUSC, LUAD, LIHC, READ, PRAD, UCEC and STAD. This suggests that PPIA may act as an important oncogene in 15 tumor types. Due to a lack of research on this topic, we further evaluated the difference in PPIA expression between the tumor and normal tissues in GC. As shown in Figs. 1B and 1C, based on GEPIA and UALCAN databases, PPIA was upregulated in GC compared to normal controls. Next, we examined the prognostic value of PPIA in GC using K-M plotter. As displayed in Fig. 1D, GC patients with high PPIA expression exhibited poor OS, PFS and PPS. This result suggests that overexpression of PPIA can predict poor prognosis in GC patients.
The relevance between PPIA expression and clinical parameters in GC patients
The UALCAN database was utilized to evaluate the relationship between PPIA expression and different clinical parameters. The findings demonstrated that elevated expression of PPIA was detected in GC tissues compared to the corresponding normal tissues (Fig. 2A). Similarly, an increased expression of PPIA was also observed in both male and female GC specimens compared to normal controls (Fig. 2B). Then, PPIA expression was highly upregulated in GC specimens based on different age groups (61–80 years in LUSC; 21–40, 41–60, 61–80 and 81–100 years in LUAD) (Fig. 2C). According to tumor grade, an increased expression level of PPIA was observed in gastric patients at grade 1, 2 and 3 (Fig. 2D). Furthermore, the upregulation of PPIA in GC was associated with nodal metastasis status, and patients with N0, N1, N2 and N3 displayed a higher PPIA expression compared to normal controls (Fig. 2E). Moreover, high expression of PPIA was also observed in GC patients with TP53 wild-type and mutant compared to normal controls (Fig. 2F).
Regarding patient’s race, high expression of PPIA was observed in Caucasians, African-American, and Asian (Supplementary Fig. 2A). PPIA was also remarkably upregulated in GC patients with and without Helicobacter pylori infection (Supplementary Fig. 2B). Moreover, PPIA was remarkably upregulated in various kinds of histological subtypes including AdenoNOS, AdenoDiffuse, AdenoSignetRing, IntAdenoNOS, IntAdenoTubular, IntAdenoMucinous, IntAdenoPapillary in GC patients compared to normal controls (Supplementary Fig. 2C). These findings indicated that PPIA expression had a close relationship with GC patients and might affect tumor progression.
Functional enrichment analysis of PPIA in GC based on TCGA database
Differentially expressed genes (DEGs) associated with PPIA in GC were recognized using TCGA database. The top 1000 differential genes related to PPIA were chosen for GO and KEGG pathway analysis. The top GO enrichment items were receptor ligand activity, G protein-coupled peptide receptor activity, cytokine activity, DNA replication origin binding, contractile fiber, I band, Z disc, muscle system process, regulation of membrane potential, muscle contraction, and transcriptional regulation involve in G1/S transition of mitotic cell cycle (Fig. 3A). The top KEGG pathways of PPIA were systemic lupus erythematosus, alcoholism, neuroactive ligand-receptor interaction, calcium signaling pathway, cell cycle, pancreatic secretion, vascular smooth muscle contraction, protein digestion and absorption, progesterone-mediated oocyte maturaction, bile secretion, fat digestion and absorption, DNA replication, ascorbate and aldarate metabolism, and renin-angiotensin system (Fig. 3B). We also implemented GSEA analysis to confirm the key pathways correlated with PPIA (Fig. 3B). The results showed that olfactory transduction and neuroactive ligand-receptor interaction were the most significantly enriched pathways (Fig. 3C).
Genetic alteration and DNA methylation analysis of PPIA
We explored genetic alteration status of PPIA in pan-cancer according to TCGA database, and there were four types of genetic alteration patterns (deep deletion, amplification, structural variant and mutation) in different tumor samples. Amplification displayed the highest alteration frequency in most tumor types and amplification was the only alteration type in GC samples (Supplementary Fig. 3A). Missense mutation was the main type of genetic alteration for PPIA (Supplementary Fig. 3B); however, we did not find any PPIA mutation sites in GC cases. To assess the correlation between PPIA genetic alteration and survival prognosis in GC patients, cBioPortal database was used to perform the prognostic analysis (Supplementary Fig. 3C). The prognosis of PPIA alteration group showed better prognosis in DFS (p = 0.0309), but not DSS (p = 0.312), progression-free survival (p = 0.623) and OS (p = 0.839). To further evaluate the mechanism of PPIA overexpression in GC, the methylation level of PPIA in GC samples (n = 395) and adjacent normal tissues was analyzed using UALCAN and DiseaseMeth version 3.0 database. However, no relationship between DNA methylation and expression of PPIA was observed (Supplementary Figs. 4A and 4B). This result suggests that the genetic alteration and DNA methylation of PPIA play a minor role in gastric carcinogenesis.
The upstream miRNAs of PPIA
It is well known that ncRNAs play an essential role in the regulation of gene expression. To confirm whether PPIA was modulated by ncRNAs, starBase 3.0 database was applied to estimate potentially upstream miRNAs of PPIA, we found a total of 33 unique miRNAs associated with PPIA, and cytoscape software was employed to draw miRNA-PPIA regulatory network (Fig. 4A). Prior studies have investigated the universal mechanism of miRNA action, and the results suggest that miRNAs is negatively correlated with gene expression. In this study, let-7c-5p, let-7e-5p and miRNA-204-5p were obviously negatively correlated with PPIA, and were selected as candidate upstream miRNAs of PPIA in GC (Fig. 4B). As shown in Fig. 4C, Supplementary Figs. 5A and 5B, let-7c-5p, let-7e-5p and miRNA-204-5p were significantly downregulated in GC. We also performed the prognosis analysis of let-7c-5p, let-7e-5p and miRNA-204-5p in GC (Fig. 4D, Supplementary Figs. 5C and 5D). It was observed that only patients with miRNA-204-5p overexpression had a good prognosis, although the results were not statistically significant. Finally, miRNA-204-5p was considered as the most appropriate regulatory miRNA of PPIA in GC.
The upstream lncRNAs of miRNA-204-5p
The starBase 3.0 database was utilized to analyze the upstream lncRNAs of miRNA-204-5p in GC. Supplementary Fig. 6 illustrates a total of 44 lncRNAs that related to miRNA-204-5p. GEPIA database was employed to display the expression and survival analyses the 44 upstream lncRNAs correlated with miRNA-204-5p. Notably, only MAlAT1, LINC01232, DHRS4-AS1 and OIP5-AS1 were markedly upregulated in GC compared with normal controls (Figs. 5A-5D). However, there were no obvious differences in DFS and OS of these lncRNAs in GC (Figs. 5E-5L). Based on the competing endogenous RNA theory, lncRNAs can upregulate mRNA expression through competitive binding to miRNA. Thus, lncRNA is negatively correlated to miRNA and positively correlated to mRNA. As shown in Table 1, only LINC01232 was negatively and positively correlated with miRNA-204-5p and PPIA, respectively. This result indicates that LINC01232 may act as a key candidate upstream lncRNA of miRNA-204-5p/PPIA in GC.
Table 1
Correlation analysis between lncRNA and miR-204-5p or lncRNA and PPIA in GC performed by starBase database.
lncRNA | miRNA | R value | P value |
MALAT1 | miR-204-5p | 0.017 | 7.41E-01 |
LINC01232 | miR-204-5p | -0.134a | 9.74E-03**a |
DHRS4-AS1 | miR-204-5p | -0.053a | 3.11E-01 |
OIP5-AS1 | miR-204-5p | 0.049 | 3.47E-01 |
lncRNA | mRNA | R value | P value |
MALAT1 | PPIA | -0.165 | 1.30E-03**a |
LINC01232 | PPIA | 0.123a | 1.68E-02*a |
DHRS4-AS1 | PPIA | -0.018 | 7.32E-01 |
OIP5-AS1 | PPIA | -0.417 | 3.33E-17***a |
aThese results are statistically significant. *p value < 0.05; **p value < 0.01; ***p value < 0.001. |
Negative correlation between PPIA and ICI in GC
PPIA encoded a member of the peptidyl-prolyl cis-trans isomerase family and had a function of cyclosporin A-mediated immunosuppression, indicating that PPIA may play a pivotal role in functionality of the immune system. Thus, TIMER database was applied to discuss the relationship between PPIA expression and ICI level (Fig. 6A). In B cell and CD4 + T cell, Arm-level deletion group, Arm-level gain group and High amplication group displayed significant changes in the ICI levels of PPIA compared to diploid/normal group. In CD8 + T cell and neutrophil cell, arm-level deletion group and arm-level gain group showed obvious changes in the ICI levels of PPIA compared to normal control. In macrophage cell and dendritic cell, only arm-level gain group exhibited an obvious change in the ICI level of PPIA compared to normal controls. PPIA expression was obviously negatively correlated to the ICI levels of CD8 + T cell, CD4 + T cell, B cell, dendritic cell, macrophage and neutrophil in GC (Figs. 6B-6G).
Correlation analysis between PPIA and immune cell biomarkers or immune checkpoints in GC
To fully explain the mechanism of PPIA in tumor immunity in GC, GEPIA database was used to confirm the relationship between PPIA expression and biomarkers of immune cells. As presented in Table 2, PPIA was obviously negatively correlated with B cell’s biomarkers (CD79A and CD19), CD4 + T cell’s biomarker (CD4), CD8 + T cell’s biomarker (CD8A), M2 macrophage’s biomarker (MS4A4A), neutrophil’s biomarkers (CCR7 and ITGAM) and dendritic cell’s biomarkers (HLA-DPB1, CD1C, ITGAX and NRP1). These results also suggested that PPIA was negatively related to ICI in GC. It is well known that CTLA-4 and PD1/PD-L1 have been demonstrated to be important checkpoints that block anti-tumor immune responses. Previous studies and this paper confirmed PPIA could act as an oncogene in various tumors and GC. To further analyze the correlation between PPIA expression and immune checkpoints in GC, TIMER database was used to perform the correlation analysis. PPIA expression was negatively related to PD1, PD-L1, and CTLA-4 in GC (Figs. 7A-7C), however, only the correlation between PD1 and PPIA was statistically significant. PPIA was obviously negatively correlated with PD1 (Figs. 7D-7F), but no statistical significance with CTLA-4 and PD-L1. Overall, the results suggest that PPIA may be participated in anti-tumor immunity progression in GC.
Table 2
Correlation analysis between PPIA and biomarkers of immune cells in GC performed by GEPIA database
Immune cell | Biomarker | R value | P value |
B cell | CD19 | -0.37 | 7.1E-15*** |
CD79A | -0.39 | 1.6E-16*** |
CD8+T cell | CD8A | -0.19 | 1.5E-04*** |
CD8B | -0.079 | 0.11 |
CD4+T cell | CD4 | -0.16 | 1.2E-03** |
M1 macrophage | NOS2 | 0.034 | 0.5 |
IRF5 | -0.017 | 0.73 |
PTGS2 | -0.035 | 0.48 |
M2 macrophage | CD163 | -0.054 | 0.28 |
VSIG4 | -0.056 | 0.26 |
MS4A4A | -0.12 | 0.012* |
Neutrophil | CEACAM8 | -0.049 | 0.32 |
ITGAM | -0.23 | 3.4E-06*** |
CCR7 | -0.36 | 1.4E-13*** |
Dendritic cell | HLA-DPB1 | -0.12 | 0.014* |
HLA-DQB1 | -0.029 | 0.56 |
HLA-DRA | -0.032 | 0.52 |
HLA-DPA1 | -0.066 | 0.18 |
CD1C | -0.36 | 8.3E-14*** |
NRP1 | -0.2 | 3.4E-05*** |
ITGAX | -0.19 | 1.5E-04*** |
*p value < 0.05; **p value < 0.01; ***p value < 0.001. |