Conservation of mononucleotide repeats within 3' and 5' untranslated regions and their instability in MSI-H colorectal cancer.

Messenger RNA contains untranslated 3' and 5' regions (3' and 5' UTRs) with sequence elements that are essential for the regulation of gene expression. A systematic search of GenBank revealed a large number of mononucleotide repeats within these UTRs. We selected 35 such mononucleotide repeats ranging in length from 15 bp to 32 bp and analysed their size in a series of 60 normal individuals. The conservation of repeats correlated inversely to their length, with longer repeats generally being more polymorphic than shorter repeats, irrespective of 3' or 5' location. Several long repeats were identified however to be monomorphic and we postulate that their conservation may be due to selective pressures relating to a possible functional role. We analysed 19 conserved UTR repeats in 117 colorectal cancers (CRC), 43 of which had defective mismatch repair characterized by widespread microsatellite instability (MSI-H). The UTR repeats were very often deleted in MSI-H tumors, with the length of deletion being proportional to the size of the repeat. Because of the high frequency of deletion observed in the conserved UTR repeats of MSI-H tumors, these could serve as a useful model for the study of possible changes in gene expression resulting from such mutations.

Identification of homozygous deletions at chromosome 16q23 in aflatoxin B1 exposed hepatocellular carcinoma.

Loss of heterozygosity (LOH) represents the most frequent genetic alteration observed in hepatocellular carcinoma (HCC). Chromosome 16q is of particular interest as it exhibits LOH in 29% of HCC tumors and is frequently lost in breast, prostate, ovarian and gastric carcinomas. We genotyped 157 HCC tumors for 17 microsatellite markers distributed on chromosome 16q and determined a common region of LOH localized between the markers D16S518 and D16S504. By refining the boundaries of two interstitial LOH and two homozygous deletions, the critical region was delimited to 180 kb between D16S3096 and D16S3029. This region is located in intron 8 of the WWOX/FOR gene, but a search for mutations in all coding exons of this gene in 27 HCC tumors and cell lines did not reveal any tumor somatic alterations. Furthermore, by RT-PCR, no abnormal transcripts of this WWOX/FOR gene was detected in nine HCC cell lines. Finally, analysis of the p53 gene mutations with the clinical parameters of all tumors revealed that the two homozygous deletions have occurred in tumors presenting a R249S mutation. Our data revealed a relationship between chromosome 16q homozygous deletions and R249S p53 mutations in tumors where the patient had been exposed to aflatoxin B1 (P=0.002). These results are consistent with a role of aflatoxin B1 in the instability of chromosome 16q at the fragile site FRA16D. However, the nature of the specific gene that is altered during hepatocarcinogenesis remains to be elucidated.

Evolution of instability at coding and non-coding repeat sequences in human MSI-H colorectal cancers.

A number of human genes containing coding mononucleotide repeat sequences are particularly prone to mutations in tumors with defects in mismatch repair (MMR) genes (MSI-H cancers). In a large series of MSI-H colorectal tumors, we looked for mutations in 25 coding repeats contained in eight genes already known to be mutated in these cancers or in 17 other genes with an expected role in carcinogenesis. Mutations were found in 19 of the 25 candidate genes. Using a maximum likelihood statistical method, they were separated into two different groups that differed significantly in their mutation frequencies, and were likely to represent mutations that do or do not provide selective pressures during MSI-H tumoral progression, respectively. Three new target genes were found (GRB-14, RHAMM, RAD50). Our results provide evidence that MSI-H tumoral progression involves the cumulative mutations of a large number of genes. For each MSI-H tumor we calculated indexes representing the number of mutations found in genes of these groups. We also evaluated a shortening index at both the Bat-25 and Bat-26 non-coding mononucleotide tracts that are known to be almost always unstable in MSI-H cancers. A significant correlation was observed between instability at both coding and non-coding repeats, suggesting that Bat-25 and Bat-26 could be used as simple phenotypical markers of the tumoral evolution. A preferential order of mutations was deduced. During this process, hMSH3 alterations, a target gene encoding for a MMR protein, was found to play an important role by increasing the instability phenomenon characterizing these cancers.

The human T-cell transcription factor-4 gene: structure, extensive characterization of alternative splicings, and mutational analysis in colorectal cancer cell lines.

The human T cell transcription factor-4 (hTCF-4) interacts functionally with beta-catenin in the Wnt signaling pathway, which regulates many developmental processes. Moreover, inappropriate reactivation of this pathway attributable to APC or beta-catenin mutations has been described in colorectal cancers. Because only the human TCF-4 cDNA sequence was known, we determined its genomic structure. A total of 17 exons, of which 5 were alternative, were identified. Moreover, four alternative splice sites were observed either experimentally or in silico by a BLAST approach in expressed sequence tag databases. The alternative use of three consecutive exons localized in the 3' part of the hTCF-4 gene changes the reading frames used in the last exon, leading to the synthesis of a number of hTCF-4 isoforms with short, medium, or long-size COOH-terminal ends. We next screened the entire hTCF-4 gene for mutations in a series of 24 colorectal cancer cell lines by denaturing gradient gel electrophoresis and/or direct sequencing. Besides an already described deletion of an A in an (A)9 coding repeat in four cases, we found DNA variants in eight cases for a total of 12 variants, of which 8 were coding. These include one frameshift mutation in the beta-catenin binding domain (exon 1), and one missense mutation in exon 4. In the remaining six cases, nonsense or frameshift mutations were localized in the 3' part of the gene. These latter alterations have as a common consequence to decrease the proportion of the long COOH-terminal hTCF-4 isoform, which contains two binding domains for c-terminal binding protein, a protein implicated in the repression of the TCF family transcriptional activity. Thus, loss of the TCF-4 capacity to interact with COOH-terminal binding protein could be an important event during colorectal carcinogenesis by modifying Wnt signaling.