Induction of prolactin-deficient variants of GH3 rat pituitary tumor cells by ethyl methanesulfonate: reversion by 5-azacytidine, a DNA methylation inhibitor

Inherited epigenetic variation--revisiting soft inheritance

Phenotypic variation is traditionally parsed into components that are directed by genetic and environmental variation. The line between these two components is blurred by inherited epigenetic variation, which is potentially sensitive to environmental inputs. Chromatin and DNA methylation-based mechanisms mediate a semi-independent epigenetic inheritance system at the interface between genetic control and the environment. Should the existence of inherited epigenetic variation alter our thinking about evolutionary change?

Human pituitary tumor-transforming gene (PTTG1) motif suppresses prolactin expression

Pituitary tumor-transforming gene (PTTG) originally isolated from GH-secreting pituitary adenoma cells causes in vitro cell transformation, in vivo tumorigenesis, and induces basic fibroblast growth factor. These functions require an intact C-terminal proline-proline-serine-proline motif. PTTG1 is abundantly expressed in human pituitary tumors and plays a role in the early stages of experimental prolactinoma formation. We now determined direct effects of PTTG1 on hormonal phenotypes of functional pituitary tumor cells. Overexpression of PTTG1 C terminus (amino acids 147-202) containing intact proline-proline-serine-proline motifs in rat prolactin (PRL)- and GH-secreting GH3 cells markedly abrogates PRL mRNA expression by more than 90% (P < 0.001) and hormone levels (P < 0.001) and PRL promoter activity (P < 0.01) compared with control vector cells or to a PTTG1 C terminus mutant (P163A, S165Q, P166L, P170L, P172A, and P173L). Wild-type PTTG1 C-terminal transfectants formed smaller (P < 0.05) sc tumors in rats compared with control or mutated PTTG1 C-terminal transfectants. Estrogen (10 nm) treatment for 48 h partially restored PRL expression in stable wild-type PTTG1 C-terminal transfectants. These results indicate that targeting PTTG1-mediated signaling alters the hormonal phenotype in pituitary cells and disrupted PTTG1 action may be a potential subcellular therapeutic tool for repressing PRL hypersecretion.

Carcinogen-induced de novo methylation in c-myc exon I

During the response of N-methyl-N'-nitro-N-nitrosoguanidine (MNNG), methylation occurred at the Hpa II site of c-myc exon I, which is located downstream of the P1 initiation site, as evidenced by the assays of Hpa II-PCR. The Hpa II spite of the 5' flanking region did not undergo methylation. UV-irradiation also led to methylation in exon I. The extent of methylation increased depending on the dose of MNNG and UV. The results suggested that methylation takes place in transcriptionally active c-myc responsible for carcinogens and is caused by mechanisms different from that of alkylation in a specific CpG site. Possible contribution of methylation to less repair found in c-myc is discussed.

Transcriptional analysis of the mts1 gene with specific reference to 5' flanking sequences

The mts1 gene is specifically expressed in certain metastatic tumors but not in their nonmetastatic counterparts. It is also expressed in several normal cell and tissue types that exhibit the ability to be motile. The gene was cloned from both mouse and human sources and the 5' flanking regions were sequenced. The sequencing data revealed a 135-base-pair region of high homology between the mouse and human mts1 gene. This homology was observed in the vicinity of the TATA box. The 5' region of the mts1 gene was also observed to have a high degree of homology to some known promoter and enhancer sequences. To determine the role this region plays in regulating the transcription of mts1, promoter analysis was performed. Sixteen constructs were prepared in which the chloramphenicol acetyltransferase gene was fused to different regions of the mouse mts1 promoter. These constructs were analyzed in transient transfection assays in two related cell lines derived from mouse mammary adenosarcomas: CSML-0, a nonmetastatic cell line with low levels of mts1 expression, and CSML-100, a metastatic cell line with high levels of mts1 expression. Results of our transient transfection assays in conjunction with results obtained from in vitro and in vivo footprinting of the promoter region show no evidence of cis-acting control elements important for the transcriptional regulation of mts1 in these cell lines. A few nucleotides upstream of the TATA box are sufficient for maximal levels of mts1 transcription. Because no cis-acting control elements were found, restriction of mts1 transcription in CSML-0 cells must exist on some other level. mts1 was found to be hypermethylated in CSML-0 cells but not in CSML-100 cells. The possible role of methylation in progression of the nonmetastatic CSML-0 adenosarcoma cell line toward the metastatic CSML-100 adenosarcoma cell line is discussed.

Resistance to interferon-alpha in a mouse B-cell lymphoma involves DNA methylation

Resistance to interferon-alpha (IFN-alpha) in the 38C13 B-lymphoma cell line results in the loss of antiviral, antiproliferative, and immune regulatory functions of IFN-alpha. Mutagenesis with ethylmethylsulfonic acid (EMS), which can induce point mutations in DNA, increases the frequency of resistance to IFN-alpha 20 to 40-fold. In contrast, treatment with 5-azacytidine, which causes hypomethylation of DNA, reduces the frequency of resistance to 5-10% of control. Furthermore, 5-azacytidine treatment reverts IFN-alpha-resistant cells to the IFN-alpha-sensitive state. Resistance to IFN-alpha occurs spontaneously at a rate of approximately 3 x 10(-6) variants/cell.generation, and is stable for more than 30 passages without selection in IFN-alpha. There is no evidence that gene amplification contributes to the high rate of resistance to IFN-alpha in these cells. These results indicate that DNA mutation and methylation are important in the development of IFN-alpha resistance in these cells.

Prolactin-deficient GH3B3 cells are defective in the utilization of the endogenous prolactin promoter yet are fully competent to initiate transcription from a transfected prolactin promoter

Transcription of the prolactin (PRL) gene has been analyzed in wild-type D6, PRL-deficient B3, and revertant r16 GH3 cells. Levels of processed nuclear transcripts from the PRL gene were substantially reduced in the deficient line compared to wild-type cells and returned to greater than wild-type levels in the revertant line. Rare PRL transcripts in the deficient line contained the same 5' end found on transcripts in wild-type and revertant cells as judged by primer extension and S1 nuclease protection assays, implying that the cells are deficient in utilization of the normal wild-type promoter. Deficient cells also contained wild-type levels of the PRL- and growth hormone-specific transcription factor pit-1/GHF-1, and no difference was found in the ability of extracts from wild-type and deficient cells to retard various restriction fragments from both the proximal and the distal PRL promoter regions. The deficient and wild-type cells were equally competent in initiating transcription from a transfected rat PRL promoter containing both the distal and proximal promoter elements. These observations imply that PRL-deficient cells are not defective in a trans-activating factor functioning on these PRL promoter fragments (trans model). Rather, inefficient use of the PRL promoter in the variant cells may reflect an increased methylation state of the PRL gene itself (cis model).

DNA methylation patterns associated with asparagine synthetase expression in asparagine-overproducing and -auxotrophic cells

In Chinese hamster ovary cells, the gene for asparagine synthetase, which spans 20 kilobase pairs, was found to contain a cluster of potential sites for CpG methylation in a 1-kilobase-pair region surrounding the first exon. Fourteen of the sites that could be assayed for methylation by MspI-HpaII digestions were found in this region, with an additional nine MspI sites spread throughout the remainder of the gene. The methylation status of the gene was analyzed in a series of cell lines that differed in the amount of asparagine synthetase activity. The level of expression showed a direct correlation with the extent of methylation of a subset of the MspI sites found in the 5' region of the gene. The rest of the gene was completely methylated in most cell lines. Wild-type cells, which expressed a basal level of asparagine synthetase activity, were partially demethylated in the 5' region. In contrast, asparagine-requiring N3 cells, which lacked detectable mRNA for asparagine synthetase, were methylated throughout the entire gene. Spontaneous revertants of strain N3, selected for growth in asparagine-free medium, exhibited extensive hypomethylation of the asparagine synthetase gene. The methylation pattern of the gene in cell lines that overproduced the enzyme was also examined. Albizziin-resistant cell lines, which had amplified copies of the gene, were extensively demethylated in the 5' region. Overexpression of asparagine synthetase in beta-aspartyl hydroxamate-resistant lines without amplified copies of the gene was also correlated with DNA hypomethylation.

Suppression of prolactin gene expression in GH cells correlates with site-specific DNA methylation

Prolactin- (PRL) producing and nonproducing subclones of the GH line of (rat) pituitary tumor cells have been compared to elucidate the regulatory mechanisms of PRL gene expression. Particular emphasis was placed on delineating the molecular basis of the suppressed state of the PRL gene in the prolactin-nonproducing (PRL-) GH subclone (GH(1)2C1). We examined six methylatable cytosine residues (5, -CCGG- and 1, -GCGC-) within the 30-kb region of the PRL gene in these subclones. This analysis revealed that -CCGG-sequences of the transcribed region, and specifically, one in the fourth exon of the PRL gene, were heavily methylated in the PRL-, GH(1)2C1 cells. Furthermore, the inhibition of PRL gene expression in GH(1)2C1 was reversed by short-term treatment of the cells with a sublethal concentration of azacytidine (AzaC), an inhibitor of DNA methylation. The reversion of PRL gene expression by AzaC was correlated with the concurrent demethylation of the same -CCGG- sequences in the transcribed region of PRL gene. An inverse correlation between PRL gene expression and the level of methylation of the internal -C- residues in the specific -CCGG-sequence of the transcribed region of the PRL gene was demonstrated. The DNase I sensitivity of these regions of the PRL gene in PRL+, PRL-, and AzaC-treated cells was also consistent with an inverse relationship between methylation state, a higher order of structural modification, and gene expression.(ABSTRACT TRUNCATED AT 250 WORDS)

DNA methylation patterns associated with asparagine synthetase expression in asparagine-overproducing and -auxotrophic cells

In Chinese hamster ovary cells, the gene for asparagine synthetase, which spans 20 kilobase pairs, was found to contain a cluster of potential sites for CpG methylation in a 1-kilobase-pair region surrounding the first exon. Fourteen of the sites that could be assayed for methylation by MspI-HpaII digestions were found in this region, with an additional nine MspI sites spread throughout the remainder of the gene. The methylation status of the gene was analyzed in a series of cell lines that differed in the amount of asparagine synthetase activity. The level of expression showed a direct correlation with the extent of methylation of a subset of the MspI sites found in the 5' region of the gene. The rest of the gene was completely methylated in most cell lines. Wild-type cells, which expressed a basal level of asparagine synthetase activity, were partially demethylated in the 5' region. In contrast, asparagine-requiring N3 cells, which lacked detectable mRNA for asparagine synthetase, were methylated throughout the entire gene. Spontaneous revertants of strain N3, selected for growth in asparagine-free medium, exhibited extensive hypomethylation of the asparagine synthetase gene. The methylation pattern of the gene in cell lines that overproduced the enzyme was also examined. Albizziin-resistant cell lines, which had amplified copies of the gene, were extensively demethylated in the 5' region. Overexpression of asparagine synthetase in beta-aspartyl hydroxamate-resistant lines without amplified copies of the gene was also correlated with DNA hypomethylation.

DNA methylation and differentiation

The methylation of specific cytosine residues in DNA has been implicated in regulating gene expression and facilitating functional specialization of cellular phenotypes. Generally, the demethylation of certain CpG sites correlates with transcriptional activation of genes. 5-Azacytidine is an inhibitor of DNA methylation and has been widely used as a potent activator of suppressed genetic information. Treatment of cells with 5-azacytidine results in profound phenotypic alterations. The drug-induced hypomethylation of DNA apparently perturbs DNA-protein interactions that may consequently alter transcriptional activity and cell determination. The inhibitory effect of cytosine methylation may be exerted via altered DNA-protein interactions specifically or may be transduced by a change in the conformation of chromatin. Recent studies have demonstrated that cytosine methylation also plays a central role in parental imprinting, which in turn determines the differential expression of maternal and paternal genomes during embryogenesis. In other words, methylation is the mechanism whereby the embryo retains memory of the gametic origin of each component of genetic information. A memory of this type would probably persist during DNA replication and cell division as methylation patterns are stable and heritable.

Cytosine methylation does not affect binding of transcription factor Sp1

DNA methylation may be a component of a multilevel control mechanism that regulates eukaryotic gene expression. We used synthetic oligonucleotides to investigate the effect of cytosine methylation on the binding of the transcription factor Sp1 to its target sequence (a G + C-rich sequence known as a "GC box"). Concatemers of double-stranded 14-mers containing a GC box successfully competed with the human metallothionein IIA promoter for binding to Sp1 in DNase I protection experiments. The presence of 5-methylcytosine in the CpG sequence of the GC box did not influence Sp1 binding. The result was confirmed using double-stranded 20-mers containing 16 base pairs of complementary sequence. Electrophoretic gel retardation analysis of annealed 28-mers containing a GC box incubated with an Sp1-containing HeLa cell nuclear extract demonstrated the formation of DNA-protein complexes; formation of these complexes was not inhibited when an oligomer without a GC box was used as a competitor. Once again, the presence of a 5-methylcytosine residue in the GC box did not influence the binding of the protein to DNA. The results therefore preclude a direct effect of cytosine methylation on Sp1-DNA interactions.

Gene reactivation: a tool for the isolation of mammalian DNA methylation mutants

We report the isolation and characterization of a mammalian strain (tsm) that has a temperature-sensitive mutation in DNA methylation. The isolation procedure was based on the observation that treatment of a CHO TK- MT- cell line with demethylating agents introduces up to 46% demethylation, resulting in phenotypic reversion and transcriptional activation of the thymidine kinase (TK) and metallothionein (MT) genes at frequencies ranging from 1% to 59%. Seven thousand individual colonies from an EMS-mutagenized CHO TK- MT- population were screened for spontaneous reversion to TK+ phenotype after treatment at 39 degrees C. Successful isolates were subsequently examined for MT+ reversion. A single clone (tsm) was obtained that showed temperature-dependent reactivation of both TK and MT genes at frequencies of 7.2 X 10(-4) and 6 X 10(-4), respectively. The tsm cells were viable at 39 degrees C and showed no increased mutation frequency. Reactivation correlated with transcriptional activation of the respective genes, whereas backreversion to the TK- phenotype was associated with transcriptional inactivation. TK- backrevertants were reactivable again with demethylating agents. Although demethylation in tsm cells was not detectable by HPLC, Southern blot analysis revealed that reactivants, irrespective of their mode of generation, showed specific demethylation of both TK and MT genes. Also, after about 150 cell generations after treatment, reactivants from both temperature-induced tsm and cells exposed to demethylating agents gained 60% and 23%, respectively, in 5-methylcytosine (5mC). It is proposed that the phenotype of tsm cells is due to a mutation involved in the regulation of DNA methylation. The further characterization of this and other mammalian mutants should help to clarify the physiological role of DNA methylation, as well as its regulation.

Alteration of the tumorigenic and metastatic properties of neoplastic cells is associated with the process of calcium phosphate-mediated DNA transfection

During the course of studies designed to assess the effect of human Ha-ras gene expression on the malignant behavior of transfected mouse tumor cells we noted that the process of Ca3(PO4)2-mediated DNA transfection was itself associated with profound alterations in tumorigenic or metastatic behavior. The cell line used as a recipient for these studies was a tumorigenic nonmetastatic CBA/J mouse mammary adenocarcinoma line called SP1. When cotransfected with plasmids containing the neo gene (pSV2neo) and the activated Ha-ras gene (pT24-c3), cells from the pooled (5-10 colonies) G418-resistant colonies gave rise to spontaneous lung metastases in 85% of mice after subcutaneous inoculation. However, we noted that 17% of control mice inoculated with G418-resistant pSV2neo-transfected SP1 cells also had lung metastases and that this number approached 100% as the inoculum comprised a greater pool size (50-100 colonies). When cell lines established from isolated pSV2neo-transfected colonies were examined, 3/16 were found to be metastatic. We also found that 3/16 clones grew slowly, or not at all, in CBA/J mice, whereas they grew readily in athymic (nude) mice. The increase in immunogenicity of two out of three of these latter clones was accompanied by expression of the class I H-2Dk major histocompatibility complex antigen that was not detectable in the parental SP1 cells. At least some of these results would appear to be due to exposure to Ca3(PO4)2 alone, as we found that it resulted in 5/20 (25%) clones manifesting metastatic properties. Our results suggest that heritable changes in malignant behavior of transfected tumor cells can be observed at high frequency subsequent to the process of Ca3(PO4)2-mediated DNA transfection, and these changes may be brought about in part by inherited disturbances in expression of recipient cell genes.

Activation of a nonexpressed hypoxanthine phosphoribosyltransferase allele in mutant H23 HeLa cells by agents that inhibit DNA methylation

HeLA H23 cells are a mutant female human tumor cell line harboring defective hypoxanthine phosphoribosyltransferase (HPRT; IMP-pyrophosphate phosphoribosyltransferase, EC 2.4.2.8) as a result of a mutation that alters the isoelectric point of the enzyme (G. Milman, E. Lee, G. S. Changas, J. R. McLaughlin, and J. George, Jr., Proc. Natl. Acad. Sci. USA 73:4589-4592, 1976). As shown by Milman et al. and confirmed by us here, rare HAT+ revertants arise spontaneously at 1.9 X 10(-8) frequency and express both mutant and wild-type polypeptides. Thus, the H23 mutant also carries a silent wild-type HPRT allele that is activated in revertants. To test whether the silent allele was activated via hypomethylation of genomic DNA, H23 cells were treated with inhibitors of DNA methylation, and revertants were scored by HAT or azaserine selection. At an optimal dose of 5 microM 5-azacytidine, the reversion frequency was increased about 50-fold when assayed by HAT selection and over 1,000-fold when assayed by azaserine selection. HAT+ and azaserine revertants were heterozygous for HPRT, expressing both wild-type and mutant HPRT polypeptides. Like spontaneous revertants, they contained active HPRT enzyme and were genetically unstable, reverting at about 10(-4) frequency. Similar results were found after treatment with N-methyl-N'-nitro-N-nitrosoguanidine, a DNA-alkylating agent and potent inhibitor of mammalian DNA methylation. By contrast, the DNA-ethylating agent, ethyl methanesulfonate (EMS), did not increase the HAT+ reversion frequency; it did, however, increase the frequency by which H23 revertants heterozygous for HPRT reverted to 6-thioguanine resistance. Of nine EMS revertants, seven lacked HPRT activity and had a substantially reduced expression of the wild-type polypeptide. These observations support the hypothesis that DNA methylation plays an important role in human X-chromosome inactivation and that EMS can inactivate gene expression by promoting enzymatic methylation of genomic DNA as found previously for the prolactin gene in GH3 rat pituitary tumor cells (R. D. Ivarie and J. A. Morris, Proc. Natl. Acad. Sci. USA 79:2967-2970, 1982; R. D. Ivarie, J. A. Morris, and J. A. Martial, Mol. Cell. Biol. 2:179-189, 1982).

Azacytidine-induced reactivation of a DNA repair gene in Chinese hamster ovary cells

Six X-ray-sensitive (xrs) strains of the CHO-K1 cell line were shown to revert at a very high frequency after treatment with 5-azacytidine. This suggested that there was a methylated xrs+ gene in these strains which was structurally intact, but not expressed. The xrs strains did not complement one another, and the locus was autosomally located. In view of the frequency of their isolation and their somewhat different phenotypes, we propose that the xrs strains are mutants derived from an active wild-type gene. However, there is in addition a methylated silent gene present in the genome. Azacytidine treatment reactivated this gene. We present a model for the functional hemizygosity of mammalian cell lines, which is based on the inactivation of genes by de novo hypermethylation. In contrast to results with xrs strains, other repair-defective lines were found not to be reverted by azacytidine.

Methylation patterns of HLA-DR alpha genes in six mononuclear cell lines

The relationship between DNA methylation and HLA-DR alpha gene expression was investigated in six mononuclear cell lines. RPMI-4265 (B cell) and HUT-78 (T cell) constitutively express HLA-DR. HL-60 (myelomonocyte) and U-937 (monocyte) can be induced to express HLA-DR. Jurkat and Molt-4 (T cells) do not and cannot be induced to express HLA-DR. Based on the known nucleotide sequence of the HLA-DR alpha gene, methylation-sensitive restriction endonucleases Msp I, Hpa II, Hha I, Ava I, Hae II, and Sma I were used to detect the CpG methylation in three regions of the HLA-DR alpha gene: the 5'flanking region, the exon 1 region, and the coding region containing exons 2, 3, 4, and 5. This precise mapping of CpG methylation yielded no correlation between DNA hypomethylation and HLA-DR alpha gene expression. In all cell lines, exon 1 region is hypomethylated, whereas 5' and coding regions are hypermethylated. Whereas hypermethylation of the coding region does not block transcription, hypomethylation of the exon 1 region may be essential but is clearly not sufficient for HLA-DR alpha gene transcription. This exon 1 region demethylation may result in an open (deoxyribonuclease I hypersensitive) chromatin conformation around the promoter where trans-acting regulatory factors presumably bind and initiate HLA-DR alpha transcription. In the course of this study, a novel Msp I polymorphism in the intron 1 of the HLA-DR alpha gene was found.

Effects of 5 azacytidine on DNA methylation and early development of sea urchins and ascidia

5-azacytidine (5-azaCR), an analogue of cytidine, inhibits nuclear DNA methylation in early sea urchin embryos. This inhibition is specific and dose-dependent. Exposure of sea urchin embryos at any stage between one-cell and blastula, to micromolar quantities of 5-azaCR invariably inhibits development beyond the blastula stage. In a substantial number of embryos arrested at the blastula stage, spicule formation proceeds although other morphological differentiation is lacking. No significant effect on development is seen if sea urchin embryos are exposed to 5-azaCR at post-blastula stages. 5-azaCR also inhibits the development of a mosaic egg such as the ascidian Phallusia mammilata at the blastula stage, indicating that both regulative (sea urchin) and mosaic (ascidian) embryos respond more or less similarly to 5-azaCR treatment.

Azacytidine-induced reactivation of a DNA repair gene in Chinese hamster ovary cells

Six X-ray-sensitive (xrs) strains of the CHO-K1 cell line were shown to revert at a very high frequency after treatment with 5-azacytidine. This suggested that there was a methylated xrs+ gene in these strains which was structurally intact, but not expressed. The xrs strains did not complement one another, and the locus was autosomally located. In view of the frequency of their isolation and their somewhat different phenotypes, we propose that the xrs strains are mutants derived from an active wild-type gene. However, there is in addition a methylated silent gene present in the genome. Azacytidine treatment reactivated this gene. We present a model for the functional hemizygosity of mammalian cell lines, which is based on the inactivation of genes by de novo hypermethylation. In contrast to results with xrs strains, other repair-defective lines were found not to be reverted by azacytidine.

Metastatic phenotype of human prostate tumor cells in athymic nude mice: alteration by exposure to ethyl methanesulfonate and "reversion" by 5-azacytidine

The human prostate tumor subline 1-LN-PC-3-1A (1-LN) is reproducibly metastatic in adult athymic nude mice. Cells surviving a brief in vitro exposure to ethyl methanesulfonate (EMS) exhibited a profound decrease in capacity for experimental lung metastasis in nude mice. Thirty days after EMS treatment, 1 X 10(6) uncloned EMS-treated 1-LN cells (1-LN-EMS-10) were injected IV into groups of 6 to 8-week-old male athymic nude mice (BALB/cAnBOM). A median of 8.5 colonies/lung was observed among 20 1-LN-EMS-10-injected mice, which was significantly different from the median of 51 colonies/lung produced among 14 1-LN-injected mice (P = 0.0002). This altered phenotype remained stable during 150 days of continuous culture. However, the 1-LN-EMS-10 cells were tumorigenic in 10/10 nude mice injected SC. Single lung tumor colonies recovered from 1-LN-EMS-10-injected mice and reinjected IV into nude mice produced medians of 32-63 colonies/lung. The altered metastatic phenotype resulting from treatment of 1-LN with EMS was reversed by exposure to a noncytotoxic dose of 5-azacytidine, but unaffected by a second exposure to EMS. Collectively these data demonstrate that the metastatic phenotype of these human tumor cells in athymic nude mice can be heritably altered by in vitro exposure to EMS and 5-azacytidine. Analysis of the mechanisms underlying these phenotypic changes may provide insight into parts of the complex process of tumor cell evolution.

Activation of a nonexpressed hypoxanthine phosphoribosyltransferase allele in mutant H23 HeLa cells by agents that inhibit DNA methylation

HeLA H23 cells are a mutant female human tumor cell line harboring defective hypoxanthine phosphoribosyltransferase (HPRT; IMP-pyrophosphate phosphoribosyltransferase, EC 2.4.2.8) as a result of a mutation that alters the isoelectric point of the enzyme (G. Milman, E. Lee, G. S. Changas, J. R. McLaughlin, and J. George, Jr., Proc. Natl. Acad. Sci. USA 73:4589-4592, 1976). As shown by Milman et al. and confirmed by us here, rare HAT+ revertants arise spontaneously at 1.9 X 10(-8) frequency and express both mutant and wild-type polypeptides. Thus, the H23 mutant also carries a silent wild-type HPRT allele that is activated in revertants. To test whether the silent allele was activated via hypomethylation of genomic DNA, H23 cells were treated with inhibitors of DNA methylation, and revertants were scored by HAT or azaserine selection. At an optimal dose of 5 microM 5-azacytidine, the reversion frequency was increased about 50-fold when assayed by HAT selection and over 1,000-fold when assayed by azaserine selection. HAT+ and azaserine revertants were heterozygous for HPRT, expressing both wild-type and mutant HPRT polypeptides. Like spontaneous revertants, they contained active HPRT enzyme and were genetically unstable, reverting at about 10(-4) frequency. Similar results were found after treatment with N-methyl-N'-nitro-N-nitrosoguanidine, a DNA-alkylating agent and potent inhibitor of mammalian DNA methylation. By contrast, the DNA-ethylating agent, ethyl methanesulfonate (EMS), did not increase the HAT+ reversion frequency; it did, however, increase the frequency by which H23 revertants heterozygous for HPRT reverted to 6-thioguanine resistance. Of nine EMS revertants, seven lacked HPRT activity and had a substantially reduced expression of the wild-type polypeptide. These observations support the hypothesis that DNA methylation plays an important role in human X-chromosome inactivation and that EMS can inactivate gene expression by promoting enzymatic methylation of genomic DNA as found previously for the prolactin gene in GH3 rat pituitary tumor cells (R. D. Ivarie and J. A. Morris, Proc. Natl. Acad. Sci. USA 79:2967-2970, 1982; R. D. Ivarie, J. A. Morris, and J. A. Martial, Mol. Cell. Biol. 2:179-189, 1982).

Ethylation of poly(dC-dG).poly(dC-dG) by ethyl methanesulfonate stimulates the activity of mammalian DNA methyltransferase in vitro

Ethylation of poly(dC-dG).poly(dC-dG) with ethyl methanesulfonate (EtMes), a known carcinogen, at increasing molar ratios of EtMes/C X G base pairs progressively stimulated the methyl-accepting ability of the DNA during in vitro methylation by partially purified rat DNA (cytosine-5)-methyltransferase (EC 2.1.1.37). Maximum stimulation was 2-fold over mock-treated DNA when 2.7% of the guanines were modified at the N-7 position, the major site of ethylation by EtMes in DNA. If a CpG site "hemiethylated" at guanine N-7 mimics a hemimethylated CpG site, we calculate that the enzyme has a relative affinity for hemiethylated CpG 18-fold above unmodified CpG. If ethylation of a dioxyphosphate oxygen of the phosphodiester bond is responsible for stimulation, the relative affinity could be much higher, up to 370-fold.

Chinese hamster cells deficient in ornithine decarboxylase activity: reversion by gene amplification and by azacytidine treatment

A group of Chinese hamster ovary (CHO) cell mutants deficient in ornithine decarboxylase (ODC) activity are described and compared to the prototype mutant reported previously (21). Although all mutants belong to the same complementation group, they can be divided into two classes: those with some residual enzyme activity and those with no activity. All mutants are putrescine auxotrophs, but they differ in their ability to utilize the enzyme's substrate, ornithine, a property which correlates with the amount of residual enzyme activity. The mutants also differ in their frequency of reversion to prototrophy. The leaky mutants revert at a high rate by overproducing a partially defective enzyme by a gene amplification mechanism similar to that leading to the ornithine analog-resistant mutants which have elevated enzyme levels. Spontaneous reversion in the null mutants is rare. However, one null mutant, which was induced with ethyl methane sulfonate and which makes ODC mRNA but no active enzyme, is nevertheless revertible with 5-azacytidine. We conclude that CHO cells are at least diploid at the ODC locus, but that only one allele is active. Further studies suggest the possibility that ethyl methane sulfonate is not just a classical mutagen but may also induce gene inactivations that are revertible by 5-azacytidine.

High-frequency induction by 5-azacytidine of proline independence in CHO-K1 cells

Proline independence in CHO-K1 Chinese hamster cells has in previous studies been characterized as an auxotrophic gene mutation. In the absence of direct proof, an alternative model must be considered, based on suppression of proline synthesis by DNA methylation changes at one or more loci concerned. This concept receives strong support from the present study, in which we show that treatment of CHO-K1 cells with 5-azacytidine induces a 10(5)-10(6) increase in background conversion to the proline-independent state. Revertants thus obtained, as well as those arising spontaneously or after treatment with ethyl methane sulfonate, are stable phenotypically in the presence or absence of proline. Proline independence in all variants examined was correlated with increased activity of pyrroline-5-carboxylate synthase. Four of five variants induced with 5-azacytidine showed simultaneous increases in activity of ornithine aminotransferase as well. Our data suggest that epigenetic, rather than genetic changes, underlie the transitions between proline dependence and independence in CHO-K1 cells.

Growth control variant cell line having increased serum requirement and decreased response to platelet-derived growth factor: reversion by 5-azacytidine

Variants of the mouse embryo fibroblast X melanoma hybrid clone 100A have been isolated by a procedure that selects against cells that are able to grow in medium containing low concentrations of serum plus insulin. Three variant clones derived from this selection were found to have a much higher serum requirement than the parental clone 100A cells, as evidenced by a very low rate of DNA synthesis and growth in medium containing low concentrations of serum. Two of the variants had approximately double the number of chromosomes as the parental cell line, while one had approximately the same number of chromosomes as the parental cells. One of the variants was very strongly reverted by 5-azacytidine but not by ethyl methanesulfonate, suggesting that it reverted by a nonmutational mechanism such as a stable change in DNA methylation. Analysis of the growth requirements in hormone-supplemented serum-free media of the 100A parent, the INS 471 variant, and revertants of the variant indicated that the variant had a specific deficiency in its growth response to platelet-derived growth factor (PDGF). PDGF dose-response curves obtained with the variant cells were shifted approximately an order of magnitude toward higher PDGF concentrations relative to PDGF dose-response curves obtained with the parental 100A cells. This quantitative increase in PDGF requirement of the INS 471 variant appears to explain the increased serum requirement of this variant. Equilibrium binding experiments performed with 125I-PDGF suggest that the variant does not have a decreased number of PDGF receptors.

Properties of asparagine synthetase in asparagine-independent variants of Jensen rat sarcoma cells induced by 5-azacytidine

Jensen rat sarcoma cells in culture require L-asparagine for growth and lack detectable levels of asparagine synthetase. Cultures exposed for 24 h to graded concentrations of 5-azacytidine give rise to asparagine-independent variants in high frequency. These prototrophs are stable phenotypically whether maintained in the presence or absence of L-asparagine. Asparagine synthetase activity in several variant clones was uniform in thermolability and several kinetic parameters, as well as in immunological properties. Parental Jensen rat sarcoma cells contained no detectable immunologically cross-reacting material. Our data suggest that transitions between asparagine dependence and independence in these cells are mediated by stable shifts in gene expression rather than by structural gene mutations.

The level of expression of the rat growth hormone gene in liver tumor cells is at least eight orders of magnitude less than that in anterior pituitary cells

Rat liver hepatoma cells (HTC) which express liver-specific gene products were assayed for the expression of the rat growth hormone (rGH) gene, which is normally expressed in anterior pituitary somatotrophs. The combination of immunoprecipitation and two-dimensional gel electrophoresis provided a highly sensitive assay for rGH synthesis at levels as low as one part in 10(9) of cell protein synthesis (or four molecules of rGH per cell). No rGH expression was detected at this level. The lack of expression in HTC cells did not derive from a deletion of the rGH gene, as shown by Southern hybridization analysis of genomic DNA. Because the gene is expressed at greater than 30% of anterior pituitary protein synthesis, differentiation regulated rGH expression by over 10(8)-fold between the two cell types. Additionally, DNA-excess solution hybridization was used to measure the level of rGH mRNA sequences. A novel and general method for preparing single-strand probes from recombinant plasmids was developed. Hybridization analyses with a sensitivity of detection of 1 part in 10(8) failed to detect any rGH RNA sequences in either the nucleus or cytoplasm of HTC cells. It is concluded, therefore, that the restriction in rGH expression in the liver tumor cells is likely to occur at the level of the transcription of the gene, and that for all practical purposes, the rGH gene is completely shut off in the hepatoma cells.

DNA modification, differentiation, and transformation

Substantial evidence has accumulated over the last 5 years that the methylation of cytosine residues in vertebrate DNA is implicated in the control of gene expression. We have used analogs of cytidine, modified in the 5 position, as specific inhibitors of DNA methylation to probe the relationship between this process and cellular differentiation. 5-Azacytidine effected marked changes in the differentiated state of cultured cells and induced the formation of biochemically differentiated muscle, fat, and chondrocytes from mouse fibroblast cell lines. Since the analog is a powerful inhibitor of DNA methylation, we suggest that this inhibition is causally related to the mechanism of phenotypic conversion. DNA extracted from cells treated with 5-azacytidine was hemimethylated and was used as an efficient acceptor of methyl groups in an in vitro reaction in the presence of eukaryotic methylases. In vitro methylation was inhibited if the substrate DNA was preincubated with a diverse range of chemical carcinogens including benzo(a)pyrene diolepoxide. Thus, chemical carcinogens may induce changes in gene expression by alteration of cellular methylation patterns. Recent experiments have also demonstrated that freshly explanted diploid fibroblasts from mice, hamsters, and humans lose substantial quantities of 5-methylcytosine during cell division and aging in culture. Taken together, these experiments suggest that the genomic distribution of 5-methylcytosine might have importance in normal differentiation and also in the aberrant gene expression found in cancer and senescence in culture.

Properties of asparagine synthetase in asparagine-independent variants of Jensen rat sarcoma cells induced by 5-azacytidine

Jensen rat sarcoma cells in culture require L-asparagine for growth and lack detectable levels of asparagine synthetase. Cultures exposed for 24 h to graded concentrations of 5-azacytidine give rise to asparagine-independent variants in high frequency. These prototrophs are stable phenotypically whether maintained in the presence or absence of L-asparagine. Asparagine synthetase activity in several variant clones was uniform in thermolability and several kinetic parameters, as well as in immunological properties. Parental Jensen rat sarcoma cells contained no detectable immunologically cross-reacting material. Our data suggest that transitions between asparagine dependence and independence in these cells are mediated by stable shifts in gene expression rather than by structural gene mutations.

Phenotypic variation associated with molecular alterations at a cluster of thymidine kinase genes

Genetic variation was studied in several mouse L cell lines containing tandemly repeated herpes simplex virus thymidine kinase (TK) genes introduced by DNA-mediated gene transfer. Variants were obtained after alternate positive and negative selection for TK expression. Three classes of molecular alteration are described. One class consisted of a concerted wave of hypermethylation affecting many sites in all or nearly all of the TK genes. This resulted in genetically stable TK- variants. Of five TK+ transformants from independent transfer experiments, only one, named HM, showed this class of methylation. Hypermethylation was a reproducible phenomenon in HM, yielding TK- variants after selection with either bromodeoxyuridine or acycloguanosine [Acyclovir or 9-(2-hydroxyethy-oxymethyl)guanine]. A second class of alteration consisted of methylation affecting some, but not all, genes in the cluster. This happened in all TK+ (HAT [hypoxanthine-aminopterin-thymidine]-resistant) cell lines investigated, and this second class of methylation was incapable of generating TK- variants. Neither type of methylation was accompanied by genomic rearrangements. The third class of molecular alteration was found among TK+ (HAT-resistant) back revertants of hypermethylated HM TK- derivatives. It consisted of a 10-fold amplification of the hypermethylated TK genes. Demethylation of hypermethylated HM variants was not observed. Thus, hypermethylation in this system can be compensated for by amplification but cannot be reversed.

Eukaryotic DNA methylation

Eukaryotic genomes contain 5-methylcytosine (5mC) as a rare base.5mC arises by postsynthetic modification of cytosine and occurs, at least in animals, predominantly in the dinucleotide CpG. The base is not distributed randomly in these genomes but conforms to a pattern. This pattern varies between taxa but appears to be inherited in a semi-conservative fashion. At the level of the genome, gross changes in the level of DNA methylation have been noted. This has encouraged speculation that the modification may play a role in cellular differentiation. Tissue-specific patterns of DNA methylation, predicted by various models of differentiation, have been found for most vertebrate genes so far examined. A correlation has emerged between the undermethylation of these regions and their transcription, but this is not always the case. While data for eukaryotic viral sequences are less equivocal, studies of this kind cannot in isolation distinguish between undermethylation being a cause or a consequence of gene activity. If it were a cause, it is probable that the demethylation of specific CpG sites would be a necessary yet not a sufficient condition for transcription to occur. The introduction of artificially methylated DNA sequences into individual eukaryotic cells by microinjection or transformation may provide the means to elucidate these questions in the future. In the meantime, the study of eukaryotic DNA methylation promises to contribute much to our understanding of the regulation of gene expression in these organisms.

The level of expression of the rat growth hormone gene in liver tumor cells is at least eight orders of magnitude less than that in anterior pituitary cells

Rat liver hepatoma cells (HTC) which express liver-specific gene products were assayed for the expression of the rat growth hormone (rGH) gene, which is normally expressed in anterior pituitary somatotrophs. The combination of immunoprecipitation and two-dimensional gel electrophoresis provided a highly sensitive assay for rGH synthesis at levels as low as one part in 10(9) of cell protein synthesis (or four molecules of rGH per cell). No rGH expression was detected at this level. The lack of expression in HTC cells did not derive from a deletion of the rGH gene, as shown by Southern hybridization analysis of genomic DNA. Because the gene is expressed at greater than 30% of anterior pituitary protein synthesis, differentiation regulated rGH expression by over 10(8)-fold between the two cell types. Additionally, DNA-excess solution hybridization was used to measure the level of rGH mRNA sequences. A novel and general method for preparing single-strand probes from recombinant plasmids was developed. Hybridization analyses with a sensitivity of detection of 1 part in 10(8) failed to detect any rGH RNA sequences in either the nucleus or cytoplasm of HTC cells. It is concluded, therefore, that the restriction in rGH expression in the liver tumor cells is likely to occur at the level of the transcription of the gene, and that for all practical purposes, the rGH gene is completely shut off in the hepatoma cells.

Phenotypic variation associated with molecular alterations at a cluster of thymidine kinase genes

Genetic variation was studied in several mouse L cell lines containing tandemly repeated herpes simplex virus thymidine kinase (TK) genes introduced by DNA-mediated gene transfer. Variants were obtained after alternate positive and negative selection for TK expression. Three classes of molecular alteration are described. One class consisted of a concerted wave of hypermethylation affecting many sites in all or nearly all of the TK genes. This resulted in genetically stable TK- variants. Of five TK+ transformants from independent transfer experiments, only one, named HM, showed this class of methylation. Hypermethylation was a reproducible phenomenon in HM, yielding TK- variants after selection with either bromodeoxyuridine or acycloguanosine [Acyclovir or 9-(2-hydroxyethy-oxymethyl)guanine]. A second class of alteration consisted of methylation affecting some, but not all, genes in the cluster. This happened in all TK+ (HAT [hypoxanthine-aminopterin-thymidine]-resistant) cell lines investigated, and this second class of methylation was incapable of generating TK- variants. Neither type of methylation was accompanied by genomic rearrangements. The third class of molecular alteration was found among TK+ (HAT-resistant) back revertants of hypermethylated HM TK- derivatives. It consisted of a 10-fold amplification of the hypermethylated TK genes. Demethylation of hypermethylated HM variants was not observed. Thus, hypermethylation in this system can be compensated for by amplification but cannot be reversed.

Phenotypic switching in GH3 rat pituitary tumor cells: linked expression of growth hormone and another hormonally responsive protein

From a phenotypically unstable, prolactin-deficient variant of GH3 rat pituitary tumor cells, variants deficient in growth hormone synthesis were isolated at high frequency (about 90%). The rate of synthesis of growth hormone (rGH) and one other hormonally responsive protein, termed p16, was substantially decreased in variant cells. Reduced synthesis of rGH was accompanied by a decrease in cytoplasmic levels of pre-rGH mRNA compared to levels found in wild-type GH3 cells. Although synthesis was decreased, inducibility of rGH and p16 synthesis by dexamethasone was unimpaired. The spontaneous reversion frequency was estimated to be near 4%. The expression of p16 was coordinately regained in an rGH-producing revertant, indicating that the expression of the two proteins is tightly coupled in GH3 cells. It was also found that p16 and p21 (another hormonally responsive protein in GH3 cells whose expression is tightly linked to that of prolactin) were detected at high levels in rat anterior pituitary. These observations suggest that, if growth hormone and p16 are encoded by distinct genes, then during the course of their evolution the two genes have developed similar genetic determinants governing basal expression and hormonal responsiveness in pituitary cells.

Asymmetrical distribution of CpG in an 'average' mammalian gene

The frequency and distribution of the rare dinucleotide CpG was examined in 15 mammalian genes. CpG is highly methylated at cytosine in mammalian DNA (1,2) and 5-methylcytosine (5mC) is thought to undergo a transition mutation via deamination to produce thymine (3). This would result in the accumulation of TpG and CpA and depletion of CpG during evolution (4). Consistent with this hypothesis, the gene sample of 26,541 dinucleotides contained CpG at 40% the frequency expected by base composition and the CpG transition products, TpG+CpA, were significantly elevated at 124% of expected random frequency. However, because CpG occurs at only 25% of expected random frequency in the genome, the sampled genes were considerably enriched in this dinucleotide. CpGs were asymmetrically distributed in sequences flanking the genes. 5'-flanking sequences were enriched in CpG at 135% of the frequency expected assuming a symmetrical distribution of all the CpGs in the sampled genes (p less than 0.01), while 3'-flanking regions were depleted in CpG at 40% of expected values (p less than 0.0001). This asymmetry may reflect the role of 5-methylcytosine in gene expression. In contrast the frequencies of GpC and GpT+ ApC did not differ significantly from that predicted by base composition and these dinucleotides were not asymmetrically distributed.