Identification of JAM2 from GEO date sets
GSE10072, GSE43458, GSE31210, and GSE32863 data sets were selected to analyze and identify the differentially expressed genes (DEGs) between normal tissues and LUAD. According to the setting criteria as p-value < 0.01 and fold change > 1 or fold change < -1, DEGs of four microarrays were identified. As shown in volcano plots, there are 846, 945, 3665 and 1334 DEGs respectively (Fig. 1A-D). Then, a Venn diagram was used to overlap among the 4 datasets, and we found 215 DEGs, among which 59 genes were downregulated, and 156 genes were upregulated (Fig. 1E). We conducted enrichment analysis of differential genes and found that genes such as COL1A1, SPP1, COL3A1, COL1A1, TGFBR2, MMP9 and JAM2 were widely enriched in signaling pathways related to tumor genesis and development (Fig. 1F). In this research, we mainly focused on the role of JAM2 in lung adenocarcinoma.
JAM2 expression levels in LUAD from TCGA.
Online database GEPIA were used to explore the expression profiles of JAM2 in LUAD. As we shown in Fig. 2A, JAM2 were downregulated in LUAD than but not in normal tissues. Consistent with its expression level in TCGA, JAM2 was relatively lower expressed (p < 0.001) in LUAD tissues against normal ones (Fig. 2B). Moreover, we investigated the expression level of JAM2 by dividing the LUAD tissues from TCGA into T stage, N stage and pathological stages. As shown in Fig. 2C-E, there were significant differences between JAM2 expression level with tumor grade, regional lymph node stage, and pathological stage. These results indicated that JAM2 may be a biomarker for LUAD progression.
Hypermethylation of JAM2 in LUAD
It is clear that hypermethylation of the host genome directly targets key tumor suppressor genes, results in gene silencing and inhibit tumorigenesis. The methylation status of JAM2 was analyzed to ascertain the downregulated mechanism in LUAD. The results demonstrated that JAM2 was highly methylated in cancer tissues than normal ones (Fig. 3A). Spearman correlation analysis of 3 DNA methyltransferases showed that JAM2 expression was negatively related to the expression of DNMT1 (r = -0.14, P = 0.001), DNMT3A (r = -0.18, P < 0.001) and DNMT3B (r = -0.263, P < 0.01) (Fig. 3B-D).
Prognostic value of JAM2 in LUAD
Furthermore, to study the association between JAM2 expression level with prognosis, we performed a survival association analysis through GEPIA. The results showed that LUAD patients with higher expression of JAM2 had longer survival times than lower ones (Fig. 4A) (p = 0.028). In the TCGA cohort, the high-JAM2 group showed a better overall survival rate compared with low-JAM2 group based the Kaplan-Meier analysis (Fig. 4B). More importantly, Univariate Cox regression analysis showed that JAM2 expression (Hazard ratio = 0.72, p = 0.039) and pathological stage (Hazard ratio = 1.57, p = 0.014) were significantly associated with the prognosis of LUAD (Fig. 4C). Multivariate Cox regression analysis confirmed that JAM2 was a protective factor of LUAD (Hazard ratio = 0.619, p = 0.005). Therefore, JAM2 could be served as an independent prognostic gene to predict the outcomes of LUAD patients (Fig. 4D).
Identification of coexpression genes of JAM2 and construction of PPI network.
To better understand the function of JAM2 in LUAD, we applied R package limma to explored genes that were aberrantly regulated in LUAD. Genes were divided into two groups according to the expression level of JAM2, the top twenty upregulated and downregulated genes were shown in heatmap (Fig. 5A). STRING was used to create a PPI network of aberrantly regulated genes (Fig. 5B). Significant modules and hub genes in the PPI network were identified by cytohubba, a plug-in of Cytoscape (3.9.1) software. A total of 10 genes were identified as hub genes (Fig. 5C).
KEGG and GO enrichment analyses of coexpression genes.
Then, GO function enrichment analysis were performed for common coexpression genes (Table 1). The common genes were enriched in many biological processes, such as epithelial cell proliferation, muscle tissue development, regulation of epithelial cell proliferation, extracellular structure organization, and extracellular matrix organization (Fig. 6A). In Cellular Component, they were mainly located in collagen-containing extracellular matrix, cell-cell junction and contractile fiber (Fig. 6B). From the molecular function, it was observed enriched in actin binding, tubulin binding, glycosaminoglycan binding and sulfur compound binding (Fig. 6C). The KEGG pathways analysis(Table 2)showed that the coexpression genes were mainly concentrated in PI3K-Akt signaling pathway, Focal adhesion, cell cycle and Dilated cardiomyopathy (Fig. 6D).
Table 1
TOP 15 GO functional enrichment analysis of co-expression genes for JAM2 in LUAD.
GO | Category | Description | p value | Count |
GO:0043062 | GO Biological Processes | extracellular structure organization | 5.31E-22 | 81 |
GO:0030198 | GO Biological Processes | extracellular matrix organization | 1.79E-21 | 80 |
GO:0045229 | GO Biological Processes | external encapsulating structure organization | 3.51E-21 | 80 |
GO:0050673 | GO Biological Processes | epithelial cell proliferation | 3.63E-16 | 90 |
GO:0050678 | GO Biological Processes | regulation of epithelial cell proliferation | 5.52E-16 | 82 |
GO:0062023 | GO Cellular Components | collagen-containing extracellular matrix | 1.20E-28 | 112 |
GO:0043292 | GO Cellular Components | contractile fiber | 1.27E-12 | 56 |
GO:0005911 | GO Cellular Components | cell-cell junction | 1.13E-11 | 88 |
GO:0030016 | GO Cellular Components | myofibril | 1.20E-11 | 53 |
GO:0030017 | GO Cellular Components | sarcomere | 1.31E-11 | 50 |
GO:0005201 | GO Molecular Functions | extracellular matrix structural constituent | 8.80E-15 | 50 |
GO:0008201 | GO Molecular Functions | heparin binding | 1.15E-11 | 44 |
GO:0005539 | GO Molecular Functions | glycosaminoglycan binding | 1.07E-10 | 52 |
GO:1901681 | GO Molecular Functions | sulfur compound binding | 5.23E-08 | 51 |
GO:0098631 | GO Molecular Functions | cell adhesion mediator activity | 3.03E-07 | 19 |
Table 2
KEGG functional enrichment analysis of co-expression genes for JAM2 in LUAD.
Term | Category | Description | Count | GeneRatio | p value | |
hsa04512 | KEGG Pathway | ECM-receptor interaction | 27 | 27/714 | 3.40E-09 | |
hsa05412 | KEGG Pathway | Arrhythmogenic right ventricular cardiomyopathy | 21 | 21/714 | 1.67E-06 | |
hsa05414 | KEGG Pathway | Dilated cardiomyopathy | 24 | 24/714 | 1.72E-06 | |
hsa04110 | KEGG Pathway | Cell cycle | 28 | 28/714 | 3.04E-06 | |
hsa05410 | KEGG Pathway | Hypertrophic cardiomyopathy | 22 | 22/714 | 6.87E-06 | |
hsa04510 | KEGG Pathway | Focal adhesion | 37 | 37/714 | 9.87E-06 | |
hsa04151 | KEGG Pathway | PI3K-Akt signaling pathway | 54 | 54/714 | 3.30E-05 | |
hsa04610 | KEGG Pathway | Complement and coagulation cascades | 19 | 19/714 | 0.000107 | |
hsa00350 | KEGG Pathway | Tyrosine metabolism | 11 | 11/714 | 0.00017 | |
hsa04914 | KEGG Pathway | Progesterone-mediated oocyte maturation | 20 | 20/714 | 0.000463 | |
GSEA Analysis of JAM2.
GSEA was conducted to explore the functions and pathways of JAM2. The MSigDB results of c2.cp.kegg.v7.1.entrez.gmt indicated that gene signatures in highly expressed JAM2 classification were mainly suppressed in “cell cycle”, “base excision repair” and “non-small cell lung cancer” gene sets (Fig. 7A-C). Considering “h.all.v7.1.entrez.gmt” as a reference, the samples of lower expressed JAM2 were mainly enriched in “DNA repair”, “PI3K-AKT-MTOR signaling” and “MTORC1 signaling” sets (Fig. 7D-E). We therefore hypothesized that JAM2 may inhibit tumors progress in LUAD.
Immune Cells Infiltration Analyses of JAM2
Immune cells play a vital role in the immune microenvironment and can affect the prognosis of various cancer. However, it is unclear whether JAM2 has impacts on the recruitment of immune cells. We evaluated the correlation between immune cell infiltration and JAM2 expression by R package ssGSEA. The scores of 24 immune cell types were calculated based on TCGA-LUAD database. Expression of JAM2 were positively with immune cells except Th2 cells (Fig. 8A). While Mask cells, NK cells, Th1 cells, iDC, Macerphages, CD8 T cells, Cytotoxic cells, and Th17 cells were more enriched in the JAM2-high subgroup, Th2 cells were more common in the JAM2-low subgroup (Fig. 8B). It has been reported that chemokines have an inhibitory effect on tumor progression, Therefore, we investigated chemokine levels in different JAM2 subgroups, and found a significant positive correlation between JAM2 with CCR2 (r = 0.437; P < 0.001), CCL14 (r = 0.682; P < 0.001) and CXCL12 (r = 0.595; P < 0.001) (Fig. 8C-E).
JAM2 is downregulated in LUAD cell lines and inhibits the invasion and migration of LUAD
To further verify the above findings, Western blot was used to investigate JAM2 expression among bronchial epithelial cells and LUAD cell lines. Consist with profiles of GEO and TCGA cohorts, JAM2 was relatively higher in bronchial epithelial cells HBE and BEAS-2B than in lung cancer cell lines, such as A549, H1299, H1650 and H460 (Fig. 9A). Then, we constructed A549 and H1299 cells that stably overexpressed JAM2, and the efficiencies of overexpression were confirmed by Western blot (Fig. 9B). Wound-healing migration assay was employed to observe whether there was an effect on the migration ability of LUAD cell lines after JAM2 overexpression. From the results, we found that the wound recovery rate was significantly reduced in after JAM2 overexpression (Fig. 9C). In addition, After JAM2 was overexpressed, migratory and invasive capabilities of A549 and H1299 cells were significantly suppressed (Fig. 9D). So JAM2 could inhibit the invasion and migration of LUAD.