Prediction And Validation Of Novel Hub Genes Associated With Liver Metastasis Of Gastric Cancer

[Bankground] Gastric cancer (GC) is the second leading cause of cancer death worldwide, although its incidence is decreasing. About 60% of new cases of GC occur in eastern Asia. Surgical resection is the most effective treatment for GC without distant metastasis. However, relapse after surgical treatment and distant metastasis contribute to the high GC-associated fatality. Understanding hub genes involved in gastric cancer (GC) metastasis could lead to effective, targeted approaches to diagnose and treat cancer metastasis.[Objective] To isolate hub genes responsible for gastric cancer metastasis, and establish novel biomarkers for the predition of gastric cancer and metastasis, and lay the groundwork for further molecular epidemiological study of gastric cancer.[Methods] Tumor tissues of 67 patients were obtained from GC patients who had undergone curative surgery at Department of General Surgery of the 1st affiliated hospital, Second Military Medical University, from January 2006 to February 2009; The synchronous liver metastasis and their matched GC from 3 patients were used for double-colored cDNA microarray assay and differentially expressed genes with fold-change 1.6 or higher in each microarray data were selected for further bioinformatics and validation analyses. a novel intergrated bioinformatics methods for the prediction of gene function through combination of microarray technology, text mining, protein-protein interaction (PPI) and genes similarity analysis. Online bioinformatics tools, such as MILANO, Chilibot, UniHI and Endeavour, were used in turn for literature ananotaions and gene founction prediction. Then, the candidate genes were validated by using real-time quantitative RT-PCR and immunohistochemistry in synchronous liver metastasis verse the paired GC from 8 other patients and in GC verse the matched gastric mucosa from 57 patients. The protein expression patterns of candidate genes also were validated by using immunohistochemistry among distance metastasis, orthotopic cancer and adjacent mucosa of GC. [Results] We selected 272 differentially expressed genes with fold-change 1.6 or higher in each microarray. KEGG pathway analysis showed that of the 13 enriched pathways, 8 were involved in cancer metastasis. Literature-based annotation with MILANO and Chilibot showed that the differentially expressed genes significantly enriched known metastasis-related genes. With the use of protein-protein interaction network, we found a sub-network significantly enriching the metastasis-related genes and hub genes. First 45 of un-annotated genes in Endeavour prioritization shared 8 hubs in the sub-network, containing NR3C1,NR4A2, HNRPA1, PSMB3, FBLN2, DARS, XAB2, and CD8A. Nine hubs,containing NR3C1, NR4A2, HNRPA1, XAB2, HSP90AA1, CCNE1, RPL17, FKBP1A, and XAB2 were then selected for validation by using quantitative RT-PCR in the sub-network. It was found that NR4A2 (P=0.001) was down-regulated, while HSP90AA1 (P=0.029) up-regulated, in synchronous liver metastasis verse the paired GC from 8 patients. NR4A2, NR3C1, ARF3, XAB2, and alternatively spliced variants of NR4A2, SP8 and SP-novel, were down-regulated (P<0.001 for each), while CCNE1 (P=0.001) up-regulated, in GC verse the matched gastric mucosa from 57 patients.Immunohistochemistry showed that HSP90AA1(P=0.043) and NR4A2(P=0.003) protein were significantly up and down-regulated in the distant metastasis when compared to the orthotopic cancer of GC, respectively. CCNE1(P<0.001) and NR3C1(P=0.005) were also significantly up and down-regulated in orthotopic cancer compared to adjacent mucosa of GC. NR4A2 protein was significantly up-regulated in normal gastric mesenchymal cells (P=0.005) and significantly down-regulated in normal gastric epithelial cells (P<0.001) when compared to primary gastric cancer tissue.[Conclusion] NR4A2 stands out as the most probable diagnostic and prognostic marker for GC. NR4A2 may play an important role in epithelial-mesenchymal transitions (EMT) of gastric cancer liver metastasis. NR3C1, ARF3, and XAB2 are novel hubs reversely associated with GC. HSP90AA1 is a potential prognostic marker, while CCNE1 a potential diagnostic marker, for GC.