p53 immunoreactivity and single-strand conformational polymorphism analysis often fail to predict p53 mutational status.

The intent of this study was to investigate the ability of p53 expression and single-strand conformational polymorphism analysis (SSCP) to predict p53 mutational status in archival, paraffin-embedded tissues of gastric cancer. We evaluated paraffin-embedded tissues from 78 patients with advanced gastric cancer. The mutational status of the p53 gene (exons 5-9) was examined by SSCP analysis and by direct sequencing. These results were compared with p53 expression as assessed by immunohistochemical analysis (IHC). We graded p53 expression on a scale from 0 to 8 on the basis of both the intensity and the number of cells staining. Overall, we detected p53 immunoreactivity in 75.6% of the gastric cases; 19 (32.2%) of these cases scored from 1 to 4, and 40 (67.8%) cases scored from 5 to 8. p53 gene mutations were detected in 18 cases (23.1%) by SSCP and in 28 cases (36%) by direct sequencing. Thus, SSCP failed to detect 38% of the mutations found by sequencing. The majority of missed mutations involved exons 7 and 8. The concordance between IHC and SSCP was 37%, and the concordance between IHC and direct sequencing was 50%. Forty-five percent of cases positive by IHC failed to show mutations in exons 5 through 9. Five percent of cases negative by IHC (4 cases) contained mutations. One had a 1-base pair insertion; one had a mutation that resulted in a stop codon; the third had a mutation in exon 8; and the fourth had a mutation in both exons 5 and 8. Our findings indicate that p53 immunoreactivity correlates with the presence or absence of gene mutations in 50% of advanced gastric cancers when exons 5 through 9 are examined and that IHC cannot be reproducibly used as a marker of mutation in the most commonly mutated exons of the p53 gene. Furthermore, the sensitivity of SSCP for detecting mutations is only 62%. Thus, SSCP analysis cannot be used reliably to screen for p53 mutations.

Definition of a DNA repair domain in the genomic region containing the human p53 gene.

The human p53 gene is repaired in UV (254 nm)-irradiated xeroderma pigmentosum group C (XP-C) cells as part of a large genomic region that is about twice the size of the gene. Surrounding genomic regions are not repaired. Through DNA cloning and measurements of DNA repair, we mapped the location of the repair domain, including the terminal regions, relative to the topological features of the gene. The domain includes only the DNA strand that is transcribed and extends in both 3' and 5' directions beyond the promoter and transcription termination sites. No transcriptional activity other than that associated with the p53 gene was detected. The results suggest that nontranscribed regions adjacent to the p53 transcribed regions are efficiently repaired in XP-C cells. This means that factors associated with transcription other than RNA polymerase II and the associated transcription repair coupling factor must also play a role in the selective repair process in XP-C cells. We also found that a DNA fragment that contains the p53 promoters is nearly twice as sensitive to cyclobutane pyrimidine dimer induction by UV irradiation than are the surrounding fragments, which have the expected sensitivity.

Identification of a large genomic region in UV-irradiated human cells which has fewer cyclobutane pyrimidine dimers than most genomic regions.

Size separation after UV-endonuclease digestion of DNA from UV-irradiated human cells using denaturing conditions fractionates the genome based on cyclobutane pyrimidine dimer content. We have examined the largest molecules available (50-80 kb; about 5% of the DNA) after fractionation and those of average size (5-15 kb) for content of some specific genes. We find that the largest molecules are not a representative sampling of the genome. Three contiguous genes located in a G+C-rich isochore (tyrosine hydroxylase, insulin, insulin-like growth factor II) have concentrations two to three times greater in the largest molecules. This shows that this genomic region has fewer pyrimidine dimers than most other genomic regions. In contrast, the beta-actin genomic region, which has a similar G+C content, has an equal concentration in both fractions as do the p53 and beta-globin genomic regions, which are A+T-rich. These data show that DNA damage in the form of cyclobutane pyrimidine dimers occurs with different probabilities in specific isochores. Part of the reason may be the relative G+C content, but other factors must play a significant role. We also report that the transcriptionally inactive insulin region is repaired at the genome-overall rate in normal cells and is not repaired in xeroderma pigmentosum complementation group C cells.