Genetic Pathways of Colorectal Carcinogenesis Rarely Involve the PTEN and LKB1 Genes Outside the Inherited Hamartoma Syndromes

Materials and Methods

Using standard methods, DNA was extracted from samples of fresh-frozen sporadic colon cancer and matched normal tissue or blood. These cases had no known family history suggestive of FAP, hereditary nonpolyposis colon cancer, or any of the hamartoma syndromes. Other standard clinicopathological data (age, grade, stage, and tumor site) were obtained from hospital records. Single-strand conformational polymorphism (SSCP) analysis was performed on the cancer samples. Published oligonucleotides and reaction conditions 4-6,15 were used to amplify specifically each exon of PTEN; for some longer exons, the products of the polymerase chain reaction were then subjected to restriction endonuclease digestion to render them optimally short for mutation detection by SSCP (AluI for exons 1, 5, 6, 7, and 9 and MboI for exon 8). New oligonucleotides were designed for exon-by-exon amplification of LKB1 (Table 1) ▶ using a protocol of 94°C for 3 minutes (one time), 35 cycles of 94°C for 1 minute/annealing temperature °C for 1 minute/72°C for 1 minute, and 72°C for 5 minutes (one time). Polymerase chain reaction products were heated to 90°C for 5 minutes and subjected to electrophoresis on a 10% acrylamide gel (30:0.8 acrylamide:bisacrylamide, 10% glycerol) under nondenaturing conditions at 20 mA for about 16 hours. DNA was detected by silver staining of gels using standard methods. For all tumors with possible mutations according to SSCP analysis, that exon was reamplified in duplicate from genomic DNA in the polymerase chain reaction, and these purified polymerase chain reaction products were sequenced in forward and reverse orientations using the Applied Biosystems, Inc. (Foster City, CA) Ready Reaction Dye Terminator Cycle Sequencing kit and the 377 Prism sequencer. All sequencing reactions were performed alongside samples with wild-type genotypes and with known mutations.

Results

For LKB1, no variant band was detected in any tumor sample on SSCP analysis. Control samples from three PJS patients with known mutations showed bandshifts. For PTEN, however, a small number of bandshifts was observed in the tumor samples using SSCP analysis. Sequencing confirmed a mutation in one tumor (1.4%), resulting from a complex change at the exon 2/intron 2 boundary. This mutation altered the “wild-type” sequence GTA AGG TAAGAAT to GTA AGA GTAATGC (where exonic sequences are in regular type and intronic sequences are in italics). This results in 1) a silent AGG→AGA Arg→Arg change in codon 53, 2) substitution of the more typical donor splice site consensus sequence GTAA for the atypical wild-type sequence TAAG, and 3) a 2-bp deletion in intron 2. It is quite possible that this change affects mRNA splicing, although no source of mRNA was available to prove this contention. Exons 2 and 3 do not constitute a mode 3 number of nucleotides, and aberrant splicing would therefore be expected to lead to a truncated protein. The exon 2/intron 2 change was not present in the germline, and the patient, a 75-year-old male with Dukes’ C colorectal carcinoma, had no features of CS, Bannayan-Zonnona syndrome, or juvenile polyposis syndrome. The mutation at codon 53 of PTEN has not been reported previously in the germline or soma. Previous studies demonstrated that this tumor showed allele loss close to PTEN 20 and did not show a truncating mutation in exon 15 or allele loss at APC. 21

Discussion

The initiating events in the pathogenesis of cancers in the hamartoma syndromes are almost certainly germline and somatic mutations at LKB1, PTEN, or related, uncharacterized genes. We have found that PTEN and LKB1 mutations in sporadic colon cancers occur at a low frequency or are absent altogether. Thus, the colon cancers in PJS and CS (and possibly some cases of juvenile polyposis syndrome) follow genetic pathways that are distinct from the majority of colorectal tumors (at least regarding their initiating events). Other workers have found APC mutations in juvenile polyps with dysplasia, 22 showing that there is partial overlap between the genetic pathways of tumorigenesis in hamartoma syndromes, FAP, and sporadic colon cancers.

Given the reported moderate sensitivity of SSCP of about 80%, 23-25 especially for detecting point mutations, we cannot exclude the possibility that mutations at PTEN or LKB1 occur in a somewhat higher proportion of colon cancers than reported here. It is noteworthy, however, that the spectra of germline and somatic mutations in PTEN and of germline mutations in PJS include small deletions that would be easily detected using SSCP analysis 4,5,8,12,19,26 ; in addition, SSCP analysis in our study detected positive control samples in both genes that resulted from point mutations. The previously unreported PTEN mutation that we detected was accompanied by allele loss and, although its effect at the protein level cannot be proven, we suspect that this mutation was selected for a role in tumorigenesis. Colon cancer may thus resemble carcinoma of the breast, in which somatic PTEN mutations occur in a small but important subset of tumors.

Further possibilities for the involvement of PTEN or LKB1 that we have not excluded are gene silencing by promoter methylation or hemi- or homozygous deletion of either locus (whether the entire gene or whole exons). The latter possibility would be consistent with the observed allele loss close to PTEN and LKB1 in colon cancers, and homozygous deletions have been observed at PTEN in a variety of tumors. 5 It remains entirely possible, however, that the allele loss close to PTEN and/or LKB1 in colon cancers _targets different loci in both cases.

There is, in general, a far from perfect association between the spectrum of tumors in Mendelian cancer syndromes and the range of sporadic tumors in which the same gene is mutated. The familial breast/ovarian cancer genes, BRCA1 and BRCA2, for example, cause cancer when mutant in the germline, but are hardly ever mutated in sporadic cancers. There is, however, evidence that BRCA1 is inactivated by promoter methylation in some sporadic breast cancers, 27 thus suggesting that defective BRCA1 can provide a selective advantage to breast tumors whether derived from the germline or soma, albeit through different mechanisms. There will be great potential interest in studying the expression of PTEN and LKB1 mRNA in colorectal tumors and in screening for mutations of LKB1 in sporadic tumors of other sites.