Article
Identifcation and Validation of EMT-immune-related Prognostic Biomarkers CMTM3 and LTBP2 in Gastric Cancer
https://doi.org/10.21203/rs.3.rs-4693527/v1
This work is licensed under a CC BY 4.0 License
posted
You are reading this latest preprint version
Gastric carcinoma
epithelial-mesenchymal transition
immune
bioinformatics analysis
biomarker
prognosis
Gastric cancer(GC) is a common gastrointestinal cancer and the second leading cause of cancer-related death worldwide. It approximately accounts for 8% of all new cancer cases yearly[1, 2]. Although the emerging treatments of GC occur, the prognosis is still poor. The search for novel valuable molecular biomarkers is significant but still in its infancy. Therefore, it is necessary to identify prognostic biomarkers and therapeutic targets[3, 4].
Epithelial-mesenchymal transition (EMT) is a cellular biological process in which epithelial cells obtain mesenchymal phenotype after epithelial characteristics are down-regulated through specific procedures[5]. EMT could result in cell migration, tumor metastasis and drug resistance. Hence, it plays an important role in facilitating GC progression[6]. The immune microenvironment is another significant factor in affecting GC progression. Malignant cells could be eliminated by an effective immune response. Nevertheless, gastric cancer cells could escape from immune surveillance by multiple mechanisms, such as deficiencies in antigen presentation mechanisms, up-regulation of immune checkpoints, and recruitment of immunosuppressive cells. These phenomena result in impaired function of immune cells and ultimately tumor progression and metastasis[7]. Many studies have found that EMT may directly or indirectly interact with immunosuppression recently[8–10]. The survival rate of gastric cancer patients may be decreased markedly if the immunosuppression and EMT occur together. So, it is significant to study the EMT-immune-related biomarkers.
An increasing number of studies have focused on the interaction between EMT and immune microenvironment, meanwhile, some EMT-immune-related biomarkers were primarily explored. Ding et al. revealed that GFPT2 as an EMT-immune-related gene was highly expressed in colon adenocarcinoma and was associated with poor pathological features and poor clinical prognosis[11]. ISM1 could promote EMT progress and result in immunosuppressive status. So, ISM1 is a critical prognostic biomarker in colorectal cancer[12]. Additionally, ILK3, KCNN4, FUT7 and other EMT-immune-related biomarkers are also targets with prognostic value in different cancers[13–15]. The identification of EMT-immune-related biomarkers in GC was far from sufficient, so, we aim to screen the valuable EMT-immune-related genes in predicting prognosis and treatment efficacy.
To sum up, when EMT and immunosuppression occur in GC patients, the risk of tumor progression, metastasis and drug tolerance will be increased compared with others, which may affect the prognosis. This study identified and validated the EMT-immune-related prognostic biomarkers named CMTM3 and LTBP2 by bioinformatics method. They could regulate the process of EMT and immune microenvironment. Moreover, CMTM3 and LTBP2 may be effective indicators in judging the therapeutic sensitivity of chemotherapy drugs and immunotherapy. So, they may be promising prognostic biomarkers and potential therapeutic targets in gastric carcinoma.
2.1. Identifying the EMT-immune-related Differential Expression Genes. The datasets GSE118916 and GSE79973 from GEO (https://www.ncbi.nlm.nih.gov/geo/) database[16] contain 15 and 10 pairs of gastric cancer and matched normal tissues. Differentially expressed genes (DEGs) between GC tissues and adjacent normal tissues were calculated by GEO2R, |logFC|>1, and adjust P value < 0.05 were the screening criteria. Furthermore, we defined immune-related genes by InnateDB (https://www.innatedb.com/)[17] and Immport (https://immport.org/shared/home) databases[18]. EMT-related genes were defined by EMTome (www.emtome.org)[19].
2.2. Screening Hub Genes by Expression, Prognosis and Clinical Correlation Analysis. To screen out the valuable targets from the EMT-immune-related DEGs, we conducted expression analyses in mRNA and protein levels. GEPIA 2.0 database (http://gepia2.cancer-pku.cn/index.html)[20] and the GSE19826 from GEO database were utilized to conduct differential expression analysis. At the protein level, we used the human protein atlas (HPA) database (https://www.proteinatlas.org)[21] to verify the expression pattern. Only the genes differentially expressed in all datasets above were defined as hub genes. The Tumor Immune Single Cell Center (TISCH) (http://tisch.comp-genomics.org/)[22] was used to study the cell-subset expression of hub genes in the tumor microenvironment. We used the GSE134520 dataset, including the following main cell types, immune cells, stromal cells, epithelial cells, and malignant cells. Furthermore, we analyzed how the hub genes affect the overall survival (OS) of GC patients by GEPIA 2.0, Kaplan-Meier plotter (https://kmplot.com/analysis/)[23] and TISIDB (http://cis.hku.hk/TISIDB/)[24] respectively. We also study the effect on first progression(FP), post progression survival(PPS) and disease free survival (DFS). Then we built receiver operating characteristic curves (ROC curves) of the hub genes by the data of stomach adenocarcinoma from The Cancer Genome Atlas TCGA data(https://portal.gdc.cancer.gov/)[25]. Furthermore, we conducted the correlation analysis with clinical and pathological indicators by the UALCAN database (http://ualcan.path.uab.edu)[26] and Gene Set Cancer Analysis(GSCA) database(http://bioinfo.life.hust.edu.cn/GSCA/#/)[27].
2.3. Expression and Survival Validation of CMTM3, LTBP2 by Immunohistochemistry Experiments. To verify the protein expression of CMTM3 and LTBP2, we collected 34 paraffin-embedded gastric cancer tissues from the pathology department. Meanwhile, 15 adjacent tissues which were paired with the gastric cancer tissues were collected. All the gastric cancer patients were primary tumors and without prior treatments. We used Anti-CMTM3 (CSB-PA597389), Anti-LTBP2 (ab121193) to do the experiments. All the sections were deparaffinized in xylene and rehydrated in ethanol. Then sodium citrate antigen retrieval solution (Solarbio) was used for heat-mediated antigen retrieval. After blocking by 3% H2O2 (30 minutes) and normal goat serum (1 hour), the tissue sections were incubated with CMTM3 (concentration 1:20), LTBP2 (1:500) antibodies overnight (4℃). We stained the tissue sections with the secondary antibody (ORIGENE SP-9001) for 20 minutes at room temperature. Then, horseradish labeled streptavidin (ORIGENE SP-9001) was utilized to incubate tissues for 15 minutes. Diaminobenzidine (ORIGENE ZLI-9018) was utilized for a peroxidase reaction, and we stained the slides with hematoxylin (LEAGENE).
CMTM3 and LTBP2 were mainly located in the cell nucleus. The integrated optical density (IOD) of all tissues was calculated by Image-Pro Plus 6.0 software and then the statistical analysis was conducted by Graphpad Prism 7.0. The studies involving human participants were reviewed and approved by the institutional review board of the second hospital of Hebei Medical University.
2.4. The Nomogram Building by CMTM3, LTBP2. The nomogram has been widely used as a predictive method in oncology in recent years. It meets requirements for an integrated model, plays important role in personalized medicine, and is convenient for clinicians to use in prognosis prediction[28]. The expression and clinical data of TCGA-STAD were downloaded, and the “rms” package[29] of R language (R 4.1.0 version) was used to construct the nomogram.
2.5. The Immune and the EMT Correlation Analysis of CMTM3, LTBP2. We evaluate the correlation between the expression level of hub genes and the immune infiltration by using the “Gene” module of TIMER 2.0 (http://timer.comp-genomics.org/)[30]. Furthermore, we used R language to conduct immune-related analysis on TCGA-STAD data by the “CIBERSORT” method. The overall immune microenvironment of all the samples was analyzed. The correlations between the hub genes and the 22 types of immune cell infiltration were carried out by the “CIBERSORT” method. Then, the TCGA-STAD data was utilized to analyze the expressional correlation between the hub genes and the immune checkpoint. We also analyzed the correlation between the hub genes and the immunosuppressive factors with the same method. TISIDB database was used to validate the correlation of hub genes with the immune infiltration and the immunosuppressive factors.
Furthermore, EMT-related analyses were conducted. We analyzed the correlation between the hub genes and the EMT markers by the TCGA-STAD data. We examined the association about hub genes and CAF-related factors in the same way. GSCA database and the TCGA-STAD data were utilized to examine the correlation with the activation of the EMT-related pathways. Finally, the correlations between the hub genes and the immune score, stromal score were analyzed by sangerbox online tools (http://sangerbox.com/Tool). All the correlation analyses above were conducted by the “spearman” method.
2.6. The Co-expression and Enrichment Analysis of CMTM3 and LTBP2. To do co-expression and enrichment analysis, the gastric cancer data of CCLE (https://sites.broadinstitute.org/ccle/)[31] database were downloaded for correlation analysis with the “spearman” method. The genes meet the following criteria were defined as co-expression genes: correlation coefficient > 0.5, P value < 0.001. Gene ontology (GO), including Biological Process(BP), Cellular Component(CC), Molecular Function(MF), were analyzed by the “ClusterProfiler” R package[32] and obtained the significantly enriched functions of the co-expression genes of CMTM3 and LTBP2 respectively (P < 0.05). We carried out the Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis in the same way to study the significantly enriched pathways. The pathway maps of critical pathways were drawn by the “pathview” package[33]. Gene Set Enrichment Analysis(GSEA) was conducted by GSEA 4.2.3 software using the CCLE data and the important pathways could be got (P < 0.05). In this way, we could verify and supplement the results of KEGG in another way.
2.7. The Genomics-related analyses of CMTM3 and LTBP2 in GC. The database GSCA was used to analyze the mutations, Copy Number Variation (CNV), and methylation of hub genes in gastric cancer. We analyzed the overall state of CMTM3 and LTBP2 mutations. Moreover, the effect on the overall survival of the hub genes’ mutation was studied. Furthermore, the influence on the expression of hub genes and the prognosis in gastric cancer by CNV and methylation were also analyzed by the GSCA database.
2.8. Chemotherapy Drugs and Immunotherapy Sensitivity Analysis. The gene expression data of NCI-60 cell line and the drug sensitivity data were downloaded from the CellMiner dataset (http://discover.nci.nih.gov/cellminer/)[34]. We removed drugs without FDA approval. The correlations between the hub genes and drug sensitivity were done by the “Pearson” method in the R language. When P < 0.01, the correlation was defined as significant. If the correlation coefficient was greater than 0 means a positive correlation with drug sensitivity. TCGA-STAD data were used for sensitivity analysis of immune-checkpoints block treatment by Tumor Immune Dysfunction and Exclusion (TIDE) algorithm[35].
3.1. EMT-immune-related Differential Expression Genes were Screened Out. Previous studies have proved that EMT and immune microenvironment were closely related, they were considered to be driving factors in tumor progression. As a result, the genes involved in the EMT process and immune infiltration were important. Based on the above theory, we intend to filter out the important EMT-immune-related biomarkers. The workflow has exhibited in Fig. 1. In this study, 1817 DEGs were calculated by the GSE118916 dataset (Fig. 2A), and 1406 DEGs were obtained by the GSE79973 dataset (Fig. 2B). 2660 immune-related genes were defined by InnateDB and Immport databases. Then, we defined 2975 EMT-related genes by the EMTome database. According to the DEGs of GSE118916, GSE79973, immune-related genes, EMT-related genes, 40 overlapped genes were got and were defined as EMT-immune-related DEGs (Fig. 2C).
3.2. CMTM3 and LTBP2 were Selected as Hub Genes among 40 EMT-immune-related DEGs. According to the results of multiple databases, only CMTM3 and LTBP2 were both up-regulated in mRNA and protein levels. CMTM3 and LTBP2 were up-regulated in the GSE19826 dataset, which was conform to the results of GSE118916 and GSE79973 (Fig. 2D). Moreover, TCGA-STAD data from the GEPIA database were used for validating the expression pattern(Fig. 2E). Figure 2F showed that the 2 genes were also higher expressed in GC at the protein level. Furthermore, as the traditional bulk profiles represent the average expression levels of the constituent cells, it could not reflct the specific expression of different cell types. So, we evaluate the expression of CMTM3 and LTBP2 at the single-cell level. The expression density of CMTM3 and LTBP2 was relatively higher in fibroblasts (Supplement Fig. 1). Above all, CMTM3 and LTBP2 were up-regulated in GC, which were filtered out to do further study.
Then, we analyzed the effect of the CMTM3 and LTBP2 on the overall survival of GC patients. CMTM3 and LTBP2 demonstrated the value in predicting prognosis in GEPIA 2.0 database (Fig. 2G). The results of the TISIDB and Kaplan-Meier plotter databases conform to the results above (Fig. 2H-2I). Moreover, CMTM3 could affect FP and PPS of GC patients according to the results of the Kaplan-Meier plotter database. Apart from the above, LTBP2 also showed an effect on DFS (Supplement Fig. 2A-2C). These results suggested the ability of CMTM3, LTBP2 to predict the survival and progression of GC. On the other hand, Supplement Fig. 2D-2F revealed that the expression of CMTM3, LTBP2 were positively correlated with the pathological grades, American Joint Council on Cancer(AJCC) stages, and lymphatic metastasis. It could be further proved that CMTM3 and LTBP2 were important indicators in affecting the progression of GC. To further study the performance in predicting survival, we drew the ROC curves and the AUC were 0.769,0.733 respectively (Fig. 2J). As a result, CMTM3 and LTBP2 were selected as promising hub genes, which were prognostic and up-regulated genes in GC among 40 EMT-immune-related DEGs.
3.3. CMTM3 and LTBP2 were Proved to be Prognostic and Up-regulated Genes by Our Follow-up and Immunohistochemistry Experiments Data. In order to further verify the expression in protein level. We used immunohistochemistry experiments to detect the protein expression of CMTM3 and LTBP2 in the GC tissues and adjacent tissues. We collected 34 gastric cancer paraffin-embedded tissues from the pathology department. Meanwhile, 15 adjacent tissues which were paired with the gastric cancer tissues were collected. The representative IHC images of CMTM3 were indicated in Fig. 3A-3B. Figure 3C showed that the expression of CMTM3 was higher in tumor tissues than in their paired adjacent tissues ( P < 0.001). Figure 3D-3E showed the representative images of LTBP2. Figure 3F showed the expression of LTBP2 was up-regulated in tumor tissues with statistical significance (P < 0.05). The ROC curves were built and the AUC of CMTM3 and LTBP2 were 0.722 and 0.806 respectively (Fig. 3G). The survival analysis showed that the high expression of LTBP2 was related to poor overall survival with statistical significance. Although the results of CMTM3 without statistical significance, but the trend that the high expression of CMTM3 was related to poor overall survival could be seen (Fig. 3H). These conform to the results of public databases. So, we could define CMTM3 and LTBP2 as hub genes to conduct the following analyses.
3.4. The Nomogram We Built Could Serve as an Important Guidance in Predicting the Survival Rate of GC. According to the results above, CMTM3 and LTBP2 were promising prognostic biomarkers in GC. To better guide the clinical process in predicting the survival of GC, we constructed a nomogram by CMTM3, LTBP2, and some clinical and pathological indicators (Fig. 4A). The nomogram could be used to predict the 1-year, 3- year, and 5-year survival of GC patients. The calibration curves indicated that the nomogram could be used as a good tool in judging prognosis in the clinic (Fig. 4B-4D ).
3.5. CMTM3 and LTBP2 May Affect the Prognosis of GC Patients by Regulating the Immune Microenvironment. To further study how CMTM3 and LTBP2 regulate the immune microenvironment, we conducted immune cell infiltrates analyses by the TIMER algorithm. CMTM3 was positively correlated with the infiltration of CD8 + T cells (R = 0.518), CD4 + T cells (R = 0.280), B cells (R = 0.105), neutrophil (R = 0.339), macrophages (R = 0.633) and dendritic cells (R = 0.540) with statistical significance (Fig. 5A). LTBP2 was positively correlated with the infiltration of CD8 + T cells (R = 0.441), CD4 + T cells (R = 0.379), B cells (R = 0.126), neutrophil (R = 0.379), macrophages (R = 0.642) and dendritic cells (R = 0.433) with statistical significance (Fig. 5B).
Furthermore, by using the CIBERSORT algorithm, the immune infiltration of each TCGA-STAD sample and the correlation among different immune cell types were shown in Supplement Fig. 3A-3B. Supplement Fig. 3C-3D showed the difference in immune infiltration between GC and normal samples. Then we studied the correlation between the expression of CMTM3, LTBP2 and the immune infiltrate by the CIBERSORT algorithm to supplement the results of the TIMER algorithm. CMTM3 was positively correlated with the infiltration of M2 type macrophages (R = 0.290), dendritic cells resting (R = 0.130), and negatively correlated with the infiltration of dendritic cells activated (R=-0.170), plasma cells (R=-0.210), CD4 + T cells memory resting (R=-0.240) with statistical significance (Fig. 5C). LTBP2 was positively correlated with the infiltration of dendritic cells resting (R = 0.180), Monocytes (R = 0.190), and negatively correlated with the infiltration of M1 type macrophages (R=-0.170), CD4 + T cells memory activated (R=-0.170), T cells follicular helper (R=-0.150), T cells gamma delta (R=-0.140) with statistical significance (Fig. 5D). Supplement Fig. 3G showed that the expression of CMTM3 and LTBP2 were positively correlated with the immune score. Above all, CMTM3 and LTBP2 were correlated with the immune infiltration, which may promote the infiltration of suppressive immune cells.
In general, the up-regulated immune checkpoints were vital factors in the suppression of the immune microenvironment. We found that the levels of CMTM3 and LTBP2 displayed remarkable associations with PDCD1, LAG3, CTLA4 and so on (Fig. 5E). Moreover, we also conducted a correlation analysis between the expression of CMTM3, LTBP2 and the key factors of immunosuppression. The results showed that the 2 genes were significantly and positively correlated with many immunosuppressive factors (Fig. 5F). The immune infiltration and the correlation analysis with factors of immunosuppression were also verified by the gene set variation analysis (GSVA) algorithm (Supplement Fig. 3E-3F). As a result, CMTM3 and LTBP2 might be immunosuppressive components in regulating GC.
3.6. CMTM3 and LTBP2 Could Influence the Prognosis of GC Patients by Regulating the EMT Process. According to the scRNA data before, CMTM3 and LTBP2 were relatively higher in fibroblasts compared to other cell subpopulations, indicating that the 2 genes have important functions in cancer-associated fibroblasts (CAFs). We also found that CMTM3 and LTBP2 were positively correlated with the key factors of CAFs (Fig. 6A). CAFs exert significant effects on tumorigenesis and development, for example promoting the initiation of EMT, boosting angiogenesis, and affecting the survival of tumor cells[36]. To sum up, CMTM3 and LTBP2 may promote the EMT process. Furthermore, to confirm the effects of CMTM3 and LTBP2 levels on EMT, we conducted the correlation analyses between the expression of 2 genes and the EMT-related factors. The results demonstrated that CMTM3 and LTBP2 were significantly correlated with the EMT process (Fig. 6B). The relevance with the stromal score, the EMT pathway activity, and the gene sets of EMT markers and the ECM-related genes could further verify the effect of CMTM3 and LTBP2 on EMT (Fig. 6C-6F). In general, CMTM3 and LTBP2 might act as EMT stimulatives in GC.
3.7. CMTM3, LTBP2 Could Regulate the Immune Microenvironment and EMT Progress by Some Pathways. Based on the results before, we found that CMTM3 and LTBP2 may affect the prognosis of GC patients by regulating the immune microenvironment and EMT progress. To further study the specific signal pathways and functions of the 2 genes, co-expression and pathway enrichment analyses were carried out. By using the gastric cancer cells data of the CCLE database, 660 and 1349 co-expression genes of CMTM3 and LTBP2 were obtained. Then, the GO and KEGG enrichment analyses were carried out. Figure 7A-7B showed that the co-expression genes of CMTM3 and LTBP2 were involved in the following functions: extracellular matrix and structure organization, and mesenchymal cell differentiation.
The co-expression genes of CMTM3 may be involved in the PI3K-Akt signaling pathway and Rap1 signaling pathway (Fig. 7C). And the co-expression genes of LTBP2 participated in the following pathways: PI3K − Akt signaling pathway, Focal adhesion, Proteoglycans in cancer, ECM − receptor interaction, Metabolism of xenobiotics by cytochrome P450, Fc gamma R − mediated phagocytosis, Drug metabolism-cytochrome P450 (Fig. 7D). The pathway map of the critical pathways were shown in Supplement Fig. 4. Furthermore, GSEA pathway analyses were conducted to refine and verify the results. The following pathways were relatively activated in the high-CMTM3 group with statistical significance (P < 0.05): Hedgehog pathway, TGF-β pathway, Focal adhesion, pathways in cancer, ECM − receptor interaction, Chemokine pathway, Wnt signal pathway, MAPK pathway (Fig. 7E). Increased LTBP2 expression was positively related to the enrichment of Focal adhesion, Chemokine pathway, ECM receptor interaction, TGF-β pathway, and pathways in cancer. To sum up, CMTM3 and LTBP2 may regulate the EMT process and immune microenvironment by the functions and pathways above. And could further affect the prognosis and drug sensitivity.
3.8. The Genomics-related analyses of CMTM3 and LTBP2 in GC. To clarify the expression regulatory mechanism of CMTM3 and LTBP2, the mutation, CNV and methylation were analyzed. The overall mutation landscape was shown in Supplement Fig. 5A-5B. Supplement Fig. 5C indicated the mutation sites of CMTM3 and LTBP2. According to the results above, we found that the mutations of LTBP2 occur more frequently. There was no statistical difference in overall survival between LTBP2 mutant samples and no-mutant samples, while there was a trend of poor OS in the mutant group. The same result could be got in the gene set (CMTM3 and LTBP2) mutation samples (Supplement Fig. 5D-5E). The overview CNV of CMTM3 and LTBP2 were shown in Supplement Fig. 5F. CMTM3 showed more amplification of copy number than LTBP2. The CNV of CMTM3 was positively correlated with the mRNA expression with statistical significance, while there was no statistical significance in the results of LTBP2 (Supplement Fig. 5G-5H). Although without statistical significance, there was a trend of poor OS in the copy number amplification group of CMTM3 and LTBP2 (Supplement Fig. 5I-5J). Furthermore, the methylation of CMTM3 and LTBP2 were negatively correlated with the mRNA expression with statistical significance (Supplement Fig. 5K-5L). Supplement Fig. 5M-5N showed the trend of poor OS in the lower methylation group. The results above proved that the expression of the three genes could be regulated by CNV, mutation, and methylation. The results of genomics-related analyses conform to the studies of us before.
3.9. CMTM3 and LTBP2 Could Regulate Chemotherapy Drugs and Immunotherapy Sensitivity. According to the results above, we know that CMTM3 and LTBP2 may promote the EMT process and regulate the immune microenvironment, as a result, they could exert important effects on the prognosis of GC. However, in addition to the prognostic value, we want to further study whether the 2 genes have the ability in affecting the sensitivity of therapy. The TIDE score was higher in the high CMTM3, and LTBP2 expression groups, and the results were statistically significant (Fig. 8A-8B). It revealed that the high CMTM3, and LTBP2 expression groups may undergo poor sensitivity to immune checkpoint blocking therapy. Furthermore, we downloaded the gene expression and drug sensitivity data of the CellMiner database to calculate the coefficient between the expression of the 2 genes and drug sensitivity by R language. 30 drugs associated with CMTM3 and LTBP2 were statistically significant (P < 0.01). The top 16 drugs by coefficient were shown in Fig. 8C. The drugs including Gemcitabine, Cladribine, Fludarabine, Cisplatin, Cytarabine, and Carboplatin may be more effective if the expression of CMTM3 was high. Lenvatinib, and Everolimus may better sensitivity if the LTBP2 is up-regulated. If the level of CMTM3 were high, the resistance of some drugs will be increased, for example, NMS-E628. And the expression level of LTBP2 was associated with increased resistance to Tamoxifen, Methotrexate, Vinorelbine, Vincristine, Nilotinib, Fluorouracil, and Crizotinib. The results above demonstrated that CMTM3 and LTBP2 may be good indicators to judge the sensitivity of a certain class of drugs, which could be seen as promising potential therapeutic targets in gastric carcinoma.
EMT and immunosuppression could appear together in multiple tumors according to previous studies. This phenomenon may accelerate the progression and metastasis of cancers and finally affect the prognosis and sensitivity of therapy[9, 37]. An accumulating number of researchers have focused on this issue and conducted some primary explorations on biomarkers related to EMT and the immune microenvironment. These studies proved that the EMT-immune-related biomarkers were potential factors in predicting the prognosis and drug sensitivity of cancer patients. However, the researches on EMT-immune-related biomarkers of gastric cancer were less.
As a result, we conducted differential expression analysis by GSE118916 data and GSE79973 data. 40 differentially expressed EMT-immune-related genes were screened out. Among them, CMTM3 and LTBP2 were verified as promising prognostic biomarkers and potential therapeutic targets in GC, which deserve further studies.
The CKLF-like MARVEL transmembrane domain-containing (CMTM) family consists of 8-member proteins (CMTM1-8), which could regulate aggressive phenotype in many cancers. Until now, how CMTM3 affects the prognosis and how it works in GC is still studied less[38, 39]. A previous study found that CMTM3 was down-regulated in gastric cancer cell lines. CMTM3 could inhibit the migration and invasion of gastric cancer cells and is associated with a favorable prognosis of gastric cancer[40]. Another study suggested that the overexpression of miR-135b-5p inhibits CMTM3 expression, and promotes gastric cancer progression and metastasis[41]. These results contend that CMTM3 seems to be a tumor suppressor. However, Liang et al. contend that CMTM3 was up-regulated in GC and had a significant association with poor overall survival[39]. This study conformed to us. We used TCGA-STAD, GSE118916, GSE19826 dataset and so on to verify that CMTM3 was up-regulated in gastric cancer tissues. And the expression level of protein by the HPA database were conform to the results above. So, we contend that CMTM3 was up-regulated in GC. Furthermore, we found that the level of CMTM3 was related to poor prognosis in GC by Kaplan-Meier plotter databases, which was conform to the results of Liang et al. To further validate the results above, we used TISIDB and GEPIA databases to conduct prognostic analyses. CMTM3 could not only influence the OS of gastric cancer, but also affect the FP and PPS. Moreover, the AUC of CMTM3 was 0.769 and it was correlated with the clinical and pathological factors. Furthermore, we collected 34 paraffin-embedded gastric cancer tissues from our hospital to do IHC experiments and collected the follow-up data of them, the results conformed to the results of public databases. These results demonstrated that CMTM3 was up-regulated in GC and associated with poor prognosis, which could predict the survival and progression of GC.
Through our screening process, CMTM3 may correlate with EMT progression and immune microenvironment. But, how it affects the 2 tumor biological processes and further regulated the prognosis of gastric cancer was still rarely studied. A study in low-grade glioma showed that CMTM3 was positively correlated with macrophages, dendritic cells, and CD4 + T cells, which is related to the immune infiltration in the glioma microenvironment, and may become a new immunotherapy target[42]. We found that CMTM3 was positively correlated with the infiltration of CD8 + T cells, CD4 + T cells, B cells, neutrophils, macrophages and dendritic cells in GC. The results were consistent with the previous studies in glioma. And the coefficient with macrophages was higher than other immune cells. By the CIBERSORT algorithm, CMTM3 was positively correlated with the infiltration of M2-type macrophages and negatively correlated with activated dendritic cells suggesting CMTM3 may promote the formation of an immunosuppression microenvironment. Furthermore, CMTM3 was significantly and positively correlated with many immunosuppressive factors and immune checkpoints. These all proved that CMTM3 could be seen as an immunosuppressional factor in GC. On the other hand, CMTM3 could also regulate the EMT process. A study on chordomas showed that CMTM3 suppressed the EGFR/STAT3 signaling pathway, which suppressed EMT progression[43]. While, Xiao et al. found a positive correlation between CMTM3 and EMT score in breast cancer cells, and with Vimentin in TNBC patients, which revealed the ability in promoting the EMT process[44]. We found that CMTM3 significantly correlated with EMT score and EMT markers. We also found that CMTM3 expression may affect some EMT-related pathways. Moreover, CMTM3 was mainly expressed in CAFs by scRNA data and positively correlated with the key factors of CAFs. According to previous studies, CAFs exert a significant effect in promoting the initiation of EMT[36]. To sum up, we could know that CMTM3 may promote the EMT process. As a result, CMTM3 could influence the survival of GC patients by regulating the EMT process and immunosuppression microenvironment.
Latent transforming growth factor (TGF) β-binding protein (LTBP)2 belongs to the LTBP/fibrillin family of extracellular matrix (ECM) proteins, which includes LTBPs 1–4 and fibrillins 1–3[45]. LTBPs play significant roles in regulating the structural integrity of the ECM and activating the latent TGF-β efficiently[46]. Previous studies have demonstrated that LTBP2 was associated with poor prognosis in many cancers, for example, colorectal cancer, pancreatic ductal adenocarcinoma and so on[47, 48]. While LTBP2 may show different functions in different cancer types. In an esophageal squamous cell carcinoma study, LTBP2 was downregulated and showed tumor-suppressive function[49]. Until now, how LTBP2 affects the prognosis and how it works in GC is still studied less. Wang et al. contend that LTBP2 could promote the migration and invasion of gastric cancer cells and may affect the OS of GC patients[50]. This study was consistent with ours. By using the same methods of CMTM3, we found that LTBP2 was up-regulated in GC and could not only affect on OS, but also FP, PPS, and DFS of GC patients according to multiple databases. Moreover, the AUC of LTBP2 was 0.733 and it was correlated with the clinical and pathological factors. Then we used our data to validate the results above, the results conformed to the public databases. These results demonstrated that LTBP2 was up-regulated in GC and associated with poor prognosis, which could be used to predict the survival and progression of GC.
According to the results above, LTBP2 may be a promising EMT-immune-related biomarker in predicting the survival of GC patients. But, the specific relationship of LTBP2 with EMT and immune microenvironment were still rarely studied. We found that LTBP2 was positively correlated with the infiltration of CD8 + T cells, CD4 + T cells, B cells, neutrophils, macrophages and dendritic cells in GC. Moreover, the expression of LTBP2 was positively correlated with the immune score, which indicates the effect on immune infiltration. Just like the results of CMTM3, the coefficient with macrophages was higher than other immune cells. Furthermore, by the CIBERSORT algorithm, LTBP2 was negatively correlated with the infiltration of M1 type macrophages, CD4 + T cells memory activated, T cells follicular helper and so on. These suggested that LTBP2 may help to form the immunosuppression microenvironment. Furthermore, LTBP2 was positively correlated with many immunosuppressive factors and immune checkpoints. These all proved that LTBP2 could be seen as an immunosuppressional factor in GC. In addition, LTBP2 could also regulate the EMT process. A study found that circEPSTI1/miR-942-5p/LTBP2 axis affects oral squamous cell carcinoma cell proliferation and invasion via the acceleration of EMT and the phosphorylation of PI3K/Akt/mTOR signaling pathway components[51]. Wan et al. contend that if LTBP2 were knockdown, the proliferation, invasion, and EMT phenotype of thyroid carcinoma cells would be inhibited[52]. In addition to the cancers above, we found that LTBP2 may affect the EMT phenotype in gastric cancer. LTBP2 was positively correlated with EMT score and EMT markers. Moreover, the expression of LTBP2 may affect some EMT-related pathways. Moreover, LTBP2 was positively correlated with the key factors of CAFs, which proved the significant effect of LTBP2 in promoting the initiation of EMT. To sum up, LTBP2 may promote the EMT process according to our results. As a result, LTBP2 could influence the survival of GC patients by regulating the EMT process and immunosuppression microenvironment.
For reasons of the foregoing, CMTM3 and LTBP2 were up-regulated in GC and may affect the prognosis by modifying the immune microenvironment and EMT process. To further study the specific signal pathways, we conducted the enrichment analysis by different methods. CMTM3 and LTBP2 were both enriched in the following pathways: PI3K − Akt signaling pathway, TGF-β pathway, Focal adhesion, Pathways in cancer, ECM − receptor interaction, Chemokine pathway. These signal pathways revealed that how the 2 genes affect the immune microenvironment and EMT process. Among these results, a large number of studies have found that the activation of the TGF-β pathway can not only promote the EMT process, but also induce immunosuppression[9, 53]. It could be inferred from our results that the up-regulated CMTM3 and LTBP2 could aggravate the EMT process by the TGF-β pathway in GC and promote metastasis. Besides, CMTM3 and LTBP2 may also influence the immune infiltration, expression of the immune checkpoints, and antigen presentation function by the TGF-β pathway. Eventually, this phenomenon will result in the formation of the immunosuppressant microenvironment. So, CMTM3 and LTBP2 may play important role in regulating the survival of GC by influencing EMT and immune simultaneously.
However, TGF-β may not be the only signaling pathway in these processes. According to previous research, we found that the PI3K − Akt signaling pathway was involved in the EMT process[54]. For example, Xu et al. contend that the PI3K/Akt signaling pathway can affect the EMT in a variety of ways to influence tumor aggressiveness[55]. This signal pathway could also regulate the immune microenvironment and leads to the occurrence of immune escape[56]. Cai et al. Found that the activation of the PI3K/Akt signaling pathway could result to M2-like macrophage polarization and help to induce the immunosuppressant microenvironment[57]. These studies conformed to our results and demonstrated the activity of the PI3K/Akt signaling pathway in regulating the immune function and EMT progression in GC. Therefore, we speculate that CMTM3 and LTBP2 may affect the progression of the immunosuppressant microenvironment and EMT mainly through the TGF-β and the PI3K/Akt signaling pathways. In the future, we will design animal experiments to confirm our hypothesis.
CMTM3 and LTBP2 were promising prognostic biomarkers in GC. While the studies about the therapy of CMTM3 and LTBP2 were still less. In this study, we found they also may regulate the sensitivity of therapy. The results of enrichment analysis showed that the drug metabolism-cytochrome P450 pathway was enriched, which could hint to us that CMTM3 and LTBP2 maybe 2 potential therapeutic targets to some extent. Furthermore, by conducting the drug sensitivity analyses, we found that up-regulated CMTM3 and LTBP2 may result in poor sensitivity to immune checkpoint blocking therapy. Moreover, the results also demonstrated that CMTM3 and LTBP2 may be good indicators to judge the sensitivity of some chemotherapy drugs. As a result, they may not only indicators in predicting the survival of GC patients, but also promising potential therapeutic targets. We also constructed a nomogram by CMTM3 and LTBP2 to better predict the survival and therapy outcomes in clinical.
There are some limitations in this study. In the future, we will further design animal and cell experiments to verify the results. And we will enlarge our samples in subsequent experiments to do further validation.
In this study, our work largely revealed the roles of CMTM3 and LTBP2 in the progression of GC, especially in the immune microenvironment, EMT process, and drug resistance, which is crucial for the development of therapeutic methods. As a result, CMTM3 and LTBP2 were significant prognostic biomarkers and potential therapeutic targets in gastric cancer, which deserve further studies.
Gastric cancer
STAD
Single-cell RNA sequencing
EMT
TCGA
IHC
DEGs
ROC curves.
Acknowledgments
We sincerely thanks to the gastric cancer patients who participated in this study.
Author s’contributions
Ning Kang conceived the idea, designed the experiments, analyzed the data and wrote the manuscript. Licui Qi collected clinical samples. Ning Kang modify and check the manuscript. All authors read and approved the final manuscript.
Funding
Not applicable
Availability of data and materials
The datasets GSE118916, GSE79973, GSE19826 are available in the GEO database (http://www.ncbi.nlm.nih.gov/geo). Meanwhile, the TCGA repository (https://portal.gdc.cancer.gov/projects/TCGA) was utilized, under the accession code: Stomach adenocarcinoma (STAD). GTEx data were used by GEPIA online database.
Ethics approval and consent to participate
The studies involving human participants were reviewed and approved by the institutional review board of the second hospital of Hebei Medical University (Approval number: 2022-R325). The informed consent was obtained from all participants. All procedures performed in this study involving human participants were in accordance with the ethical standards of the institutional research committee and with the 1964 Helsinki Declaration and its later amendments.
Consent for publication
Not applicable.
Competing interests
The authors declare that there are no conflicts of interest regarding the publication of this study.
No competing interests reported.
Supplementary Information See Figures S1-S5 in the “Supplemental Figure.pdf” for comprehensive image analysis. Figure S1:The expression analysis on scRNA-seq data (GSE134520) by TISCH database. Figure S2: The prognosis, clinical pathological correlation analysis of CMTM3 and LTBP2. Figure S3: The immune-related analyses of CMTM3 and LTBP2. Figure S4: The critical pathway maps of KEGG analysis. Figure S5: Mutation, copy number variation, and methylation analysis of CMTM3 and LTBP2 in gastric cancer.
posted
You are reading this latest preprint version