H19 Imprinting Control Region Methylation Requires an Imprinted Environment Only in the Male Germ Line

Generation of mutant mice.All animal research was done according to NIH and Public Health Service (PHS) policy and was approved by the Eunice Kennedy Shriver NICHD Animal Care and Use Committee. The generation of Afp-A, Afp-D, H19R (45), and H19-BAC1 transgenic (26) animals (Fig. 1A and B) has been described previously. To generate the Afp-DCK insertion, a 7.3-kb BglII-SalI fragment carrying the H19ICR and an additional downstream sequence including the entire H19 promoter and mRNA-encoding region was targeted to the XbaI site at −0.9 kb relative to the Afp transcriptional start site (Fig. 1B). The CD3-CMG insertion carries the H19ICR as a 2.4-kb BglII-BglII fragment that has been targeted to the BpuAI site between the CD3 gamma and CD3 delta genes (Fig. 1C). Animals with germ line transmission of the Afp-DCK or CD3-CMG insertion were mated to EIIa-cre transgenic mice (30) to excise the Neo^r cassette that was used for positive selection during embryonic stem cell mutagenesis.

Southern hybridization.Genomic DNA was isolated from tissues of adult mice by phenol-chloroform extraction, and 30 μg was digested with SacI. Half of the SacI-digested DNA was incubated further with the methylation-sensitive enzyme AciI. SacI and SacI-plus-AciI-digested DNAs were electrophoresed in a 0.8% agarose gel, blotted onto a nylon membrane, and probed with a ³²P-labeled 1.6-kb KpnI-HindIII fragment internal to the H19ICR (Fig. 2A).

• Open in new tab
• Download powerpoint

FIG. 2.

DNA methylation at the endogenous and ectopic H19ICRs in somatic tissues and in testis. DNAs isolated from kidney or testis were digested with SacI (−) or SacI plus AciI (+) and then analyzed by Southern blotting. The identity of the ICR insertion and its parental origin are indicated above the lanes. (A) The 3.8- and 6.7-kb SacI fragments carrying the H19ICR (thickened line) at the endogenous Igf2/H19 locus (top line) and at the CD3-CMG chromosome (bottom line) are depicted. The arrowhead above the top line indicates the polymorphic SacI site that distinguishes wild-type domesticus (Dom) and castaneus (Cas) H19 alleles, cleaving castaneus H19 alleles into 2.5-kb (Cas1) and 1.3-kb (Cas2) SacI fragments. AciI restriction sites within the SacI fragments are indicated by vertical lines. The 1.6-kb KpnI-HindIII probe used to identify the ICR is indicated. (B) By crossing CD3-CMG animals with DIS7CAS mice (19) carrying the endogenous H19ICR on a castaneus allele, we were able to use SacI digestion to distinguish between the three H19ICRs present in these progeny, i.e., the ICR inserted at CD3 and the maternal and paternal endogenous H19ICRs. The ICR insertion is always of the same parental origin as the endogenous domesticus (Dom) allele. (C) Methylation of the H19ICR insertion in Afp-DCK and H19-BAC1 mice was analyzed in animals homozygous for the H19^Δ13 mutation, which deletes the entire H19 gene and 10 kb of upstream sequence, including the endogenous H19ICR. Thus, the insertion is the only H19ICR in these animals.

Embryo collection.Adult female mice were treated for superovulation (43) and mated to adult males homozygous for the CD3-CMG insertion. Female mice were sacrificed at embryonic day 1 (E1) or E3.5 to collect zygotes or morulae and blastocysts, respectively. To collect zygotes, matings were set up so that the endogenous maternal allele was DIS7CAS (19), harboring single nucleotide polymorphisms (SNPs) specific to the castaneus mouse strain, and the endogenous paternal allele was H19^Δ13 (31). Morulae and blastocysts were homozygous for H19^Δ13 on chromosome 7.

Bisulfite sequencing.DNA methylation in mature sperm, testis, and preimplantation mouse embryos was determined as previously described (21, 35). Briefly, mature sperm was incubated in 500 μl lysis buffer (10 mM Tris-HCl, pH 8, 10 mM EDTA, pH 8, 1% SDS [wt/vol]), 0.8 mg/ml proteinase K, 40 mM dithiothreitol (DTT), and 0.04 mg/ml glycogen at 55°C overnight. Individual pools of embryos were mixed with 100 μl lysis buffer (without DTT), 2 mg/ml proteinase K, and 0.2 mg/ml glycogen and incubated at 55°C for 5 h. Genomic DNA was recovered from sperm and embryos by phenol-chloroform extraction, using the Phase Lock Gel Heavy system (PLG 1.5 ml; Eppendorf North America), and dissolved in sterile water. Genomic testis DNA was isolated by traditional phenol-chloroform extraction. Genomic DNA equivalent to 100 ng of sperm and testis DNA and 2 to 17 blastocysts, 0.5 to 6 morulae, and 7 to 25 zygotes was embedded in low-melting-point agarose beads, and up to seven individual beads per 800 μl 2.5 M bisulfite-hydroquinone solution were incubated at 50°C for 4 h. Single beads were then subjected to nested PCR. Primers were newly designed or adopted from the work of Park et al. (45) and Tremblay et al. (57) and used to amplify five consecutive regions within the H19ICR. For region 1 (GenBank accession no. NT_039448; upper strand), the primers were Bis-CD3-CMG for19, 5′-TGA TTT TTT TGT TGA ATT TGG GGT AT-3′; Bis-CD3-CMG rev19, 5′-AAA ACT TAA CTC ATT CCC TA-3′ (400 bp); Bis-CD3-CMG for21, 5′-TTG GGG TAT TTA AAG TTT TGT T-3′; and Bis-CD3-CMG rev21, 5′-ATC CCA CAT ACT TTA TCA TA-3′ (304 bp). For region 2 (accession no. NT_039448; lower strand), the primers were BDMRTR3, 5′-CTA CCC AAA AAA TAT ATA TTA TAC CAC CCC-3′; BDMRTF7, 5′-ATA TGG TTT ATA AGA GGT TGG AA-3′ (507 bp); BDMRTR4, 5′-CCC TTA TAA ATC ATT AAA TAC TAT ACC TAA-3′; and BDMRTF8, 5′-TAT TTG TGT TTT TGG AGG GGG TT-3′ (450 bp). For region 3 (accession no. U19619; upper strand), the primers were BMsp4t1, 5′-GGA ATT TTA TAT TAA GTT TTG GGT GT-3′; BHha4t2, 5′-AAC CCC CTC CAA AAA CTC AAA T-3′ (455 bp); BMsp4t2, 5′-AGT TAA AAT TAA TTG AAG AGG-3′; and BHha4t3, 5′-ATT CCA ACC TCT TAT AAA CCA TAT-3′ (406 bp). For region 4 (accession no. U19619; upper strand), the primers were BHha2t1, 5′-ATA GTT ATG GGT TTT ATG AGG-3′; BMsp3t2, 5′-CCT CTT CAA TTA ATT TTA ACT-3′ (448 bp); BHha2t2, 5′-AGG GGT TTA TGT TAG TTT TTG ATA A-3′; and BMsp3t, 5′-ACA CCC AAA ACT TAA TAT AAA ATT CC-3′ (403 bp). For region 5 (accession no. U19619; upper strand), the primers were BMsp2t1, 5′-GAG TAT TTA GGA GGT ATA AGA ATT-3′; BHha1t3, 5′-ATC AAA AAC TAA CAT AAA CCC CT-3′ (471 bp); BMsp2t2, 5′-GTA AGG AGA TTA TGT TTA TTT TTG G-3′; and BHha1t4, 5′-CCT CAT TAA TCC CAT AAC TAT-3′ (423 bp). Nested primers amplifying control regions within the Snrpn (420 bp) and Rasgrf1 (284 bp) genes were adopted from the work of Lucifero et al. (36) and Li et al. (33), respectively; they read as follows: Snrpn outside forward, 5′-TAT GTA ATA TGA TAT AGT TTA GAA ATT AG-3′; Snrpn outside reverse, 5′-AAT AAA CCC AAA TCT AAA ATA TTT TAA TC-3′; Snrpn inside forward, 5′-AAT TTG TGT GAT GTT TGT AAT TAT TTG G-3′; Snrpn inside reverse, 5′-ATA AAA TAC ACT TTC ACT ACT AAA ATC C-3′; Rasgrf1 outside forward, 5′-TAA TTT TAG GTG TAG AAT ATG GGG TTG-3′; Rasgrf1 outside reverse, 5′-TAA AAA AAC AAA AAC AAC AAT AAC AAC TAA AAC AAA AAC AA-3′; Rasgrf1 inside forward, 5′-TAG AGA GTT TAT AAA GTT AG-3′; and Rasgrf1 inside reverse, 5′-ACT AAA ACA AAA ACA ACA-3′. PCR conditions for all primers were adopted from the work of Tremblay et al. (57). PCR products were cloned into the pCRII vector of the TOPO-TA cloning system (Invitrogen), and individual clones were sequenced.

Parent-of-origin-specific methylation of CD3-CMG H19ICR insertion on chromosome 9.At its endogenous location, the H19ICR is methylated on the paternally but not on the maternally inherited chromosome. This parent-of-origin difference arises via methylation of the ICR during spermatogenesis and is then maintained in all somatic tissues after fertilization (58). Likewise, an insertion of a second copy of the 2.4-kb H19ICR at the Igf2/H19 locus (H19R) (Fig. 1A) becomes normally methylated in both somatic tissue and testes (45). When the same 2.4-kb H19ICR element is inserted at the Afp locus (Afp-A) (Fig. 1B) on chromosome 5, it becomes methylated successfully upon paternal inheritance in somatic tissue, but this methylation is acquired only after fertilization. That is, when inserted on chromosome 5, all 65 of the assayed CpGs are unmethylated in sperm (45). In this study, we wanted to address the generality of these findings and also wanted to better understand the genetic basis for acquisition of DNA methylation at the ICR. We therefore established and characterized several novel mouse strains.

We first generated a knock-in mouse model (CD3-CMG) which carries the 2.4-kb H19ICR between the CD3 gamma and CD3 delta genes on mouse chromosome 9 (Fig. 1C). CD3 maps at 26 cM, in a nonimprinted region of the mouse genome. The only chromosome 9 genes known to be imprinted are Rasgrf1 (encoding RAS protein-specific guanine nucleotide-releasing factor 1) and A19, which map at 50 cM (12). Therefore, by choosing the CD3 locus, we were able to study the autonomous function of the H19ICR, excluding any likely interactions between the ICR and a neighboring imprinting regulatory element. The CD3 locus does not harbor any CpG islands, based on a computer search using the EMBL-EBI CpGPlot computer program (http://www.ebi.ac.uk/Tools/emboss/cpgplot/ ). The search was based on the definition of a CpG island by Takai and Jones (53). The closest CpG islands are located 60 kb upstream and 157 kb downstream of the inserted H19ICR. CD3 gamma contains 97 CpG dinucleotides and CD3 delta contains 61 CpG dinucleotides, scattered throughout their 10.8-kb and 4.5-kb gene bodies, respectively.

Using Southern hybridization, we studied parent-of-origin DNA methylation patterns of the CD3-CMG insertion in thymus, kidney, testis, and mature sperm. Five animals carried the CD3-CMG insertion on the maternal allele, and five animals carried the insertion on the paternal allele. Our data show that the H19ICR inserted into the CD3 locus is unmethylated on the maternal allele and methylated on the paternal allele in kidney DNA (Fig. 2B) and in the thymus (data not shown), just like the endogenous ICR. Thus, the 2.4-kb element is sufficient to effectively mark parental origin at this ectopic locus. Note that the DNA methylation patterns seen in Fig. 2 are dependent only upon the parent of origin and are entirely independent of the grandparental origin of the ICR insertion (data not shown).

At the endogenous ICR, the paternal allele-specific DNA methylation was already established during spermatogenesis, and the ICR was completely refractory to digestion with the methylation-sensitive enzyme AciI (see “Dom” and “Cas” bands in Fig. 2B). In contrast, the CD3-CMG insertion was digested by AciI, indicating that the ICR is relatively unmethylated in sperm (Fig. 2B). Thus, unlike the case for the endogenous ICR but like that for the ICR insertion at the Afp locus (45), methylation of the CD3-CMG insertion seen in somatic tissue was not simply maintenance of the methylation patterns created in sperm.

We did, however, note a 3.6-kb fragment in testis, but not in kidney (Fig. 2B), that is consistent with partial methylation of the CD3-CMG insertion in sperm. To test this hypothesis and to assess DNA methylation in all CpGs across the ICR, we performed bisulfite sequencing. Genomic DNA was isolated from the testes and mature sperm from four animals homozygous for the CD3-CMG insertion on chromosome 9 and the H19^Δ13 deletion on chromosome 7. H19^Δ13 is a 13-kb chromosomal deletion at the Igf2/H19 locus that removes the entire H19 gene and 10 kb of additional upstream sequences, including the entire H19ICR (31). Thus, the ectopic ICR insertion is the only copy of the H19ICR in these animals. Following bisulfite treatment, five regions within the H19ICR insertion were PCR amplified to characterize 59 of 67 CpG dinucleotides within the ICR (Fig. 3). Regions 1 and 2, normally proximal to the H19 promoter, showed only minimal methylation. However, promoter-distal regions 4 and 5 were heavily methylated on at least some chromosomes. This finding indicates that the 2.4-kb H19ICR becomes partially methylated in testis when inserted at the CD3 location and thus is somewhat different in its behavior from the Afp-A insertion reported by Park et al. (45), where 65 of the 67 CpGs analyzed were unmethylated in testes and sperm. Thus, the methylation of the H19ICR in sperm is affected by its neighboring sequences. At its endogenous location or when inserted just 12 kb downstream of its normal location (H19R) (Fig. 1A), all CpGs are methylated in sperm, and this methylation is maintained during embryogenesis. In contrast, when the H19ICR is inserted at normally nonimprinted sites, much (CD3 insertion) or all (Afp insertions) of the CpG methylation is acquired only after fertilization.

• Open in new tab
• Download powerpoint

FIG. 3.

DNA methylation of individual CpG dinucleotides within the ectopic H19ICR at CD3 in sperm (Sp) and testis (T). Genomic DNAs from animals carrying only the ectopic H19ICR in their genomes were analyzed by bisulfite sequencing. Numbers 1 to 5 indicate regions within the ICR that were individually amplified by PCR. Region 1 shows the most proximal and region 5 shows the most distal region relative to the endogenous H19 promoter. Each lane of circles represents one individual clone. Lanes drawn for different regions do not originate from the same chromosome. Open circles represent unmethylated CpGs, and closed circles represent methylated CpGs. Arrows beneath individual CpGs indicate AciI restriction sites harboring the given CpG within their sequence. The Snrpn and Rasgrf1 DMRs were used as positive controls to test the accuracy of the bisulfite assay. In sperm, the Snrpn DMR (36) is unmethylated and the Rasgrf1 DMR (33) is methylated.

Additional H19 sequences can rescue H19ICR methylation in sperm.We next asked whether adding more sequence from the endogenous imprinted environment would rescue methylation of an exogenous H19ICR in testis. Using the Afp-D mouse model (Fig. 1B), we had already noted that an additional sequence upstream of the H19ICR did not help to methylate the ICR in testis (45). We therefore searched for a candidate sequence downstream of the endogenous H19ICR. The H19 promoter harbors a G box which is very highly conserved in mammals but whose function remains unclear, because targeted deletion of the element on chromosome 7 does not result in any loss-of-imprinting phenotype (56). The H19 promoter also overlaps a conserved CpG island (Fig. 1A). We considered that these elements might function to efficiently target DNA methyltransferases during spermatogenesis.

Therefore, we generated the Afp-DCK mouse by inserting the H19ICR and 5 kb of additional downstream sequence, including the H19 promoter's G box and CpG island and the entire H19 gene body upstream of Afp (Fig. 1B). Parent-of-origin-specific methylation of the ectopic ICR was analyzed in mice in the H19^Δ13/H19^Δ13 background. In somatic tissue, the Afp-DCK insertion, like the Afp-A, Afp-D, and CD3-CMG insertions, was unmethylated on the maternal allele and methylated on the paternal allele (Fig. 2C, top panel). In testis (Fig. 2C, bottom left panel) and sperm (data not shown), however, the maternally inherited Afp-DCK insertion remained completely unmethylated. Thus, the additional 5 kb of sequence downstream of the endogenous H19ICR did not rescue the H19ICR methylation in testis.

We did note, however, that the paternally inherited Afp-DCK insertion became heavily methylated in a small but visible portion of testicular (Fig. 2C, bottom right panel) and sperm (data not shown) cells.

Finally, we analyzed the ICR's methylation pattern in testes isolated from animals homozygous for a bacterial artificial chromosome (BAC) transgene (H19-BAC1) (26) (Fig. 1A) carrying the H19 sequence from kb −7 to +140. This fragment is, of course, too large to target to a specific locus and so was generated by random insertion of a single-copy BAC. We had previously shown that H19 expression from this transgene is maternal allele specific (26). The H19ICR within this transgene becomes specifically methylated on the paternal allele in somatic tissue (data not shown). In Fig. 2C, we show that this transgene's ICR was fully methylated in testis (Fig. 2C). Thus, when sufficient sequence downstream of the H19 gene body is included, ectopic H19ICR insertions behave like the endogenous element.

Is DNA methylation on ectopic H19ICR inserts maintained after fertilization?The mammalian genome becomes demethylated after fertilization, except for methylation imprints that have been established in the germ line (40). DMRs that are not ICRs, such as the DMRs upstream of and within the Igf2 gene, do not escape the demethylation activity, and their methylation patterns are removed from the parental alleles (16). To test whether the H19ICR inserted at CD3 maintains its methylation acquired during spermatogenesis after fertilization and therefore behaves like the endogenous ICR on chromosome 7, we studied its methylation patterns in zygotes, morulae, and blastocysts (Fig. 4). We assayed the 3′ end of the ICR (region 5 in Fig. 3), which includes 15 CpGs and is the most methylated region in sperm. We found that the methylation patterns in sperm, morulae, and blastocysts were not distinguishable. Thus, the CD3-CMG insertion does escape the genomewide demethylation wave that occurs after fertilization and does indeed behave like the endogenous ICR.

• Open in new tab
• Download powerpoint

FIG. 4.

DNA methylation patterns of the exogenous H19ICR at CD3 in zygotes and in E3.5 morulae and blastocysts. The DNA region analyzed by bisulfite sequencing in this study covers 15 CpGs within the H19ICR, from kb −4.0 to −3.5, and includes two CTCF sites. This region corresponds to the promoter-distal region 5, whose methylation patterns in sperm and testis are shown in Fig. 3. Zygotes, morulae, and blastocysts had paternal alleles inherited from CD3-CMG fathers. Morulae and blastocysts were homozygous for the H19^Δ13 allele. Thus, the ectopic ICR was the only H19ICR in these genomes. Zygotes were in an H19⁺ (cas)/H19^Δ13 background. Clones associated with the ectopic H19ICR were identified using single nucleotide polymorphisms that distinguish castaneus and domesticus H19ICRs. Each circle represents one CpG, and each lane represents one clone. Unmethylated CpGs and methylated CpGs are depicted as open circles and filled circles, respectively.