Somatic Cell and Molecular Genetics, Vol. 17, No. 1, 1991, pp. 35-47

DNA Methylation of Two X Chromosome Genes in Female Somatic and Embryonai Carcinoma Cells Molly H. Bartlett, 1 Chaker N. Adra, ~Jonggwang Park, 2 Verne M. Chapman, 2 and Michael W. McBurney I IDepartments of Medicine and Biology, Universityof Ottawa, 451 Smyth Road, Ottawa, Ontario, Canada, KIH 8M5; and 2Department of Molecular Biology, Roswell Park Memol~al bTstitute, Buffalo, New York 14263 Received 30 May 1990--Final 20 September 1990

Abstract--The extent of methylation of DNA sequences" upstream and within the two X-linked genes, Pgk-1 and Hprt, was analyzed in male and female somatic cells and in female embryona! carcinoma cells cmTying either two active X chromosomes (Xa) or one active and one inactive X chromosome (Xi). Sites upstream and within the first intron of both Pgk-1 and Hprt were heavily methylated on the Xi in somatic cells and in embryonal carcinoma cells with an Xi. Reactivation of this Xi was accompanied by extensive demethylation of these sites. 1n female embryonal carcinoma cells with two active X chromosomes, one X inactivates during differentiation in culture; however; methylation dM not occur during differentiation, consistent with the idea that DNA methylation does not play a role in the initiation of X inactivation but may be involved in maintaining inactivation of those genes on the Xi.

INTRODUCTION

number of sites in and around X-linked genes are methylated on the Xi but not methylated on the Xa (14-18). Embryonal carcinoma (EC) cells offer useful culture systems with which to investigate the mechanisms of X chromosome inactivation (19). Some female EC lines, such as P10 cells, carry two fully active X chromosomes (20, 21). During differentiation of these cells in culture, one X in each of these ceils inactivates (20, 22). Other female EC lines, such as C86, carry one Xa and one Xi (23). In C86 cells, treatment with DNA demethylating agents can induce expression of genes from the Xi (9, 10). We set out to investigate the relationship between gene activity and methylation in female EC cells whose X chromosomes had been inactivated or reactivated by the above manipulations.

DNA methylation was postulated to play an important role in the process of X chromosome inactivation based on the hereditary nature of this type of DNA modification (1, 2) and on the frequent correlation between gene methylation and inactivity (3, 4). Three lines of evidence are consistent with this model: (1) treatment of humanrodent hybrid cells (5-8) or certain mouse embryonal carcinoma (EC) cells (9, 10) with the DNA demethylating agent 5-azacytidine (5AC) results in expression of certain genes from the inactive X chromosome (Xi), (2) the Hprt gene from the Xi is inactive while the same gene from the active X chromosome (Xa) is active in DNA-mediated gene transfer experiments (6, 11-13), and (3) a 35

0740-7750/91/0100-0035506.50/0© 1991 Plenum Publishing Corporation

36

MATERIALS AND M E T H O D S

Cell Lines and Cell Culture Conditions. The cell culture methods we used are described in Rudnicki and McBurney (24). All cells were cultured in alpha minimal essential medium (Gibco, Mississauga, Canada), which contained 7.5% calf serum and 2.5% fetal calf serum (Bocknek Laboratories Inc., Rexdale, Ontario, Canada). The cells were maintained in plastic tissue culture dishes, which were placed at 37°C in a 5% CO 2 incubator. They were subcultured every 48 h or less by first washing the monolayer in phosphate-buffered saline (PBS) (0.8% NaC1, 0.2% KC1, 0.02% KH2PO4 and 0.115% NaHPO4), and then incubating for 5 min in T-EDTA (1 mM EDTA and 0.025% trypsin in PBS). The cells detached from the dishes and were dispersed by vigorous pipetting. They were then counted and replated at a concentration of 105 cells/ml into fresh medium. A number of different lines of embryonal carcinoma cells were used in these experiments. All of the lines had neardiploid chromosome numbers and karyotypes but differed in the number and genetic activity of their sex chromosomes. P19 cells (25) are XY and served as a control for a single active X chromosome. P10 cells have two active X chromosomes (20). Pl0 (XO) were derived from P10 following the loss of one of the X chromosomes (22). C86AGM2 (10) is a subclone of C86 cells and has one active and one inactive X chromosome (23). Following treatment with 5-azacytidine, clones of C86AGM2 were isolated that express genes on both X chromosomes (10). The clones used in this study are called ACla, AClb, AC2a, AC2b, and AC3b (10). C86 (XO) is a derivative of C86 that had spontaneously lost the Xi. C100 and C145 are two other lines with two X chromosomes in which one is genetically inactive (9).

Bartlett et al.

DNA Isolation and Electrophoresis. Genomic DNA was extracted from embryohal carcinoma cells by a modification of the procedure by Blin and Stafford (26). Cultures were rinsed in PBS and then lysed in a solution of 0.5% SDS, 10 mM NaC1, 10 mM EDTA, and 10 mM Tris HC1 (pH 8.0) containing 100 ~g/ml proteinase K. The ceils were scraped from the dishes and incubated at 37°C for 2 h with gentle shaking. The cell suspension was extracted with phenolchloroform (1:1), treated with 50 ~g/ml RNase and incubated overnight at 37°C. The lysate was extracted again with phenolchloroform and the DNA was precipitated in two volumes of 95% ethanol containing 2% potassium acetate. The DNA was spooled and redissolved in 10 mM Tris HC1 (pH 8.0) and 1 mM EDTA. DNA was extracted from mouse organs by placing the freshly dissected organs in ice-cold PBS, breaking the tissue up using a rubber spatula, and filtering to produce a single cell suspension. These cells were washed several times in PBS and then treated with the method described above. Plasmid DNAs were prepared by a modification of the method described by Birnboim and Doly (27). Restriction fragments of DNAwere isolated by electrophoresis into troughs cut into the agarose gels (28). The diluted DNA fragments were recovered and concentrated by passage through NACS columns (BRL, Gaithersburg, Maryland). Plasmid DNA fragments were labeled for use as hybridization probes by nick translation or by the method of Feinberg and Vogelstein (29). DNA was digested with restriction endonucleases under the conditions recommended by the manufacturers. DNA was usually digested with 5-10 units/p~g of enzyme for at least 6 h. Completeness of the restriction enzyme digestions was tested and confirmed by mixing pSP64 DNA with an aliquot of each reaction mixture. The di-

DNA methylation and X chromosome inactivation

37

region upstream of the human Pgk-1 gene were methylated on the Xi but were unmethylated on the Xa. More recent studies of this region in clones of h u m a n - r o d e n t hybrids in which the human Pgk-1 gene was or was not expressed suggested that demethylation of these and a number of additional sites is correlated with gene activity (31). We recently cloned the mouse Pgk-1 gene and found that the sequence of the upstream regions of the mouse and human genes were very homologous (32). We set out to investigate the degree of methylation of this upstream region in mouse Pgk-1 genes present on the Xa and Xi. A region of approximately 700 bp upstream of exon 1 of the mouse Pgk-1 gene is rich in GC and contains seven MspI or HpaII recognition sites (Fig. 1). To examine the extent of D N A methytation in this region, D N A from the tissue or cells to be tested was digested with BamHI and PvuII together with either MspI or HpaII. Plasmid D N A (pSP64) was mixed with the cell D N A and the profile of plasmid D N A checked on ethidium bromide-stained gels to ensure that all digests were complete. Independent prepRESULTS arations of D N A were tested to verify that DNA Methylation of Pgk-1 Gene the observations were reproducible. Promoter. Keith et al. (30) found that all D N A from digested samples was sepaeight sites they could assess in the CG-rich rated by electrophoresis in agarose gets,

gested D N A was electrophoresed in 1% agarose gels or in a mixture of 4% NuSieve agarose (FMC Rockland, Maine) and 2% B R L agarose (Bethesda Research Laboratories) in Tris acetate E D T A buffer containing 0.2 mg/ml ethidium bromide. The D N A was blotted onto Biodyne membranes as described by Maniatis et al. (28) and crosslinked by U V light. Prehybridization was carried out in 50% formamide, 5 x SSPE, 2.5 x Denhardt's solution, 0.1% SDS, 0.2 mg/ml denatured salmon testis D N A (1 x SSPE is 0.15 M NaC1, 0.01 M Na2H2PO4, 0.001 M EDTA). Hybridization was carried out for 8 h at 42°C in the same solution containing radioactive probes labeled with 32p. Blots were washed at 65°C in 0.1 x SSC-0.1% SDS (1 x SSC is 0.15 M NaCI-0.015 M sodium citrate). Blots were exposed to Kodak XAR-5 film at - 7 0 ° C with two lanex regular intensifying screens. Scanning densitometry was performed on the autoradiograms to assess the proportion of hybridizing material in the bands representing D N A fragments uncut by methylation-sensitive restriction enzymes.

B

Pv

Hha I Hpa ml 425

112

240

193

100 bp

t------4 Fig. 1. Restriction map of the 1106-bp BamItI-PvuII fragment containing the 5' end of the mousePgk-Igene. The

first exon is indicated by a black box. Restriction sites are: BamHI, B; PvuII, Pv; HhaI, circles above the line; HpaII, boxes belowthe line. The numbers above and belowthe line indicate the size in base pairs of the expected HhaI and HpalI fragments, respectively.The second HpatI site from the left is, in fact, two nearly adjacent sites (32).

38

blotted, and probed with the radiolabeled BamHI-PvulI fragment isolated from the mouse Pgk-1 gene. BamHI + PvuII-digested DNA yielded the expected 1106-bp fragment, which was approximately twice as abundant in female as in male tissues (Fig. 2, lanes 1, 2, 7, 8). DNA digested with MspI + BamHI + PvuII yielded four of the expected bands of 425,240, 193, and 112 bp. The three other expected fragments were 60, 59, and 14 bp, but they were not detected likely because these small fragments adhered poorly to the nylon membranes and/or hybridized to only

Bartlett et al.

low levels of radioactive probe (Fig. 2, lanes 3-6). Male somatic DNA digested with HpalI + BamHI + 1N'ulI yielded identical patterns to those seen with MspI, indicating that few if any of the seven HpalI sites were methylated on the Xa. HpalI digests of female somatic DNA yielded the same pattern as seen for MspI digests except for the appearance of four novel bands marked with asterisks in lanes 6 and 12 in Fig. 2. There was no detectable band at 1106 bp in these HpalI digests, indicating that complete

Fig. 2. Southern analysis showing differential methylation of HpaII sites in the 5' end of the Pgk-1 gene. Spleen or kidney DNA (15 mg) was digested with BamHI + PvulI (lanes 1, 2, 7, and 8), with BamHI + PvulI + Mspl (lanes 3, 4, 9, and 10), or BamHI + PvulI + HpalI (lanes 5, 6, 11, and 12). The DNA was electrophoresed, blotted, and probed with labeled DNA covering the 1106 BamHI-PvulI fragment. The numbers on the left denote the fragment sizes detected. Male and female sources of DNA are noted on the top of each lane along with MspI (M) or HpalI (H). Asterisks indicate novel bands present only in female DNA digested with HpalI.

DNA methylation and X chromosome inactivation

methylation of the seven HpaII sites occurs on the Xi with very low frequency. The combined intensities of the four novel bands was less than that of the four bands of 425, 240, 193, and 112 bp, and these four latter bands appeared more intense in female than in male DNA. Thus, methylation of the seven HpaII sites on the Xi of somatic female D N A appears to be present but at a relatively modest level in murine somatic cells. Recent work on female mouse embryos (33) suggests that the most proximal HpaII site upstream of Pgk-1 becomes extensively methylated shortly after X inactivation is known to occur. Thus in the mouse Pgk-1 gene promoter, methylation of only a subset of HpalI sites shows a strong correlation with genetic inactivity. HhaI is another methylation-sensitive restriction enzyme and nine sites with its recognition sequence are present in the GC island 5' of the Pgk-1 gene (Fig. 1). Only the three largest HhaI fragments were detected (Fig. 3). Male and female somatic cell DNA yielded identical fragments (roughly twice as intense in female as in male DNA), indicating an undetectably low level of HhaI methylation of both Xa and Xi. Methylation in Body ofPgk-I Gene. The Pgk-1 gene consists of 11 exons spread over approximately 16 kb of DNA. Two alleles are known, Pgk-lb and Pgk-la, and the partial restriction maps of these two genes is shown in Fig. 4 (32, 34). To examine the methylation of sequences within the body of the gene, we used AvaI and XhoI, two methylationsensitive restriction enzymes with 6-bp recognition sequences. The XhoI site (CTCGAG) is a subset of sites recognized by AvaI (CPyCGPuG). As shown in Fig. 4, there are six AvaI sites in the 15-kb BamHI fragment containing the first eight exons of Pgk-lb and seven Ava! sites in the corresponding 20-kb BamHI fragment from Pgk-la. Four of these Avat sites also are recognized by XhoI. Four of the AvaI sites are present in intron I,

39

Fig. 3. Lack of methylation of the HhaI sites on the inactive X chromosome. Genomic DNA (15 mg) was digested to completion with BamHI + PvuII + HhaI, blotted, and hybridized as described previouslyin the legend of Fig. 2. Lane 1, male mouse kidney;2, female mouse kidney; 3, C86AGM2; 4, CI00; 5, C145. The femaleEC cells in lanes 3--5all carryan Xi.

perhaps indicating that intron I contains a region of high GC content (35). To determine the extent of methylation of the XhoI sites, genomic DNAs were digested to completion with BamHI + XhoI, separated by electrophoresis in agarose gels, blotted to nylon membranes, and hybridized with probe I or probe II, sequences that derive from the upstream and from an internal region of the Pgk-1 gene respectively (Fig. 4). The results are shown in Fig. 5 for DNA derived from male and female somatic tissues. The Southern blots are shown along with schematics of the fragments observed. When hybridized with probe I (Fig. 5,

40

B a r t l e t t et al.

1

site

2

3

5

4

1 kb

I

I

III

°

"[J Ill 1

II

I

t

0

0

II

ill

pgk-lb

IT

I

J

6

2

3

4

,,. pgk-la

1 5

O_ 6

Fig. 4. Restriction map of the methylation-sensitive AvaI and XhoI sites in a 20-kb region of the X chromosome containing the 5' region of the Pgk-1 gene. The upper map represents Pgk-lb and the lower Pgk-la. The approximate locations of the first eight exons of Pgk-1 are shown as black boxes below the Pgk-lb map. Sites 0-6 represent the seven AvaI sites in the BamHI fragment carrying the mouse Pgk-1 genes. The wavy lines labeled I and II represent the two probes used for hybridization studies. The histograms below the map show the approximate proportions of each site, which are methylated on the Xa (open bars) and on the Xi (black bars).

upper panel), male DNA yielded a strong 1.5-kb fragment and a weak 1.7-kb band, indicating that there is very little methylation of site 2 on the Xa chromosome. In female DNA, additional bands of 3.5 and 11 kb also were seen, suggesting that on the Xi there is extensive methylation of sites 2 and 3 and partial methylation of site 4. The intensity of hybridization to the 3.5- and 11-kb bands was roughly equal to that of the 1.5-kb band, suggesting that sites 2 and 3 were almost completely methylated on the Xi. When the same blot was hybridized with probe II, a number of bands appeared in both male and female DNA (Fig. 5, lower panel). Male DNA yielded strong hybridization signals to fragments of 11.5 and 7.4 kb and a weaker band at 9.2 kb. Since site 2 is almost entirely unmethylated in Xa DNA (Fig. 5, upper panel), our interpretation of these fragments, as shown in the schematic, suggests that on the Xa, site 4 is methytated

at a relatively low level and site 6 is more extensively methylated. Female DNA gave patterns of hybridization similar to those of male DNA, except that the 9.2-kb fragment was more abundant, suggesting that the Xi was methylated more frequently at site 4 than the Xa. We believe that the Xi did not yield an 11.5-kb band of the composition supposed for the Xa but instead yielded an 11.0-kb fragment (not resolved from the 11.5-kb band from the Xa) shown in the schematic. This reasoning derives from the upper panel of Fig. 5, in which it is clear that the Xi yielded no fragments larger than 11 kb, suggesting that site 6 on the Xi was methylated to only a very low level. The extent of methylation of AvaI sites was determined from Southern blots of DNA digested to completion with BamHI + AvaI. The upper panel of Fig. 6 shows a blot hybridized with probe I. Male DNA yielded a single 1.1-kb fragment, indicating that the Xa

DNA methylation and X chromosome inactivation

41

7.4, 9.2, and 11 kb, as seen in Fig. 5, lower panel. Lanes 5 and 6 in Fig. 6 contained DNA from female mice carrying two X chromosomes with the Pgk-la allele. The weak 4.5-kb band indicates that site 0 is heavily methylated on both Xa and Xi. The distribution of other bands is consistent with methylation of sites 1, 2, and 3 on the Xi, as was seen in the Pgk-lb gene in female mice. A summary of the levels of methylation observed for each of the seven AvaI sites is shown in Fig. 4. Sites 1, 2, and 3, which are clustered in the 5' end of intron I, appear to show a strong correlation between methylation and gene inactivity. Conversely, site 6 appears to show the inverse correlation; it is more methylated on the Xa than on the Xio Sites 0 and 5 are heavily methylated on both the Xa and Xi. Fig. 5. Methylation of XhoI sites in Pgk-lb genes on the Xa and Xi. DNA from male and female mouse kidney, Methylation of Pgk-1 in C86AGM2 Cells. spleen, and liver was digested with BamHI + XhoI, The two female EC lines, C86AGM2 and electrophoresed, blotted, and hybridized to probe I C100, carry two X chromosomes but one is (upper panel) or probe II (lower panel). The restriction map on the left shows the XhoI sites (2, 3, 4, and 6) and genetically inactive (10, 23). There was no the fragments being detected. Each band is labeled by its detectable methylation in DNA of the Pgk-1 approximate size in kilobase pairs. The wavy lines promoters in these cells following HhaI indicate the probes. The hybridizing fragment of about 11 kb in the lower panel is from two sources; it is an digestion (Fig. 3); however, methylation of ll.5-kb fragment from the Xa in male and female DNA some HpaII sites was evident in the DNA of and an ll-kb fragment from the Xi in female DNA. both cell lines (Fig. 7). 5AC treatment of C86AGM2 cells results in reactivation of the Xi at high frequency (10). DNA from cells chromosome always carries an unmethylated with a reactivated Xi was not detectably site 1. Female DNA yielded additional bands methylated in the Pgk-1 promoter (Fig. 7, of hybridizing material at 1.5 and 3.5 kb, lane 8). Retinoic acid (RA) treatment of indicating extensive methylation of sites 1, 2, C86AGM2 cells results in differentiation, and 3 on the Xi. The sum of the hybridization and such cells are resistant to 5AC-induced signals to bands of 1.5, 3.5, and 11 kb was reactivation of genes on the Xi (9). DNA roughly equal to that of the 1.1 band, from differentiated C86AGM2 cells was suggesting a high degree of methylation of sites 1, 2, and 3 on the Xi. A band at 11 kbp methylated to an extent similar to that of the was seen in female spleen DNA, indicating C86AGM2 parent (Fig. 7, lane 7). DNA from these female EC cells also methylation of sites 1-5 on some Xis. The lower panel of Fig. 6 is the same were analyzed for methylation of sites in blot hybridized with probe II. There is only a intron I of Pgk-t. Sites within intron t are very weak 3.6-kb band in both male and heavily methylated on the Xi in C86AGM2 female DNA, indicating that there is exten- cells (Fig. 8, lanes C). Clones with a sive methylation of site 5 on both Xa and Xi. reactivated Xi had tess extensive methylation The majority of hybridization was to bands of of intron I (Fig. 8, lanes D-G). There was

42

Bartlett et al.

Fig. 6. Methylation of AvaI sites in Pgk-1 genes on the Xa and Xi. DNA from mice carrying the Pgk-lb allele was isolated from male and female spleen and liver (lanes 1-4), digested with BamHI + AvaI, electrophoresed, blotted, and hybridized with probe I (upper panel) or probe II (lower panel). The detected fragments and their approximate sizes in kilobase pairs are shown on the left. Lanes 5 and 6 show DNA from the spleen and kidney of female mice homozygous forPgk-la. The fragments detected are shown below the map on the right.

some clone-to-clone variation regarding the extent of this methylation, and the reduced methylation was most evident in the clone shown in Fig. 8, lane F. Methylation of Hprt in C86AGM2 Cells. As we found for Pgk-1, a number of sites in intron I of the mouse Hprt gene have been shown to be heavily methylated on the Xi but unmethylated on the Xa (18). To assess the extent of methylation of these sites in female EC cells, the DNA from these cells was digested with TaqI or EcoRI along with one of the methylation-sensitive enzymes AvaI, HpaII, or HaeII. The blots were hybridized with an intron I probe, pHPTX 13in-1 (36) to

assess the sites shown in Fig. 9. Representative results are shown in Fig. 10 and indicate that the Xi in C86AGM2 cells is essentially 100% methylated at sites 2, 3, 4, and 5. The eight HpaII sites in the promoter and exon 1 region were also heavily methylated. Thus, the Hprt gene on the Xi in C86AGM2 cells resembles that in female somatic cells (18). The extent of methylation of Hprt intron I sites was then examined in clones derived from 5AC-treated C86AGM2 cells with reactivated Xi chromosomes. In each clone the level of methylation of sites 2, 4, and 5 was reduced to less than 10%. (The low level of methylation of sites 2 and 4 precludes

43

DNA methylation and X chromosome inactivation

Fig. 7. Methylation of the HpaII sites in the 5' end of the Pgk-1 gene in EC cells. Genomic DNA (15 rag) was digested to completion with MspI (lanes 1-4) or HpaII (lanes 5-8) in combination with BamHI and PvuII. Southern blotting and hybridization were carried out as described in the legend of Fig. 2. Lane A, C86AGM2; B, C100; C, differentiated C86AGM2 cells; D, AC2a, C86AGM2 cells that have been treated with 5-azacytidine and carry a reactivated X chromosome. The stars in lanes 5, 6, and 7 identify bands from methylated DNA

X chromosome carries a different Pgk-1 allele. Sites 1 and 2 are unmethylated on bothPgk-1 alleles (Fig. 8, lanes H + I). Site 0 appears to be only partially methylated on the Pgk-la allele in these cells. Following RA treatment, P10 cells differentiate and one X chromosome in each cell becomes inactivated within the first three to four days following R A exposure (20, 22). However, even after 13 days, there was no detectable methylation of Pgk-1 sites 1, 2, or 3 in these differentiated cells (Fig. 8, lanes J and K). Similarly, P10 cells had unmethylated Hprt genes, and differentiation did not result in the appearance of methytated residues at sites 2, 4, or 5 (Fig. 10A, lane 11; Fig. 10B, lanes 3, 4, 12). Thus, although X inactivation apparently occurs rapidly following RAinduced differentiation of P10 cells, methylation of DNA in intron I sites of the Pgk-1 and Hprt genes does not occur, suggesting that, at least in this cell culture system, X inactivation and DNA methylation are not causally linked.

(see Fig. 2). DISCUSSION determination of methylation levels of sites 1 and 3). The TaqI + MspI digests (Fig. 10, lanes 2-5) were done as controls and yielded the expected 4.2-kb fragment except in C86AGM2 cells and these same cells treated with 5AC. In these two cases (lanes 2 and 3), two equal bands of 4.6 and 4.2 kb were present, suggesting that the Xi in C86AGM2 cells is methylated on both C residues of the MspI site (CCGG) labeled site 4 in Fig. 9. C86AGM2 cells exposed to 5AC transiently express Hprt from the Xi (9, 10) and significant cleavage of sites 2, 4, and 5 was evident (Fig. 10A, lanes 3 and 7; Fig. 10B, lanes 8 and 16), suggesting that these sites became demethylated by the 5AC exposure. Methylation of Pgk-1 and Hprt in PIO Cells. In P10 cells both X chromosomes are known to be genetically active (20) and each

Both the mouse Pgk-I and Hprt genes contain restriction sites within their first introns which, in somatic cells, are unmethytated on the Xa but methylated on the Xi. In both Pgk-1 and Hprt, methylation of these sites does not occur in cultures of differentiating P 10 cells. Since X chromosome inactivation occurs early during the differentiation of these and similar EC cells (20-22, 37), the absence of methylation indicates that at least at those sites investigated, DNA methylation appears to be a consequence rather than a cause of genetic inactivation of the X chromosome. Similar conclusions were drawn by Lock et al. (37) regarding studies of the Hpt¢ gene methylation in embryos. In addition, the work of Migeon's group on methylation of X-linked genes in marsupial and eutherian mammals is consistent with the idea that DNA methylation plays a role in

44

Bartlett et al.

Fig. 8. Methylation of XhoI and AvaI sites in Pgk-1genes on the Xi of female EC cells. Genomic DNA was digested with BamHI + XhoI (lanes 1-10) or BamHI + Aval (lanes 11-21), electrophoresed, blotted, and hybridized to probe L Lane A, P19; B, C86(XO); C, C86AGM2; D, ACIa; E, AClb; F, AC2a; G, AC2b; H, P10(XO); I, P10; J, P10 cells that had been differentiated in retinoic acid for 10 days; K, P10 cells that had been differentiated for 13 days, and L, spleen DNA from a female mouse homozygousfor Pgk-la. The DNA in lanes D-G is from clones of 5AC-treated C86AGM2 cells with a reactivated Xi. C86(XO) cells are a derivative of C86 cells that have lost the Xi, and P10(XO) are derived from P10 cells followingloss of the X chromosome bearing the Pgk-lballele. maintenance rather than initiation of inactivation (38). The D N A sequence of the region u p s t r e a m of the mouse Pgk-1 gene shows significant sequence identity with the analogous region of the h u m a n Pgk-1 gene (32). Keith et al. (30) showed that the Xi in h u m a n somatic D N A was almost completely methylated in a region upstream of the Pgk-1 gene, while our results with the mouse gene indicated a m o r e modest degree of methylation in this region. The high degree of sequence conservation in the p r o m o t e r suggests that it contains important cis-regulatory elements. Since the extent of overall D N A methylation is not conserved, it seems possible that methylation of this region may not play an important role in regulating the activity of the Pgk-1 gene. It is perhaps

noteworthy that there are sites in h u m a n and rodent D N A in which methylation is more extensive on the Xa than the Xi (18, 39). Site 6 in the Pgk-1 gene is an example. The relationship between Xi and D N A methylation has b e e n repeatedly demonstrated for housekeeping genes, those with G C islands and S p l binding sites in their p r o m o t e r regions. For tissue-specific genes that generally have no S p l binding sites in their promoters, this relationship often does not hold (40, 41). While some transcription factors bind only to unmethylated D N A sequences (42), it may be significant that Spl binds methylated D N A as well as unmethylated D N A (43) and a n u m b e r of proteins that bind the h u m a n Pgk-1 p r o m o t e r bind both methylated and unmethylated D N A (44).

45

DNA methylation and X c h r o m o s o m e inactivation R

T

(

i

100

M

M M M MMM

M

AH

II I,.!ll.,I

I

il

M

T

H

I I

/

R

__A

) =

I

i i

!

50 4

= =

i

E

o --

m

I

~

2

:

m _ _ 3

m _ _

4

~m 5

Fig. 9. Restriction map of the 5' end of the mouse Hprtgene showing the methylation-sensitive MspI (M), AvaI (A), and HaelI (H) sites. The EcoRI (R) and TaqI (T) sites are separated by 6.3 and 5.3 kb, respectively. Exon 1 of the Hprtgene is shown as a box below the restriction line. The lower part of the figure shows the extent of methylation of each site on the Xa (filled bars) and Xi (open bars) as determined for EC cell cultures (see Fig. 10), Site 1 represents eight MspI sites and its status could not be determined on the Xa.

A nuclear activity that binds methytated GC islands both in vitro (45) and in vivo (46) has been described recently. However, this activity is significantly lower in EC cells than in other cell types. Whereas the GC rich region in and 5' to exon I of Hprt is heavily methylated on the Xi of C86AGM2 and female somatic cells, the corresponding GC rich region of Pgk-1 is less extensively methylated, suggesting that methylation of the GC islands in the promoter regions of X-linked genes may not be important for regulation. The strong correlation between methylation and gene inactivity for Hprt and Pgk-1 occurs in sites within intron I of both genes, suggesting that this region, rather than the promoter, might be where the important regulators of activity are located for these genes. Clones of C86AGM2 cells selected for

the expression of Hprt usually have both X chromosomes replicate in synchrony with the autosomes (10). This fact, along with the elevated levels of expression of other X-linked genes, suggested that reactivation of the entire Xi occurs with high frequency. The level of expression of Hprt from this Xi varies considerably from clone to clone (10), but in all clones it appears that the intron I sites have become demethylated. Demethylation of intron I sites in Pgk-1 also occurred but to a variable extent, perhaps reflecting the variable levels of Pgk-1 mRNA present in these clones (10). If DNA methylation is not responsible for the initiation of X chromosome inactivation, it is perhaps pertinent to ask why known DNA demethylating agents such as 5-azacytidine are effective at reactivating genes from the Xi. Perhaps DNA methylation is impor-

46

Bartlett et al.

mechanism for inactivation (or activation) of X-linked genes might exist similar to the myoD-induced activation of muscle-specific genes.

ACKNOWLEDGMENTS Fig. 10. The Hprt gene is heavily methylated on the Xi but only slightly methylated when expressed. (A) DNAs were digested with TaqI alone (lane 1), TaqI + MspI (lanes 2-5), or TaqI + HpalI (lanes 6-11) before electrophoresis, blotting, and hybridization to the intron I probe for HUt, pHPTh 13 in-I (36). Lanes 1, 2, and 6, C86AGM2; 3 and 7, C86AGM2 treated for 24 h with 5AC; 4 and 8, ACla; 5, AC2a; 9, AC2b; 10, AC3b; 11, P10 differentiated by treatment with retinoic acid. The 5.3-kb band is the TaqI fragment, the 4.2-kb band indicates digestion at site 4, the band of about 4.6 kb in lanes 2, 3, and 6 indicates no digestion at site 4 but digestion at site 1. (B) DNAs were digested with EcoRI + AvaI (lanes 1-9) or EcoRI + HaeII (lanes 10-17) before analysis as above. Lanes 1 and 10, female somatic DNA; 2 and 11, male somatic DNA; 3, PI0 cells differentiated with retinoic acid; 4 and 12, P10; 5 and 13, AC2b; 6 and 14, AC2a; 7 and 15, ACla; 8 and 16, C86AGM2 cells treated with 5AC; 9 and 17, C86AGM2. The 6.3-kb band is the EcoRI fragment, the 4.4-kb band in lanes I-9 results from digestion at site 2, the 4.4-kb band in lanes 10-17 results from digestion at site 3 but not 5, and the 4.1-kb band in lanes 10-17 results from digestion at site 5. ACla, AC2a, AC2b, and AC3b are clones of C86AGM2 cells isolated following 5AC treatment and selection for expression of the HUt gene on the Xi (10).

tant for maintaining the Xi in the inactive state as has been suggested (38). There are, however, some curious inconsistencies with this idea. For example, the 5AC-induced reactivation of Xi genes occurs efficiently only in rodent cells and is particularly effective only in tumorigenic cells (47). X chromosome reactivation occurs in cell hybrids formed between certain EC and somatic cells (48) and it occurs naturally during oogenesis (49, 50). We favor explanations for X chromosome inactivation based on DNA binding of transcription activators and/or repressors (19). Some transcription factors such as myoD (51, 52) regulate their own synthesis and are induced by 5AC. If 5AC induces activation or inactivation of a transcription factor specific for X-linked genes, a

This work was supported by grants from the Medical Research Council of Canada and the National Institutes of Health. M.W.McB. is a Terry Fox Cancer Research Scientist of the NCIC.

LITERATURE CITED 1. Holliday, R., and Pugh, J.E. (1975). Science 187:226-232. 2. Riggs, A.D. (1975). Cytogenet. Cell Genet. 14:9-25. 3. Cedar, H. (1988). Cell 53:3-4. 4. Monk, M. (1986). BioEssays 4:204-208. 5. Mohandas, T., Sparkes, R.S., and Shapiro, LJ. (1981). Science 211:393-396. 6. Lester, S.C., Korn, N.J., and DeMars, R. (1982). Somat. Cell Genet. 8:265-284. 7. Graves, J.A.M. (1982). Exp. CellRes. 14:99-105. 8. Hors-Cayla, M.C., Heurtz, S., and Frezal, J. (1983). Somat. Cell Genet. 9:645-657. 9. Paterno, G.D., Adra, C.N., and McBurney, M.W. (1985). Mol. Cell. Biol. 5:2705-2712. 10. Hockey, A.J., Adra, C.N., and McBurney, M.W. (1989). Somat. Cell Mol. Genet. 15:421--434. 11. Liskay, R.M., and Evans, R.J. (1980). Proe. Natl. Acad. Sci. U.S.A. 77:4895-4898. 12. Chapman, V.M., Kratzer, P.G., Sircusa, L.O., Quarantillo, B.A., Evans, R., and Liskay, R.M. (1982). Proc. Natl. Acad. Sci. U.S.A. 79:5357-5361. 13. Venolia, L., Gartler, S.M., Wasserman, E.R., Yen, P., Mohandas, T., and Shapiro, L.J. (1982). Proc. Natl. Acad. Sci. U.S.A. 79:2352-2354. 14. Yen, P.H., Patel, P., Chinault, A.C., Mohandas, T., and Shapiro, L.J. (1984). Proc. Natl. Acad. Sci. U.SM. 81:1754-1763. 15. Wolf, S.F., Jolly, D.J., Lunnen, K.D., Friedman, T., and Migeon, B.R. (1984). Proc. Natl. Acad. Sci. U.S.A. 81:2806-2810. 16. Toniolo, D., D'Urso, M., Martini, G., Persico, G., Tufano, V., Battistuzzi, G., and Luzzatto, L. (1984). EMBOJ. 3:1987-1995. 17. Lindsay, S., Monk, M., Holliday, R., Huschtecha, L., Davies, K.E., Riggs, A.D., and Flavell, R.A. (1985). Ann. Hum. Genet. 48:115-127. 18. Lock, L.F., Melton, D.W., Caskey, C.T., and Martin, G.R. (1986). MoL Cell. Biol. 6:914-924. 19. McBurney, M.W. (1988). BioEssays 9:85-88. 20. McBurney, M.W, and Strutt, B.J. (1980). Cell 21:357-364.

DNA methylation and X chromosome inactivation

21. Martin, G.R., Epstein, CJ., Travis, B., Tucker, G., Yatsiv, S., Martin, D.M., Clift, S., and Cohen, S. (1978). Nature 271:329-333. 22. Paterno, G.D., and McBurney, M.W. (1985).J. Cell Sci. 75:149-163. 23. McBurney, M.W., and Adamson, E.D. (1976). Cell 9:57.90. 24. Rudnicki, M.A., and McBurney, M.W. (1987). In Teratocarcinomas and Embryonic Stem Cells, A Practical Approach, (ed.) Robertson, EJ. IRL

Press, Oxford, pp. 19-49. 25. McBurney, M.W., and Rogers, B.J. (1982). Dev. Biol. 89:503-508. 26. Blin, N., and Stafford, D.W. (1976). Nucleic Acids Res. 3:2303-2308. 27. Birnboim, H.C., and Doly, J. (1979). Nucleic Acids Res. 3:2303-2308. 28. Maniatis, T., Fritsch, T., andSambrook, J. (1982). Molecular Cloning: A Laboratory Manual, (Cold Spring IIarbor Laboratory, Cold Spring Harbor, New York). 29. Feinberg, A.P., and Vogelstein, B. (1983). Anal. Biochem. 132:6-7. 30. Keith, D.tI., Singer-Sam, J., and Riggs, A.D. (1986). MoL Cell. Biol. 6.'4122-4125. 31. Hansen, R.S., and Gartler, S.M. (1990). Proc. Natl. Acad. Sci. U.S.A. 87:4174-4t78. 32. Adra, C.N., Boer, P.H., and McBurney, M.W. (1987). Gene 60:65-74. 33. Singer-Sam, J., Grant, M., LeBon, J.M., Okuyama, K., Chapman, V., Monk, M., and Riggs, A.D. (1990). Mol. Cell. Biol. 10:4987-4989. 34. Boer, P.H, Potten, H., Adra, C.N., Jardine, YL, Mullhofer, G., and McBurney, M.W. (1990). Biochem. Genet. 28: 299-308. 35. Bird, A.P. (1986). Nature 321.:209-213.

47

36. Melton, D.W., Konecki, D.S., Brennand, J., and Caskey, C.T. (1984). Proc. Natl. Acad. Sci. U.S.A. 81:2147-2t51. 37. Lock, L.F., Takagi, N., and Martin, G.R. (1987). Cell 48:36-46. 38. Kaslow, P.C., and Migeon, B.R. (1987). Proc. Natl. Acad. Sci. U.S.A. 84:6210-6214. 39. Starr, T., and Wood, S. (1988).Am. J. Hum. Genet. 42:26%280. 40. Cullen, D.R., Hubberman, P., Kaslow, D.C., and Migeon, B.R. (1986). EMBO J. 5:2223-2229. 41. Mullins, L.J., Veres, G., Caskey, C.T., and Chapman, V. (1987). MoL Cell. Biol. 7:3916-3922. 42. Watt, F., and Molloy, P.L (1988). Genes Dev. 2:1136-1143. 43. Holler, M., Westin, G, Jiricny, J., and Schaffner, W. (1988). Genes Dev. 2:1127-1135. 44. Yang, T.P., Singer-Sam, J., Flores, J.C., and Riggs, A.D. (1988). Somat. Cell MoL Genet. 14:461-472. 45. Meehan, R.R., Lewis, J.D., McKay, S., Kleiner, E.L., and Bird, A.P. (1989). Cell 58:499-507. 46. Antequera, F., Macleod, D., and Bird, A.P. (1989). Cell 58:50%517. ' 47. Grant, S.G., and Worton, R.G. (1989). MoL Cell. Biol. 9:1635-1641. 48. Takagi, N., Yoshida, M.A, Sugawara, W., and Sasaki, M. (1983). Ceil 34:1053-1062. 49. Gartler, S.M., Andina, R., and Gant, N. (1975). Exp. Cell Res. 91:454-457. 50. Kratzer, P.G., and Chapman, V.M. (1981). Proc. Natl. Acad. Sci. U.S.A. 78:3093-3097. 51. Davis, R.L, Weintraub, H, and Lassar, A.B. (1987). Cell 51:987-1000. 52. Thayer, M.J., Tapscott, SJ., Davis, R.L., Wright, W.E., Lassar, A.B., and Weintraub, H. (1989). Cell 58:241-248.