Differential Reactivity of the RatS100A4(p9Ka) Gene to Sodium Bisulfite Is Associated with Differential Levels of the S100A4 (p9Ka) mRNA in Rat Mammary Epithelial Cells
1999; Elsevier BV; Volume: 274; Issue: 4 Linguagem: Inglês
10.1074/jbc.274.4.2483
ISSN1083-351X
AutoresDongsheng Chen, Philip S. Rudland, Hailan Chen, Roger Barraclough,
Tópico(s)Epigenetics and DNA Methylation
ResumoElevated intracellular levels of S100A4, an S100-related calcium-binding protein, induce metastatic capability in benign mammary tumor-derived epithelial cells and in transgenic mice bearing oncogene-induced benign mammary tumors. TheS100A4(p9Ka) gene in rat mammary epithelial cells expressing low levels of S100A4 yields a reduced number of fragments upon digestion with the methylation-sensitive restriction enzyme, HpaII, compared with the gene from high S100A4-expressing cells. Genomic sequencing of two potential regulatory elements in the S100A4 gene, an intronic enhancer and TATA box region, revealed that in low S100A4-expressing cells, most cytosine bases exhibited high levels of resistance to conversion to thymine by sodium bisulfite. In derivative cell lines, which express high levels of S100A4, only a small number of cytosine bases were resistant to treatment with sodium bisulfite. In contrast, cytosine bases in the DNA surrounding an upstream regulatory region, which binds inhibitory GC factor in the low-expressing cell lines, are sensitive to conversion to thymine by sodium bisulfite in both low- and high-expressing cell lines. The results suggest that the rat S100A4 gene is maintained in a different state in the low-expressing cell lines and that this state might be a consequence of the pattern of methylation in this regulated gene that does not contain a CpG island. Elevated intracellular levels of S100A4, an S100-related calcium-binding protein, induce metastatic capability in benign mammary tumor-derived epithelial cells and in transgenic mice bearing oncogene-induced benign mammary tumors. TheS100A4(p9Ka) gene in rat mammary epithelial cells expressing low levels of S100A4 yields a reduced number of fragments upon digestion with the methylation-sensitive restriction enzyme, HpaII, compared with the gene from high S100A4-expressing cells. Genomic sequencing of two potential regulatory elements in the S100A4 gene, an intronic enhancer and TATA box region, revealed that in low S100A4-expressing cells, most cytosine bases exhibited high levels of resistance to conversion to thymine by sodium bisulfite. In derivative cell lines, which express high levels of S100A4, only a small number of cytosine bases were resistant to treatment with sodium bisulfite. In contrast, cytosine bases in the DNA surrounding an upstream regulatory region, which binds inhibitory GC factor in the low-expressing cell lines, are sensitive to conversion to thymine by sodium bisulfite in both low- and high-expressing cell lines. The results suggest that the rat S100A4 gene is maintained in a different state in the low-expressing cell lines and that this state might be a consequence of the pattern of methylation in this regulated gene that does not contain a CpG island. 5-azacytidine 6-azacytidine double-distilled water DNA methyltransferase base pair(s) kilobase (pair) polymerase chain reaction S-adenosyl-l-homocysteine. Altered DNA methylation has been associated with carcinogenesis in a number of experimental systems (1Counts J. Goodman J. Cell. 1995; 83: 13-15Abstract Full Text PDF PubMed Scopus (211) Google Scholar, 2Zingg J.-M. Jones P. Carcinogenesis. 1997; 18: 869-882Crossref PubMed Scopus (280) Google Scholar). Hypomethylation of DNA in tumor cells might result in the elevated expression of tumor-promoting genes, whereas hypermethylation of DNA might be involved in reducing the expression of tumor-suppressing genes (1Counts J. Goodman J. Cell. 1995; 83: 13-15Abstract Full Text PDF PubMed Scopus (211) Google Scholar). Furthermore, it has been suggested that 5-methylcytosine might contribute to mutations in DNA as a result of its chemical conversion to thymine (1Counts J. Goodman J. Cell. 1995; 83: 13-15Abstract Full Text PDF PubMed Scopus (211) Google Scholar). Understanding the methylation status of genes thought to be involved in the development of cancer is therefore of some importance.S100A4(p9Ka), a small calcium-binding protein, is present at a relatively low level in a number of normal differentiated cell types (3Gibbs F. Barraclough R. Platt-Higgins A. Rudland P. Wilkinson M. Parry E. J. Histochem. Cytochem. 1995; 43: 169-180Crossref PubMed Scopus (58) Google Scholar). Cultured rat or mouse mammary epithelial cells exhibit a low level of S100A4 and its mRNA, whereas closely related metastatic epithelial cells (4.Dunnington, D. J., The Development and Study of Single Cell-cloned Metastasizing Mammary Tumour Cell Systems in the RatPh.D. thesis, 1984, University of London.Google Scholar, 5Ebralidze A. Tulchinsky E. Grigorian M. Afanayeva A. Senin V. Revazova E. Lukanidin E. Genes Dev. 1989; 3: l086-l093Crossref Scopus (311) Google Scholar) and derivative elongated cells (6Barraclough R. Dawson K.J. Rudland P.S. Eur. J. Biochem. 1982; 129: 335-341Crossref PubMed Scopus (33) Google Scholar, 7Barraclough R. Dawson K.J. Rudland P.S. Biochem. Biophys. Res. Commun. 1984; 120: 351-358Crossref PubMed Scopus (16) Google Scholar, 8Barraclough R. Kimbell R. Rudland P. Nucleic Acids Res. 1984; 21: 8097-8114Crossref Scopus (32) Google Scholar) exhibit high levels of S100A4 and its mRNA. Experimental elevation of the level of S100A4(p9Ka) in benign tumor-derived mammary epithelial cells induces the metastatic phenotype in the cells (9Davies B. Davies M. Gibbs F. Barraclough R. Rudland P. Oncogene. 1993; 8: 999-1008PubMed Google Scholar). In transgenic mice, multiple rat (10Davies M. Rudland P. Robertson L. Parry E. Jolicoeur P. Barraclough R. Oncogene. 1996; 13: 1631-1637PubMed Google Scholar) or mouse (11Ambartsumian N. Grigorian M. Larsen F. Karlstrom O. Sidenius N. Rygaard J. Georgiev G. Lukanidin E. Oncogene. 1996; 13: 1621-1630PubMed Google Scholar) S100A4 transgenes confer the metastatic phenotype on oncogene-induced (10Davies M. Rudland P. Robertson L. Parry E. Jolicoeur P. Barraclough R. Oncogene. 1996; 13: 1631-1637PubMed Google Scholar) or spontaneous (11Ambartsumian N. Grigorian M. Larsen F. Karlstrom O. Sidenius N. Rygaard J. Georgiev G. Lukanidin E. Oncogene. 1996; 13: 1621-1630PubMed Google Scholar) benign mammary tumors, respectively. In view of the metastasis-inducing properties of S100A4 in epithelial cells, it is important to understand mechanisms whereby its level is kept low in these cells.The rat S100A4(p9Ka) gene contains acis-acting sequence that binds a GC factor-like repressor protein and exerts a repressive effect on the transcription of theS100A4(p9Ka) gene in low S100A4(p9Ka)-expressing rat benign mammary tumor cells but not in high-expressing metastatic or myoepithelial-like rat mammary epithelial cells (12Chen D. Davies M. Rudland P. Barraclough R. J. Biol. Chem. 1997; 272: 20283-20290Abstract Full Text Full Text PDF PubMed Scopus (20) Google Scholar). Hypermethylation is also believed to play a role in the transcriptional silencing of the mouse (mts1) and human (CAPL) S100A4genes (13Pedrocchi M. Schäfer B. Mueller H. Eppenberger U. Heizmann C. Int. J. Cancer. 1994; 57: 684-690Crossref PubMed Scopus (132) Google Scholar, 14Tulchinsky E. Ford H. Kramerov D. Reshetnyak E. Grigorian M. Zain S. Lukanidin E. Proc. Natl. Acad. Sci. U. S. A. 1992; 89: 9146-9150Crossref PubMed Scopus (40) Google Scholar, 15Tulchinsky E. Grigorian M. Tkatch T. Georgiev G. Lukanidin E. Biochim. Biophys. Acta. 1995; 1261: 243-248Crossref PubMed Scopus (43) Google Scholar). To find out whether DNA methylation modulates transcription of the S100A4 gene in rat cells, the pattern and degree of methylation of genomic DNA from high or low S100A4(p9Ka)-expressing rat mammary tumor-derived cell lines have been investigated using sodium bisulfite-mediated genomic sequencing.DISCUSSIONThe rat S100A4(p9Ka) gene (12Chen D. Davies M. Rudland P. Barraclough R. J. Biol. Chem. 1997; 272: 20283-20290Abstract Full Text Full Text PDF PubMed Scopus (20) Google Scholar, 20Barraclough R. Savin J. Dube S. Rudland P. J. Mol. Biol. 1987; 198: 13-20Crossref PubMed Scopus (78) Google Scholar) does not contain a CpG island located between 2,696 base pairs upstream and 2,474 base pairs downstream of the start site of transcription, in common with the human (30Engelkamp D. Schäfer B.W. Mattei M.G. Erne P. Heizmann C.W. Proc. Natl. Acad. Sci. U. S. A. 1993; 90: 6547-6551Crossref PubMed Scopus (174) Google Scholar) and mouse (14Tulchinsky E. Ford H. Kramerov D. Reshetnyak E. Grigorian M. Zain S. Lukanidin E. Proc. Natl. Acad. Sci. U. S. A. 1992; 89: 9146-9150Crossref PubMed Scopus (40) Google Scholar) S100A4 genes. However, the rat S100A4(p9Ka) gene shows a 10–20-fold increase in expression between low-expressing benign rat mammary epithelial (Rama 37CL-A3) and high-expressing derivative elongated, myoepithelial-like (Rama 37-E8), or metastatic (Rama 800) epithelial mammary tumor cells (4.Dunnington, D. J., The Development and Study of Single Cell-cloned Metastasizing Mammary Tumour Cell Systems in the RatPh.D. thesis, 1984, University of London.Google Scholar, 6Barraclough R. Dawson K.J. Rudland P.S. Eur. J. Biochem. 1982; 129: 335-341Crossref PubMed Scopus (33) Google Scholar, 7Barraclough R. Dawson K.J. Rudland P.S. Biochem. Biophys. Res. Commun. 1984; 120: 351-358Crossref PubMed Scopus (16) Google Scholar). UsingMspI/HpaII digestion and Southern blotting analyses, different patterns of methylation of theS100A4(p9Ka) gene were revealed between the low-expressing Rama 37CL-A3 cell line on the one hand and the high expressing, Rama 37-E8 and Rama 800 cell lines on the other. Treatment of the Rama 37CL-A3 cells with either of two demethylating agents, 5-aza-C or SAH, significantly raises the level of p9Ka mRNA in this cell line, strongly suggesting that methylation of cytosine bases is, at least in part, responsible for repressing the production of S100A4(p9Ka) mRNA. The fact that the methylation inhibitors do not increase significantly the level of S100A4(p9Ka) mRNA in the high-expressing Rama 800 cells suggests that in the Rama 37CL-A3 cells the methylation inhibitors are affecting directly the methylation of the S100A4 gene and are not acting indirectly, for example, by affecting the expression of another gene product.Mapping by genomic sequencing of individual protected cytosine bases indicated that cytosines in the TATA box and putative enhancer-containing DNA regions of theS100A4(p9Ka) gene were highly protected from the action of sodium bisulfite only in the low S100A4(p9Ka)-expressing cell line but not in the two high S100A4(p9Ka)-expressing cell lines. In these regions of the S100A4(p9Ka) gene, resistance to conversion of cytosine bases to uracil by sodium bisulfite was found infrequently in CpG dinucleotides due to their paucity (no CpG in the TATA box region, 2 CpGs in the intron region). In contrast, protection of cytosine bases from reaction with bisulfite is more commonly observed in CpNpG trinucleotides (8 in the TATA box region and 6 in the intron region). Methylation of CpNpG trinucleotides has been shown to occur in mammalian cells using the bisulfite modification procedure, but generally only a proportion of the molecules examined were methylated (31Clark S. Harrison J. Frommer M. Nat. Genet. 1995; 10: 20-27Crossref PubMed Scopus (227) Google Scholar). In the present experiments, all randomly picked clones gave the same result.In the Rama 37CL-A3 cells, protection from sodium bisulfite is also evident in sequences CpCpCpApA, CpCpTpA, CpCpCpTpG, CpCpCpApT, CpTpCpTpT, CpApA, CpApCpApA, CpTpC, CpTpT, and CpTpA (16 such sequences in the TATA box region and 8 in the intron region). It is possible that the protection afforded to the cytosine bases is due to their methylation. A low level of similar unconverted cytosines has been reported recently (31Clark S. Harrison J. Frommer M. Nat. Genet. 1995; 10: 20-27Crossref PubMed Scopus (227) Google Scholar); however, it was not possible to distinguish a low background of unreacted cytosine bases from a low level of non-symmetrical methylation in the plasmid-based experimental system employed. Although in genes that possess discrete CpG islands, it is well established that cytosine methylation occurs predominantly in CpG dinucleotides; few genes that do not possess CpG islands have so far been studied. In the genome as a whole, methylation of CpA, CpT, and CpC dinucleotides has been detected by nearest-neighbor analysis (32Woodcock D. Crowther P. Diver W. Biochem. Biophys. Res. Commun. 1987; 145: 888-894Crossref PubMed Scopus (105) Google Scholar). Although a greater proportion of CpG dinucleotides are methylated due to the low overall occurrence of this dinucleotide in the genome, over 50% of the methylation of the genome as a whole occurs in CpA, CpT, and CpC dinucleotides (32Woodcock D. Crowther P. Diver W. Biochem. Biophys. Res. Commun. 1987; 145: 888-894Crossref PubMed Scopus (105) Google Scholar). However, it is not known whether these methylated dinucleotides are associated with any particular genes. Similar, diverse, non-symmetric sites of methylation have been described previously as “densely methylated islands” at origins of replication associated with ribosomal protein S14 and thedhfr locus in Chinese hamster ovary cells (33Tasheva E. Roufa D. Mol. Cell. Biol. 1994; 14: 5636-5644Crossref PubMed Scopus (53) Google Scholar). However, it has been suggested that these densely methylated islands of methylated cytosine residues might be artifactual, arising from incomplete denaturation of the DNA prior to treatment with sodium bisulfite, at least at the dhfr locus (34Rein T. Natale D. Gartner U. Niggemann M. DePamphilis M. Zorbas H. J. Biol. Chem. 1997; 272: 10021-10029Abstract Full Text Full Text PDF PubMed Scopus (22) Google Scholar). In the present experiments, it is unlikely that the unusual pattern of methylation arises from incomplete denaturation of the DNA. Neither of two closely related cell lines, Rama 37-E8 nor Rama 800, showed the same pattern of protection of cytosine bases as Rama 37CL-A3 cells, even though treatment of their DNA with sodium bisulfite was carried out under identical conditions and at the same time as the DNA from the Rama 37CL-A3 cells. The lack of protection of cytosines in the upstream region of the DNA from the Rama 37CL-A3 cells further argues against a generalized technical failure to denature the DNA prior to reaction with sodium bisulfite. However, it is possible that the DNA in the TATA box and intron regions of the S100A4(p9Ka) gene in the Rama 37CL-A3 cells is in some way different from that isolated from the Rama 37-E8 and Rama 800 cells, such that it is not completely denatured under conditions that denature the DNA from Rama 37-E8 and Rama 800 cells (24Barraclough R. Kimbell R. Rudland P.S. J. Cell. Physiol. 1987; 131: 393-401Crossref PubMed Scopus (29) Google Scholar). Based on the derepressive effect of methylation inhibitors on S100A4(p9Ka) mRNA levels and the results of the digestions withMspI and HpaII, the reduced accessibility of cytosines to bisulfate in the Rama 37CL-A3 DNA is likely to be associated with methylation. However, it cannot be ruled out that the reduced accessibility of cytosine bases to sodium bisulfite in the Rama 37CL-A3 cells is a consequence of CpG methylation outside the regions of the genome analyzed in the present experiments.The region surrounding the upstream cis-acting inhibitory sequence of the S100A4(p9Ka) gene (12Chen D. Davies M. Rudland P. Barraclough R. J. Biol. Chem. 1997; 272: 20283-20290Abstract Full Text Full Text PDF PubMed Scopus (20) Google Scholar) exhibits little or no protection of cytosine bases in all three cell lines. In the Rama 37CL-A3 cells, there is a clear contrast between the density of protected cytosine bases observed in this region (0/39) and the density of protected cytosine bases in the TATA box (21/27) and intron regions (16/16) of the rat S100A4(p9Ka) gene. It is possible that the unprotected nature of this region of the gene in all three cell lines is associated with the binding of the inhibitory GC factor (12Chen D. Davies M. Rudland P. Barraclough R. J. Biol. Chem. 1997; 272: 20283-20290Abstract Full Text Full Text PDF PubMed Scopus (20) Google Scholar). In the Rama 37-E8 and Rama 800 cells, the level of GC factor mRNA is lower than in the Rama 37CL-A3 cells (12Chen D. Davies M. Rudland P. Barraclough R. J. Biol. Chem. 1997; 272: 20283-20290Abstract Full Text Full Text PDF PubMed Scopus (20) Google Scholar).Treatment of Rama 37CL-A3 cells with demethylating agents, but not an inactive analogue, reduces the number of methylcytosine residues in theS100A4(p9Ka) gene region of the Rama 37CL-A3 cells and causes an approximately 5–6-fold increase in the steady state level of S100A4(p9Ka). However, although the endogenous S100A4(p9Ka) mRNA in the Rama 37CL-A3 cells can be increased to a certain extent, neither of the two demethylating agents was able to increase the level of this mRNA to more than half that of its relative level in the Rama 800 cells. Since the inhibitors have no effect on S100A4(p9Ka) mRNA level in at least one high-expressing cell line (Rama 800), it is possible that this lack of full induction represents the presence of inhibitory levels of GC factor which serve to limit the expression of the S100A4(p9Ka) gene in Rama 37CL-A3 cells, and suggests that methylation of positive regulatory elements cooperates with an inhibitory factor to completely inhibit the production of S100A4 mRNA.Altered levels of DNA methyltransferases and/or patterns of methylation are thought to contribute to the formation of cancer. Since it is likely that cytosine methylation of theS100A4(p9Ka) gene plays a role in the suppression of transcription of rat and mouse S100A4(p9Ka), and in view of the metastasis-inducing properties of human S100A4 (35Lloyd B. Platt-Higgins A. Rudland P. Barraclough R. Oncogene. 1998; 17: 465-473Crossref PubMed Scopus (101) Google Scholar), it is possible that inadvertent hypomethylation of theS100A4(p9Ka) gene in human breast epithelial cells might contribute to the malignant progression of some breast cancers. Altered DNA methylation has been associated with carcinogenesis in a number of experimental systems (1Counts J. Goodman J. Cell. 1995; 83: 13-15Abstract Full Text PDF PubMed Scopus (211) Google Scholar, 2Zingg J.-M. Jones P. Carcinogenesis. 1997; 18: 869-882Crossref PubMed Scopus (280) Google Scholar). Hypomethylation of DNA in tumor cells might result in the elevated expression of tumor-promoting genes, whereas hypermethylation of DNA might be involved in reducing the expression of tumor-suppressing genes (1Counts J. Goodman J. Cell. 1995; 83: 13-15Abstract Full Text PDF PubMed Scopus (211) Google Scholar). Furthermore, it has been suggested that 5-methylcytosine might contribute to mutations in DNA as a result of its chemical conversion to thymine (1Counts J. Goodman J. Cell. 1995; 83: 13-15Abstract Full Text PDF PubMed Scopus (211) Google Scholar). Understanding the methylation status of genes thought to be involved in the development of cancer is therefore of some importance. S100A4(p9Ka), a small calcium-binding protein, is present at a relatively low level in a number of normal differentiated cell types (3Gibbs F. Barraclough R. Platt-Higgins A. Rudland P. Wilkinson M. Parry E. J. Histochem. Cytochem. 1995; 43: 169-180Crossref PubMed Scopus (58) Google Scholar). Cultured rat or mouse mammary epithelial cells exhibit a low level of S100A4 and its mRNA, whereas closely related metastatic epithelial cells (4.Dunnington, D. J., The Development and Study of Single Cell-cloned Metastasizing Mammary Tumour Cell Systems in the RatPh.D. thesis, 1984, University of London.Google Scholar, 5Ebralidze A. Tulchinsky E. Grigorian M. Afanayeva A. Senin V. Revazova E. Lukanidin E. Genes Dev. 1989; 3: l086-l093Crossref Scopus (311) Google Scholar) and derivative elongated cells (6Barraclough R. Dawson K.J. Rudland P.S. Eur. J. Biochem. 1982; 129: 335-341Crossref PubMed Scopus (33) Google Scholar, 7Barraclough R. Dawson K.J. Rudland P.S. Biochem. Biophys. Res. Commun. 1984; 120: 351-358Crossref PubMed Scopus (16) Google Scholar, 8Barraclough R. Kimbell R. Rudland P. Nucleic Acids Res. 1984; 21: 8097-8114Crossref Scopus (32) Google Scholar) exhibit high levels of S100A4 and its mRNA. Experimental elevation of the level of S100A4(p9Ka) in benign tumor-derived mammary epithelial cells induces the metastatic phenotype in the cells (9Davies B. Davies M. Gibbs F. Barraclough R. Rudland P. Oncogene. 1993; 8: 999-1008PubMed Google Scholar). In transgenic mice, multiple rat (10Davies M. Rudland P. Robertson L. Parry E. Jolicoeur P. Barraclough R. Oncogene. 1996; 13: 1631-1637PubMed Google Scholar) or mouse (11Ambartsumian N. Grigorian M. Larsen F. Karlstrom O. Sidenius N. Rygaard J. Georgiev G. Lukanidin E. Oncogene. 1996; 13: 1621-1630PubMed Google Scholar) S100A4 transgenes confer the metastatic phenotype on oncogene-induced (10Davies M. Rudland P. Robertson L. Parry E. Jolicoeur P. Barraclough R. Oncogene. 1996; 13: 1631-1637PubMed Google Scholar) or spontaneous (11Ambartsumian N. Grigorian M. Larsen F. Karlstrom O. Sidenius N. Rygaard J. Georgiev G. Lukanidin E. Oncogene. 1996; 13: 1621-1630PubMed Google Scholar) benign mammary tumors, respectively. In view of the metastasis-inducing properties of S100A4 in epithelial cells, it is important to understand mechanisms whereby its level is kept low in these cells. The rat S100A4(p9Ka) gene contains acis-acting sequence that binds a GC factor-like repressor protein and exerts a repressive effect on the transcription of theS100A4(p9Ka) gene in low S100A4(p9Ka)-expressing rat benign mammary tumor cells but not in high-expressing metastatic or myoepithelial-like rat mammary epithelial cells (12Chen D. Davies M. Rudland P. Barraclough R. J. Biol. Chem. 1997; 272: 20283-20290Abstract Full Text Full Text PDF PubMed Scopus (20) Google Scholar). Hypermethylation is also believed to play a role in the transcriptional silencing of the mouse (mts1) and human (CAPL) S100A4genes (13Pedrocchi M. Schäfer B. Mueller H. Eppenberger U. Heizmann C. Int. J. Cancer. 1994; 57: 684-690Crossref PubMed Scopus (132) Google Scholar, 14Tulchinsky E. Ford H. Kramerov D. Reshetnyak E. Grigorian M. Zain S. Lukanidin E. Proc. Natl. Acad. Sci. U. S. A. 1992; 89: 9146-9150Crossref PubMed Scopus (40) Google Scholar, 15Tulchinsky E. Grigorian M. Tkatch T. Georgiev G. Lukanidin E. Biochim. Biophys. Acta. 1995; 1261: 243-248Crossref PubMed Scopus (43) Google Scholar). To find out whether DNA methylation modulates transcription of the S100A4 gene in rat cells, the pattern and degree of methylation of genomic DNA from high or low S100A4(p9Ka)-expressing rat mammary tumor-derived cell lines have been investigated using sodium bisulfite-mediated genomic sequencing. DISCUSSIONThe rat S100A4(p9Ka) gene (12Chen D. Davies M. Rudland P. Barraclough R. J. Biol. Chem. 1997; 272: 20283-20290Abstract Full Text Full Text PDF PubMed Scopus (20) Google Scholar, 20Barraclough R. Savin J. Dube S. Rudland P. J. Mol. Biol. 1987; 198: 13-20Crossref PubMed Scopus (78) Google Scholar) does not contain a CpG island located between 2,696 base pairs upstream and 2,474 base pairs downstream of the start site of transcription, in common with the human (30Engelkamp D. Schäfer B.W. Mattei M.G. Erne P. Heizmann C.W. Proc. Natl. Acad. Sci. U. S. A. 1993; 90: 6547-6551Crossref PubMed Scopus (174) Google Scholar) and mouse (14Tulchinsky E. Ford H. Kramerov D. Reshetnyak E. Grigorian M. Zain S. Lukanidin E. Proc. Natl. Acad. Sci. U. S. A. 1992; 89: 9146-9150Crossref PubMed Scopus (40) Google Scholar) S100A4 genes. However, the rat S100A4(p9Ka) gene shows a 10–20-fold increase in expression between low-expressing benign rat mammary epithelial (Rama 37CL-A3) and high-expressing derivative elongated, myoepithelial-like (Rama 37-E8), or metastatic (Rama 800) epithelial mammary tumor cells (4.Dunnington, D. J., The Development and Study of Single Cell-cloned Metastasizing Mammary Tumour Cell Systems in the RatPh.D. thesis, 1984, University of London.Google Scholar, 6Barraclough R. Dawson K.J. Rudland P.S. Eur. J. Biochem. 1982; 129: 335-341Crossref PubMed Scopus (33) Google Scholar, 7Barraclough R. Dawson K.J. Rudland P.S. Biochem. Biophys. Res. Commun. 1984; 120: 351-358Crossref PubMed Scopus (16) Google Scholar). UsingMspI/HpaII digestion and Southern blotting analyses, different patterns of methylation of theS100A4(p9Ka) gene were revealed between the low-expressing Rama 37CL-A3 cell line on the one hand and the high expressing, Rama 37-E8 and Rama 800 cell lines on the other. Treatment of the Rama 37CL-A3 cells with either of two demethylating agents, 5-aza-C or SAH, significantly raises the level of p9Ka mRNA in this cell line, strongly suggesting that methylation of cytosine bases is, at least in part, responsible for repressing the production of S100A4(p9Ka) mRNA. The fact that the methylation inhibitors do not increase significantly the level of S100A4(p9Ka) mRNA in the high-expressing Rama 800 cells suggests that in the Rama 37CL-A3 cells the methylation inhibitors are affecting directly the methylation of the S100A4 gene and are not acting indirectly, for example, by affecting the expression of another gene product.Mapping by genomic sequencing of individual protected cytosine bases indicated that cytosines in the TATA box and putative enhancer-containing DNA regions of theS100A4(p9Ka) gene were highly protected from the action of sodium bisulfite only in the low S100A4(p9Ka)-expressing cell line but not in the two high S100A4(p9Ka)-expressing cell lines. In these regions of the S100A4(p9Ka) gene, resistance to conversion of cytosine bases to uracil by sodium bisulfite was found infrequently in CpG dinucleotides due to their paucity (no CpG in the TATA box region, 2 CpGs in the intron region). In contrast, protection of cytosine bases from reaction with bisulfite is more commonly observed in CpNpG trinucleotides (8 in the TATA box region and 6 in the intron region). Methylation of CpNpG trinucleotides has been shown to occur in mammalian cells using the bisulfite modification procedure, but generally only a proportion of the molecules examined were methylated (31Clark S. Harrison J. Frommer M. Nat. Genet. 1995; 10: 20-27Crossref PubMed Scopus (227) Google Scholar). In the present experiments, all randomly picked clones gave the same result.In the Rama 37CL-A3 cells, protection from sodium bisulfite is also evident in sequences CpCpCpApA, CpCpTpA, CpCpCpTpG, CpCpCpApT, CpTpCpTpT, CpApA, CpApCpApA, CpTpC, CpTpT, and CpTpA (16 such sequences in the TATA box region and 8 in the intron region). It is possible that the protection afforded to the cytosine bases is due to their methylation. A low level of similar unconverted cytosines has been reported recently (31Clark S. Harrison J. Frommer M. Nat. Genet. 1995; 10: 20-27Crossref PubMed Scopus (227) Google Scholar); however, it was not possible to distinguish a low background of unreacted cytosine bases from a low level of non-symmetrical methylation in the plasmid-based experimental system employed. Although in genes that possess discrete CpG islands, it is well established that cytosine methylation occurs predominantly in CpG dinucleotides; few genes that do not possess CpG islands have so far been studied. In the genome as a whole, methylation of CpA, CpT, and CpC dinucleotides has been detected by nearest-neighbor analysis (32Woodcock D. Crowther P. Diver W. Biochem. Biophys. Res. Commun. 1987; 145: 888-894Crossref PubMed Scopus (105) Google Scholar). Although a greater proportion of CpG dinucleotides are methylated due to the low overall occurrence of this dinucleotide in the genome, over 50% of the methylation of the genome as a whole occurs in CpA, CpT, and CpC dinucleotides (32Woodcock D. Crowther P. Diver W. Biochem. Biophys. Res. Commun. 1987; 145: 888-894Crossref PubMed Scopus (105) Google Scholar). However, it is not known whether these methylated dinucleotides are associated with any particular genes. Similar, diverse, non-symmetric sites of methylation have been described previously as “densely methylated islands” at origins of replication associated with ribosomal protein S14 and thedhfr locus in Chinese hamster ovary cells (33Tasheva E. Roufa D. Mol. Cell. Biol. 1994; 14: 5636-5644Crossref PubMed Scopus (53) Google Scholar). However, it has been suggested that these densely methylated islands of methylated cytosine residues might be artifactual, arising from incomplete denaturation of the DNA prior to treatment with sodium bisulfite, at least at the dhfr locus (34Rein T. Natale D. Gartner U. Niggemann M. DePamphilis M. Zorbas H. J. Biol. Chem. 1997; 272: 10021-10029Abstract Full Text Full Text PDF PubMed Scopus (22) Google Scholar). In the present experiments, it is unlikely that the unusual pattern of methylation arises from incomplete denaturation of the DNA. Neither of two closely related cell lines, Rama 37-E8 nor Rama 800, showed the same pattern of protection of cytosine bases as Rama 37CL-A3 cells, even though treatment of their DNA with sodium bisulfite was carried out under identical conditions and at the same time as the DNA from the Rama 37CL-A3 cells. The lack of protection of cytosines in the upstream region of the DNA from the Rama 37CL-A3 cells further argues against a generalized technical failure to denature the DNA prior to reaction with sodium bisulfite. However, it is possible that the DNA in the TATA box and intron regions of the S100A4(p9Ka) gene in the Rama 37CL-A3 cells is in some way different from that isolated from the Rama 37-E8 and Rama 800 cells, such that it is not completely denatured under conditions that denature the DNA from Rama 37-E8 and Rama 800 cells (24Barraclough R. Kimbell R. Rudland P.S. J. Cell. Physiol. 1987; 131: 393-401Crossref PubMed Scopus (29) Google Scholar). Based on the derepressive effect of methylation inhibitors on S100A4(p9Ka) mRNA levels and the results of the digestions withMspI and HpaII, the reduced accessibility of cytosines to bisulfate in the Rama 37CL-A3 DNA is likely to be associated with methylation. However, it cannot be ruled out that the reduced accessibility of cytosine bases to sodium bisulfite in the Rama 37CL-A3 cells is a consequence of CpG methylation outside the regions of the genome analyzed in the present experiments.The region surrounding the upstream cis-acting inhibitory sequence of the S100A4(p9Ka) gene (12Chen D. Davies M. Rudland P. Barraclough R. J. Biol. Chem. 1997; 272: 20283-20290Abstract Full Text Full Text PDF PubMed Scopus (20) Google Scholar) exhibits little or no protection of cytosine bases in all three cell lines. In the Rama 37CL-A3 cells, there is a clear contrast between the density of protected cytosine bases observed in this region (0/39) and the density of protected cytosine bases in the TATA box (21/27) and intron regions (16/16) of the rat S100A4(p9Ka) gene. It is possible that the unprotected nature of this region of the gene in all three cell lines is associated with the binding of the inhibitory GC factor (12Chen D. Davies M. Rudland P. Barraclough R. J. Biol. Chem. 1997; 272: 20283-20290Abstract Full Text Full Text PDF PubMed Scopus (20) Google Scholar). In the Rama 37-E8 and Rama 800 cells, the level of GC factor mRNA is lower than in the Rama 37CL-A3 cells (12Chen D. Davies M. Rudland P. Barraclough R. J. Biol. Chem. 1997; 272: 20283-20290Abstract Full Text Full Text PDF PubMed Scopus (20) Google Scholar).Treatment of Rama 37CL-A3 cells with demethylating agents, but not an inactive analogue, reduces the number of methylcytosine residues in theS100A4(p9Ka) gene region of the Rama 37CL-A3 cells and causes an approximately 5–6-fold increase in the steady state level of S100A4(p9Ka). However, although the endogenous S100A4(p9Ka) mRNA in the Rama 37CL-A3 cells can be increased to a certain extent, neither of the two demethylating agents was able to increase the level of this mRNA to more than half that of its relative level in the Rama 800 cells. Since the inhibitors have no effect on S100A4(p9Ka) mRNA level in at least one high-expressing cell line (Rama 800), it is possible that this lack of full induction represents the presence of inhibitory levels of GC factor which serve to limit the expression of the S100A4(p9Ka) gene in Rama 37CL-A3 cells, and suggests that methylation of positive regulatory elements cooperates with an inhibitory factor to completely inhibit the production of S100A4 mRNA.Altered levels of DNA methyltransferases and/or patterns of methylation are thought to contribute to the formation of cancer. Since it is likely that cytosine methylation of theS100A4(p9Ka) gene plays a role in the suppression of transcription of rat and mouse S100A4(p9Ka), and in view of the metastasis-inducing properties of human S100A4 (35Lloyd B. Platt-Higgins A. Rudland P. Barraclough R. Oncogene. 1998; 17: 465-473Crossref PubMed Scopus (101) Google Scholar), it is possible that inadvertent hypomethylation of theS100A4(p9Ka) gene in human breast epithelial cells might contribute to the malignant progression of some breast cancers. The rat S100A4(p9Ka) gene (12Chen D. Davies M. Rudland P. Barraclough R. J. Biol. Chem. 1997; 272: 20283-20290Abstract Full Text Full Text PDF PubMed Scopus (20) Google Scholar, 20Barraclough R. Savin J. Dube S. Rudland P. J. Mol. Biol. 1987; 198: 13-20Crossref PubMed Scopus (78) Google Scholar) does not contain a CpG island located between 2,696 base pairs upstream and 2,474 base pairs downstream of the start site of transcription, in common with the human (30Engelkamp D. Schäfer B.W. Mattei M.G. Erne P. Heizmann C.W. Proc. Natl. Acad. Sci. U. S. A. 1993; 90: 6547-6551Crossref PubMed Scopus (174) Google Scholar) and mouse (14Tulchinsky E. Ford H. Kramerov D. Reshetnyak E. Grigorian M. Zain S. Lukanidin E. Proc. Natl. Acad. Sci. U. S. A. 1992; 89: 9146-9150Crossref PubMed Scopus (40) Google Scholar) S100A4 genes. However, the rat S100A4(p9Ka) gene shows a 10–20-fold increase in expression between low-expressing benign rat mammary epithelial (Rama 37CL-A3) and high-expressing derivative elongated, myoepithelial-like (Rama 37-E8), or metastatic (Rama 800) epithelial mammary tumor cells (4.Dunnington, D. J., The Development and Study of Single Cell-cloned Metastasizing Mammary Tumour Cell Systems in the RatPh.D. thesis, 1984, University of London.Google Scholar, 6Barraclough R. Dawson K.J. Rudland P.S. Eur. J. Biochem. 1982; 129: 335-341Crossref PubMed Scopus (33) Google Scholar, 7Barraclough R. Dawson K.J. Rudland P.S. Biochem. Biophys. Res. Commun. 1984; 120: 351-358Crossref PubMed Scopus (16) Google Scholar). UsingMspI/HpaII digestion and Southern blotting analyses, different patterns of methylation of theS100A4(p9Ka) gene were revealed between the low-expressing Rama 37CL-A3 cell line on the one hand and the high expressing, Rama 37-E8 and Rama 800 cell lines on the other. Treatment of the Rama 37CL-A3 cells with either of two demethylating agents, 5-aza-C or SAH, significantly raises the level of p9Ka mRNA in this cell line, strongly suggesting that methylation of cytosine bases is, at least in part, responsible for repressing the production of S100A4(p9Ka) mRNA. The fact that the methylation inhibitors do not increase significantly the level of S100A4(p9Ka) mRNA in the high-expressing Rama 800 cells suggests that in the Rama 37CL-A3 cells the methylation inhibitors are affecting directly the methylation of the S100A4 gene and are not acting indirectly, for example, by affecting the expression of another gene product. Mapping by genomic sequencing of individual protected cytosine bases indicated that cytosines in the TATA box and putative enhancer-containing DNA regions of theS100A4(p9Ka) gene were highly protected from the action of sodium bisulfite only in the low S100A4(p9Ka)-expressing cell line but not in the two high S100A4(p9Ka)-expressing cell lines. In these regions of the S100A4(p9Ka) gene, resistance to conversion of cytosine bases to uracil by sodium bisulfite was found infrequently in CpG dinucleotides due to their paucity (no CpG in the TATA box region, 2 CpGs in the intron region). In contrast, protection of cytosine bases from reaction with bisulfite is more commonly observed in CpNpG trinucleotides (8 in the TATA box region and 6 in the intron region). Methylation of CpNpG trinucleotides has been shown to occur in mammalian cells using the bisulfite modification procedure, but generally only a proportion of the molecules examined were methylated (31Clark S. Harrison J. Frommer M. Nat. Genet. 1995; 10: 20-27Crossref PubMed Scopus (227) Google Scholar). In the present experiments, all randomly picked clones gave the same result. In the Rama 37CL-A3 cells, protection from sodium bisulfite is also evident in sequences CpCpCpApA, CpCpTpA, CpCpCpTpG, CpCpCpApT, CpTpCpTpT, CpApA, CpApCpApA, CpTpC, CpTpT, and CpTpA (16 such sequences in the TATA box region and 8 in the intron region). It is possible that the protection afforded to the cytosine bases is due to their methylation. A low level of similar unconverted cytosines has been reported recently (31Clark S. Harrison J. Frommer M. Nat. Genet. 1995; 10: 20-27Crossref PubMed Scopus (227) Google Scholar); however, it was not possible to distinguish a low background of unreacted cytosine bases from a low level of non-symmetrical methylation in the plasmid-based experimental system employed. Although in genes that possess discrete CpG islands, it is well established that cytosine methylation occurs predominantly in CpG dinucleotides; few genes that do not possess CpG islands have so far been studied. In the genome as a whole, methylation of CpA, CpT, and CpC dinucleotides has been detected by nearest-neighbor analysis (32Woodcock D. Crowther P. Diver W. Biochem. Biophys. Res. Commun. 1987; 145: 888-894Crossref PubMed Scopus (105) Google Scholar). Although a greater proportion of CpG dinucleotides are methylated due to the low overall occurrence of this dinucleotide in the genome, over 50% of the methylation of the genome as a whole occurs in CpA, CpT, and CpC dinucleotides (32Woodcock D. Crowther P. Diver W. Biochem. Biophys. Res. Commun. 1987; 145: 888-894Crossref PubMed Scopus (105) Google Scholar). However, it is not known whether these methylated dinucleotides are associated with any particular genes. Similar, diverse, non-symmetric sites of methylation have been described previously as “densely methylated islands” at origins of replication associated with ribosomal protein S14 and thedhfr locus in Chinese hamster ovary cells (33Tasheva E. Roufa D. Mol. Cell. Biol. 1994; 14: 5636-5644Crossref PubMed Scopus (53) Google Scholar). However, it has been suggested that these densely methylated islands of methylated cytosine residues might be artifactual, arising from incomplete denaturation of the DNA prior to treatment with sodium bisulfite, at least at the dhfr locus (34Rein T. Natale D. Gartner U. Niggemann M. DePamphilis M. Zorbas H. J. Biol. Chem. 1997; 272: 10021-10029Abstract Full Text Full Text PDF PubMed Scopus (22) Google Scholar). In the present experiments, it is unlikely that the unusual pattern of methylation arises from incomplete denaturation of the DNA. Neither of two closely related cell lines, Rama 37-E8 nor Rama 800, showed the same pattern of protection of cytosine bases as Rama 37CL-A3 cells, even though treatment of their DNA with sodium bisulfite was carried out under identical conditions and at the same time as the DNA from the Rama 37CL-A3 cells. The lack of protection of cytosines in the upstream region of the DNA from the Rama 37CL-A3 cells further argues against a generalized technical failure to denature the DNA prior to reaction with sodium bisulfite. However, it is possible that the DNA in the TATA box and intron regions of the S100A4(p9Ka) gene in the Rama 37CL-A3 cells is in some way different from that isolated from the Rama 37-E8 and Rama 800 cells, such that it is not completely denatured under conditions that denature the DNA from Rama 37-E8 and Rama 800 cells (24Barraclough R. Kimbell R. Rudland P.S. J. Cell. Physiol. 1987; 131: 393-401Crossref PubMed Scopus (29) Google Scholar). Based on the derepressive effect of methylation inhibitors on S100A4(p9Ka) mRNA levels and the results of the digestions withMspI and HpaII, the reduced accessibility of cytosines to bisulfate in the Rama 37CL-A3 DNA is likely to be associated with methylation. However, it cannot be ruled out that the reduced accessibility of cytosine bases to sodium bisulfite in the Rama 37CL-A3 cells is a consequence of CpG methylation outside the regions of the genome analyzed in the present experiments. The region surrounding the upstream cis-acting inhibitory sequence of the S100A4(p9Ka) gene (12Chen D. Davies M. Rudland P. Barraclough R. J. Biol. Chem. 1997; 272: 20283-20290Abstract Full Text Full Text PDF PubMed Scopus (20) Google Scholar) exhibits little or no protection of cytosine bases in all three cell lines. In the Rama 37CL-A3 cells, there is a clear contrast between the density of protected cytosine bases observed in this region (0/39) and the density of protected cytosine bases in the TATA box (21/27) and intron regions (16/16) of the rat S100A4(p9Ka) gene. It is possible that the unprotected nature of this region of the gene in all three cell lines is associated with the binding of the inhibitory GC factor (12Chen D. Davies M. Rudland P. Barraclough R. J. Biol. Chem. 1997; 272: 20283-20290Abstract Full Text Full Text PDF PubMed Scopus (20) Google Scholar). In the Rama 37-E8 and Rama 800 cells, the level of GC factor mRNA is lower than in the Rama 37CL-A3 cells (12Chen D. Davies M. Rudland P. Barraclough R. J. Biol. Chem. 1997; 272: 20283-20290Abstract Full Text Full Text PDF PubMed Scopus (20) Google Scholar). Treatment of Rama 37CL-A3 cells with demethylating agents, but not an inactive analogue, reduces the number of methylcytosine residues in theS100A4(p9Ka) gene region of the Rama 37CL-A3 cells and causes an approximately 5–6-fold increase in the steady state level of S100A4(p9Ka). However, although the endogenous S100A4(p9Ka) mRNA in the Rama 37CL-A3 cells can be increased to a certain extent, neither of the two demethylating agents was able to increase the level of this mRNA to more than half that of its relative level in the Rama 800 cells. Since the inhibitors have no effect on S100A4(p9Ka) mRNA level in at least one high-expressing cell line (Rama 800), it is possible that this lack of full induction represents the presence of inhibitory levels of GC factor which serve to limit the expression of the S100A4(p9Ka) gene in Rama 37CL-A3 cells, and suggests that methylation of positive regulatory elements cooperates with an inhibitory factor to completely inhibit the production of S100A4 mRNA. Altered levels of DNA methyltransferases and/or patterns of methylation are thought to contribute to the formation of cancer. Since it is likely that cytosine methylation of theS100A4(p9Ka) gene plays a role in the suppression of transcription of rat and mouse S100A4(p9Ka), and in view of the metastasis-inducing properties of human S100A4 (35Lloyd B. Platt-Higgins A. Rudland P. Barraclough R. Oncogene. 1998; 17: 465-473Crossref PubMed Scopus (101) Google Scholar), it is possible that inadvertent hypomethylation of theS100A4(p9Ka) gene in human breast epithelial cells might contribute to the malignant progression of some breast cancers. We thank the Cancer and Polio Research Fund for providing running expenses to enable this work to be carried out. We thank Maureen Wilde for photography.
Referência(s)