Genes involved in senescence and immortalization.

Senescence is now understood to be the final phenotypic state adopted by a cell in response to several distinct cell physiological processes, including proliferation, oncogene activation and oxygen free radical toxicity. The role of telomere maintenance in immortalization and the roles of p16(INK4A), p19(ARF), p53 and other genes in senescence are being further elucidated. Significant progress continues to be made in our understanding of cellular senescence and immortalization.

The INK4A/ARF locus and its two gene products.

The INK4A/ARF locus on chromosome 9 is one of the sites mutated most frequently in human cancer. Two genes comprising overlapping reading frames encoding p16(INK4a) and p19(ARF) have been discovered at this locus and, remarkably, both play an important role in regulating cell growth, survival and senescence.

Efficiency alleles of the Pctr1 modifier locus for plasmacytoma susceptibility.

The susceptibility of BALB/c mice to pristane-induced plasmacytomas is a complex genetic trait involving multiple loci, while DBA/2 and C57BL/6 strains are genetically resistant to the plasmacytomagenic effects of pristane. In this model system for human B-cell neoplasia, one of the BALB/c susceptibility and modifier loci, Pctr1, was mapped to a 5.7-centimorgan (cM) chromosomal region that included Cdkn2a, which encodes p16(INK4a) and p19(ARF), and the coding sequences for the BALB/c p16(INK4a) and p19(ARF) alleles were found to be polymorphic with respect to their resistant Pctr1 counterparts in DBA/2 and C57BL/6 mice (45). In the present study, alleles of Pctr1, Cdkn2a, and D4Mit15 from a resistant strain (BALB/cDAG) carrying DBA/2 chromatin were introgressively backcrossed to the susceptible BALB/c strain. The resultant C.DAG-Pctr1 Cdkn2a D4Mit15 congenic was more resistant to plasmacytomagenesis than BALB/c, thus narrowing Pctr1 to a 1.5-cM interval. Concomitantly, resistant C57BL/6 mice, from which both gene products of the Cdkn2a gene have been eliminated, developed pristane-induced plasma cell tumors over a shorter latency period than the traditionally susceptible BALB/cAn strain. Biological assays of the p16(INK4a) and p19(ARF) alleles from BALB/c and DBA/2 indicated that the BALB/c p16(INK4a) allele was less active than its DBA/2 counterpart in inducing growth arrest of mouse plasmacytoma cell lines and preventing ras-induced transformation of NIH 3T3 cells, while the two p19(ARF) alleles displayed similar potencies in both assays. We propose that the BALB/c susceptibility/modifier locus, Pctr1, is an "efficiency" allele of the p16(INK4a) gene.

Characterization of Sp1, AP-1, CBF and KRC binding sites and minisatellite DNA as functional elements of the metastasis-associated mts1/S100A4 gene intronic enhancer.

The mts1/S100A4 gene encodes a small acidic calcium-binding protein that is expressed in a cell-specific manner in development, tumorigenesis and certain tissues of adult mice. A composite enhancer that is active in murine mammary adenocarcinoma cells was previously identified in the first intron of the mts1/S100A4 gene. Here we present a detailed analysis of the structure and function of this enhancer in the Mts1/S100A4-expressing CSML100 and non-expressing CSML0 mouse adenocarcinoma cell lines. In CSML100 cells the enhancer activity is composed of at least six cis-elements interacting with Sp1 and AP-1 family members and CBF/AML/PEBP2 and KRC transcription factors. In addition, a minisatellite-like DNA sequence significantly contributes to the enhancer activity via interaction with abundant proteins, which likely have been described previously under the name minisatellite-binding proteins. Extensive mutational analysis of the mts1/S100A4 enhancer revealed a cooperative function of KRC and the factors binding minisatellite DNA. This is the first example of an enhancer where two nuclear factors earlier implicated in different recombination processes cooperate to activate transcription. In Mts1/S100A4-negative CSML0 cells the strength of the enhancer was 7- to 12.5-fold lower compared to that in CSML100 cells, when referred to the activities of three viral promoters. In CSML0 cells the enhancer could be activated by exogenous AP-1 and CBF transcription factors.

Genotype-phenotype relationships in U.S. melanoma-prone families with CDKN2A and CDK4 mutations.

BACKGROUND: Two genes have been implicated in the development of cutaneous malignant melanoma (CMM). CDK4 (the gene encoding cyclin-dependent kinase 4, an oncogene) has exhibited germline mutations found in only three melanoma-prone families to date. CDKN2A is a tumor suppressor gene that encodes p16 (which inhibits activity of the cyclin D1-CDK4 complex) with germline mutations detected in 10%-25% of melanoma-prone families, some of whom are also prone to pancreatic cancer. METHODS: We compared 104 CMM patients from 17 CDKN2A families and 12 CMM case subjects from two CDK4 families. We used nonparametric statistics to test for differences in median age at first CMM diagnosis, numbers of CMMs, and numbers of nevi. The three recurrent mutations were haplotyped. All P values were two-sided. RESULTS: The median age at CMM diagnosis (P =.70) and the median numbers of CMMs (P =.73) did not differ between CMM case subjects from CDKN2A versus CDK4 families. Assessment of CMM case subjects from CDKN2A families with and without pancreatic cancer revealed no statistically significant differences in median age at diagnosis (P =.80) or in tumor number (P =.24). There was, however, a statistically significant difference in age-adjusted median numbers of nevi (P =.004), and CMM case subjects from CDKN2A families without pancreatic cancer had greater numbers of nevi. Recurrent CDKN2A mutations were a change from valine to aspartic acid at codon 126 (n = 3) and from glycine to tryptophan at codon 101 (n = 3). Six CDKN2A families had pancreatic cancer. Both CDK4 families carried a mutation resulting in an arginine-to-cysteine substitution at codon 24. Analyses of recurrent CDKN2A and CDK4 mutations suggested common haplotypes. CONCLUSIONS: The recurrent CDKN2A mutations were observed in families with and without pancreatic cancer, which suggests that other factors may be involved in the development of pancreatic cancer. Despite hypothetical differences in the mechanisms of action between CDKN2A and CDK4, clinical factors were indistinguishable between CMM case subjects from CDKN2A versus CDK4 families.

CDKN2A/p16 is inactivated in most melanoma cell lines.

The CDKN2A gene maps to chromosome 9p21-22 and is responsible for melanoma susceptibility in some families. Its product, p16, binds specifically to CDK4 and CDK6 in vitro and in vivo, inhibiting their kinase activity. CDKN2A is homozygously deleted or mutated in a large proportion of tumor cell lines and some primary tumors, including melanomas. The aim of this study was to investigate the involvement of CDKN2A and elucidate the mechanisms of p16 inactivation in a panel of 60 cell lines derived from sporadic melanomas. Twenty-six (43%) of the melanoma lines were homozygously deleted for CDKN2A, and an additional 15 (25%) lines carried missense, nonsense, or frameshift mutations. All but one of the latter group were shown by microsatellite analysis to be hemizygous for the region of 9p surrounding CDKN2A. p16 was detected by Western blotting in only five of the cell lines carrying mutations. Immunoprecipitation of p16 in these lines, followed by Western blotting to detect the coprecipitation of CDK4 and CDK6, revealed that p16 was functionally compromised in all cell lines but the one that carried a heterozygous CDKN2A mutation. In the remaining 19 lines that carried wild-type CDKN2A alleles, Western blot analysis and immunoprecipitation indicated that 11 cell lines expressed a wild-type protein. Northern blotting was performed on the remaining eight cell lines and revealed that one cell line carried an aberrantly sized RNA transcript, and two other cell lines failed to express RNA. The promoter was found to be methylated in five cell lines that expressed CDKN2A transcript but not p16. Presumably, the message seen by Northern blotting in these cell lines is the result of cross-hybridization of the total cDNA probe with the exon 1beta transcript. Microsatellite analysis revealed that the majority of these cell lines were hemi/homozygous for the region surrounding CDKN2A, indicating that the wild-type allele had been lost. In the 11 cell lines that expressed functional p16, microsatellite analysis revealed loss of heterozygosity at the markers immediately surrounding CDKN2A in five cases, and the previously characterized R24C mutation of CDK4 was identified in one of the remaining 6 lines. These data indicate that 55 of 60 (92%) melanoma cell lines demonstrated some aberration of CDKN2A or CDK4, thus suggesting that this pathway is a primary genetic target in melanoma development.

CDKN2A germ-line mutations in individuals with multiple cutaneous melanomas.

Germ-line CDKN2A mutations are present in some kindreds with hereditary cutaneous melanoma, and in Sweden a founder mutation with an extra arginine in codon 113 (113insR) has been identified. We screened 80 individuals with at least two primary cutaneous melanomas, who were identified mainly by a search of a regional cancer registry, for germ-line CDKN2A mutations. In nine patients, CDKN2A alterations that may contribute to melanoma predisposition were detected. In six individuals with a family history of melanoma, the 113insR founder mutation was present. One patient, who also had a family history of melanoma, had a 24-bp deletion that included codons 62-69. An in vitro binding assay established that the resulting mutant p16 protein was unable to bind cyclin-dependent kinase 4 and cyclin-dependent kinase 6. Two patients without a family history of melanoma had CDKN2A alterations: (a) one had a mutation in the 5' noncoding sequence (-14C/T); and (b) the other had an insertion of an extra T in codon 28, which results in a stop signal in codon 43. The median age at diagnosis of the first melanoma was significantly lower, the number of primary melanomas was significantly higher, and the presence of a family history of melanoma was significantly more common in patients with CDKN2A mutations than in those without germ-line mutations. The proportion of CDKN2A mutation carriers was significantly higher among patients treated for three or more primary melanomas compared with those with two tumors only. We conclude that mutation screening of individuals with multiple primary melanomas is a useful strategy to identify new melanoma kindreds with CDKN2A germ-line mutations.

Increased loss of chromosome 9p21 but not p16 inactivation in primary non-small cell lung cancer from smokers.

Epidemiological studies have demonstrated a causal association between tobacco use and carcinoma of the lung, and some genetic targets of the carcinogens in cigarette smoke have been defined recently. We further examined the effect of cigarette smoking on the frequency of allelic losses on chromosome 9p21 and the incidence of p16 inactivation. Chromosomal loss at 9p21-24 was determined by microsatellite analysis using 14 markers in 47 patients with non-small cell lung cancer. In addition, p16 gene inactivation was determined by DNA sequence analysis, methylation-specific PCR, and immunohistochemistry. Tumors from a group of nonsmokers (n = 14) were compared with tumors from a group of smokers (n = 33) matched for cell type, tumor stage, and gender. Allelic loss encompassing the p16 locus was present significantly (P = 0.01) more often in smokers (23 of 33 smokers, 70%) than in nonsmokers (4 of 14 nonsmokers, 28%). No significant differences in the frequency of p16 inactivation were observed between smokers and nonsmokers (45% versus 36%). However, homozygous deletion of the p16 gene locus and point mutation of p16 gene were only observed in tumors from smokers, whereas the p16 gene was inactivated in tumors from nonsmokers only through promoter hypermethylation. Thus, inactivation of the p16 gene is a common event in all non-small cell lung cancer, but the mechanism of gene alteration differs between smokers and nonsmokers. The significant link between tobacco and loss of the p16 locus identifies additional genetic targets of smoking in the pathogenesis of lung cancer.

The transcriptional repressor of p16/Ink4a, Id1, is up-regulated in early melanomas.

The helix-loop-helix transcription factor Id1 coordinates cell growth and differentiation pathways within mammalian cells and has been implicated in regulating G(1)-S phase cell cycle transitions. Recently Id1 has been shown to repress Ets- and E-protein-mediated transactivation of p16/Ink4a. Because the p16/Ink4a protein has been demonstrated to be inactivated in subsets of familial and sporadic melanomas, we sought to determine whether Id1 regulation of p16/Ink4a expression might be involved in the development of this human tumor. Here we evaluate 21 melanocytic lesions at various stages of malignant progression from common melanocytic nevi to metastatic melanomas and examine these lesions for Id1 and p16/Ink4a expression. We demonstrate that Id1 expression correlates with loss of p16/Ink4a expression in melanoma in situ; however, more advanced stages of melanoma do not express Id1 except within perivascular regions, despite overall decreased p16/Ink4a expression in these lesions. Microdissected lesions were evaluated for p16/Ink4a sequence, and invasive melanomas that did not express Id1 were found to have sustained inactivating p16/ink4a mutations. These data suggest a role for Id1 in regulating p16/Ink4a expression in early melanomas and demonstrate that later genetic changes may provide for irreversible loss of p16 expression in advanced stages of this tumor.

Accelerated age-related CpG island methylation in ulcerative colitis.

CpG island hypermethylation is a mechanism of gene silencing that can be usurped by neoplastic cells to inactivate undesirable genes. In the colon, hypermethylation often starts in normal mucosa as a function of age and is markedly increased in cancer. To test the hypothesis that subjects at increased risk of colon cancer have higher levels of methylation in their nonneoplastic mucosa, we studied methylation patterns of five genes in the normal and dysplastic mucosa of patients with ulcerative colitis (UC), a condition associated with a marked increased risk of colon cancer. One gene (Mlh1) was unmethylated in all tissues examined. All four remaining genes had low but detectable levels of methylation in the epithelium of UC patients without evidence of dysplasia, and this methylation was not different from non-UC controls. By contrast, all four genes were highly methylated in dysplastic epithelium from high-grade dysplasia (HGD)/cancer patients with UC; methylation in HGD versus controls averaged 40.0% versus 7.4% (P = 0.00003) for ER, 44.0% versus 3.0% (P < 0.00003) for MYOD, 9.4% versus 2.4% (P = 0.03) for p16 exon 1, and 57.5% versus 30.6% (P = 0.01) for CSPG2. Importantly, three of the four genes were also highly methylated in the normal appearing (nondysplastic) epithelium from these same HGD/cancer patients, indicating that methylation precedes dysplasia and is widespread in these patients. Compared with controls, methylation averaged 20.1% versus 7.2% (P = 0.07) for ER, 18.4% versus 3.0% (P < 0.008) for MYOD, and 7.9% versus 2.4% (P = 0.007) for p16 exon 1. These results are consistent with the hypothesis that age-related methylation marks (and may lead to) the field defect that reflects acquired predisposition to colorectal neoplasia. Furthermore, the data suggest that chronic inflammation is associated with high levels of methylation, perhaps as a result of increased cell turnover, and that UC can be viewed as resulting in premature aging of colorectal epithelial cells.

Genetic and epigenetic alterations in normal bladder epithelium in patients with metachronous bladder cancer.

Mechanisms for multifocal bladder carcinogenesis remain unclear. To see whether normal mucosa had already acquired genetic or epigenetic changes, we examined loss of heterozygosity (LOH) at 10 microsatellite loci and methylation of the p16(INK4) CpG island in multiple tumors and pathologically normal mucosa in six patients with bladder cancer. Either LOH or methylation was detected in 77% of samples of normal epithelium, and LOH detected in samples of normal epithelium was also observed in most tumor samples. This result indicated that a population of cells in morphologically normal epithelium possessed genetic or epigenetic aberrations in common with bladder cancer, which might provide a ground for multiple tumorigenesis.

Genetic alterations in gastrinomas and nonfunctioning pancreatic neuroendocrine tumors: an analysis of p16/MTS1 tumor suppressor gene inactivation.

Neoplasms of the endocrine pancreas are extremely rare, and molecular mechanisms influencing their development are poorly understood. Nevertheless, gastrinomas have become a paradigm for the study of hormonally active tumors. In the present study, 12 gastrinoma and nonfunctioning pancreatic neuroendocrine tumor specimens were evaluated for genetic alterations of the p16/MTS1 tumor suppressor gene. DNA extracted from microdissected portions of paraffin-embedded tumor sections were examined for mutations and homozygous deletions using "Cold" single-strand conformation polymorphism and semiquantitative PCR-based analyses, respectively. Samples were also analyzed for the presence of 5' CpG island hypermethylation using methylation-specific PCR. The p16/MTS1 gene was found to be homozygously deleted in 41.7% of tumors and methylated in 58.3%, but no mutations were identified by single-strand conformation polymorphism analyses. Overall, 91.7% of the specimens demonstrated inactivating alterations in p16/MTS1. These data suggest that transcriptional silencing of p16/MTS1 is a frequent event in these rare and poorly understood tumors.

Promoter hypermethylation patterns of p16, O6-methylguanine-DNA-methyltransferase, and death-associated protein kinase in tumors and saliva of head and neck cancer patients.

Aberrant promoter hypermethylation is common in head and neck cancer and may be useful as a marker for cancer cells. We examined whether cells with tumor-specific aberrant DNA-methylation might be found in the saliva of affected patients. We tested 30 patients with primary head and neck tumors using methylation-specific PCR searching for promoter hypermethylation of the tumor suppressor gene p16 (CDKN2A), the DNA repair gene O6-methylguanine-DNA-methyltransferase (MGMT) and the putative metastasis suppressor gene death-associated protein kinase (DAP-K). Aberrant methylation of at least one of these genes was detected in 17 (56%) of 30 head and neck primary tumors; 14 (47%) of 30 at p16, 10 (33%) of 30 at Dap-K and 7 (23%) of 30 at MGMT. In 11 (65%) of 17 methylated primary tumors abnormal methylated DNA was detected in the matched saliva samples. Abnormal promoter methylation in saliva DNA was found in all tumor stages and more frequently in tumors located in the oral cavity. Moreover, none of the saliva from patients with methylation-negative tumors displayed methylation of any marker. Of 30 saliva samples from healthy control subjects (15 smokers and 15 nonsmokers), only one sample from a smoking patient was positive for DNA methylation at two target genes. Detection of aberrant promoter hypermethylation patterns of cancer-related genes in saliva of head and cancer patients is feasible and may be potentially useful for detecting and monitoring disease recurrence. Long-term longitudinal studies are needed to evaluate this approach for early detection of head and neck cancer in at-risk populations.

Tumor suppressor genes in the 9p21 gene cluster are selective targets of inactivation in neuroendocrine gastroenteropancreatic tumors.

Functional inactivation of the Rb and p53 pathways appears to be a rite of passage for all cancerous cells. However, p53 and Rb alterations are rare events in neuroendocrine gastroenteropancreatic (GEP) tumors. The CDKN2 locus on chromosome 9p21 sits at the nexus of both pathways harboring tumor suppressor genes, which restrain cell growth by affecting the function of pRb and p53. Therefore, we analyzed the implication of their inactivation in 37 primary neuroendocrine GEP tumors and two cell culture models. RT-PCR analysis revealed loss of expression of at least one of the tumor suppressor genes CDKN2A/p16, CDKN2B/p15, and CDKN2D/p14 with distinct genetic profiles, most frequently in nonfunctional pancreatic tumors (57%) and small intestinal carcinoids (44%), and less commonly in insulinomas (30%) and gastrinomas (22%). DNA analysis and methylation-specific PCR attributed loss of expression to either homozygous deletion or 5'CpG island hypermethylation. 5-Aza-2-deoxycytidine treatment reversed CDKN2A/p16 and CDKN2B/p15 silencing with concurrent growth restraint. Thus, tumor suppressor genes localized in the 9p21 gene cluster are specific targets of inactivation in neuroendocrine GEP tumors, and demethylating agents might hold promise for selective therapy.

CpG island methylation in premalignant stages of gastric carcinoma.

There are limited reports on methylation analysis of the premalignant lesions of gastric carcinoma thus far. This is despite the fact that gastric carcinoma is one of the tumors with a high frequency of CpG island hypermethylation. To determine the frequency and timing of hypermethylation during multistep gastric carcinogenesis, non-neoplastic gastric mucosa (n = 118), adenomas (n = 61), and carcinomas (n = 64) were analyzed for their p16, human Mut L homologue 1 (hMLH1), death-associated protein (DAP)-kinase, thromobospondin-1 (THBS1), and tissue inhibitor of metalloproteinase 3 (TIMP-3) methylation status using methylation-specific PCR. Three different classes of methylation behaviors were found in the five tested genes. DAP-kinase was methylated at a similar frequency in all four stages, whereas hMLH1 and p16 were methylated in cancer samples (20.3% and 42.2%, respectively) more frequently than in intestinal metaplasia (6.3% and 2.1%, respectively) or adenomas (9.8% and 11.5%, respectively). However, hMLH1 and p16 were not methylated in chronic gastritis. THBS-1 and TIMP-3 were methylated in all stages but showed a marked increase in hypermethylation frequency from chronic gastritis (10.1% and 14.5%, respectively) to intestinal metaplasia (34.7% and 36.7%, respectively; P < 0.05) and from adenomas (28.3% and 26.7%, respectively) to carcinomas (48.4% and 57.4%, respectively: P < 0.05). The hMLH1, THBS1, and TIMP-3 hypermethylation frequencies were similar in both intestinal metaplasia and adenomas, but the p16 hypermethylation frequency tended to be higher in adenomas (11.5%) than in intestinal metaplasia (2.1%; P = 0.073). The average number of methylated genes was 0.6, 1.1, 1.1, and 2.0 per five genes per sample in chronic gastritis, intestinal metaplasia, adenomas, and carcinomas, respectively. This shows a marked increase in methylated genes from non-metaplastic mucosa to intestinal metaplasia (P = 0.001) as well as from premalignant lesions to carcinomas (P = 0.002). These results suggest that CpG island hypermethylation occur early in multistep gastric carcinogenesis and tend to accumulate along the multistep carcinogenesis.

Allelic deletions of cell growth regulators during progression of bladder cancer.

Cell growth regulators include proteins of the p53 pathway encoded by the genes CDKN2A (p16, p14arf), MDM2, TP53, and CDKN1A (p21) as well as proteins encoded by genes like RB1, E2F, and MYCL. In the present study we investigated allelic deletions of all these genes in each recurrent bladder tumor from well-defined clinical material with more than 3 years of follow-up. We followed three groups (22 or 23 patients/group) of patients with: (a) recurrent noninvasive tumors (Ta); (b) primary muscle-invasive tumors (T2-T4); and (c) progressing tumors (Ta/T1 --> T2/T4). We found a significant difference in the numbers of gene loci hit by deletions muscle-invasive versus noninvasive tumors (P = 0.0000002), with the genes most often hit by deletions in muscle-invasive tumors being TP53, RB1, and MYCL. A number of novel findings were made. Losses of MYCL and RB1 alleles were more pronounced in patients having concomitant field disease because 11 of 14 informative cases showed losses compared with 3 of 8 cases without field disease. A more pronounced deletion of TP53 (P = 0.002) and RB1 (P = 0.02) was found in the progressing tumor group compared with the recurrent noninvasive group, and, finally, the combined loss of TP53 and RB1 was present only in the progressing tumor or muscle-invasive groups. Deletion of two or more loci in TP53, MYCL, RB1, and CDKN2A was found in 10 patients in the progressing tumor group and in only 1 patient in the recurrent noninvasive group (P = 0.004). The data demonstrate that a characteristic difference between recurrent noninvasive and recurrent progressing bladder tumors is loss of cell cycle-regulatory genes in the latter group.

Aberrant promoter methylation of multiple genes in non-small cell lung cancers.

Aberrant methylation of CpG islands acquired in tumor cells in promoter regions is one method for loss of gene function. We determined the frequency of aberrant promoter methylation (referred to as methylation) of the genes retinoic acid receptor beta-2 (RARbeta), tissue inhibitor of metalloproteinase 3 (TIMP-3), p16INK4a, O6-methylguanine-DNA-methyltransferase (MGMT), death-associated protein kinase (DAPK), E-cadherin (ECAD), p14ARF, and glutathione S-transferase P1 (GSTP1) in 107 resected primary non-small cell lung cancers (NSCLCs) and in 104 corresponding nonmalignant lung tissues by methylation-specific PCR. Methylation in the tumor samples was detected in 40% for RARbeta, 26% for TIMP-3, 25% for p16INK4a, 21% for MGMT, 19% for DAPK, 18% for ECAD, 8% for p14ARF, and 7% for GSTP1, whereas it was not seen in the vast majority of the corresponding nonmalignant tissues. Moreover, p16INK4a methylation was correlated with loss of p16INK4a expression by immunohistochemistry. A total of 82% of the NSCLCs had methylation of at least one of these genes; 37% of the NSCLCs had one gene methylated, 22% of the NSCLCs had two genes methylated, 13% of the NSCLCs had three genes methylated, 8% of the NSCLCs had four genes methylated, and 2% of the NSCLCs had five genes methylated. Methylation of these genes was correlated with some clinicopathological characteristics of the patients. In comparing the methylation patterns of tumors and nonmalignant lung tissues from the same patients, there were many discordancies where the genes methylated in nonmalignant tissues were not methylated in the corresponding tumors. This suggests that the methylation was occurring as a preneoplastic change. We conclude that these findings confirm in a large sample that methylation is a frequent event in NSCLC, can also occur in smoking-damaged nonmalignant lung tissues, and may be the most common mechanism to inactivate cancer-related genes in NSCLC.

Fields of aberrant CpG island hypermethylation in Barrett's esophagus and associated adenocarcinoma.

Esophageal adenocarcinoma (EAC) is thought to develop through a multistage process in which Barrett's metaplasia progresses through low- and high-grade dysplasia to invasive cancer. Transcriptional silencing of tumor suppressor genes by promoter CpG island hypermethylation has been observed in many types of human cancer. Analysis of CpG island hypermethylation in EAC has thus far been limited to the CDKN2A (p16) gene. In this study, we extend the methylation analysis of EAC to include three other genes, APC, CDH1 (E-cadherin), and ESR1 (ER, estrogen receptor alpha), in addition to CDKN2A. Molecular analysis can provide insight into the complex relationships between tissues with different histologies in Barrett's esophagus and associated adenocarcinoma. Therefore, we have mapped the spatial distribution of methylation patterns in six esophagectomy cases in detail. Hypermethylation of the four CpG islands was analyzed by the MethyLight technique in 107 biopsies derived from these six patients for a total of 428 methylation analyses. Our results show that normal esophageal squamous epithelium is unmethylated at all four CpG islands. CDH1 is unmethylated in most other tissue types as well. Hypermethylation of ESR1 is seen at high frequency in inflammatory reflux esophagitis and at all subsequent stages, whereas APC and CDKN2A hypermethylation is found in Barrett's metaplasia, dysplasia, and EAC. When it occurs, hypermethylation of APC, CDKN2A, and ESR1 is usually found in a large contiguous field, suggesting either a concerted methylation change associated with metaplasia or a clonal expansion of cells with abnormal hypermethylation.

Patterns of CDKN2A gene loss in sequential oral epithelial dysplasias and carcinomas.

The CDKN2A gene locus encodes two different proteins derived from alternative splicing. p16 (exons 1alpha, 2, and 3) acts as a G1 cell cycle regulator, and p14ARF (exons 1beta, 2, and 3) acts to modulate MDM2-mediated degradation of p53. Inactivation of p16 is a common finding in many cancers; however, there is little data on CDKN2A gene abnormalities in oral precancer. In this longitudinal study, we examined changes in the CDKN2A gene locus in sequential epithelial dysplasias and oral carcinomas from 11 patients. Genomic DNA was extracted from laser-microdissected lesional tissue, and exons 1alpha, 1beta, and 2 were analyzed by duplex PCR. Immunohistochemistry was done to identify p16 and p14ARF protein expression. Two adjacent polymorphic microsatellite markers were used for allelotyping. Homozygous deletion of exon 1alpha was identified in 2 of 17 (12%) precancerous lesions. Loss of either exon 1alpha, exon 2, or both was seen in seven of nine (78%) carcinomas. In five of these carcinomas, there was loss of only exon 1alpha. No case showed deletion of exon 1beta. In 5 of 11 patients, microsatellite markers showed differing patterns of allelic imbalance in the precancerous lesions and the subsequent carcinoma, suggesting a complex genetic pattern of progression from dysplasia to carcinoma. We conclude that during oral carcinogenesis homozygous deletion of exon 1alpha of the CDKN2A gene is common but that deletion of exon 2 and 1beta is less frequent. Moreover, our results suggest that the progression from oral precancer to cancer, in some cases, is more complex genetically than predicted by linear models of carcinogenesis.

Analysis of the CDKN2A, CDKN2B and CDK4 genes in 48 Australian melanoma kindreds.

Germline mutations within the cyclin-dependent kinase inhibitor 2A (CDKN2A) gene and one of its targets, the cyclin dependent kinase 4 (CDK4) gene, have been identified in a proportion of melanoma kindreds. In the case of CDK4, only one specific mutation, resulting in the substitution of a cysteine for an arginine at codon 24 (R24C), has been found to be associated with melanoma. We have previously reported the identification of germline CDKN2A mutations in 7/18 Australian melanoma kindreds and the absence of the R24C CDK4 mutation in 21 families lacking evidence of a CDKN2A mutation. The current study represents an expansion of these efforts and includes a total of 48 melanoma families from Australia. All of these families have now been screened for mutations within CDKN2A and CDK4, as well as for mutations within the CDKN2A homolog and 9p21 neighbor, the CDKN2B gene, and the alternative exon 1 (E1beta) of CDKN2A. Families lacking CDKN2A mutations, but positive for a polymorphism(s) within this gene, were further evaluated to determine if their disease was associated with transcriptional silencing of one CDKN2A allele. Overall, CDKN2A mutations were detected in 3/30 (10%) of the new kindreds. Two of these mutations have been observed previously: a 24 bp duplication at the 5' end of the gene and a G to C transversion in exon 2 resulting in an M531 substitution. A novel G to A transition in exon 2, resulting in a D108N substitution was also detected. Combined with our previous findings, we have now detected germline CDKN2A mutations in 10/48 (21%) of our melanoma kindreds. In none of the 'CDKN2A-negative' families was melanoma found to segregate with either an untranscribed CDKN2A allele, an R24C CDK4 mutation, a CDKN2B mutation, or an E1beta mutation. The last three observations suggest that these other cell cycle control genes (or alternative gene products) are either not involved at all, or to any great extent, in melanoma predisposition.