- Email: [email protected]
Abnormal methylation of estrogen receptor gene and reduced estrogen receptor RNA levels in human endometrial carcinomas
002%4731/89 $3.00 + 0.00 Copyright 0 1989 Pergamon Press plc
J.sreroid Biochem. Vol.32,No. lA,pp.14, 1989 Printed in Great Britain. All rights reserved
ABNORMAL METHYLATION OF ESTROGEN RECEPTOR GENE AND REDUCED ESTROGEN RECEPTOR RNA LEVELS IN HUMAN ENDOMETRIAL CARCINOMAS ROBERTA PIvAt, VIJAY L. KUMARII, STEFANIAHANAUt, ANNA PAOLA RIMONDI~, SAVERIO PANSIN& GIOACCHINO MOLLICA~ and LAURA DEL SENNo*t' *Centro Studi Biochimici delle patologie del genoma Umano, TInstitute of Biochemistry and IInstitute of Obstetrics and Gynecology, University of Ferrara, $Division of Anatomopathology, Hospital of Rovigo, Italy and 11 Laboratoire de GBnttique Moltculaire des Eucaryotes du CNRS, Institut de Chimie Biologique, 67085 Strasbourg Cedex, France (Received 14 May 1988) Summary-Demethylation of specific sites or restricted genomic regions has been reported to correlate with gene activation and also with carcinogenesis. As abnormal expression of Estrogen Receptor (ER) could be involved in the genesis or progression of tumors in estrogen target tissues, the methylation of ER gene has been compared in 8 endometrial carcinomas and 29 normal endometria. In order to look for a correlation between methylation and expression, levels of ER
RNA were also measured. While the 5’ region of ER gene was found to be demethylated in both normal and carcinomatous tissues, there was demethylation of some specific sites in the internal part of the gene only in the carcinomas examined. In addition, in the carcinomatous tissues the levels of ER RNA were low, indicating that an increase of ER gene hypomethylation does not raise, and even may reduce, the ER expression in endometrium. The abnormal undermethylation observed in ER gene appears to be unrelated to genera1 DNA hypomethylation which is frequently present in neoplastic tissues; nor has it been found in ER DNA isolated from breast carcinomas. These data strongly support the hypothesis that such a methylation is specifically related to endometrial
transformation and therefore it can be considered an additional marker of this disease.
such a pattern changes with the endometrial transformation. In addition, levels of ER-RNA in various endometrial samples have been measured to see whether a correlation exists between ER DNA methylation, expression and cell transformation.
lNTRODUCl'lON A variety of changes in the genome has been reported
in abnormal cell growth [l, 21. These alterations include point mutations, deletions, chromosomal translocations, amplifications and variations in DNA methylation [ 141. DNA methylation is a DNA modification that occurs at cytosine level, mostly distributed in CpG doublets [S]. It is a dynamic process that changes depending on the DNA region, cell type and differentiation [6]. Experimental data suggest that hypomethylation of a specific DNA region is associated with transcriptional activity [S, 61. In addition, general and gene specific hypomethylation have been reported to correlate closely with carcinogenesis [3,4]. This paper deals with the methylation and expression of the Estrogen Receptor (ER) DNA in neoplastic endometria. Recently, we studied the methylation pattern of the human estrogen receptor DNA in normal endometria and in this tissue the gene appeared hypomethylated in its 5’ portion, but not in its internal part [7]. We have investigated whether
EXPERIMENTAL Tissue samples
The study included 42 biopsy specimens histologically confirmed as proliferative (14), secretive (15), hyperplastic (5) and adenocarcinomatous endometria(8). They were frozen in liquid nitrogen and stored at -80°C. DNA and RNA preparation Tissue cytoplasms and nuclei were obtained as described by White and Bancroft[8]. DNA was obtained from proteinase K treated nuclei by phenol
DNA was cleared, 10 fig, with restriction enzymes in the incubation conditions suggested by the manufacturers (Biolabs). DNA restriction fragments were electrophoresed on agarose gel 0.8%, and transferred to nitrocellulose filters as described [9].
‘Correspondence to: Laura de1 Senno, Istituto di Chimica Biologica, Universitl 44100 Ferrara, Italy.
degli Studi, V. L. Borsari 46, I
ROBERTA
E PGHER
ER
?
PBcap
I
cDNA
pOR3
,
,
e
1Kb
PI
ai.
MspI can cleave when the internal cytosine is methylated (CYZGG) [3, 1I]. Differences between Hpall and Mspf restriction patterns are therefore a measure of the extent of CCGG methylation. The ER specific DNA restriction fragments were identified by using both a cDNA probe and a 5’ genomic probe, shown in Fig. I. The MspI and HpaII restriction pattern of the ER DNA in different human endometrial samples are shown in Fig. 2. When hybridized to the 5’ probe, four DNA fragments of 1.6, 0.7, 0.5 and 0.3 Kb in size were detected after MspI digestion of all DNA samples examined. The same pattern was observed after HpaII digestion of normal and neoplastic DNAs (Fig. 2A). The same results were found in all the endometrial samples investigated which indicates that the CCGG sites of the 5’ region of the ER gene are extensively demethylated in both normal and neoplastic endometrium. After hybridization to the cDNA probe which recognizes internal gene sequences, four ER specific DNA fragments of 11, 5.8, 3.5 and 1.6 Kb in size were generated by MspI in endometrium (Fig. 2B) and in all tissues examined (data not shown). The 1.6 Kb ER DNA fragment is different from that identified by the pGHER 1 and is located inside the gene, as previously reported [7]. Hpaii digestion of endometrium DNA gave rise to some fragments larger in size than MspI bands. Only in the neoplastic samples was the 1.6 Kb band found and this finding was observed in all the 8 adencarcinomas examined (data not shown). In the two hyperplastic endometria examined no differences from normal were observed. The ER sequences recognized by the cDNA probe are therefore only partially demethylated in normal endometrium, but are undermethylated to a greater degree in adenocarcinoma samples. Similar conclusions were reached by analysing HpaII digestions performed in combination with PvuII and BamHI restriction enzymes (data not shown). In order to test whether hypomethylation of ER
E
.............. I Ilk-----l
1
&VA
Fig. 1. Map of HpaII~MspI (0) and HhaI (Cl) restriction sites in the human ER cDNA sequence. The enzymes recognize the CCGG and GCGC respectively [I 11. Coding region is indicated as a box. The pOR3 cDNA and pGHER1 genomic probes are included. PvuII (P), BamHI (B) and EcoRI (E) restriction sites are reported in the genomic probe. Nucleic acid hybridization
Southern blots were hybridized in formamidedextran sulphate mixture [l 1] containing 32P multiprime labelled cDNA and 5’ estrogen receptor probes (pOR3 and pGHER1) [7,10] and genomic Ha-rasl oncogene probe (T24C3) [I I]. Cytoplasmic dot hybridization
Scalar dilutions of the cytoplasmic samples were spotted onto nitrocellulose filters after formaldehyde denaturation [8]. Filters were hybridized to estrogen receptor cDNA and to ribosomal RNA probes pXCR7 [12] respectively. The relative amount of the two RNAs was measured by spectrophotometric scanning of the autoradiographic signals. RESULTS
The methylation of ER gene has been evaluated by using the HpaII and Mspi restriction enzymes. Both of them recognize the sequence CCGG, but only B
A pGHER
C Ha-m-l
oOR3
1
r-.
-
123 It 5.8 -
3.5 -
ll-
4
-12
6-
-6 2"
3.5-2.5
1.6 -
Mspl
HpolI
1.6-
Mspl
HpolI
Mspl
HpoIl
Fig. 2. Methylation of HpaII sites in the ER gene. Southern blots of one MspI digest and HpaII digests from endometrial carcinomas [l-3], proliferative phase endometrium [4] secretory phase endometrium [5] and hyperplastic endometrium [6] after hybridization with the pGHER1 probe (A) and with the pOR3 probe (B). (C) Methylation pattern of Ha-ras-1 oncogene in three endometrial carcinomas [l-3] and in a normal endometrium [4]. The hybridization was performed with the T24-C3 probe. Different electrophoretic migrations are shown with their relative Kb sizes.
-to -6.:
Methylation of ER gene in endometrial cancers pGHER 1
2
1
pOR 3 -3
1
2
3
Cl1
1.6 -6
-3.5 0.7
0.5 0.3 -1.6
HpaII
HpalI
Fig. 3. Methylation of HpaII sites in the 5’ region (pGHER1) and inside the ER gene (pOR3). Southern blots of HpaII digests from normal white blood cells [l], breast
carcinoma [2] and endometrial carcinoma [3]. DNA observed in endometrial carcinomas is also present in a different gene, the methylation status of the Ha-ras-1 gene, found to be demethylated in other tumors [ 13-151, was studied. After hybridization of HpaII digested DNAs (Fig. 2C), various methylation patterns, even lower than normal, were observed in the three carcinomas examined. In addition, they were different from those expected in a highly hypomethylated state. As the hypomethylation of ER gene could be related only to the carcinogenesis of estrogen target tissues, the HpaII pattern of ER gene in breast and endometrial carcinomas was compared. Different restriction patterns were found in breast and endometrial samples, both in the 5’ and internal region of ER gene (line 2 and 3 of Fig. 3). This latter region appeared highly methylated in the breast carcinoma as well as in the wbc from the same subject (line 1 of Fig. 3). The same results were observed in various wbc and breast carcinoma samples (data not shown). As DNA methylation has been correlated with expression in several genei [S, 6, 161, we studied ER RNA levels in the normal and neoplastic endometrial samples. ER gene expression was measured by cytoplasmic dot blot hybridizations, as shown in Fig. 4. ER RNA levels in each sample were normalized for their ribosomal RNA content. The distribution of ER RNA levels appeared to be different in normal and in neoplastic samples. Values were highest in the proliferative cases (very high in only two samples), reduced to approximately half in secretory phase endometria, and reduced even more in adenocarcinomas. DISCUSSION
The extent of DNA methylation and expression of the ER gene was measured in a variety of endometrial
3
conditions. The methylation analysis with the MspI and HpaII restriction enzymes shows that specific CCGG sites internal to the gene are undermethylated only in neoplastic samples (Fig. 2B). This finding could be compatible with an extensive DNA hypomethylation in such carcinomas as is found in several neoplastic tissues and cell lines [3,4, 13-15, 171. However, we have shown that in these endometrial carcinomas the extensive undermethylation of ER gene, is not apparent in a different DNA region. In particular, in the same samples, Ha-ras-1 DNA, in which there is low methylation in several cancers [13-151, does not appear to be undermethylated to same extent. In addition, this abnormal hypomethylation in the internal part of ER gene is specifically related to endometrial and not to other adenocarcinomas, including breast, as shown in line 2 of Fig. 3. In the latter neoplastic tissues we did find hypomethylation, but restricted to the 5’ flanking regions of the gene (manuscript in preparation). No demethylation is observed in tissues which are not target for estrogen such as white blood cells (line 1 of
A p
SHC
B
P
S
H
C
Fig. 4. (A) Dot-spot hybridization of formaldehyde-treated cytoplasmic preparations from neoplastic (C), hyperplastic (H) proliferative phase (P) and secretory phase- @-endometrium with DORA and nXCR7 orobes. The dilutions used (1, l/2, l/4. .\ are indicated. (B)i’he autoradiograms were scanned through a spectrophotometer and the pOR3 values were expressed relative to the rRNA content (pOR3/pXCR7).
ROBERTAPIVA et al
A
Fig. 3). Overall this result favours a possible role of methylation in the control of steroid receptor genes which are reported to be finely regulated [18]. In previous studies of normal tissue ER DNA methylation was lower in endometrium, in which the gene is maximally expressed, than in breast [7]. This suggested that ER gene expression is inversely related to methylation. Therefore it was expected that the high level of undermethylation observed in neoplastic endometria samples would be associated with an increase in ER gene expression. However, the opposite was found, the hypomethylation present in neoplastic sample is not associated with high gene expression, and in fact, what is more, in these samples the expression is reduced (Fig. 4). As the map of CCGG sites inside the genomic sequences is not yet available, we cannot say whether the regions containing the HpaII demethylated sites are important or not for the ER gene expression. Nevertheless, the low levels of ER RNA in tumors, either related simply to a loss of differentiative features occurring in tumors or to abnormal ER gene methylation, indicate deregulation of the ER gene expression. Concerning this issue it has been published that in carcinomatous cells the distribution of ER is highly heterogeneous [ 191. This could be due either to a cell cycle restricted expression of ER gene or to activation in a sub-population of cells of a ER-independent growth associated to a decrease of ER RNA content. In conclusion, the results presented here provide evidence that the extensive hypomethylation of ER gene is a DNA modification related to endometrial carcinoma and that it could be considered as an additional marker of endometrial transformation. Endometrial hyperplasia and dysplasia frequently develop into autonomous cancer growth; despite the fact that we could not find differences from normal in the two hyperplastic samples examined, it might be of interest to look for changes in ER DNA methylation in preneoplastic conditions which might foretell the possibility of malignant change. Acknowledgements-We thank D. Buzzoni for preparing the labelled probes, R. Gambari for helpful suggestions and B. Anderson for checking the manuscript. pOR3 has been kindly provided by P. Chambon and the pXCR7 by I. Bozzoni. This work is supported by grants from Regione Veneto, from Ricerca Finalizzata della Regione EmiliaRomagna and from Progetto Finalizzato Oncologia No. 87.01260.44.
REFERENCES
1. Nishimura S. and Sekiya T.: Human cancer and cellular oncogenes. Biochem. J. 243 (1987) 313-327. 2. Bishop J. M.: Cellular oncogenes and retroviruses. A. Rev. Biochem. 52 (1983) 301-354.
3. Feimberg A. P. and Vogelstein B.: Hypomcthylation distinguishes genes of some human cancers from their normal counterparts. Nature 301 (1983) 89-92. D.: Alteration of 4. Boehm T. L. J. and Drahovskv enzymatic methylation of DNA cytosines by chemical carcinogens: a mechanism involved in the initiation of carcinogenesis. J. natn Cancer Inst. 71 (1983) 4291133. 5. Doerfler W.: DNA methylation and gene activity. A. Reo. Biochem. (1983) 93~-124. expression and 6. Jaenisch R. and Jahner D.: Methylation, chromosomal position of genes in mammals. Biochem. biophys. Acta 782 (1984) 1-9. 7 Piva R., Kumar V. L., Hanau S., Maestri I., Rimondi A. P., Luppi L., Pansini F. S.. Mollica G., Chambon P. and del Senno L.: Expression linked undermethylation of CpG sites in the human estrogen receptor gene. Eur. J. Biochem. Submitted for publication. F. C.: Cytoplasmic dot 8 White B. A. and Bancroft hybridization. Simple analysis of relative mRNA levels in multiple small cell or tissue samples. J. biol. Chem. 257 (1982) 8569-8572. J.: Molecular 9 Maniatis T., Fritsh E. F. and Sambrook Cloning. A Laborafory Manual. Cold Spring Harbor Laboratory Publications, New York (1982). 10 Green S., Walter P., Kumar V., Krust A., Bornert J. M., Argos P. and Chambon P.: Human oestrogen receptor cDNA: sequence, expression and homology to v-erb-A. Nature 320 (1896) 134-139. I1 Santos E., Tronick S. R., Aaronson S. A.. Pulciani S. and Barbacid M. : T24 human bladder carcinoma oncogene is an activated form of the normal human homologue of BALB- and Harvey-MSV transforming genes. Nature 298 (1982) 343~ 347. 12 del Senno L.. Degli Uberti E., Rossi M., Buzzoni D.. Barbieri R., Rossi P.. Patracchini P., Bernardi F., Marchetti G., Conconi F. and Gambari R.: Identification of a c-myc oncogene lacking the exon 1 in the normal cells of a patient carrying a-thyroid carcinoma. FEBS Le/l. 1% (1986) 296300. 13 Feinberg A. P. and Vogelstein B.: Hypomethylation of ras oncogenes in primary human cancers. Biochem. biophys. Res. Commun. 111 (1983) 47754. 14 Piva R., Barbieri R., Castagnoli A., Rimondi A. P., de1 Senno L., Luppi L. and Gambari R.: Methylation of human Ha-ras-I oncogene in primary tumors, tumor cell lines and normal white blood cells. Eur. J. Cell Biol. 42 (1986) 25. 15 Barbieri R., Piva R., Buzzoni D., Volinia S. and Gambari R.: Clustering of undermethylated CCGG and GCGC sequences in the 5’ region of the Ha-ras-1 oncogene of human leukemic KS62 cells. Biochem. biophys. Res. Commun. 145 (1987) 96104. 16. Philipsen J. N. J., Gruber M. and Geert A. B.: Expression-linked demethylation of 5-methylcytosines in the chicken vitellogenin gene region. Biochim. biophys. Acfa 826 (1985) 186-194. 17. de1 Senno L., degli Uberti E., Bernardi F., Marchetti G., Perrotta C.. Gambari R.. Barbieri R.. Buzzoni D.. Pansini G. and Conconi F.: c-myc oncogene alterations in human thyroid carcinomas. Cancer Detect. 10 (1987) 1599166. 18. Beato M.: Induction of transcription by steroid hormones. Biochim. biophys. Acta 910 (1987) 95-102. 19. Castagnetta L., Lo Cast M., Mercadante T., Polite L., Cowan S. and Leake R. E.: Intratumoral variation of estrogen receptor status in endometrial cancer. Br. J. Cancer 42 (1983) 261-265.
J.sreroid Biochem. Vol.32,No. lA,pp.14, 1989 Printed in Great Britain. All rights reserved
ABNORMAL METHYLATION OF ESTROGEN RECEPTOR GENE AND REDUCED ESTROGEN RECEPTOR RNA LEVELS IN HUMAN ENDOMETRIAL CARCINOMAS ROBERTA PIvAt, VIJAY L. KUMARII, STEFANIAHANAUt, ANNA PAOLA RIMONDI~, SAVERIO PANSIN& GIOACCHINO MOLLICA~ and LAURA DEL SENNo*t' *Centro Studi Biochimici delle patologie del genoma Umano, TInstitute of Biochemistry and IInstitute of Obstetrics and Gynecology, University of Ferrara, $Division of Anatomopathology, Hospital of Rovigo, Italy and 11 Laboratoire de GBnttique Moltculaire des Eucaryotes du CNRS, Institut de Chimie Biologique, 67085 Strasbourg Cedex, France (Received 14 May 1988) Summary-Demethylation of specific sites or restricted genomic regions has been reported to correlate with gene activation and also with carcinogenesis. As abnormal expression of Estrogen Receptor (ER) could be involved in the genesis or progression of tumors in estrogen target tissues, the methylation of ER gene has been compared in 8 endometrial carcinomas and 29 normal endometria. In order to look for a correlation between methylation and expression, levels of ER
RNA were also measured. While the 5’ region of ER gene was found to be demethylated in both normal and carcinomatous tissues, there was demethylation of some specific sites in the internal part of the gene only in the carcinomas examined. In addition, in the carcinomatous tissues the levels of ER RNA were low, indicating that an increase of ER gene hypomethylation does not raise, and even may reduce, the ER expression in endometrium. The abnormal undermethylation observed in ER gene appears to be unrelated to genera1 DNA hypomethylation which is frequently present in neoplastic tissues; nor has it been found in ER DNA isolated from breast carcinomas. These data strongly support the hypothesis that such a methylation is specifically related to endometrial
transformation and therefore it can be considered an additional marker of this disease.
such a pattern changes with the endometrial transformation. In addition, levels of ER-RNA in various endometrial samples have been measured to see whether a correlation exists between ER DNA methylation, expression and cell transformation.
lNTRODUCl'lON A variety of changes in the genome has been reported
in abnormal cell growth [l, 21. These alterations include point mutations, deletions, chromosomal translocations, amplifications and variations in DNA methylation [ 141. DNA methylation is a DNA modification that occurs at cytosine level, mostly distributed in CpG doublets [S]. It is a dynamic process that changes depending on the DNA region, cell type and differentiation [6]. Experimental data suggest that hypomethylation of a specific DNA region is associated with transcriptional activity [S, 61. In addition, general and gene specific hypomethylation have been reported to correlate closely with carcinogenesis [3,4]. This paper deals with the methylation and expression of the Estrogen Receptor (ER) DNA in neoplastic endometria. Recently, we studied the methylation pattern of the human estrogen receptor DNA in normal endometria and in this tissue the gene appeared hypomethylated in its 5’ portion, but not in its internal part [7]. We have investigated whether
EXPERIMENTAL Tissue samples
The study included 42 biopsy specimens histologically confirmed as proliferative (14), secretive (15), hyperplastic (5) and adenocarcinomatous endometria(8). They were frozen in liquid nitrogen and stored at -80°C. DNA and RNA preparation Tissue cytoplasms and nuclei were obtained as described by White and Bancroft[8]. DNA was obtained from proteinase K treated nuclei by phenol
DNA was cleared, 10 fig, with restriction enzymes in the incubation conditions suggested by the manufacturers (Biolabs). DNA restriction fragments were electrophoresed on agarose gel 0.8%, and transferred to nitrocellulose filters as described [9].
‘Correspondence to: Laura de1 Senno, Istituto di Chimica Biologica, Universitl 44100 Ferrara, Italy.
degli Studi, V. L. Borsari 46, I
ROBERTA
E PGHER
ER
?
PBcap
I
cDNA
pOR3
,
,
e
1Kb
PI
ai.
MspI can cleave when the internal cytosine is methylated (CYZGG) [3, 1I]. Differences between Hpall and Mspf restriction patterns are therefore a measure of the extent of CCGG methylation. The ER specific DNA restriction fragments were identified by using both a cDNA probe and a 5’ genomic probe, shown in Fig. I. The MspI and HpaII restriction pattern of the ER DNA in different human endometrial samples are shown in Fig. 2. When hybridized to the 5’ probe, four DNA fragments of 1.6, 0.7, 0.5 and 0.3 Kb in size were detected after MspI digestion of all DNA samples examined. The same pattern was observed after HpaII digestion of normal and neoplastic DNAs (Fig. 2A). The same results were found in all the endometrial samples investigated which indicates that the CCGG sites of the 5’ region of the ER gene are extensively demethylated in both normal and neoplastic endometrium. After hybridization to the cDNA probe which recognizes internal gene sequences, four ER specific DNA fragments of 11, 5.8, 3.5 and 1.6 Kb in size were generated by MspI in endometrium (Fig. 2B) and in all tissues examined (data not shown). The 1.6 Kb ER DNA fragment is different from that identified by the pGHER 1 and is located inside the gene, as previously reported [7]. Hpaii digestion of endometrium DNA gave rise to some fragments larger in size than MspI bands. Only in the neoplastic samples was the 1.6 Kb band found and this finding was observed in all the 8 adencarcinomas examined (data not shown). In the two hyperplastic endometria examined no differences from normal were observed. The ER sequences recognized by the cDNA probe are therefore only partially demethylated in normal endometrium, but are undermethylated to a greater degree in adenocarcinoma samples. Similar conclusions were reached by analysing HpaII digestions performed in combination with PvuII and BamHI restriction enzymes (data not shown). In order to test whether hypomethylation of ER
E
.............. I Ilk-----l
1
&VA
Fig. 1. Map of HpaII~MspI (0) and HhaI (Cl) restriction sites in the human ER cDNA sequence. The enzymes recognize the CCGG and GCGC respectively [I 11. Coding region is indicated as a box. The pOR3 cDNA and pGHER1 genomic probes are included. PvuII (P), BamHI (B) and EcoRI (E) restriction sites are reported in the genomic probe. Nucleic acid hybridization
Southern blots were hybridized in formamidedextran sulphate mixture [l 1] containing 32P multiprime labelled cDNA and 5’ estrogen receptor probes (pOR3 and pGHER1) [7,10] and genomic Ha-rasl oncogene probe (T24C3) [I I]. Cytoplasmic dot hybridization
Scalar dilutions of the cytoplasmic samples were spotted onto nitrocellulose filters after formaldehyde denaturation [8]. Filters were hybridized to estrogen receptor cDNA and to ribosomal RNA probes pXCR7 [12] respectively. The relative amount of the two RNAs was measured by spectrophotometric scanning of the autoradiographic signals. RESULTS
The methylation of ER gene has been evaluated by using the HpaII and Mspi restriction enzymes. Both of them recognize the sequence CCGG, but only B
A pGHER
C Ha-m-l
oOR3
1
r-.
-
123 It 5.8 -
3.5 -
ll-
4
-12
6-
-6 2"
3.5-2.5
1.6 -
Mspl
HpolI
1.6-
Mspl
HpolI
Mspl
HpoIl
Fig. 2. Methylation of HpaII sites in the ER gene. Southern blots of one MspI digest and HpaII digests from endometrial carcinomas [l-3], proliferative phase endometrium [4] secretory phase endometrium [5] and hyperplastic endometrium [6] after hybridization with the pGHER1 probe (A) and with the pOR3 probe (B). (C) Methylation pattern of Ha-ras-1 oncogene in three endometrial carcinomas [l-3] and in a normal endometrium [4]. The hybridization was performed with the T24-C3 probe. Different electrophoretic migrations are shown with their relative Kb sizes.
-to -6.:
Methylation of ER gene in endometrial cancers pGHER 1
2
1
pOR 3 -3
1
2
3
Cl1
1.6 -6
-3.5 0.7
0.5 0.3 -1.6
HpaII
HpalI
Fig. 3. Methylation of HpaII sites in the 5’ region (pGHER1) and inside the ER gene (pOR3). Southern blots of HpaII digests from normal white blood cells [l], breast
carcinoma [2] and endometrial carcinoma [3]. DNA observed in endometrial carcinomas is also present in a different gene, the methylation status of the Ha-ras-1 gene, found to be demethylated in other tumors [ 13-151, was studied. After hybridization of HpaII digested DNAs (Fig. 2C), various methylation patterns, even lower than normal, were observed in the three carcinomas examined. In addition, they were different from those expected in a highly hypomethylated state. As the hypomethylation of ER gene could be related only to the carcinogenesis of estrogen target tissues, the HpaII pattern of ER gene in breast and endometrial carcinomas was compared. Different restriction patterns were found in breast and endometrial samples, both in the 5’ and internal region of ER gene (line 2 and 3 of Fig. 3). This latter region appeared highly methylated in the breast carcinoma as well as in the wbc from the same subject (line 1 of Fig. 3). The same results were observed in various wbc and breast carcinoma samples (data not shown). As DNA methylation has been correlated with expression in several genei [S, 6, 161, we studied ER RNA levels in the normal and neoplastic endometrial samples. ER gene expression was measured by cytoplasmic dot blot hybridizations, as shown in Fig. 4. ER RNA levels in each sample were normalized for their ribosomal RNA content. The distribution of ER RNA levels appeared to be different in normal and in neoplastic samples. Values were highest in the proliferative cases (very high in only two samples), reduced to approximately half in secretory phase endometria, and reduced even more in adenocarcinomas. DISCUSSION
The extent of DNA methylation and expression of the ER gene was measured in a variety of endometrial
3
conditions. The methylation analysis with the MspI and HpaII restriction enzymes shows that specific CCGG sites internal to the gene are undermethylated only in neoplastic samples (Fig. 2B). This finding could be compatible with an extensive DNA hypomethylation in such carcinomas as is found in several neoplastic tissues and cell lines [3,4, 13-15, 171. However, we have shown that in these endometrial carcinomas the extensive undermethylation of ER gene, is not apparent in a different DNA region. In particular, in the same samples, Ha-ras-1 DNA, in which there is low methylation in several cancers [13-151, does not appear to be undermethylated to same extent. In addition, this abnormal hypomethylation in the internal part of ER gene is specifically related to endometrial and not to other adenocarcinomas, including breast, as shown in line 2 of Fig. 3. In the latter neoplastic tissues we did find hypomethylation, but restricted to the 5’ flanking regions of the gene (manuscript in preparation). No demethylation is observed in tissues which are not target for estrogen such as white blood cells (line 1 of
A p
SHC
B
P
S
H
C
Fig. 4. (A) Dot-spot hybridization of formaldehyde-treated cytoplasmic preparations from neoplastic (C), hyperplastic (H) proliferative phase (P) and secretory phase- @-endometrium with DORA and nXCR7 orobes. The dilutions used (1, l/2, l/4. .\ are indicated. (B)i’he autoradiograms were scanned through a spectrophotometer and the pOR3 values were expressed relative to the rRNA content (pOR3/pXCR7).
ROBERTAPIVA et al
A
Fig. 3). Overall this result favours a possible role of methylation in the control of steroid receptor genes which are reported to be finely regulated [18]. In previous studies of normal tissue ER DNA methylation was lower in endometrium, in which the gene is maximally expressed, than in breast [7]. This suggested that ER gene expression is inversely related to methylation. Therefore it was expected that the high level of undermethylation observed in neoplastic endometria samples would be associated with an increase in ER gene expression. However, the opposite was found, the hypomethylation present in neoplastic sample is not associated with high gene expression, and in fact, what is more, in these samples the expression is reduced (Fig. 4). As the map of CCGG sites inside the genomic sequences is not yet available, we cannot say whether the regions containing the HpaII demethylated sites are important or not for the ER gene expression. Nevertheless, the low levels of ER RNA in tumors, either related simply to a loss of differentiative features occurring in tumors or to abnormal ER gene methylation, indicate deregulation of the ER gene expression. Concerning this issue it has been published that in carcinomatous cells the distribution of ER is highly heterogeneous [ 191. This could be due either to a cell cycle restricted expression of ER gene or to activation in a sub-population of cells of a ER-independent growth associated to a decrease of ER RNA content. In conclusion, the results presented here provide evidence that the extensive hypomethylation of ER gene is a DNA modification related to endometrial carcinoma and that it could be considered as an additional marker of endometrial transformation. Endometrial hyperplasia and dysplasia frequently develop into autonomous cancer growth; despite the fact that we could not find differences from normal in the two hyperplastic samples examined, it might be of interest to look for changes in ER DNA methylation in preneoplastic conditions which might foretell the possibility of malignant change. Acknowledgements-We thank D. Buzzoni for preparing the labelled probes, R. Gambari for helpful suggestions and B. Anderson for checking the manuscript. pOR3 has been kindly provided by P. Chambon and the pXCR7 by I. Bozzoni. This work is supported by grants from Regione Veneto, from Ricerca Finalizzata della Regione EmiliaRomagna and from Progetto Finalizzato Oncologia No. 87.01260.44.
REFERENCES
1. Nishimura S. and Sekiya T.: Human cancer and cellular oncogenes. Biochem. J. 243 (1987) 313-327. 2. Bishop J. M.: Cellular oncogenes and retroviruses. A. Rev. Biochem. 52 (1983) 301-354.
3. Feimberg A. P. and Vogelstein B.: Hypomcthylation distinguishes genes of some human cancers from their normal counterparts. Nature 301 (1983) 89-92. D.: Alteration of 4. Boehm T. L. J. and Drahovskv enzymatic methylation of DNA cytosines by chemical carcinogens: a mechanism involved in the initiation of carcinogenesis. J. natn Cancer Inst. 71 (1983) 4291133. 5. Doerfler W.: DNA methylation and gene activity. A. Reo. Biochem. (1983) 93~-124. expression and 6. Jaenisch R. and Jahner D.: Methylation, chromosomal position of genes in mammals. Biochem. biophys. Acta 782 (1984) 1-9. 7 Piva R., Kumar V. L., Hanau S., Maestri I., Rimondi A. P., Luppi L., Pansini F. S.. Mollica G., Chambon P. and del Senno L.: Expression linked undermethylation of CpG sites in the human estrogen receptor gene. Eur. J. Biochem. Submitted for publication. F. C.: Cytoplasmic dot 8 White B. A. and Bancroft hybridization. Simple analysis of relative mRNA levels in multiple small cell or tissue samples. J. biol. Chem. 257 (1982) 8569-8572. J.: Molecular 9 Maniatis T., Fritsh E. F. and Sambrook Cloning. A Laborafory Manual. Cold Spring Harbor Laboratory Publications, New York (1982). 10 Green S., Walter P., Kumar V., Krust A., Bornert J. M., Argos P. and Chambon P.: Human oestrogen receptor cDNA: sequence, expression and homology to v-erb-A. Nature 320 (1896) 134-139. I1 Santos E., Tronick S. R., Aaronson S. A.. Pulciani S. and Barbacid M. : T24 human bladder carcinoma oncogene is an activated form of the normal human homologue of BALB- and Harvey-MSV transforming genes. Nature 298 (1982) 343~ 347. 12 del Senno L.. Degli Uberti E., Rossi M., Buzzoni D.. Barbieri R., Rossi P.. Patracchini P., Bernardi F., Marchetti G., Conconi F. and Gambari R.: Identification of a c-myc oncogene lacking the exon 1 in the normal cells of a patient carrying a-thyroid carcinoma. FEBS Le/l. 1% (1986) 296300. 13 Feinberg A. P. and Vogelstein B.: Hypomethylation of ras oncogenes in primary human cancers. Biochem. biophys. Res. Commun. 111 (1983) 47754. 14 Piva R., Barbieri R., Castagnoli A., Rimondi A. P., de1 Senno L., Luppi L. and Gambari R.: Methylation of human Ha-ras-I oncogene in primary tumors, tumor cell lines and normal white blood cells. Eur. J. Cell Biol. 42 (1986) 25. 15 Barbieri R., Piva R., Buzzoni D., Volinia S. and Gambari R.: Clustering of undermethylated CCGG and GCGC sequences in the 5’ region of the Ha-ras-1 oncogene of human leukemic KS62 cells. Biochem. biophys. Res. Commun. 145 (1987) 96104. 16. Philipsen J. N. J., Gruber M. and Geert A. B.: Expression-linked demethylation of 5-methylcytosines in the chicken vitellogenin gene region. Biochim. biophys. Acfa 826 (1985) 186-194. 17. de1 Senno L., degli Uberti E., Bernardi F., Marchetti G., Perrotta C.. Gambari R.. Barbieri R.. Buzzoni D.. Pansini G. and Conconi F.: c-myc oncogene alterations in human thyroid carcinomas. Cancer Detect. 10 (1987) 1599166. 18. Beato M.: Induction of transcription by steroid hormones. Biochim. biophys. Acta 910 (1987) 95-102. 19. Castagnetta L., Lo Cast M., Mercadante T., Polite L., Cowan S. and Leake R. E.: Intratumoral variation of estrogen receptor status in endometrial cancer. Br. J. Cancer 42 (1983) 261-265.