| With the accomplishment of Human Genome Project, some new technologies, such as microarray and Bioinformatics, are more and more widely applied. These technologies also shed new light on cancer research. Our lab constructed a homemade cDNA array platform and used it to analyze the expression pattern of hepatocellular carcinoma. Some genes involved in HCC carcinogenesis were discovered and further studied. By utilizing public databases and Bioinformatics methods, we extended our research to other fields of cancer research. My dissertation focuses on these two topics.The first part of my dissertation described homemade cDNA array construction and its application on HCC. CDNA clones were obtained from libraries of hypothalamus-pituitary-adrenal, myeloma, normal liver and HCC tissue sources. After clustering, 14,088 independent genes or EST were selected out. PCR products of these clones were concentrated and spotted on membrane array. Employing such cDNA array, the expression profiles in paired clinical hepatocellular carcinoma(HCC) and the distal non-tumorous liver tissues were studied. Despite of the heterogeneity among the clinical samples, 72 genes (including 34 novel genes) down-regulated and 84 genes (including 48 novel genes) up-regulated in over half of the cancer samples were identified. It was conspicuous that 21 out of 42 down-regulated known genes were regulated by a group of liver-enriched transcription factors (LETFs). The expression level of CCAAT/enhancer-binding protein (C/EBP) α was down-regulated in HCC by Northern blotting, whereas other LETFs were over-expressed. It was suggested these LETFs might be involved in carcinogenesis. This part of work was cooperated with Dr. Liang Xu.Analysis of array data leaded to a small GTPase of Rho family, Cdc42, which was demonstrated as up-regulated in 65% HCC samples by RT-PCR, Northern blotting and immunohistological staining. The EGFR expression was also moderately over-expressed in about 60% HCC samples by immunostaining. EGF might induce invasion of the tumor cells through Rho GTPase; however studies on Cdc42 in this process were still very limited. Both transfection of dominate negative Cdc42 (N17Cdc42) and treated with EGFR inhibitor, AG1478, to HCC cell, BEL-7404, would block the EGF induced filopodia formation and Matrigel invasion. Over-expression of Cdc42 increased the sensitivity of HCC cells to lower EGF concentration. It was suggested that up-regulated Cdc42 pathway might be involved in EGFR-mediated tumor cells invasion. However, constitutively activated Cdc42 (V12Cdc42) could not independently cause filopodia formation and Matrigel penetration. Furthermore, the expression level and activated level of Cdc42 will not be altered by EGF stimulation. These results proposed that cross talk between Cdc42 and other pathways might be involved in EGF induced invasion.The second part of my dissertation is about prediction for cancer related alternative splicing using public EST database. Several databases have been published to predict alternatively spliced mRNAs through analyzing the exon linkage relationship by alignment of ESTs to the genome sequence. EST database gives detailed information about cDNA libraries and makes it possible to predict tissueexpression through counting EST numbers. We developed a program, Alternative Splicing Assembler (ASA), and a database, Biosino Alternative Splicing Database (BASD), to look for splicing variants related to cancers. Of 4,322 genes screened, 3,490 (81%) were observed of at least one alternatively spliced variants. Using Fisher's exact test, 2,149 (8%) out of 26,812 predicted variants included in BASD, might be tumor-associated. By RT-PCR, 11 out of 13 novel spliced variants and 8 out of 9 variants' tissue specificity, especially in the HCC and lung cancer, were confirmed. Proper mining this database may be helpful in cancer studies. This part of work was cooperated with Dr. Xin Zhang of Shanghai Jiaotong University.
|
PDF Full Text Request