Hypomethylation of DNA derived from purified human erythroid cells correlates with gene activity of the beta-globin cluster

Analysis of methylation at the beta-globin gene cluster was carried out on DNA derived from nucleated RBCs (orthochromatic normoblasts) isolated from peripheral blood of patients with beta-thalassemia major or other congenital hemolytic anemia after splenectomy. A procedure to separate these normoblasts from the other nucleated cells of the peripheral blood was developed, providing us with a convenient source of DNA for investigating parameters related to human erythroid differentiation. Blood samples were obtained from six adult patients who express their gamma-globin genes at different levels. Inverse correlation between methylation and gene activity was consistently observed for five of the eight sites analyzed. A site 3' to the beta gene was always unmethylated, two sites flanking the epsilon gene were always found to be methylated, and two sites 5' to the two gamma genes, G gamma and A gamma, were hypomethylated in correlation with gamma gene activity of the individual patients. A site 5' to the delta gene was unmethylated in normoblasts as well as in WBC. No apparent relation between hypomethylation and gene activity was observed for two additional sites. The results suggest that methylation at specific chromosomal locations participate in genetic regulation of the beta- like globin genes in humans.     

Pattern of methylation of two genes coding for housekeeping functions.

The distribution of sites that can be methylated was analyzed in the Chinese hamster adenine phosphoribosyl-transferase (aprt) gene and the patterns of methylation of this gene and the mouse dihydrofolate reductase (dhfr) gene were studied by using CpG restriction enzymes. Both genes were found to be unmethylated completely at their 5'-end region and methylated heavily throughout the rest of the gene. Because the hamster aprt gene can be inhibited by DNA methylation in vivo, the results suggest that 5' undermethylation of this gene may be a necessary condition for its expression. The pattern of methylation of each of these two genes was similar in sperm and all other somatic tissue DNAs. This is in contrast to many tissue-specific genes that were found to be highly methylated in sperm DNA and undermethylated in the tissue in which they are expressed. This result is consistent with the fact that both aprt and dhfr are key enzymes in the biosynthesis of nucleotides and therefore expected to be synthesized in all cells.     

Human beta-globin gene expression in transgenic mice is enhanced by a distant DNase I hypersensitive site.

Several lines of evidence suggest that erythroid-specific DNase I hypersensitive sites (HS) located far upstream of the human beta-globin gene are important in regulating beta-globin gene expression. We used the polymerase chain reaction technique to amplify and clone an 882-base-pair DNA fragment spanning one of these HS, designated HSII, which is located 54 kilobases upstream of the beta-globin gene. The cloned HSII fragment was linked to a human beta-globin gene in either the genomic (HSII-beta) or antigenomic (HSII-beta) orientation. These two constructs and a beta-globin gene alone (beta) were injected into fertilized mouse eggs, and expression was analyzed in liver and brain from day-16 transgenic fetuses. Five of 7 beta-transgenic fetuses expressed human beta-globin mRNA, but the level of expression per gene copy was low, ranging from 0.93 to 22.4% of mouse alpha-globin mRNA (average 9.9%). In contrast, 11 of 12 HSII-beta transgenic fetuses expressed beta-globin mRNA at levels per gene copy ranging from 31.3 to 336.6% of mouse alpha-globin mRNA (average 139.5%). Only three fetuses containing intact copies of the HSII-beta construct were produced. Two of three expressed human beta-globin mRNA at levels per gene copy of 179.2 and 387.1%. Expression of human beta-globin mRNA was tissue-specific in all three types of transgenic fetuses. These studies demonstrate that a small DNA fragment containing a single erythroid-specific HS can stimulate high-level human beta-globin gene expression in transgenic mice.     

Murine erythroleukemia cell differentiation: DNase I hypersensitivity and DNA methylation near the globin genes

The sensitivity to digestion by DNase I of chromatin containing the alpha- and beta(major)-globin genes and the pattern of DNA methylation near these genes was examined during hexamethylenebisacetamide (HMBA)-mediated erythroid differentiation of murine erythroleukemia cells (MELC). In uninduced and induced cells, the chromatin regions containing the alpha- and beta-(major)-globin genes are more sensitive to digestion by DNase I than is the region containing an immunoglobulin gene (Igalpha) not expressed during erythroid differentiation. However, at low concentrations of DNase I, a 6- to 10-fold increase in site-specific cleavages was generated in chromatin regions near both the alpha- and beta(major)-globin genes in cells induced to differentiate by HMBA. The DNase I hypersensitive site near the beta(major)-globin gene maps to a small region near the 5' terminus of the gene. No detectable change in the pattern of DNA methylation around either the alpha- or beta-globin genes was observed during HMBA-mediated erythroid differentiation. Of the potentially methylated sites assayed and mapped near the beta(major)-globin gene, one site is fully methylated, one is partially methylated, and one is unmethylated both in uninduced and induced cells. Many (but not all) sites assayed near the alpha-globin genes are unmethylated in both uninduced and induced cells. These results show that specific alterations of chromatin structure occur during MELC differentiation and suggest that these changes may not involve alterations in the pattern of DNA methylation.     

Methylation of foreign DNA sequences in eukaryotic cells.

The herpesvirus thymidine kinase gene has been used to introduce foreign DNA sequences into mouse L cells by DNA-mediated gene transfer. These inserted genes were then assayed for methylation at the specific sequence C-C-G-G by using the restriction enzyme isoschizomers Hpa II and Msp I. Despite the fact that 70% of the cellular C-C-G-G sites are methylated, herpesvirus sequences, plasmid DNA, and growth hormone gene DNA were found to remain unmethylated in 90% of the clones that contain these genes. DNA that had been methylated in vitro with Hpa II methylase was also inserted into L cells. The presence of this modification in the vector DNA did not, however, guarantee that these sequences remained methylated in the recipient clones. Only 10% of all transformed clones were found to contain methylated C-C-G-G sequences in the vector DNA, and these modifications were stable for 25-50 generations. Hha I and Mbo I were used to probe for methyl groups at these restriction sites, but none of the inserted sequences acquired these modifications. These results are discussed in relation to various models put forth to explain the process of methylation in eukaryotic cells.     

Nucleotide-sequence-specific de novo methylation in a somatic murine cell line.

DNA fragments encoding the mouse steroid 21-hydroxylase (C21 or Cyp21A1) gene are de novo methylated when introduced into the mouse adrenocortical tumor cell line Y1 by DNA-mediated gene transfer. Although CCGG sequences within the C21 gene are de novo methylated, CCGG sites within flanking vector sequences, other mammalian gene sequences driven by the C21 promoter, and the neomycin-resistance gene, which was cotransfected with the C21 gene, do not become methylated. At least two separate signals for de novo methylation are encoded within the gene since three fragments derived from the C21 gene were methylated de novo. Specific de novo methylation of C21-derived sequences does not occur in L cells or Y1 kin8 cells; this suggests that the cellular factors needed for de novo methylation of the C21 gene are not ubiquitous. Most DNA sequences are not de novo methylated when introduced into somatic cells and DNA sequences other than the C21 gene are not de novo methylated when introduced into Y1 cells. Several groups have suggested that de novo methylation occurs in early embryonic cells and that somatic cells strictly maintain their methylation pattern by a semiconservative methyltransferase. Our results suggest that de novo methylation of specific nucleotide sequences can occur in some mammalian somatic cells.     

Mechanistic aspects of genome-wide demethylation in the preimplantation mouse embryo.

Gene-specific methylation patterns in mammals play a role in a variety of biological processes in the embryo and adult tissues. These patterns are established during embryo development by a process that involves genome-wide demethylation in the morula and de novo methylation in the pregastrula. To elucidate the mechanism of demethylation in the early mouse embryo, we have injected mouse zygotes with gene sequences that were methylated in vitro by Hpa II methylase and analyzed the methylation status of specific sites in blastocyst DNA. Because it had been propagated in Escherichia coli, the DNA used for these injections was also methylated at adenine residues in GATC sites. This allowed us to eliminate fully methylated, unintegrated DNA by Dpn I digestion and fully unmethylated, integrated DNA that underwent several rounds of replication by Mbo I digestion. The integrated, originally injected DNA strands were in a hemimethylated state and survived this treatment. The methylation status of Hpa II sites in these molecules was analyzed by Hpa II digestion of the genomic DNA isolated from blastocysts, followed by PCR amplification using appropriate primers. The results demonstrate that demethylation is achieved by an active mechanism and that specific sites in imprinted genes escape demethylation, maintaining a methylated state throughout preimplantation development.     

In vitro methylation of the hamster adenine phosphoribosyltransferase gene inhibits its expression in mouse L cells.

The effect of DNA methylation on the expression of the hamster adenine phosphoribosyltransferase (aprt) gene in mouse cells has been examined. This gene was methylated in vitro at all of its C-C-G-G sites by using Hpa II methylase and was inserted into mouse Ltk- aprt- L cells by cotransformation, with the herpes virus thymidine kinase gene as a selectable vector. Whereas clones carrying unmethylated aprt sequences were found to have an aprt+ phenotype as shown by their ability to grow in azaserine-containing medium, almost all clones carrying methylated aprt sequences were shown to be phenotypically aprt-. Blot hybridization analysis demonstrated that both the methylated and unmethylated aprt sequences were integrated into the cellular genome to the same extent and that the in vitro modification was stably maintained in these cells for many generations. When clones containing methylated aprt genes were exposed to conditions that select for the expression of the aprt gene, a low frequency of reversion to the aprt+ phenotype was observed. In all of these clones, this reversion was accompanied by reorganization and undermethylation of the aprt sequences. These results show that the expression of certain genes may be inhibited by site-specific methylation of these sequences and suggest that methylation may play a direct role in the regulation of gene expression.     

Inhibition of promoter activity by methylation: possible involvement of protein mediators.

To study the relationship between DNA methylation and promoter activity we have methylated in vitro the promoters of the mouse metallothionein I gene and the herpes simplex virus thymidine kinase gene. We have transiently transfected these promoters fused to the human growth hormone in their methylated or unmethylated state into mouse L or F9 cells. Promoters methylated by methylase (M.) Hpa II and M.Hha I caused inhibition of reporter gene expression in L cells but not in F9 cells, while methylation of all CpGs by M.Sss I caused inhibition in both cell lines. Repression of promoter activity by M.Hpa II and M.Hha I methylation, but not by M.Sss I methylation, could be alleviated by cotransfection with an excess of untranscribable DNA methylated with M.Sss I. The methylated sites in nuclei isolated from the transfected L cells, but not F9 cells, were found to be protected from Msp I digestion. Taken together these results suggest that a factor present in L cells and missing in F9 cells mediates the methylation-directed inhibition of promoter activity. The ability of methylated DNA to overcome the inhibition seems to reflect competition for the mediator factor. Interestingly, treatment with Zn2+ ions brought about activation of the methylated promoter of the metallothionein gene. Similarly, butyrate could override the repression of the thymidine kinase methylated promoter. These activations were not accompanied by demethylation of the promoter or displacement of the mediator factor.     

DNA methylation in chicken alpha-globin gene expression.

We have investigated certain specific methylation sites of the chicken alpha-globin gene cluster in DNA from embryonic and adult erythroid cells as well as from brain and sperm cells. Eight contiguous DNA fragments of the alpha-globin gene cluster were subcloned from a recombinant lambda phage. The subclones were used as probes to map all the Msp I/Hpa II and Hha I sites in the unmethylated cloned DNA and specific sites of methylation in and around the alpha-globin gene cluster in chromosomal DNA. The data show that sperm DNA is totally methylated at these restriction sites in the globin gene region, as is brain DNA, with some exceptions. Interestingly, the methylation status of specific sites 5' to the coding sequences is correlated with expression of the embryonic or adult alpha-globin genes in different stages of erythroid development. Some sites showing partial methylation, however, do not conform to the model that transcribed genes are unmethylated or undermethylated. We also find a well-defined 3.5-kilobase region of DNA 5' to the alpha-globin gene cluster in which all C-C-G-G sites are resistant to Msp I digestion in all tissues. This "Msp block" is presumably caused by 5-MeCpC methylation.     

Site-specific cleavage of DNA at 8- and 10-base-pair sequences.

A method is described for cutting DNA at specific sites that are 8 and 10 base pairs long. The DNA is first treated with a specific methylase, either the restriction-modification enzyme M. Taq I, which converts the 4-base sequence T-C-G-A to T-C-G-mA, or the similar enzyme M. Cla I, which converts the 6-base sequence A-T-C-G-A-T to A-T-C-G-mA-T. The DNA is then cleaved with Dpn I, a restriction endonuclease that recognizes the sequence G-mA-T-C. Dpn I is unique in that it cuts only DNA that is methylated at adenine in both strands of its recognition sequence. In DNAs that are not otherwise methylated at adenine in both strands of the sequence G-A-T-C, cleavage by Dpn I occurs only at the following sequences: in the case of M. Taq I methylation, 5' T-C-G-mA - T-C-G-mA 3' 3' mA-G-C - T-mA-G-C - T 5'; in the case of M. Cla I methylation, 5' A - T-C-G-mA - T-C-G-mA-T 3' 3' T-mA-G-C - T-mA-G-C - T-A 5'. Specific cutting and cloning at these methylase/Dpn I-generated sites is shown experimentally. Further, we describe how the above technique can be extended to generate Dpn I cleavage sites of up to 12 base pairs. In DNA that contains equal amounts of each base distributed at random, 8- and 10-base-pair recognition sequences occur, on the average, approximately once every 65,000 and 1,000,000 base pairs, respectively. Potential applications, including the development of cloning vectors and a rapid method for chromosome walking, are discussed.