ABSTRACT: Originally linked with development and differentiation, the role of microRNAs (miRNAs) in many important biological processes has emerged and dysregulation of miRNA expression has been connected with cancer pathogenesis. The association between abnormal expression levels of miRNAs and colon cancer has been mainly demonstrated in primary tumors; more recently, non-overlapping sets of oncomirs, tumour suppressor miRNAs and metastamirs have been associated with distinct stages of colo-rectal cancer (CRC) progression. In an attempt to identify changes in both miRNA and gene expression levels along the colon mucosa - primary tumor - liver metastasis- sequence and to classify miRNAs into distinct functional networks according to the expression of their anti-correlated and differentially expressed target genes, we analyzed a large set of sample-matched miRNA and mRNA expression profiles, including the complete non tumor tissue, primary carcinoma and liver metastasis sequence from 8 patients. Our study shows that the largest changes in miRNA and gene expression levels occur in normal mucosa to tumor transition and are almost stably maintained in the subsequent tumor to metastasis transition. Only a few miRNAs were differentially expressed between liver metastasis and primary carcinoma; however, miRNA expression profiles classified better primary tumors and metastases compared to mRNA profiles. By integration of miRNAs and target genes expression data, we were able to infer regulatory networks modulated by miRNAs during colorectal tumorigenesis and metastatic process, and associate them to the affected biological pathways. In particular, we identified a combination of interconnected miRNAs, up- or down-regulated during tumorigenesis, which are organized into distinct sub-networks, each of which includes several regulatory relationships with differentially expressed genes. Among them, the relationship between miR-182, up-regulated in colon cancer, and the anti-correlated ENTPD5 gene was identified for the first time and confirmed in an independent set of samples. We determined miRNA expression profile for 78 samples comprising 23 normal adjacent mucosa (N), 31 primitive colorectal cancer (T) and 24 liver metastases (M), obtained from 45 patients who underwent surgery at University of Padova (Surgery Branch, Department of Oncological and Surgical Sciences) between March 1994 and September 2008. This dataset included 24 samples belonging to 8 patients with matched samples (N, T and M) from the same patients. We also examined the expression profiles for 80 samples, comprising 23 N, 30 T, 27 M, and including 27 per-patient matched samples from 9 patients. Considering both miRNAs and genes expression experiments, we obtained a set of 77 samples (23N, 30T, 24M) with matched miRNAs and genes expression data from the same biological sample, to reconstruct post-transcriptional regulatory networks.
