| Colorectal cancer (CRC) is one of the most popular cancers today. The mortality and morbidity of the colorectal cancer was in third place in advanced countries. In china, mortality and morbidity of the colorectal cancer has changed from fifth and sixth place in 1980s to fourth and fifth place, and about 200,000 people died of colon cancer. Liver metastasis is the main reasons of therapy failure. The overall five year survival of colon cancer with liver metastasis is below 40%, however, colon cancer patient without liver metastasis is over 70%. Therefore, it is really important that early detection and treatment of liver metastasis especially heterogeneous liver metastasis of colon cancer to improve prognosis of colon cancer patients. Colon cancer liver metastasis is a complicated process with multiple factors, steps, genes involved in it. These characteristics made it possible to screen and identify molecular markers for predicting liver metastasis and personal treatment target.In this study, we established three models which was used for screening, and identifying molecular markers for predicting colon cancer liver metastasis. From the model one, we retrieved the public cancer microarray data from Oncomine database (www.oncomine.org). Firstly, we systematically analyzed three datasets including Graudens, Bittner, ki containing 134 cases of colon cancer primary tissues with liver metastasis versus 362 cases of colon cancer primary tissues without liver. Apart from the stastical analytical methods and fold change, we also chose differential expression analysis using whether commercial antibody was available, function and related lituratures. In the end, we selected 6 candidate genes including OPN, BNIP3, ARG1, MAP2, MAP1A and SNCG for further analysis.Considering that from the first model we just identified cancer cells with metastatic potential can leave the primary site and invade blood vessel, but we hypothesis that not all those cell in the blood can survive in a target organ and form metastatic foci. To increase the predictive sensitivity of markers, we established another model. In this model, we co-culturing colon cancer cell sw1116 with HLSECs, then selected a subfraction of sw1116 cells that had a strong ability to interact with human liver sinusoidal endothelial cells and collected the colon cancer cells. Remaining cells were resubjected to this selection process for 21 rounds. The subline was designated as SW1116P21 (the 21th cycle with HLSEC). Comparing with sw1116, sw1116P21 showed high proliferation rate, high adhesion with HLSEC, more colony formation, acceleration of subcutaneous tumor formation and had a high propensity to metastasize to liver. SW1116P21 cells caused significantly more visible liver metastatic foci than sw1116 cells. Then we analysed the gene expression profiles of sw1116p21, sw1116 and found 200 different genes including 102 upregulated genes and downregulated 98. Application of similar source method in model one, we acquired CCL2, Cyr61, Axl, GalectinBP4 as the candidate predictive molecular markers. To increase predictive sensitivity and accuracy, we also compared different genes between colon cancer primary tissues and paired liver metastasis tissues using gene array. Applying SAM analysis, we got 44 significant upregulated genes in liver metastasis foci. From them, we selected P-cadherin, OPN, SNCG as the candidate predictive molecular markers.In the end, we got 11 candidate genes from those models and test their expression in colon tissues. Firstly, we evaluated the candidate proteins by immunohistochemistry in 40 cases including 20 CRC-M0 and 20 CRC-M1, which were randomly selected in the 117 training set. T-test analysis revealed that the expression levels in malignant cells of four markers were significantly greater in the metastasis set than in the control set. Therefore, four markers were selected for further immunohistochemical analysis using totally 245 (including 40 in initial screen and 205 other cases) cases of colorectal carcinoma. Significant differences were seen in the staining, by antibodies to P-cadherin, OPN, SNCG and CCL2 of tissue in the metastatic versus the nonmetastatic groups. Logistic regression analysis was performed on the 117 cases of colorectal carcinoma to calculate the respective significance of the individual candidate markers for liver metastasis prediction. OPN, SNCG and CCL2 showed predictive significance and increased the metastatic ratio, while P-cadherin was excluded because of P value (>0.05). To further evaluate the markers that reached statistical significance in logistic regression analysis, OPN, SNCG and CCL2 were tested for leave-one-out cross-validation. Leave-one-out validation revealed that the combination of OPN, SNCG and CCL2 yielded the most satisfactory sensitivity (90.5%) and specificity (90.7%). A test course was then performed with an independent set of 128 cases, including 78 cases of CRC-M1 (30 synchronous cases and 48 heterochronous cases) and 50 cases of CRC-M0. Using the three biomarkers, the diagnostic assay could correctly classify 43 of 50 CRC-M0s and 69 of 78 CRC-M1s. The sensitivity and specificity were 88.4% and 86%, respectively. Moreover, of the 48 cases of heterochronous metastasis included in the CRC-M1 cases, this combination could correctly classify 43 cases and the sensitivity and specificity were 89.5% and 86%. The immunoreactivity of all three markers in tissues was significantly correlated with serum CEA level, TNM stage and liver metastasis. The expression of OPN and CCL2 was further correlated with lymph node metastasis and depth of invasion. SNCG expression also showed correlation with tumor size.In the second part of this paper, we studied P-cadherin which highly expressed in the colon cancer tissue with liver metastasis and liver metastasis foci. Then we designed functional studies as follows to reveal the role of P-cadherin in colon cancer progression, related function as well as molecular mechanism. Functional analysis by siRNA-mediated silencing revealed that P-cadherin decreased proliferation, capabilities in wound healing, haptotatic migration and colony formation increased homotypic adhesion. In vivo assay that inhibiton of P-cadherin expression also decreased tumor growth and liver metastasis of Lovo cells. Furthermore, ectopic expressions of P-cadherin in 3T3 cells promoted proliferation, soft agar colony formation and tumor formation. Subsequent mechanistic studies showed that P-cadherin was strongly associated with expression ofβ-catenin and cytoplasmic or nuclear accumulation and played important role in malignant transformation, proliferation. P-cadherin expression also reduced expression of E-cadherin and resulted in increased migration.Inclusion, for one thing, from three models we selected 11 candidated gene and identified three biomarkers which could predict heterogenous colon cancer liver metastasis. For another, we deeply studied the function and molecular mechanism of P-cadherin and revealed that P-cadherin contributed to colon cancer liver metastasis by downregulating E-cadherin expression and affectingβ-catenin expression and location.Therefore, P-cadherin may be a valuable target for the treatment of colon carcinoma liver metastasis.
|
PDF Full Text Request