Differential mRNA expression of the human DNA methyltransferases (DNMTs) 1, 3a and 3b during the G(0)/G(1) to S phase transition in normal and tumor cells.

DNA methylation is essential for mammalian development, X-chromosome inactivation, and imprinting yet aberrant methylation patterns are one of the most common features of transformed cells. One of the proposed causes for these defects in the methylation machinery is overexpression of one or more of the three known catalytically active DNA methyltransferases (DNMTs) 1, 3a and 3b, yet there are clearly examples in which overexpression is minimal or non-existent but global methylation anomalies persist. An alternative mechanism which could give rise to global methylation errors is the improper expression of one or more of the DNMTs during the cell cycle. To begin to study the latter possibility we examined the expression of the mRNAs for DNMT1, 3a and 3b during the cell cycle of normal and transformed cells. We found that DNMT1 and 3b levels were significantly downregulated in G(0)/G(1)while DNMT3a mRNA levels were less sensitive to cell cycle alterations and were maintained at a slightly higher level in tumor lines compared to normal cell strains. Enzymatic activity assays revealed a similar decrease in the overall methylation capacity of the cells during G(0)/G(1)arrest and again revealed that a tumor cell line maintained a higher methylation capacity during arrest than a normal cell strain. These results reveal a new level of control exerted over the cellular DNA methylation machinery, the loss of which provides an alternative mechanism for the genesis of the aberrant methylation patterns observed in tumor cells.

Establishment of long-term in vitro cultures of human ovarian cystadenomas and LMP tumors and examination of their spectrum of expression of matrix-degrading proteinases.

OBJECTIVES: To obtain long-term cultures of ovarian cystadenomas and ovarian tumors of low malignant potential (LMP) displaying gene expression patterns similar to those found in vivo and test the hypothesis that such cultures would express different levels of matrix-degrading proteinases than cultured ovarian carcinomas. METHODS: Transfection with an adenoviral expression vector for simian virus 40 (SV40) large T antigen was used to establish long-term cultures of the above tumors. Levels of expression of various genes were evaluated using molecular biological and immunohistochemical approaches. Zymography and reverse zymography were used to examine the activity of various metalloproteinases and plasminogen activators (PA). Two-sided P values for differences in plasminogen activator expression between different cell types were evaluated by Fisher's exact test. RESULTS: Long-term cultures derived from cystadenomas and LMP tumors were obtained which formed colonies on semisolid supports, but were not tumorigenic in nude mice. The cultured cells expressed keratin, estrogen receptor, gonadotropin receptors, BRCA1, and originated from monoclonal populations. There was no apparent association between the malignant phenotype and the expression of either matrix metalloproteinases or tissue inhibitors of metalloproteinases. However, a correlation was seen between this phenotype and expression of urokinase (uPA) and tissue type (tPA) plasminogen activators (P = 0.08 and 0.02 respectively). CONCLUSIONS: The above cell strains provide a useful model for investigating various aspects of the biology of benign ovarian tumors, including their response to steroid and gonadotropin hormones, and the role of specific proteinases in the acquisition of invasive and metastatic abilities.

Potential role of the inactivated X chromosome in ovarian epithelial tumor development.

BACKGROUND: Ovarian epithelial tumors can be divided into subcategories often regarded as different stages of neoplastic transformation. Cystadenomas belong to the least aggressive subgroup and are noninvasive and nonmetastatic. Ovarian tumors of low malignant potential (LMP) are intermediate between cystadenomas and carcinomas and show markedly reduced invasive and metastatic abilities. Invasion and metastasis are the hallmarks of carcinomas, which constitute the most aggressive subgroup and can be further subdivided into different grades. PURPOSE: We performed comparative allelotype analyses of ovarian cystadenomas, LMP tumors, and carcinomas, reasoning that such analyses could provide clues about the molecular determinants of their phenotypic differences. Because we realized that allelic losses involving the X chromosome might be associated with LMP tumor development, we determined whether such losses were interstitial and whether they involved the active or the inactive X chromosome. METHODS: Frequencies of loss of heterozygosity (LOH) at specific loci in every chromosomal arm were determined in 16 ovarian cystadenomas, 23 ovarian LMP tumors, 15 low-grade ovarian carcinomas, and 35 high-grade ovarian carcinomas by use of either the polymerase chain reaction (PCR) or Southern blot analyses. We took advantage of the fact that DNA methylation is an important mechanism of X-chromosome inactivation to determine whether losses involving the X chromosome were in the active or the inactive copy. We analyzed the methylation status of retained alleles on the X chromosome by determining whether they could be amplified by PCR after digestion with the methylation-sensitive restriction endonuclease Hpa II. RESULTS: High-grade carcinomas contained frequent(>50%) LOH in four autosomal chromosome arms, i.e., 6q, 13q, 17p, and 17q. Except for 13q, these same chromosomal arms showed frequent LOH in low-grade carcinomas. LOH in autosomal chromosomes was comparatively rare in LMP tumors and was absent in cystadenomas. In contrast, half (eight of 16) of LMP tumors informative for a locus in the proximal portion of chromosome Xq showed LOH at that locus. These losses were the result of interstitial deletions in six of the eight cases and involved the inactive copy of the X chromosome exclusively. Similar losses in the X chromosome were not seen in either cystadenomas or low-grade carcinomas. CONCLUSIONS AND IMPLICATIONS: LOH at multiple loci is associated with the development of ovarian carcinomas but not with the development of cystadenomas and LMP tumors. However, the integrity of a locus in chromosome Xq that possibly escapes X-chromosome inactivation is important for the control of LMP tumor development. The fact that this locus does not appear to be involved in the genesis of low-grade carcinomas suggests that LMP tumors are not precursors of such carcinomas.

Genetic disparity between morphologically benign cysts contiguous to ovarian carcinomas and solitary cystadenomas.

BACKGROUND: Ovarian carcinomas occasionally contain large, histologically benign cysts contiguous to the clearly malignant areas (cystadenocarcinomas). The question of whether such cysts are remnants of pre-existing benign tumors (cystadenomas) or constitute integral components of the carcinomas is important in clarifying the role of cystadenomas in ovarian carcinogenesis. It is also important for our general understanding of tumor heterogeneity, a phenomenon thought to result from the gradual accumulation of genetic abnormalities in initially homogeneous tumors. This question is also pertinent to the clinical management of ovarian cystadenomas, which are frequent in women of childbearing age and are usually treated surgically based on the possibility that they may give rise to carcinomas. PURPOSE: Reasoning that molecular markers of ovarian malignancy would be confined to the histologically malignant portions of cystadenocarcinomas if the morphologically benign portions are in fact pre-existing typical cystadenomas, we sought to verify that mutations in the p53 tumor suppressor gene are markers of malignancy in ovarian tumors and to determine the distribution of such mutations in cystadenocarcinomas. METHODS: We used immunohistochemical and DNA-sequencing techniques to analyze 46 ovarian carcinomas, 21 ovarian tumors of low malignant potential, and 16 solitary cystadenomas for the presence of p53 mutations. We then used similar techniques to examine the distribution of such mutations in different portions of cystadenocarcinomas. The observed differences in mutation frequencies were analyzed by the two-tailed Fisher's exact test. RESULTS: Mutations in the p53 gene were present in 24 (52%) of the 46 carcinomas, but they were absent in the 21 tumors of low malignant potential (P < .0001) and the 16 solitary cystadenomas (P = .0002). Six of six cystadenocarcinomas with p53 mutations showed the presence of the same mutations in the adjacent, histologically benign cysts. The mutations were seen not only in cells immediately adjacent to the carcinomas, but also throughout the morphologically benign cysts. Twenty (83%) of the 24 cases showing mutation of one p53 allele also showed loss of genetic heterozygosity, suggesting that the other p53 allele was deleted. Such allelic loss, if present in morphologically malignant portions of cystadenocarcinomas, was also observed in the contiguous cysts. CONCLUSIONS: Ovarian carcinomas can be distinguished from ovarian cystadenomas and tumors of low malignant potential by p53 mutations. The fact that the mutations were present in histologically benign cysts contiguous to ovarian carcinomas suggests that such cysts are not typical cystadenomas and may carry a genetic predisposition to carcinogenesis that is not present in ordinary cystadenomas.

Loss of heterozygosity on chromosome 13 is common only in the biologically more aggressive subtypes of ovarian epithelial tumors and is associated with normal retinoblastoma gene expression.

We examined the frequencies of loss of heterozygosity on chromosome 13 in 77 primary ovarian epithelial tumors subdivided into cystadenomas, tumors of low malignant potential, low grade carcinomas, and high grade carcinomas. Such losses were found in approximately 50% of high grade carcinomas but were not detected in any of the other tumor subtypes (P < 0.0001), suggesting a strong association between these abnormalities and the high grade carcinoma phenotype. The tumors were also examined for abnormalities in expression of the retinoblastoma susceptibility gene (RB). This was assessed by immunohistochemical staining of archival tumor sections with a polyclonal antibody directed against both the phosphorylated and the underphosphorylated forms of the RB protein. Most of the tumors, including those with allelic deletions on chromosome 13, showed normal RB nuclear protein staining patterns. We conclude that loss of RB expression is not a frequent event in primary ovarian carcinomas and that this gene is probably not the target of the frequent allelic deletions on chromosome 13 found in high grade ovarian carcinomas.

3 distinct regions of chromosome-6 are targets of loss of heterozygosity in human ovarian carcinomas.

We performed mapping studies of allelic deletions on chromosome 6 in 52 ovarian carcinomas using 20 polymorphic markers assigned to various regions of both chromosomal arms. Loss of heterozygosity was seen in 28 (55%) tumors. The allelic losses, however, did not involve the whole chromosome in 18 of those tumors, allowing further localization of the targeted chromosomal region(s). Three different chromosomal regions appeared involved based on the distribution of partial deletions, 2 on the long arm and one on the short arm. One of the 2 regions on the long arm was confined to band 6q27 whereas the other was between the Arg1 and estrogen receptor loci. The targeted region on the short arm was bounded by the D6S89 and D6S248 loci. These results suggest the presence of at least 3 different tumor suppressor genes important for the control of ovarian tumorigenesis on chromosome 6.

Histologically benign or low-grade malignant tumors adjacent to high-grade ovarian carcinomas contain molecular characteristics of high-grade carcinomas.

It is presently not clear if ovarian carcinomas arise de novo or from benign precursors (cystadenomas) and if high-grade malignant tumors (carcinomas) develop from preexisting low-grade carcinomas. The presence of allelic losses on chromosome 11p15.5 distinguishes high-grade ovarian carcinomas from either low-grade carcinomas or cystadenomas. We therefore examined the distribution of such losses in different parts of heterogeneous tumors showing mixed histological grades or showing adjacent large histologically benign neoplasms. The results showed that all neoplastic areas, including those that were histologically benign or compatible with low-grade carcinomas, contained allelic losses at the above locus. This suggests that the morphologically less aggressive portions of these heterogeneous tumors were not typical cystadenomas or low-grade carcinomas and contained molecular abnormalities indicative of at least a predisposition to the high-grade carcinoma phenotype.

Mutations in an alternatively spliced exon of h-ras are not associated with loss of heterozygosity on chromosome 11p15.5 in ovarian tumorigenesis.

There is evidence that an alternatively spliced exon of the H-ras gene, called idx is associated with down-regulation of H-ras activity. We tested the hypothesis that mutations in this exon play an important role in the development of ovarian carcinomas because loss of heterozygosity at the H-ras locus is frequently observed in these tumors. The idx sequence of 26 different ovarian carcinomas was amplified by PCR and the products were analyzed for possible mutations by single-stranded conformation polymorphism and DNA sequencing. The results showed no idx mutations, even in tumors with a demonstrable loss of one H-ras allele.

Culture of human fetal ovarian epithelium in a chemically-defined, serum-free medium: a model for ovarian carcinogenesis.

We report on of the possibility to culture normal human fetal ovarian epithelium in vitro under chemically defined conditions. Cell growth was dependent upon the presence of epidermal growth factor, insulin, and hydrocortisone in the culture medium but was unaffected by the presence of estrogen, progesterone, follicle stimulating hormone, or Müllerian inhibiting Substance. Growth could be maintained for up to 6 passages in culture. The possibility of culturing these cells in serum-free medium may represent an attractive model to investigate the biology of normal ovarian epithelium as well as its role in human ovarian tumorigenesis.