DNA Methylation Precedes Chromatin Modifications under the Influence of the Strain-Specific Modifier Ssm1

Transgenic mice.D2 and B6 strain mice transgenic for the HRD construct were used for all chromatin immunoprecipitations (ChIPs) from adult tissue. The HRD construct (8) consists (from 5′ to 3′) of the immunoglobulin (Ig) μ heavy chain enhancer, the mouse metallothionein-1 promoter, Vkappa and Jkappa Ig recombination signal sequences flanking a rat preproinsulin initiation codon, the Escherichia coli xanthine-guanine phosphoribosyl transferase (GPT) coding sequence, and simian virus 40 (SV40) splice and polyadenylation sequences (Fig. 1). The transgenic line 342, used in these studies, contains seven head-to-tail tandem integrations of the HRD transgene. The locus of integration is unknown but is the same for the B6 and D2 mice, since the same transgene was propagated in both strains.

Embryonic stem cells.Embryonic stem (ES) cells were derived from the inner cell mass of D2 and B6 HRD transgenic mice (34). In this study, we used the D2 line awD1 and the B6 line awB2. The latter already has extensive methylation of the HRD transgene, but DNA methylation increases with differentiation (34). To prevent differentiation, the cells were grown at 37°C on gamma-irradiated STO feeder cells in the presence of leukemia inhibitory factor at 10³ U/ml (28). To differentiate the cells, leukemia inhibitory factor was removed and the ES cells were replated without STO feeder cells. Cells were allowed to differentiate into embryoid bodies over periods of 1 to 4 weeks.

Chromatin preparation from adult mice.Chromatin was prepared from the livers of adult mice (4, 20, 21). The livers were dissected and nuclei prepared by Dounce homogenization. The nuclei were fixed in 1% formaldehyde for 10 min at 37°C (cross-linking was quenched by the addition of glycine to 125 mM at room temperature). The nuclear slurry was sonicated and centrifuged (SW50.1 rotor at 31,000 rpm for 20 h at 20°C) through a CsCl gradient (three steps: 1.35 g/ml, 1.5 g/ml, and 1.75 g/ml). The gradient was separated into 10 fractions, and those containing the chromatin were pooled. The cross-linked chromatin was dialyzed against Tris-EDTA overnight in the presence of RNase A (25 ng/μl).

Chromatin preparation from ES cells.Chromatin was prepared from both undifferentiated and differentiated ES cells as described by Upstate Biotechnology, Inc. (Lake Placid, NY). Briefly, chromatin was cross-linked by adding formaldehyde directly to cell culture medium (to a final concentration of 1%) for 10 min at 30°C. Cells were washed with phosphate-buffered saline (containing protease inhibitors), removed from the tissue culture plate by trypsinization, and pelleted by centrifugation. Cells were then lysed in sodium dodecyl sulfate lysis buffer and finally dialyzed to Tris-EDTA in the presence of RNase A (25 ng/μl).

Chromatin immunoprecipitation.Before the ChIP assay was performed, chromatin was broken into fragments from 100 to 500 bp by sonication or micrococcal nuclease digestion. Chromatin IP was performed as described by Upstate Biotechnology, Inc. (Lake Placid, NY). Cross-linked chromatin was diluted and precleared with salmon sperm-protein A agarose. One microgram of a specific antibody was added to the precleared chromatin and incubated at 4°C overnight with rotation (all antibodies were from Upstate Biotechnology, Inc., except anti-Brg1 from Santa Cruz [Santa Cruz, CA]; the anti-methylated H3 K9 and H3 K4 antibodies are against the dimethylated forms [catalog no. 07-212 and no. 07-030, respectively]). The antibody complex was recovered by using salmon sperm-protein A agarose. The histone-DNA-antibody-protein A agarose complex was washed several times, and the histone-DNA complex was eluted with 1% sodium dodecyl sulfate and 0.1 M NaHCO₃. Cross-linking was reversed by heating the chromatin complex to 65°C for 4 h. DNA was purified by proteinase K digestion (at a final concentration of 100 μg/ml) for 1 h at 50°C, followed by phenol-chloroform extraction and ethanol precipitation. The resulting DNA was analyzed by PCR amplification. Input DNA was acquired by purifying DNA from chromatin as described above, only without immunoprecipitation.

PCR assay of immunoprecipitated chromatin.DNA recovered from immunoprecipitation was analyzed by PCR amplification. Three primer pairs were used to amplify three different parts of the HRD transgene (Fig. 1). The PMT product comprises the 3′ end of the HRD metallothionein promoter and the 5′ end of the Vkappa recombination signal sequence. The GPT PCR product comprises a central portion of the HRD E. coli xanthine-guanine phosphoribosyl transferase sequence. The SV40 product is located in the HRD SV40 polyadenylation region. DNA was purified from immunoprecipitated chromatin and then made into four serial fourfold dilutions. Each of these DNA preparations was then amplified by PCR in parallel in order to accurately determine the amount of immunoprecipitated chromatin. Input DNA was amplified by a method identical to that used for DNA obtained by immunoprecipitation.

Quantification of chromatin immunoprecipitation.Immunoprecipitated DNA was amplified by PCR and then electrophoresed through an 8% polyacrylamide gel. Resulting bands were quantified from at least triplicate ChIP analyses with at least two different chromatin preparations, using the program NIH Image 1.63, downloaded from National Center for Biotechnology Information. Quantification results were normalized to the amount of input DNA and expressed as fractions of input.

Southern blotting.DNA from transgenic mice and ES cells was analyzed for CpG methylation by restriction digestion followed by Southern blotting. HpaII and BamHI digestion was used for methylation analysis. Control digests were performed with MspI and BamHI. Restricted DNA was electrophoresed through 1% agarose gels and then transferred to a Hybond N membrane. The DNA was fixed to the membrane by UV irradiation. Radiolabeled [α-³²P]dCTP probe to the HRD GPT region was synthesized by random priming.

RT-PCR for HRD transcripts.RNA was purified from livers, ES cell cultures, and embryos by using the RNeasy kit (QIAGEN, Valencia, CA). The cDNA was made by priming with random hexamers and then amplified by using primers specific for the GPT region of HRD (the same primers used to amplify the GPT region in the chromatin immunoprecipitation assays). STO feeder cells (which do not contain the transgene) and water were used as negative controls. β-Actin expression was used to normalize the amounts of cDNA in each of the samples. Experiments with controls without reverse transcriptase (RT) were carried out to rule out DNA contamination.

Bisulfite sequencing.To study the methylation patterns of the HRD transgene in the undifferentiated ES cells and adult livers obtained from the B6 and D2 strains of transgenic mice, we used the bisulfite genomic sequencing technique (5). We used the plasmid carrying the HRD transgene, which was originally used to create the HRD transgenic line as a control for the bisulfite technique.

For the PCR amplification, nested PCRs were necessary to produce enough product from the genomic DNA. The outer pair of primers used for the adult B6 and DBA/2 livers was 5′GTATTTTTTAGGGTGGGTTT3′ and 5′CACCACTACTCCCATTCATC3′, and the inner pair of primers was 5′GTGGTGGATGTTTGGTGGAG3′ and 5′CAAAACCCACTCATAAATCC3′. For the undifferentiated cells, we could amplify only the GPT region of the transgene in two halves. The first half was amplified with 5′GTATTTTTTAGGGTGGGTTT3′ and 5′AAAACTATTATAACCCACCTAAAAT3′ as the outer pair of primers and 5′GTGGTGGATGTTTGGTGGAG3′ and 5′ TGGTTGTTAATTTTTTTAAT3′ as the inner pair of primers. The second half of the transgene was amplified with 5′ATTGATGATTTGGTGGATATT3′ and 5′CACCACTACTCCCATTCATC3′ as the outer pair of primers and 5′TGTATTTTGTTATTATTTTT3′ and 5′CAAAACCCACTCATAAATCC3′ as the inner pair of primers. The PCRs in the two halves included a central overlap of 162 nucleotides.

PCR products were electrophoresed on a 1% agarose gel and isolated by using a fragment isolation kit (QIAGEN, Valencia, CA). PCR fragments were subcloned into the TOPO-TA pCR2.1 vector (Invitrogen, Carlsbad, CA) and sequenced by using the M13 forward primer.

The HRD transgene is silenced when carried in adult B6 strain mice relative to D2 mice.Earlier results had shown that the HRD transgenes display complete, integration site-independent DNA methylation when carried in a B6 background (7). However, when in a D2 strain background, the transgene is not methylated. Moreover, under the influence of the strain-specific modifier Ssm1, HRD appears to be functionally silenced in B6 mice and active in D2 mice (8, 9, 11, 33). Given the interrelationship of DNA methylation and chromatin (17, 18, 30), we investigated chromatin modifications of the HRD transgene and the temporal relationship of strain-specific DNA methylation and chromatin modifications.

Chromatin from the livers of adult mice was analyzed by ChIP using antibodies specific to various chromatin modifiers. All experiments were done at least in triplicate, and results were compared to results from experiments with input DNA as well as albumin and β-actin controls (genes that would not be expected to display any strain-specific difference in chromatin structure or DNA methylation). First, we examined histone H3 K9 methylation. Methylation of this residue is strongly correlated with silenced/heterochromatic regions of the genome in many different systems (22, 26). Following ChIP with an antibody specific to H3 dimethyl-K9, the HRD transgene shows much stronger H3 K9 methylation when carried in a B6 background relative to a D2 background at the PMT, GPT, and SV40 regions (Fig. 1 and 2A). This is interesting not only because it shows that the transgene, when carried in B6 mice, is silenced at the level of chromatin structure and DNA methylation, but also because the chromatin changes appear to cover the whole of the transgene, not merely the area near the regulatory regions, Eμ, and PMT (Fig. 1).

• Open in new tab
• Download powerpoint

FIG. 2.

ChIP assays with adult liver. Chromatin from B6 and D2 strain mice was immunoprecipitated with the indicated antibody, and DNA was amplified with primers to the PMT, GPT, and SV40 regions of the HRD transgene (see Materials and Methods). Grey bars, B6; white bars, D2. (A) Histone H3 K9 methylation; (B) MeCP2 binding; (C) histone H3 acetylation (K14 and K9); (D) histone H3 acetylation (K14 only); (E) histone H3 K4 methylation; (F) Brg1 binding.

These findings suggested that there was an overall strain-specific difference in the HRD chromatin structure. This was supported by ChIP with antibodies to the methylated DNA binding protein MeCP2. MeCP2 showed robust binding to the PMT and GPT regions of HRD when carried by the B6 strain, but much less was detected bound in the D2 strain (Fig. 2B). The low levels of MeCP2 binding to the SV40 region in both strains are probably due to the fact that there is only one CpG site in the entire SV40 splice and polyadenylation region.

We next investigated activating chromatin modifications that might be carried by the transgene when integrated into the D2 strain genome. Using a polyclonal antibody to histone H3 acetylated at both the K9 and K14 positions of its N-terminal tail, ChIP experiments showed hyperacetylation in a D2 strain (relative to a B6 strain) at three HRD transgene regions (Fig. 2C). However, the PMT region showed acetylation in both D2 and B6 strains. The latter was unexpected, since the PMT region was highly methylated in the B6 strain at H3 K9 in the associated H3 (Fig. 2A). Virtually identical results were obtained with an antibody specific to the N-terminal tail of H3 acetylated at only K14 (Fig. 2D). While the GPT and SV40 regions are both hyperacetylated at K14 only in the D2 strain, the PMT region again appears acetylated in both the B6 and D2 strains. It was possible that the PMT results might be caused by the amplification of endogenous PMT sequence. However, this is unlikely for three reasons. First, while the 5′ primer for the PMT product is from the HRD sequence shared with the endogenous murine metallothionein promoter, the 3′ primer is rooted in the Vkappa recombination signal sequence of the transgene. Second, with the transgene-specific primer pair, no PMT product is produced when amplification is attempted from nontransgenic mice (data not shown). Finally, when the PMT product is isolated and sequenced, its sequence is exactly as expected for amplification from the HRD transgene (data not shown). Therefore, the PMT results do appear to be genuine. Given that the metallothionein promoter is functional in hepatocytes, it is possible that histone acetylases are recruited to the PMT region when in this environment, despite the overall silenced state of the transgene. This conclusion is supported by the absence of PMT acetylation in 4-week-differentiated ES cells (see Fig. 4). While the PMT results seem to be true outliers, the GPT and SV40 results further support a pattern of HRD silencing in the adult B6 mouse and activation in the adult D2 mouse.

Unexpectedly, H3 K4 was methylated to similar degrees in both strains (Fig. 2E). This is correlated with euchromatin and inversely correlated with the presence of methylated histone H3 K9. We have no explanation for this finding.

Since the MeCP2 binding and covalent histone modifications strongly supported a strain-specific chromatin structure for the HRD transgene, we also examined the binding of chromatin remodeling proteins. With an antibody for Brg1 (an ATPase subunit of the mammalian SWI/SNF complex), strong Brg1 binding was found in all regions of the transgene in the D2 strain. In contrast, binding was weak in all regions of the B6 transgene (Fig. 2F).

As controls for the ChIP assays, the β-actin and albumin genes were analyzed. Neither of these genes would be expected to show any strain-specific difference in chromatin structure. Amplified regions of both genes showed equivalent levels of H3 K9 methylation and H3 acetylation in both strains (Fig. 3A and B). This supports the conclusion that the HRD transgene displays a true strain-specific difference in chromatin structure.

• Open in new tab
• Download powerpoint

FIG. 3.

ChIP controls (β-actin and albumin genes). ChIP with the indicated antibodies is shown. Diff-28d, 4-week-differentiated ES cells; Undiff, undifferentiated ES cells.

With the exception of PMT acetylation, all elements of chromatin analyzed in the adult liver show that there is a strain-specific difference in the chromatin structure of the HRD transgene. Overall, the transgene appears to be heterochromatic and silenced when carried in the B6 background. In contrast, HRD is euchromatic and active in the D2 background.

Four-week-differentiated embryonic stem cells recapitulate the strain-specific HRD chromatin structural changes seen in the adult, but 1-week-differentiated cells show essentially no strain-specific difference.We examined the development of the strain-specific chromatin modifications over developmental time. First, differentiated HRD transgenic embryonic stem cells from the D2 or B6 strain were analyzed. Following differentiation into embryoid bodies for either 4 weeks or 1 week, chromatin was extracted from the ES cells and analyzed by ChIP.

Results from ES cells differentiated for 4 weeks recapitulate those seen with adult tissue. H3 lysine-9 methylation is very evident in the PMT, GPT, and SV40 regions of the B6 HRD transgene (Fig. 4A). However, this modification is undetectable when the transgene is carried in D2 cells (Fig. 4A). H3 acetylation (using an antibody specific for histone H3 acetylated at both K9 and K14) is much stronger in all three regions of the D2 transgene (Fig. 4B). In contrast to the adult liver (Fig. 2), in the differentiated B6 ES cells, H3 in PMT is not hyperacetylated. Finally, Brg1 binding is much stronger in all assayed regions of HRD when the transgene is carried in D2 differentiated ES cells (Fig. 4C). However, there is no strain difference for chromatin patterns of β-actin and albumin genes (Fig. 3C and D). Thus, fully (4-week) differentiated ES cells generally recapitulate the strain-specific chromatin pattern differences seen in the HRD transgene of adult mice.

• Open in new tab
• Download powerpoint

FIG. 4.

ChIP assays with 4-week- and 1-week-differentiated ES cells. Chromatin from B6 and D2 strain differentiated embryonic stem cells was immunoprecipitated with the indicated antibodies. See the legend to Fig. 2.

Because strain-specific structural changes of the HRD transgene appeared to be complete by 4 weeks of differentiation in culture, we investigated an earlier stage of ES cell differentiation. In ES cells differentiated for only 1 week, the central GPT region of the HRD transgene was analyzed. This region is of specific importance because when it is deleted, the HRD transgene no longer shows strain-specific methylation and GPT alone shows the strain difference (6). At the 1-week stage of differentiation, the results differ from those of the adult stage. H3 K9 methylation appears to be identical for B6 and D2 ES cells (Fig. 4D). K4 methylation and acetylation of H3 are also similar (Fig. 4D). In contrast, MeCP2 binding does appear to be slightly (but reproducibly) greater in the B6 differentiated ES cells (Fig. 4D). Therefore, the chromatin structures of the transgenes appear to be similar for the B6 and D2 strains after 1 week of differentiation (with the possible exception of MeCP2 binding). Also of interest is the fact that the transgenes appear to be at an intermediate chromatin structural state in ES cells of both mouse strains. While silencing marks are found (i.e., histone 3 K9 methylation and MeCP2 binding), activating marks are also present (histone 3 K4 methylation and acetylation).

Undifferentiated ES cells show an intermediate HRD chromatin structure that does not differ by strain.As was the case for the adult, H3 lysine-9 methylation was readily evident in the PMT, GPT, and SV40 regions of the HRD transgene when it was carried in B6 undifferentiated ES cells (Fig. 5A). However, in contrast to those of the adult, all regions of D2 strain HRD also showed H3 K9 hypermethylation (Fig. 5A).

• Open in new tab
• Download powerpoint

FIG. 5.

ChIP assays with undifferentiated ES cells. Chromatin from B6 and D2 strain undifferentiated embryonic stem cells was immunoprecipitated with the indicated antibodies. See the legend to Fig. 2.

Also, covalent modifications associated with euchromatin were equivalent in both B6 and D2 undifferentiated ES cells. H3 acetylation and Brg1 binding are clearly seen in both B6 and D2 GPT and SV40 regions (Fig. 5C and D). The PMT regions of both B6 and D2 appear to be relatively free of any of these modifications (Fig. 5B, C, and D), and K4 methylation seems to be restricted to the SV40 region, but again, the modifications do not differ in a strain-specific manner. As expected, the β-actin and albumin genes show the same chromatin patterns in B6 and D2 cells (Fig. 3E and F). Therefore, the analysis of undifferentiated ES cells leads to the conclusion that the strain specificity of the HRD chromatin structure is likely established after the undifferentiated ES cell stage (after the blastocyst stage in the developing embryo).

HRD transgene methylations differ in B6 and D2 ES cells before chromatin change is evident.Earlier work has shown that the strain-specific transgene methylation increases between the undifferentiated and differentiated ES cell stages (34). In order to clarify the temporal relationship between DNA methylation and chromatin structural change, the methylation statuses of the HRD transgenes in B6 and D2 mice, and also in ES cells at various stages, were assayed (Fig. 6). Restriction digestion by BamHI followed by either HpaII (a methylation-sensitive enzyme) or MspI (a methylation-insensitive enzyme; blots not shown) gives a characteristic band pattern on a Southern blot. The upper band represents a full-length BamHI/BamHI fragment (Fig. 1) that is not restricted at any internal HpaII sites (and therefore indicates DNA methylation). Two lower bands represent efficient HpaII digestion. Southern blots reveal that both adult tissue and ES cells at various times of differentiation have negligible amounts of complete HRD methylation in the D2 strain (Fig. 6A). In contrast, B6 adult tissue and ES cells at every stage of differentiation show significant complete DNA methylation. However, partially methylated sequences are present in undifferentiated ES cells and up to 12 days after in vitro differentiation. Development in vitro is slow compared to development in vivo, in which complete HRD methylation is already seen in 6.5-day embryos (34). Reprobing the blot with mitochondrial DNA which does not undergo CpG methylation (14) shows that all the DNAs are equally digestible by HpaII (Fig. 6B).

• Open in new tab
• Download powerpoint

FIG. 6.

Southern blot of B6 and D2 DNAs to determine DNA methylation. DNA was digested with BamHI and then HpaII. The blots were probed for GPT (Fig. 1) (A) and mitochondrial DNA (no methylated CpG) (B). me, BamHI (B2 and B3 [Fig. 1]) fragment given when all HpaII sites are CpG methylated and therefore not digested by HpaII; un, bands given when most HpaII sites are unmethylated and therefore cut by HpaII. The size markers (left) are 2, 1.65, 1.4, 1, 0.75, 0.5, 0.4, and 0.3 kb (from top to bottom).

To confirm the strain-specific difference in the methylation patterns of the HRD transgene, the sodium bisulfite technique was employed. This technique allows the analysis of every CpG dinucleotide present in the GPT region. The GPT region was chosen because it is the only sequence in the HRD transgene that is essential for strain-specific DNA methylation (6). In the B6 undifferentiated ES cells, 45% of the CpGs in the first half of the GPT region were methylated, as opposed to 15% in that of the D2 strain (Fig. 7 and Table 1). Similarly, 26% of the CpGs were methylated in the second half of the B6 GPT region, while only 14% were methylated in that of the D2 strain. In the adult livers, the difference between the two strains was more dramatic. In the B6 adult liver, 93% of the CpGs were methylated, while the D2 strain showed methylation of only 10% of its CpGs. This clearly confirms that there is a significant level of CpG methylation in the B6 strain compared to that in the D2 strain already at the undifferentiated ES cell stage. However, some of the B6 sequences of the undifferentiated ES cells showed as little CpG methylation as or less CpG methylation than some D2 sequences (Fig. 7A), suggesting that Ssm1-dependent DNA methylation is initiated at the blastocyst stage.

• Open in new tab
• Download powerpoint

FIG. 7.

Bisulfite analysis of DNA methylation. Each horizontal line represents the sequence from an individual bacterial colony carrying the region of the transgene being analyzed. Each vertical bar represents a methylated CpG dinucleotide, and a lack of methylation at a CpG dinucleotide is depicted by the absence of a bar. All CpG dinucleotides present in the analyzed sequence are shown above the data. (A) Analysis of CpG methylation in the GPT region of the HRD transgene in B6 and D2 undifferentiated ES cells. The analyzed region was PCR amplified in two halves; hence, the GPT region is indicated as two sets of sequences. (B) Analysis of CpG methylation in the GPT region of the HRD transgene in B6 and D2 adult liver. For the adult tissues, it was possible to PCR amplify the entire GPT region. (C) Analysis of CpG methylation in the GPT region of the HRD plasmid.

Given that chromatin structural differences are undetectable at the undifferentiated stage and barely present after 1 week of differentiation, it appears that the DNA methylation of the HRD transgene precedes its chromatin structural modification.

Transcription of the HRD transgene.In a previous study, no transcripts from the HRD transgene were found in adult B6 mice, whereas HRD was transcribed in adult D2 mice (33). Surprisingly, despite the significant levels of DNA methylation of HRD in the undifferentiated and differentiated B6 ES cells, HRD was transcribed in these cells up to day 7 (Fig. 8). Representing the more mature embryonic stage, RNA from day 10 embryos was analyzed. No expression was seen in B6 embryos, while D2 embryos continued to express HRD (Fig. 8) (reproducing the findings from adult B6 and D2 mice [33]). Thus, it appears that the repression of transcription in B6 may be delayed until the chromatin has adopted an inactive state.

• Open in new tab
• Download powerpoint

FIG. 8.

HRD transcription. HRD transcription was assayed by RT-PCR at various developmental stages of B6 and D2 (10-day in vivo embryos, 7- and 3.5-day in vitro differentiated ES cells, and undifferentiated ES cells). STO (feeder cell) and water are negative controls. (A) GPT; (B) β-actin.