| Tumor metastasis is strongly associated with the overall survival of cancer patients. Although a number of genes have been reported to be involved in tumor metastasis, the molecular regulation mechanism remains unclear. There are no clinically practical biomarkers for metastasis prediction of lung cancer, hitherto.In present study, we accomplished the transcriptome analysis of totally 82 primary tumors derived from the patients with squamous cell carcinoma (SCC) of the lung, among which there were 43 with lymph node metastasis (pN+) diagnosed histopathologically and 39 without lymph node invasion (pNO). Statistical analysis identified a total of 1027 genes that demonstrated differential expression between the two groups (p?0.05). A set of signaling pathways related to cancer metastasis had been screened out by annotating the gene list using BioCarta database. Co-expression network associated with those pathways were generated among the differentially expressed genes that were significantly correlated (p?0.001) with those genes located in the selected pathways. The co-expression networks were further validated by biological experiments in vitro.Following the information contained in the co-expression molecular network, dlkl gene was picked up and its role in tumor metastasis was investigated. RT-PCR analysis showed that dlkl was aberrantly expressed in 36.7% none small cell lung cancers. A cell model that stably expressed exogenous dlkl was established following that the dlkl gene was cloned into a eukaryotic expression vector and then transfected into the lung cancer cells H520. In vitro study showed that the abnormal expression of dlkl could up-regulate the expression of cyclin B1 and finally accelerated the proliferation of the cells. On the other hand, the expression of dlkl could elevate the invasion ability of the cells but had no effect of the migration. Furthermore, dlkl could stimulate the expression and activity of MMP9 through the NOTCH signal pathway and might function as a positive regulator of metastasis. Meanwhile, we also found that the secretive DLK1 protein could induce the expression of itself through a paracrine mechanism.Besides the molecular mechanism study of cancer metastasis, we also constructed a decision tree model to assist clinical diagnosis of lymph node metastasis of early stage patients with SCC in lung. To construct this mathematical model, a total of 23 proteins were examined by immunohistochemical (IHC) analysis on the primary tumor tissues from 319 patients with SCC in lung. With the IHC data, GLM analysis was applied followed by the recursive partitioning decision tree analysis, and finally a tree model with 8 protein markers was generated. In the test cohort, the tree model had an accuracy of 76.6%, and the specificity and sensibility was 76.9% and 76.0% respectively, compared to the histopathology diagnosis.Taking all together, based on high-throughput microarray and system biology methodology, we constructed a series of tumor metastasis-related co-expression network, which might infer quite a lot to the molecular regulation mechanism of metastasis. Taking those hints from the molecular networks constructed, dlkl was selected and its potential mechanism on regulation of tumor metastasis was revealed through the experiments in vitro. Meanwhile, a decision tree model (with 8 protein markers) was constructed to estimate risk of lymphatic metastasis in pNO SCC of lung, which could be clinically feasible and practical as it uses IHC data from primary tumor as input. |
PDF Full Text Request