A Systemic Analysis Of The Colonic Adenoma-Normal SSH Library And Identification Of Differentially Expressed Genes In Colorectal Cancer

Colorectal cancer (CRC) is a common gastrointestinal malignancy which threatens the health of human. In developed countries, its mortality and incidence had only changed little since great efforts had been taken in the fields of cancer research. In American, the incidence and mortality of colorectal carcinoma still ranks the 3rd among all the cancers in both the male and the female in 2007. In China, the incidence of colorectal carcinoma is increasing gradually and also ranks the 3rd place in all the malignancy. We will encounter the fastigium of colorectal carcinoma in decades due to the changes of life style and diet elements.Carcinogenesis of colorectal cancer is believed to be a long-term and multi-step process with multi-gene changes. The adenoma-carcinoma sequence is now widely accepted as the major pathway for the development of colorectal cancer in the general population. Colorectal adenomas are recognized as the precursor lesions for most cases of colorectal cancer. Removal of these adenomas resulted in a markedly lower colorectal cancer occurrence than without polypectomy. However, the early genetic events in the processes of adenoma initiation and development still remain to be elucidated. Study on differentially expressed genes between adenoma and normal mucosa is of great significance not only in deciphering the underlying molecular biology but also in discovering potential biomarkers, which are crucial to early cancer detection. Therefore, we constructed three subtracted cDNA libraries by SSH (suppressive subtraction hybridization) in 1999. They were colonic adenocarcinoma-normal mucosa cDNA subtracted library (subtracted T-N library), colonic adenoma-normal mucosa cDNA subtracted library (subtracted A-N library) and colonic adenocarcinoma-adenoma cDNA subtracted library (subtracted T-A library).By Virtual Northern Blot and RNA dot blot, we had confirmed some differentially expressed genes. Subsequently, all the 109 cDNA clones in A-N library were sequenced and the whole profile of genes was analyzed by a constructed bioinformatic analysis platform and web-based Webgestalt tool. We validated 4 genes upregulated in colorectal cancers and investigated two of them exhaustively. By this means, we hope to reveal the early molecular events and markers for early diagnose and therapy of colorectal cancer.Firstly, in order to achieve the precise profile for the 109 cDNA clones in A-N library effectively, we constructed a bioinformatic analysis platform to cluster the EST to candidate genes automatically. The platform include two distinct software packages, the nuclear acid sequence analysis package (software registration number: 2005RS01351) and GetUni package (software registration number: 2005SR10081).We downloaded non-redundant nucleotide database and UniGene database and then successfully got 62 candidate genes by the platform. The information for 62 genes, such as Gene Name,Gene Symbol,Chromosomal Location and hits were output in Excel format and saved on local computer. Subsequently, overall features for the A-N library were analyzed by Gene Ontology WebGestalt analytic tool (WEB-based GEne SeT AnaLysis Toolkit, http:// bioinfo.vanderbilt.edu/gotm/). GOTM analysis demonstrated that at the default level 4 of biological processes, these genes were relatively enriched in cellular metabolism, macromolecular metabolism and primary metabolism especially involving 6 ribosomal protein genes. Moreover, they were enriched in molecular functions of in purine nucleotide binding, or in cellular processes of intracellular organelle, cytoplasm and intrinsic to membrane. Fifty genes were found to be involved in 38 KEGG pathways, strikingly, of which, 6 genes were associated with ribosome pathway. Twenty-night genes participated in 25 Biocarta pathways. Unexpectedly, 6 genes, major histocompatibility complex, class II, DR alpha (HLADR-A), chemokine (C-C motif) ligand 11 (CCL11), Beta-2-microglobulin (B2M) and transducer of ERBB2, 1 (TOB1) were involved in immune-related pathways such as IL5 Signaling Pathway, Antigen Processing and Presentation and the Role of Eosinophils in the Chemokine Network of Allergy.Next, the most abundant genes in A-N library REG4,PERP,YWHAZ and KLF4 were validated by SYBR Green Q-PCR in 8 colonic normal mucosa-adenoma-carcinoma (N-A-T)paired samples and 49 colonic normal-carcinoma(N-T) paired samples. We found REG4 overexpressed in 7 adenomas and 23 carcinomas compared to the normal mucosa(medium change folds: 2.263 for adenoma vs normal mucosa, 0.982 for carcinomas vs normal mucosa). Especially in 6 N-A-T paired samples, REG4 was extraordinarily upregulated in adenomas compared to the normal mucosa and carcinomas. REG4 significantly overexpressed in adenomas than those in paired normal mucosa (p=0.036). PERP overexpressed in 6 adenomas and 39 carcinomas compared to the normal mucosa with statistic significance (medium change folds: 2.146 for adenoma vs normal mucosa, 1.941 for carcinomas vs normal mucosa) (p=0.036, p=0.000). In 5 N-A-T paired samples, PERP was extraordinarily upregulated in adenomas compared to the normal mucosa and carcinomas. YWHAZ overexpressed in 5 adenomas and 31 carcinomas (medium change folds: 1.339 for adenoma vs normal mucosa, 1.200 for carcinomas vs normal mucosa with p=0.03). It was extraordinarily upregulated in adenomas in 4 N-A-T paired samples compared to the normal mucosa and carcinomas. However, we found KLF4 overexpressed only in 2 adenomas and 4 carcinomas and the downregulation of its expression in carcinomas were significant statistically compared to the normal mucosa and adenomas (medium change folds: 0.521 for adenoma vs normal mucosa, 0.230 for carcinomas vs normal mucosa with p=0.000).By Q-PCR results for the 4 genes and the previous validation of 6 ribosomal protein genes and C6orf37, we successfully validated the overexpressed genes in A-N library.Then, we focused on PERP, one of the overexpressed genes in both adenomas and carcinomas. It was newly found downregulated in the cancer cell lines with high metastasis ability and may be the effector of p53.We carried out an elaborate analysis on its expression and function in colorectal cancers. By tissue microarray (TMA) combined with immunohistochemistry (IHC) or RNA in site hybridization (ISH) assays, we detected its mRNA and protein levels in colorectal tissues. The PERP mRNA located in the lower zones was higher than those in the upper zones in normal mucosa. It was positive in 62.5% normal mucosa, 97.6% adenomas and 88.9% primary carcinomas. PERP protein was positive in 67.7% normal mucosa, 84.2% adenomas and 81.5% primary carcinomas. Both the mRNA and protein levels of PERP were significantly higher in adenomas and carcinomas compared to the normal mucosa (p<0.001, p<0.001).For the relationship between the expression of PERP and p53, the p53 protein was detected on the same TMA samples by IHC. It shows that p53 was positive in 3.17% normal mucosa, 78.1% adenomas and 71.9% primary carcinomas. It was significantly overexpressed in adenomas and carcinomas compared to the normal mucosa (p=0.000, p=0.000). Subsequent correlation analysis showed that both PERP mRNA and protein levels were correlated with the p53 protein level significantly (r=0.452, p=0.000 for PERP mRNA level and p53 protein level; r=0.341, p=0.000 for PERP protein level and p53 protein level). In order to reveal the possible function of PERP in colorectal cancer, we selected two CRC cell lines without or with low endogenous PERP expression for transfection assays. The constructed expression vector PcDNA3.1-PERP CDs was transfected into LoVo and RKO cell lines. After confirmation the PERP expression in transfectants by RT-PCR, we tested the biological effects. Growth proliferation assay using MTT showed that PERP could promote the growth of CRC cells. A significant increase in growth of LoVo-PERP (p=0.000, versus control) and RKO-PERP transfectants (p=0.058, versus control) were observed in the 5th day. Moreover, PERP facilitated migration of LoVo cell. However, PERP did not affect the cell cycle and apoptosis of both cell lines detected by flow cytometric analysis of PI staining and annexin V/PI double staining.Meanwhile, we noticed reverse result of KLF4 that overexpressed in our A-N library but was downregulated in N-A-T samples. It was just a bias from construction of A-N library with only single patient. However, recently, KLF genes family became hotspot in cancer researches because their important roles in cell differentiation and mature. The proliferation-differentiation dynamics of colonic crypts provides us an excellent model to observe cell proliferation and differentiation. The significant downregulation of KLF4 expression in N-A-T sequence supported not only its important role in the homeostasis and maintenance of colonic mucosa but also the longstanding hypothesis that "tumorigenesis is reversal of cell differentiation or dedifferentiation". All of these attracted us continuing to carry out a large clinicopathological survey exhaustively on expression of KLF4 in colorectal tissues.Using immunohistochemistry, we systematically analyzed distinct features of KLF4 expression in colorectal normal mucosa, adenomas and carcinomas. The colorectal epithelium follows the paradigms of stem cell biology. With a unique topology, it roughly constitutes a two-dimensional structure: the proliferative crypts (lower 1/2 crypt) and the differentiated villi (upper 1/2 crypt). Hence, in the current study, we divided the normal or dysplastic glands into two equal crypt parts, the upper and lower zone, to evaluate the staining distribution patterns.The glandular epithelial cells in all normal mucosa showed moderate to strong nuclear KLF4 staining. Intriguingly, KLF4+ cells were predominantly located in the upper zones, significantly higher than those in the lower zones (Mean±S.D.: upper 35.7%±19.0% versus lower 11.4%±9.8%; p<0.01). As in the normal mucosa, KLF4+ cells were also more common in the upper zones of dysplastic crypts of colorectal adenoma than those in the lower zones (Mean±S.D.: 19.7%±20.5% versus 9.0%±13.4%, p<0.01).Moreover, KLF4 expression was significantly down-regulated in colorectal adenomas than those in the peri-adenomatous normal mucosa (Mean±S.D.: 15.0%±16.3% in adenomas versus 30.6%±13.5% in normal mucosa, p<0.01). In sharp contrast, KLF4+ cells were diffusely distributed along the whole dysplastic glands in colorectal carcinomas although KLF4 staining was lost in most cancers (Mean±S.D.:3.0%±6.72%). As compared with those in the individual-matched normal mucosa adjacent to the excision margin, KLF4+ cells decreased significantly in colorectal cancers (Mean±S.D.: 2.0%±4.6% in cancers versus 23.4%±11.5% in normal mucosa, p<0.01). Collectively, KLF4 were significantly downregulated in N-A-T sequence (Mean±S.D.:23.5%±13.0% in normal mucosa versus 15.0%±16.3% in adenomas versus 3.0%±6.72% in carcinomas,p<0.01).KLF4 protein was positive in all 73 normal tissues, 28/48 (58.3%) adenomas and 38/129 (29.5%) primary carcinomas. KLF4 protein expression was significantly lower in colorectal adenomas or carcinomas than that in normal mucosa (p<0.01). We found that the rate of positive KLF4 protein staining cells was higher in adenomas with low-grade dysplasia (26/39) than that in adenomas with high-grade dysplasia (2/9) (p<0.01).KLF4 expression was more common in colorectal cancer patients with younger age (≤60 years) or patients with less tumour buddings(<15) (p<0.01, p<0.01). The overall 5-year survival rate was 67.9% (19/28) in KLF4 positive colorectal cancers and 51.7% (30/58) in KLF4 negative ones (p=0.172) . Accumulated 5-year survival rate (74%) was higher in KLF4 positive colorectal cancers than that in negative ones(63%)(p=0.379).KLF4 positive colorectal cancer patients with lymph node metastasis had a trend towards better overall survival than KLF4 negative patients with lymph node metastasis did (p=0.252).For the mechanism of KLF4 downregulation and its possible roles in cell differentiation in colorectal cancer, we treated colorectal cancer cells SW480, SW620, RKO by 5-aza-dC and butyrate sodium. The mRNA level of KLF4 increased after 5-aza-dC or butyrate sodium treatment. The medium change folds of 5-aza-dC treatment cells over control were 1.14 in SW480, 5.49 in SW620 and 2.10 in RKO. In contrast, butyrate sodium induced a moderately elevated expression level of KLF4 in colorectal cancers with medium fold changes of 1.36 in SW480, 1.57 in SW620 and 1.47 in RKO. KLF4 expression were significantly upregulated after 5-aza-dC treatment in SW620(p<0.01).In conclusion, we demonstrated that:(1) The energetic bioinformatic analysis platform was efficient in clustering EST to candidate genes. Combined with web-based WebGestalt analysis, it enabled not only the understanding of early molecular events in colorectal cancers but also the discovering of potential biomarkers, which are crucial to early cancer detection.(2) PERP significantly overexpressed in colorectal adenomas and carcinomas. Its Overexpression may correlate with the mutant p53 in colorectal carcinomas. PERP could promote the growth proliferation and migration in LoVo and RKO cell lines. (3) KLF4 was drastically downregulated in colorectal adenomas and carcinomas via possible epigenetic modifications. It is associated with terminal differentiation in colorectal epithelium. Loss of KLF4 protein expression might contribute to assessing prognosis in colorectal cancer with lymph node metastasis.(4) Suppressive subtraction hybridization (SSH) is still helpful in identification of differentially expression genes. It facilitates us in deciphering the underlying molecular biology and in discovering potential biomarkers for early diagnosis, therapy and prognosis.