3.1 Identification of differentially expressed genes in GC

The flow chart of this study is shown in Fig. 1. We explored gene expression information of GC from GEO database and obtained the genomic profiles of GSE19826, GSE54129 and GSE79973. Volcano plots were generated to visualize the expression patterns of these differentially expressed genes in these databases (Fig. 2A). Venn diagram software was next used to detect the commonly up-regulated genes in these datasets (Fig. 2B). In total, 26 genes were identified to be significantly highly expressed in gastric tumor tissues compared with normal controls (Table S2). Interestingly, these highly expressed genes except retinoic acid receptor responder 1 (RARRES1) was significantly associated with poor overall survival in patients with GC (Figs. S1 and 2). Of these genes, we pay attention to WISP1 which belongs to CCN family and plays crucial roles in many physiological and pathological processes.

Fig. 1

Flow chart of data collection, analysis and validation

Fig. 2

Gene ontology enrichment analysis and the genetic alterations of the CCN family genes. A Volcano plots of the differentially expressed genes in the GSE19826, GSE54129 and GSE79973 databases. B Venn diagram to show the commonly up-regulated genes. C Significantly enriched gene ontology terms of CCN family. D The genetic alteration profiles of the six CCN family genes in STAD from cBioportal. E Genetic alteration frequency of CCN family genes in STAD using cBioPortal. F Relative expression differences of CCN family genes in STAD from HAP database

3.2 Functional enrichment analysis of CCN family genes in GC

To better understand the functions of CCN family, we performed the functional enrichment analysis using DAVID 6.8. As shown in Fig. 2C, among all the significant enriched functions in the biological process (BP) category, negative regulation of cell death, regulation of cell growth, signal transduction, reactive oxygen species metabolic process, and angiogenesis were involved in the tumorigenesis and progression of GC. Cell–cell signaling and cell adhesion were associated with tumor microenvironment. The proteinaceous extracellular matrix was the only significant enriched item in the cell component (CC) category. In the molecular function (MF) category, the CCN family members were mainly involved in integrin binding and heparin binding activities. Table 1 showed the specific genes involved in these biological activities. Therefore, the function of CCN family is regulating cell–cell and cell-matrix interaction and in turn modulating cell growth.

Table 1 Gene ontology analysis of CCN family genes

We also performed Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses and found no significant pathway was enriched in the CCN family. This limited information may be attributable to insufficient studies on CCN family. We next conducted the functional enrichment analysis with Metascape database. The functions of CCN family genes were mainly enriched in naba extracellular matrix glycoproteins pathway (Table S3).

3.3 The genetic alterations of CCN family members in patients with GC

We next explored the genetic alterations of CCN family members in 412 patient samples with stomach adenocarcinoma via cBioportal database. Stomach adenocarcinoma (STAD) is the most common histological type (~ 95%) of GC [7]. In detail, CCN1, CCN2, CCN3, CCN4, CCN5, and CCN6 were altered in 0.5, 1.7, 8, 9, 6, and 2.9% of these samples (Fig. 2D). WISP1 (CCN3) was frequently altered CCN family gene with an alteration rate of 8%. More than 30% of GC patients showed genetic aberrations of CCN family genes and the most common alteration was mRNA deregulation including mRNA high and mRNA low (Fig. 2E). Additionally, we compared the mRNA level among CCN family members in STAD via Human Protein Atlas (HPA) database, the mRNA level of WISP2 was quite low (Fig. 2F).

3.4 The expressions of CCN family members in patients with GC

We explored the expression levels of CCN family genes in STAD and normal tissues with GEPIA. Based on the data form GEPIA, only the expression of WISP1 in GC tissues was significantly elevated while the expression of WISP2 were significantly reduced (Fig. 3). The expressions of CTGF, NOV and WISP3 were slightly increased in tumor tissues without a significant p-value.

Fig. 3

Gene expression of CCN family genes in STAD. Comparison of the mRNA expression of CYR61 (A), CTGF (B), NOV (C), WISP1 (D), WISP2 (E), and WISP3 (F) in gastric tumor tissue (left plot n = 408) and normal tissue (right plot n = 211) from GEPIA database

In addition, we analyzed the transcriptional levels of CCN family members in GC and normal gastric tissues using ONCOMINE database. The transcription levels of CYR61, CTGF, NOV, WISP1, and WISP3 in different types of GC tissues were significantly elevated compared with normal gastric tissues in these datasets shown in Table 2. Three datasets including Cui dataset [11], Derrico dataset [12] and Wang dataset [13], suggested that the mRNA expressions of CTGF, WISP1 and WISP3 in GC were prominently more than 1.5-fold higher than in normal gastric controls. The transcriptional level of WISP2 in GC tissues exhibited no significant alteration in all the datasets [11,12,13,14]. The possible explanation for the different results in these databases might be the different samples included and detection methods selected. We considered excluding several CCN family members such as WISP2 from further analysis, however, we preferred to identify the potential of the whole CCN family as effective prognostic biomarker in GC.

Table 2 The elevated mRNA levels of CCN family members in different types of GC (ONCOMINE)

We then assessed the alteration of CCN family expressions during the gastric tumor progression, finding a significant relationship between the expression of NOV and WISP1 and pathological stage (Fig. 4). With the progression of the gastric tumor, the expression of NOV and WISP1 increased, suggesting that NOV and WISP1 might play a pivotal role in the promotion and progression of GC. In addition, by analyzing the protein expressions of CCN family form HPA database, we found that only WISP1 expression was increased in gastric tumor tissues (Table S4). This result mainly due to insufficient sample size and ineffectiveness of antibodies.

Fig. 4

Correlation between different expressed CCN family genes and the pathological stage of patients with STAD. A CYR61, B CTGF, C NOV, D WISP1, E WISP2, F WISP3. *p < 0.05

3.5 The prognostic value of CCN family genes in patients with GC

We analyzed the association between the mRNA expressions of CCN family genes and clinical outcome in patients with GC. The patients had been divided into high and low expression groups in the Kaplan–Meier Plotter database. Patients with high expressed CTGF was significantly related to better overall survival, whereas patients with high transcriptional levels of NOV, WISP1, WISP2, and WISP3 were remarkably associated with shorter overall survival (Fig. 5; CYR61, HR = 1.1, 95% CI 0.93–1.3, p = 0.28; CTGF, HR = 0.72, 95% CI 0.6–0.87, p = 0.00064; NOV, HR = 1.55, 95% CI 1.31–1.84, p = 3.4e−07; WISP1, HR = 1.49, 95% CI 1.19–1.87, p = 0.00049; WISP2, HR = 1.45, 95% CI 1.19–1.78, p = 0.00021; WISP3, HR = 1.45, 95% CI 1.21–1.74, p = 4e−05). The prognostic value of differentially expressed CCN family genes in the first progressive survival was also assessed. Similar with the result in overall survival, high expressed CTGF and low expressed NOV, WISP1, WISP2, and WISP3 were significantly associated with longer first progressive survival (Fig. 6; CYR61, HR = 0.83, 95% CI 0.66–1.03, p = 0.094; CTGF, HR = 0.72, 95% CI 0.58–0.89, p = 0.0028; NOV, HR = 1.4, 95% CI 1.31–1.74, p = 0.002; WISP1, HR = 1.6, 95% CI 1.25–2.05, p = 0.00015; WISP2, HR = 1.44, 95% CI 1.14–1.81, p = 0.0021; WISP3, HR = 1.38, 95% CI 1.13–1.7, p = 0.0016).

Fig. 5

Overall survival analysis of CCN family genes in patients with GC. Survival curves of CYR61 (A), CTGF (B), NOV (C), WISP1 (D), WISP2 (E), and WISP3 (F) analyzed via the Kaplan–Meier plotter database

Fig. 6

Survival analysis of CCN family genes in patients with diffuse GC. First progressive survival curves of CYR61 (A), CTGF (B), NOV (C), WISP1 (D), WISP2 (E), and WISP3 (F) analyzed via the Kaplan–Meier plotter database

According to the Lauren classification, GC is separated into diffuse and intestinal types histologically [15]. Therefore, we investigated the prognostic value of CCN family genes in these two cancer types. Subgroup analyses indicated that higher expression levels of all the CCN family members were correlated with poorer overall survival (OS) in patients with intestinal type (Fig. S3; CYR61, HR = 2.25, 95% CI 1.63–3.11, p = 4e−07; CTGF, HR = 1.53, 95% CI 1.11–2.1, p = 0.0084; NOV, HR = 3.17, 95% CI 2.05–4.9, p = 4.7e−08; WISP1, HR = 1.65, 95% CI 1.14–2.4, p = 0.0077; WISP2, HR = 2.01, 95% CI 1.46–2.77, p = 1.3e−05; WISP3, HR = 2.05, 95% CI 1.45–2.9, p = 3.7e−05). And elevated mRNA expressions of CTGF, NOV and WISP2 were associated with poorer OS in patients with diffuse type (Fig. S4; CYR61, HR = 1.3, 95% CI 0.9–1.89 p = 0.16; CTGF, HR = 1.45, 95% CI 1.03–2.04, p = 0.031; NOV, HR = 2.05, 95% CI 1.41–2.97, p = 0.00011; WISP1, HR = 1.39, 95% CI 0.98–1.96, p = 0.064; WISP2, HR = 1.66, 95% CI 1.1–2.51, p = 0.015; WISP3, HR = 0.78, 95% CI 0.55–1.09, p = 0.14).

3.6 Inhibition of WISP1 dampens proliferation and migration of GC cells

Our data suggested that targeting WISP1 could be a promising strategy for the treatment of GC. Therefore, we preliminarily investigated the effects of WISP1 expression level on the proliferation and migration of GC cells. We used RNA interference technology to downregulate WISP1 expression. After cellular transfection of WISP1-siRNA and WISP1-negative control (NC), WSIP1 expression in each cell lines were detected using western blot. As shown in Fig. 7A, WISP1 expression was significantly lower in the WISP-siRNA group compared with the control group. Decreased WISP1 exhibited no affection on cell activity (Fig. 7B). To determine the effect of WISP1 on the proliferation and migration of GC cells, colony formation and scratch test were performed. The activity of colony formation in the WISP1-siRNA group was significantly attenuated compared with the WISP1-NC groups (Fig. 7C). In addition, the cell scratch distance in the WISP1-siRNA group was significantly larger compared with the WISP1-NC groups after 24 h (Fig. 7D).

Fig. 7

Inhibition of WISP1 dampens proliferation and migration of GC cells. A siRNA lowers WISP1 expression in HGC-27 and SNU-1 cells. B apoptosis of GC cells after transfection for 48 h. C Colony formation of GC cells after transfection. D Scratch test results in SNU-1 cells after transfection. The experiment was repeated 3 times

3.7 Key transcription factors of CCN family genes in GC

We further evaluated possible transcription factor targets of CCN family using TRRUST database. Unfortunately, the most valuable member WISP1 was not searched in TRRUST. CYR61, CTGF, NOV, and WISP2 were included in this database. We found that two transcription factors including Wilms tumor 1 (WT1) and signal transducer and activator of transcription 3 (STAT3) were involved in the regulation of CCN family genes (Table 3). Specifically, WT1 was the key transcription factor for CTGF and NOV. STAT3 was the key transcription factor for CYR61 and CTGF.

Table 3 Key regulated factor of CCN family members in GC (TRRUST)