Sox-Oct motifs contribute to maintenance of the unmethylated H19 ICR in YAC transgenic mice

Abstract

Abnormal methylation at the maternally inherited H19 imprinted control region (H19 ICR) is one of the causative alterations leading to pathogenesis of Beckwith–Wiedemann syndrome (BWS). Recently, it was shown in human BWS patients, as well as mouse cell culture experiments, that Sox-Oct motifs (SOM) in the H19 ICR might play a role in protecting the maternal ICR from de novo DNA methylation. By grafting a mouse H19 ICR fragment into a human β-globin yeast artificial chromosome (YAC) followed by analysis in transgenic mice (TgM), we showed previously that the fragment carried sufficient information to establish and maintain differential methylation after fertilization. To examine possible functions of the SOM in the establishment and/or maintenance of differential methylation, two kinds of YAC-TgM were generated in this study. In the ΔSOM TgM, carrying the mouse H19 ICR bearing an SOM deletion, a maternally inherited transgenic ICR exhibited increased levels of methylation around the deletion site, in comparison to the wild-type control, after implantation. In the λ + CTCF + b (LCb) TgM, carrying a 2.3 kb λ DNA fragment supplemented with the fragment b including the SOM and four CTCF binding sites, maternally and some of the paternally inherited LCb fragments were significantly less methylated when compared with a control λ + CTCF fragment that was supplemented only with additional CTCF sites; the λ + CTCF was substantially methylated regardless of the parent of origin after implantation. These results demonstrated that the SOM in the maternal H19 ICR was required for maintaining surrounding sequences in the unmethylated state in vivo.

INTRODUCTION

Mono-allelic gene expression (genomic imprinting) at the insulin-like growth factor 2 (Igf2)/H19 locus is controlled by an intergenic H19 ICR (imprinting control region) sequence (Fig. 1A), that is highly methylated only after paternal transmission. This differential (imprinted) DNA methylation pattern is established during gametogenesis and is maintained after fertilization, against the genome-wide demethylation and de novo methylation activities during pre- and post-implantation periods, respectively. The mouse H19 ICR has four binding sites for CTCF, an enhancer-blocking insulator protein, binding ability of which is sensitive to DNA methylation of its recognition sequence (1). Therefore, in somatic cells, the CTCF-bound, hypo-methylated maternal H19 ICR insulates the distal Igf2 gene from activation by the 3′ enhancer, located downstream of the H19 gene, resulting exclusively in H19 gene expression. In contrast, a hyper-methylated paternal H19 ICR is permissive for Igf2-enhancer communication, while it epigenetically silences the more proximal H19 gene promoter (2,3).

Figure 1.

Generation of YAC-TgM. (A) Genomic structure of the mouse Igf2/H19 locus (top). The Igf2 and H19 genes (open boxes) are ∼70 kb apart on mouse chromosome 7, and expression of both genes requires the shared 3′enhancer (shaded box). The H19 ICR (2.4 kb BglII fragment) (31) is located approximately from −2 to −4 kb relative to the transcription initiation site of the H19 gene. In the enlarged map, CTCF-binding sites (numbered from 1 to 4) and fragment b are shown as solid rectangles. (Bottom) The nucleotide sequences of the fragment b (19). The box encloses a deleted Sox-Oct motif sequence (SOM, 37 bp), in which Sox (single line) and Oct (double line) binding sites are underlined (20). (B) Schematic representation of the LCb/ΔSOM YAC transgene. The positions of the β-like globin genes (open boxes) are shown relative to the LCR (shaded box). SfiI restriction enzyme sites are shown as vertical lines. Probes (solid rectangles) used for long-range structural analyses shown in (C) and expected restriction fragments with their sizes are shown. The enlarged map shows the detailed structure of the LCb/ΔSOM fragment. The loxP5171 and loxP2272 sequences are indicated as solid and open triangles, respectively. CTCF-binding sites and the fragment b are shown as solid rectangles and the deleted SOM as a solid, inverted triangle. (C) Long-range structural analysis of the transgenes in the LCb/ΔSOM YAC-TgM. DNA from thymus cells was digested with SfiI in agarose plugs, separated by pulsed-field gel electrophoresis and Southern blots were hybridized separately to probes in (B). (D) Cre-loxP recombination in the YAC-TgM. (Left) In vivo cre-loxP recombination in the parental LCb/ΔSOM transgene generates either ΔSOM or LCb daughter transgenes. Positions of BamHI (B) restriction enzyme sites and expected restriction fragments with their sizes are shown. Recombination taking place between a set of loxP sites concomitantly removes additional loxP site and no further recombination occurs. (Right) Tail DNA from each YAC-TgM sub-lines was digested with BamHI, separated on agarose gel, and subjected to Southern blotting and hybridization to the HS1B probe (solid rectangles).

In order to assess the mechanisms controlling methylation imprinting establishment during gametogenesis and its protection against global demethylation/methylation activities during early embryogenesis, we inserted a 2.9 kb DNA fragment encompassing the mouse H19 ICR into a 150 kb yeast artificial chromosome (YAC) carrying the whole human β-globin locus and then generated transgenic mice (TgM). Upon paternal transmission, the transgenic H19 ICR was preferentially methylated in somatic cells, demonstrating that the 2.9 kb fragment carries sufficient information to establish and maintain paternal-allele-specific methylation (4). Unexpectedly, however, the fragment was not methylated in sperm, suggesting that the H19 ICR was molecularly distinguished by an unknown epigenetic mark other than DNA methylation during gametogenesis and that, after fertilization, the mark would be translated into a differential methylation pattern. Because the paternally inherited transgenic H19 ICR acquired DNA methylation by the blastocyst stage (5), we speculated that this post-fertilization methylation activity might be responsible for protecting the imprinted, endogenous H19 ICR from genome wide DNA demethylation after fertilization (6,7).

On the other hand, protection-against-methylation activity borne by the maternally inherited H19 ICR is also essential to maintain its differential methylation status. In addition to executing imprinted expression at the Igf2/H19 locus, CTCF is also required to maintain the hypomethylated state of the maternal H19 ICR in somatic cells. When the CTCF binding sites were disrupted in the endogenous (8–11) and transgenic H19 ICRs (4xMut) (5), aberrant DNA methylation was observed post-implantation in the maternally inherited mutant ICRs. To test if the CTCF sites were sufficient to maintain the hypomethylation status of the transgene, we introduced four copies of the CTCF recognition motif into a CpG-rich 2.3 kb λ bacteriophage DNA fragment (termed λ + CTCF) and generated YAC-TgM. Unexpectedly, the maternally inherited λ + CTCF fragment acquired DNA methylation after implantation. When the λ + CTCF sequence was inserted into the middle of the H19 ICR fragment, it was hypomethylated after maternal transmission (12). These results revealed that CTCF sites were not sufficient to confer methylation protective activity, and that the 2.9 kb H19 ICR harbors additional cis sequences that must be contained within this fragment.

Beckwith–Wiedemann syndrome (BWS: OMIM #130650), often characterized by macrosomia, macroglossia and abdominal wall defects, is caused by genetic and epigenetic mutations in the human chromosome 11p15.5, where the imprinted KIP2/LIT1 and IGF2/H19 gene clusters are located. Although it is known that hypermethylation of the H19 ICR and ectopic Igf2 expression in the maternal allele are associated with some (∼10%) cases of BWS, how such aberrant DNA methylation emerges is not completely clear. As predicted from its protection-against-methylation activity, CTCF-binding site mutations have been identified in BWS cases bearing H19 ICR hypermethylation (13–16). Recently, mutations in non-CTCF site, i.e. Oct or Sox pluripotency factor binding sites within the maternally inherited H19 ICR of some BWS patients have been reported (16–18). These clinical findings also demonstrated that CTCF-binding is essential, but not sufficient to protect the H19 ICR from de novo DNA methylation.

In search for DNA demethylation activity within the mouse H19 ICR fragment, Hori et al. (19,20) prepared sub-fragments of the H19 ICR, CpG-methylated them in vitro and then introduced them individually into epiblast-like P19 embryonal carcinoma cells. One of the fragments (‘b’) was found to be demethylated when introduced into that cell line, suggesting that the 164 bp fragment b carries an intrinsic DNA demethylation activity. Within the ‘b’ sequence, two copies of tandemly arrayed Sox-Oct binding motifs (SO motifs; SOM) were found, to which Oct-1, Oct-3/4 and Sox-2 factors bound in vitro. Importantly, deletion of these motifs from the fragment abrogated its demethylation activity, indicating that these cis motifs are required for the activity (19,20).

In this study, we generated two different YAC-TgM, loss and gain of function models, to test for possible in vivo functions of the SOM/fragment b in the establishment and/or maintenance of the hypomethylation status of the maternal H19 ICR. The results demonstrated that the SOM (in the fragment b of the maternal H19 ICR) plays an essential role in protecting surrounding sequences from de novo DNA methylation in vivo.

RESULTS

Generation of YAC-TgM

In BWS patients displaying aberrant maternal H19 ICR methylation, mutations in the Sox or Oct binding sites have been identified (16–18). In the mouse H19 ICR, tandemly arrayed Sox and Oct binding sites are present in the middle of the H19 ICR fragment (Fig. 1A) (19,20). To examine whether these motifs are essential to establish and/or maintain the differential methylation status of the mouse H19 ICR, we deleted 37 bp containing the SOM sequence from the 2.4 kb ICR fragment (ΔSOM, Fig. 1A and B). At the same time, to determine whether the fragment b contributes to the establishment and/or maintenance of hypomethylation in the H19 ICR, we introduced the fragment b (164 bp, including the SOM and its flanking sequences, Fig. 1A) (20) into the λ + CTCF segment (λ + CTCF + b or LCb, Fig. 1B) (12). To reduce the time required to obtain mouse lines carrying single-copy, intact transgenes, two fragments (ΔSOM and LCb) were individually floxed using hetero-specific loxP sequences (21) and combined to employ a co-placement strategy (22), and then introduced 3′ to the LCR in the human β-globin YAC (Fig. 1B). TgM were generated by pronuclear injection of the purified YAC DNA into fertilized eggs. Intact, single-copy transgene carriers were identified by long-range structural analyses of the genomic DNA using pulsed-field gel electrophoresis (Fig. 1C). Parental YAC-TgM lines (No. 1 and 112) were crossed with Cre-TgM to initiate in vivo cre-loxP recombination, which generated daughter lines carrying either the ΔSOM or LCb transgene at the identical chromosomal integration site (Fig. 1D).

DNA methylation status in the ΔSOM TgM

The methylation status of the ΔSOM fragment (Fig. 2A), as well as the control ICR’ fragment (Fig. 2D), was first determined by Southern blotting, in which methylation status of individuals were separately examined. Both fragments were hypo- (Fig. 2B and E) and hyper-methylated (Fig. 2C and F) in maternally and paternally inherited alleles, respectively, i.e. they were differentially methylated. However, when compared with the control ICR′ fragment (Fig. 2E), maternally inherited ΔSOM fragment acquired more methylation (Fig. 2B), extending from around CTCF site 1 through 4.

Figure 2.

DNA methylation status in the ΔSOM and ICR′ YAC-TgM analyzed by Southern blotting. (A and D) Partial restriction enzyme map of the β-globin YAC transgene with the inserted ΔSOM (A) or ICR’ (D) fragments. Methylation-sensitive HhaI sites in the DraI fragment are displayed as vertical lines beneath the map. HS1B’ probe used for Southern blot analysis in (B, C, E, F and H) is shown as a filled rectangle. Expected fragments after enzyme digestion are shown as horizontal lines with their sizes. (B, C, E and F) DNA methylation status of the ΔSOM (lines 1 and 112 in B and C) or ICR’ (lines 24 and 29 in E and F) transgenes. Tail DNA from YAC-TgM (1- to 2-weeks-old) that inherited the transgenes either maternally (B and E) or paternally (C and F) was digested with DraI alone (D) or DraI + HhaI (D + H) and the blots were hybridized with the α-³²P-labeled HS1B’ probe. Asterisks indicate the positions of methylated, undigested fragments. (G) Pedigree depicting maternal inheritance of the ΔSOM transgene. Non-transgenic male and transgenic female (hemi-zygous) individuals are denoted as open squares and solid circles, respectively. (H) Methylation status of the ΔSOM fragment in TgM depicted in (G) was determined by Southern blotting as described earlier.

The methylation status of the ΔSOM fragment (Fig. 3A) was then determined by bisulfite sequencing and was compared with that of the control ICR′ fragment (Fig. 3F). In somatic cells, the paternally inherited ΔSOM sequence was highly methylated (lines 1 and 112 in Fig. 3B), as was seen in the control TgM (lines 24 and 29 in Fig. 3G), demonstrating that the SOM is not essential for the paternal H19 ICR to acquire DNA methylation. In contrast, the maternally inherited fragment was aberrantly methylated, especially in the vicinity of the mutated sequences (Fig. 3C), when compared with the control lines (Fig. 3H). We next examined the methylation status of CpG sequences adjacent to the deletion sequence in ΔSOM fragment in blastocysts (Fig. 3D) and oocytes (Fig. 3E). At both developmental stages, the sequences were infrequently methylated, demonstrating that the maternally inherited ΔSOM fragment acquired its aberrant DNA methylation pattern during the post-implantation period. These results demonstrated that the SOM is required to fully protect its surrounding sequences from de novo DNA methylation, at least after the implantation stage.

Figure 3.

DNA methylation status in the ΔSOM and ICR′ YAC-TgM analyzed by bisulfite sequencing. (A and F) Partial maps of the β-globin YAC transgene with the inserted ΔSOM (A) or ICR’ (F) fragments. Four regions (I, II, III and IV) analyzed by bisulfite sequencing in (B–E, G and H) are shown as gray bars beneath each map. (B, C, G and H) Tail somatic DNA from ΔSOM (lines 1 and 112 in B and C) or ICR’ (lines 24 and 29 in G and H) YAC-TgM, inheriting the transgenes either paternally (B and G) or maternally (C and H) was pooled, digested with XbaI and treated with sodium bisulfite. The numbers in parentheses are ID (in Fig. 2) of samples pooled. The regions I, II, III or IV were amplified by nested PCR, which were subcloned and then sequenced. Here and in the following figures, methylated and unmethylated CpG motifs are shown as filled and open circles, respectively, and each horizontal row represents a single DNA template molecule. Gray and solid bars indicate the location of the CTCF-binding and fragment b or the SOM sites, respectively. The positions of CpG motifs in the HhaI sites used for Southern blot analysis in Figure 2 are shown by vertical lines in B and C. Distance between the HhaI and deletion sites are: a; 0.59 kb, b; 0.18 kb, c; 0.35 kb, d; 0.62 kb, e; 1.0 kb. Asterisks (in G and H) indicate the CpG sites that are present in the ICR’, but not in the ΔSOM sequences. (D) For analysis of the blastocyst cells, embryos at 3.5 days postcoitum that inherited the ΔSOM transgenes maternally (lines 1 and 112) were embedded in agarose beads and treated with sodium bisulfite. Individual beads were separately and directly used to amplify region II of the transgenes by nested PCR. The PCR products were subcloned and sequenced. The results from single beads are presented together in a cluster. (E) For analysis of female germ cells, MII oocytes of ΔSOM YAC-TgM (lines 1 and 112) were embedded in agarose beads and treated with sodium bisulfite. The beads were directly used to amplify region II of the transgene by nested PCR. The PCR products were subcloned and sequenced. (I) Tail somatic DNA from 4xMut YAC-TgM, inheriting the transgene maternally (lines 74 and 228) (5), was analyzed by bisulfite sequencing and the methylation status of the region II was determined.

Anticipation in the aberrant DNA methylation was not seen in the ΔSOM TgM

Recently, Berland et al. (18) reported a human case with a gradual increase in DNA methylation at the maternal H19 ICR over generations, in which single nucleotide mutation was found in Oct-binding site. Accordingly, BWS features grew in severity in the later generations. We therefore examined methylation status of the maternally inherited ΔSOM fragment over several generations (Fig. 2G and H). Southern blot analysis of the tail DNA demonstrated that anticipation of the phenotype was not seen at least in our transgenic mouse context.

DNA methylation status in the 4xMut TgM

We previously generated a human β-globin YAC-TgM carrying the 2.9 kb H19 ICR with all four of the CTCF binding sites mutated (4xMut) (5). In this mutant mouse, the H19 ICR sequences acquired methylation after implantation stages, even when maternally inherited, indicating that the CTCF sites are indispensable for maintaining the unmethylated state of the ICR also in the transgenic context. Because the ‘b’ sequence is intact in the 4xMut transgene, we examined the methylation status of sequences surrounding the SOM in maternally inherited mutant H19 ICR. As shown in Figure 3I, the region was highly methylated, suggesting that the fragment b requires CTCF binding sites to protect surrounding sequences from de novo DNA methylation.

Maternal DNA methylation status in the LCb (λ + CTCF + b) TgM

We next examined whether the fragment b was capable of protecting heterologous CpG-rich sequences from de novo DNA methylation in reconstruction assay system. Bacteriophage λ DNA inserted into the YAC and then incorporated into TgM acquired DNA methylation irrespective of its parental origin (23). Even when CTCF sites were introduced at the same location as in the H19 ICR, the λ DNA sequence (λ + CTCF, Fig. 4A) was substantially methylated even after maternal transmission (Figs 4B and 5B) (12), demonstrating that CTCF sites alone were not sufficient to confer anti-methylation protective activity. Because the λ + CTCF sequence was methylated neither in blastocysts (Fig. 5C) nor oocytes (Fig. 5D), it was assumed that the DNA methylation of the maternal allele was acquired after implantation. When the ‘b’ sequence, in addition to the CTCF sites, was introduced into the λ fragment (LCb, Fig. 4D) and analyzed by Southern blotting (Fig. 4D), maternally inherited LCb (Fig. 4E) was less efficiently methylated than the control λ + CTCF fragment (Fig. 4B).

Figure 4.

DNA methylation status in the λ + CTCF and LCb YAC-TgM analyzed by Southern blotting. (A and D) Partial restriction enzyme maps of the β-globin YAC transgene with the inserted λ + CTCF (A) or LCb (D) fragments. Methylation-sensitive BstUI sites in the BamHI fragment are displayed as vertical lines beneath the map. λ probe used for Southern blot analysis in (B, C, E and F) is shown as a filled rectangle. The dotted line indicates that this probe hybridizes to non-continuous regions of the LCb transgene. (B, C, E and F) DNA methylation status of the λ + CTCF (lines 2619 and 2653 in B and C) or LCb (lines 1 and 112 in E and F) transgenes. Tail DNA from YAC-TgM (1- to 2-weeks-old) that inherited the transgenes either maternally (B and E) or paternally (C and F) was digested with BamHI alone (B) or BamHI + BstUI (B + B) and the blots were hybridized with the α-³²P-labeled λ probe. Asterisks indicate the positions of methylated, undigested fragments.

Figure 5.

DNA methylation status in the λ + CTCF and LCb YAC-TgM analyzed by bisulfite sequencing (maternal allele). (A and E) Partial maps of the β-globin YAC transgene with the inserted λ + CTCF (A) or LCb (E) fragments. Three regions (V, VI, and VII) analyzed by bisulfite sequencing in (B–D and F) are shown as gray bars beneath each map. (B and F) Tail somatic DNA from λ + CTCF (lines 2619 and 2653 in B) or LCb (lines 1 and 112 in F) YAC-TgM, inheriting the transgenes maternally was analyzed as described in the legend to Figure 3. The numbers in parentheses are ID (in Fig. 4) of samples pooled. Regions V, VI or VII were amplified by nested PCR. The position of the BstUI sites used for Southern blot analysis in Figure 4 are shown by arrowheads. (C) For analysis of the blastocyst cells, embryos at 3.5 days postcoitum that inherited the λ + CTCF transgenes maternally (lines 2619 and 2653) were analyzed as described in the legend to Figure 3D. Region VI of the transgene was amplified by nested PCR. (D) For analysis of female germ cells, MII oocytes of λ + CTCF YAC-TgM (lines 2619 and 2653) were analyzed as described in the legend to Figure 3E. A part of region VI (VI’ in A) of the transgene was amplified by nested PCR. Asterisks in B, C and D indicate the CpG sites unique to the λ + CTCF sequences.

Because our Southern blotting strategy mainly examines methylation status around CTCF sites 1 and 3/4 of the transgene (Fig. 4A and D), we next pooled representative samples and analyzed them by bisulfite sequencing (Fig. 5). Methylation levels of the CpG sequences in the LCb fragment around CTCF sites 1 (sequenced region V in Fig. 5E) and 3/4 (region VII), as well as those in the λ + CTCF (12), were consistent with those determined by Southern blotting (Fig. 4), i.e. the former was significantly lower than the latter. Strikingly, methylation level around the insertion site of the ‘b’ sequence in the LCb fragment (sequenced region VI in Fig. 5E) was much lower than that of the corresponding region (region VI in Fig. 5A) in the λ + CTCF fragment (Fig. 5B and F).

These results demonstrated that the ‘b’ sequence of the H19 ICR bears an activity that protects the λ DNA sequence from de novo DNA methylation, at least after implantation.

Paternal DNA methylation status in the LCb (λ + CTCF + b) TgM allele

In order to examine the allele specificity of the protection-against-methylation activity exhibited by the fragment b, we next analyzed the methylation status of the λ DNA sequences in the paternally inherited allele. When analyzed by Southern blotting, some of the paternally inherited λ + CTCF fragments were highly methylated and others were partially methylated (Fig. 4A and C), while most of the maternally inherited λ + CTCF fragments were partially methylated (Fig. 4B). Bisulfite sequencing of the pooled samples revealed that both categories of paternally inherited samples were significantly methylated around the sequenced region VI (Fig. 6A and B). This DNA methylation pattern must have been acquired after the implantation stages since the sequence was not methylated in blastocysts (Fig. 6C). Next, methylation status of the paternally inherited LCb fragment was determined. Southern blot analysis of 23 samples (in total in lines 1 and 112) identified hyper- (9/23), partially (12/23) and hypo-methylated (2/23) sequences (Fig. 4F). We then pooled representative samples from each category of methylation pattern and analyzed them by bisulfite sequencing (Fig. 6D and E). The results clearly revealed that the hypo- and partially methylated samples were significantly less methylated, especially around the insertion site of the ‘b’ sequence, when compared with the control λ + CTCF fragment (Fig. 6B). Therefore, we concluded that the protection-against-methylation activity of the fragment b was allele-non-specific, at least in the transgenic context.

Figure 6.

DNA methylation status in the λ + CTCF and LCb YAC-TgM analyzed by bisulfite sequencing (paternal allele). (A and D) Partial maps of the β-globin YAC transgene with the inserted λ + CTCF (A) or LCb (D) fragments. Three regions (V, VI, and VII) analyzed by bisulfite sequencing in (B, C and E) are shown as gray bars beneath the each map. (B and E) Tail somatic DNA from λ + CTCF (lines 2619 and 2653 in B) or LCb (lines 1 and 112 in E) YAC-TgM, inheriting the transgenes paternally was analyzed as described in the legend to Figure 3. Representative samples from each category of methylation level, determined by Southern blotting in Figure 4, were pooled and individually subjected to bisulfite DNA sequencing. The numbers in parentheses are ID (in Fig. 4) of samples pooled. h, hyper-methylated; p, partially methylated in (B). (C) For analysis of the blastocyst cells, embryos at 3.5 days postcoitum that inherited the λ + CTCF transgenes paternally (lines 2619 and 2653) were analyzed as described in the legend to Figure 3D. Region VI of the transgene was amplified by nested PCR.

DISCUSSION

The results reported here clearly demonstrate that the SOM in the H19 ICR was essential to fully protect the maternal H19 ICR from de novo DNA methylation after implantation. We have previously shown that CTCF sites were also essential to maintain the hypomethylation status of the H19 ICR in YAC-TgM at the same developmental stage (5). To ask whether CTCF and Sox/Oct factors play redundant roles or if they function cooperatively (24), we examined the methylation levels in 4xMut YAC-TgM. Because the CpG sites surrounding the intact SOM were densely methylated in the maternal allele in somatic cells of these TgM (Fig. 3I), the ‘b’ sequence was not capable of protecting nearby CpG sequences from post-implantation de novo DNA methylation in the absence of CTCF binding sites. Furthermore, a maternally inherited λ + CTCF sequence (a λ DNA fragment carrying four CTCF binding sites) was significantly methylated in the YAC-TgM (Fig. 5B), indicating that CTCF sites alone cannot fully protect nearby sequences from DNA methylation. In addition, we have previously shown that the λ + CTCF sequence became protected from de novo DNA methylation when H19 ICR fragments were ligated on both ends (12). Therefore, it was suggested that cis elements exist in the H19 ICR that fulfill their function in cooperation with the CTCF sites, and we infer that the SOM in the fragment b is one such element. It has been reported that Oct4 is capable of forming a nucleosome-free region around its binding site, which then facilitates the recruitment of other factors to the region (25). CTCF may therefore be more readily accessible to the H19 ICR through the aid of Oct family factor binding. Alternatively, Oct factors may directly bind to CTCF (24) and recruit it to nearby CTCF binding sites in the H19 ICR (26).

Interestingly, we recovered several hypo-methylated λ sequences in paternally inherited LCb TgM (Figs 4F and 6E), which we never found among the λ + CTCF (without the ‘b’ sequence) YAC-TgM (Fig. 4C) (12). It is therefore conceivable that protection-against-methylation activity of the fragment b is not strictly allele-specific. However, since a paternally inherited wild-type H19 ICR (carrying the ‘b’ sequence) is highly methylated, a mechanism to ensure hypermethylation of the paternal H19 ICR, but not the LCb, must exist. CTCF is able to bind to an unmethylated, but not to a methylated H19 ICR sequence, and that this binding is essential to protect the sequences from global DNA methylation after implantation. We showed here that the ‘b’ sequence is not capable of exerting its protection-against-methylation function in the absence of nearby CTCF-binding sites in the 4xMut TgM (Fig. 3I). Taken together, CTCF may bind to a LCb sequence that is not yet methylated at the blastocyst stage, even after paternal transmission (Fig. 6C), and the ‘b’ sequence protects the fragment from de novo DNA methylation. On the other hand, because the paternally inherited H19 ICR is highly methylated at the blastocyst stage, CTCF cannot bind to the fragment and ‘b’ sequence may not be able to exert its function. Alternatively, Hori et al. (20) reported that Sox-Oct binding to the SOM itself was also sensitive to the methylation of the CpG dinucleotides within the Sox-recognition motif in vitro. It is therefore possible that the ‘b’ sequence of the H19 ICR is capable of sensing its methylation status at the blastocyst stage and exerts its function ‘allele-specifically’. In addition, the paternal H19 ICR as a whole may have dominant methylation activity over protection-against-methylation activity of the fragment b. In this case, the LCb did not exhibit significant differential methylation characteristics, probably because it lacks this dominant, paternal allele-specific methylation deposition activity.

Ideraabdullah et al. (27) deleted a 0.9 kb fragment, including the ‘b’ sequence, from the endogenous mouse H19 ICR (H19^ICRΔIVS) and found that differential methylation of the remaining H19 ICR sequences was properly established and maintained. Several explanations could account for the different effects on the methylation status of the mouse H19 ICR exerted by that 0.9 kb deletion and the 37 bp deletion in the ΔSOM mutation examined here. First, the smaller mutations sometimes have more significant effect than do the larger mutations, depending on the spatial arrangement of the remaining functional elements nearby. For example, the enhancer activity of the human β-globin LCR, which comprises of five core hypersensitive (HS) sites, was more severely compromised when ‘a core HS site’ (ca. 200 bp), rather than ‘a core site plus its flanking sequences’ (several kb), was deleted (28). This phenotype was interpreted to mean that the latter deletion allowed the formation of a smaller yet functional LCR complex while the former resulted in collapse of the complex because of remnant flanking sequences. In addition in the human BWS cases, while 1.4–1.8 kb microdeletions within the maternal IC1 (imprinting center 1; the human ortholog of the H19 ICR) allele were associated with its hypermethylation status, a 2.2 kb deletion was associated with stochastic DNA methylation (29). Because interaction between adjacent CTCF sites of the mouse H19 ICR has been demonstrated (30), the spacing between any remaining CTCF sites may be important for the function of the maternal H19 ICR, i.e. closely spaced CTCF sites in the H19^ICRΔIVS allele could compensate for the loss of the Sox/Oct motif in maintaining its hypomethylation status. Our observation that introduction of the ‘b’ sequence between two CTCF sites in the middle of the λ + CTCF fragment conferred demethylation activity to it (i.e. λ + CTCF + b) is consistent with this notion. A second possibility is that the 0.9 kb sequences harbor dormant DNA methylation activity that is normally masked by the demethylation activity of the maternal SOM, although such an activity has not been demonstrated in the H19 ICR. Finally, it is formally possible that the SOM mutation may have different effects on DNA methylation status in the transgenic versus endogenous H19 ICR context.

Curiously, extent of aberrant DNA methylation in human BWS patients with Oct-site mutation seems wider than that in our mutant animals (16,17). This difference may stem from the species difference, such as presence (in human) (18) or absence (in mouse, Fig. 2G and H) of the anticipation phenomenon or the numbers of Oct and Sox sites. It is also possible, in human BWS patients that only a case with severe epi-genotype (aberrant DNA methylation) has been preferentially found because of its significant symptoms. Experiment with the human locus transgene, such as YAC- or BAC-TgM, may disclose underlying molecular basis for the difference.

Differential methylation status can be generated by either protecting one allele from allele-non-specific de novo DNA methylation or by actively depositing methylation onto one allele, although the two are not mutually exclusive. We propose that, in pre-implantation stages, the paternal H19 ICR may be in dynamic equilibrium between global demethylation and H19 ICR-specific, post-fertilization methylation that we and others have described (4,31,32). On the other hand, both alleles are subject to de novo DNA methylation after implantation stages. Allele-specific (e.g. CTCF sites, because of its methylation-sensitive DNA binding property) and allele-non-specific anti-methylation protective activities (e.g. Oct/Sox sites in the fragment b) may cooperatively function to ensure differential methylation status at these stages and at those sites.

MATERIALS AND METHODS

Co-placement target vector (pHS1/loxP-5171-B-2272-5171-G-2272)

The backbone plasmid for the targeting construct, pHS1 (33), carried a human β-globin HS1 DNA fragment (from 13299 to 14250 nt; HUMHBB, GenBank) in the context of pRS306. The following double-stranded DNA (only the upper strand sequences are shown) fragment was subcloned into the HindIII site (at 13769) of the pHS1 to generate pHS1/loxP5171-2272: 5′-AAGCTTATAACTTCGTATAGTACACATTATACGAAGTTATGGATCCTAGGATCCATAACTTCGTATAGGATACTTTATACGAAGTTATAAGCTT-3′ (HindIII-loxP5171-BamHI-AvrII-BamHI-loxP2272-HindIII; enzyme sites are underlined and loxP sequences italicized). Next, the following two oligonucleotides were phosphorylated, annealed and ligated in the forward orientation with BamHI-digested pHS1/loxP5171-2272 to generate BamHI and BglII sites in between two loxP sites of the pHS1/loxP5171-2272: 5′-GATCCCAGATCT-3′ and 5′-GATCAGATCTGG-3′ (BglII). The plasmid was then digested with BamHI/BglII and ligated with the following double-stranded DNA (upper strand sequences): 5′- GGATCCCGGGGTACCGATAAATAAGCTTATAACTTCGTATAGGATACTTTATACGAAGTTATCCCGGGGATATCATAACTTCGTATAGTACACATTATACGAAGTTATGTTAACTACATCAGATCT-3′ (BamHI-SmaI-KpnI-HindIII-loxP2272-SmaI-EcoRV-loxP5171- HpaI-BglII). The resultant co-placement target vector, pHS1/loxP-5171-B-2272-5171-G-2272, carried 5′-loxP5171-BamHI-loxP2272-loxP5171-BglII-loxP2272-3′ sequences.

ΔSOM fragment

An H19 ICR DNA fragment with a 37 bp deletion containing the Sox-Oct motifs was generated by PCR with following primers using the murine H19 ICR DNA as a template; del-SOM-5S: 5′-TTCTGGCATCGAACCACATGCACT-3′ and del-SOM-3A: 5′-GAATCACTTAAGGAACCGCCAACAAGAAAGTC TGGTGATTTGC-(deletion)-ATCTCCGGCTCAGGGCT-3′. Following BspEI and AflII digestion, the mutant fragment was replaced with the corresponding portion of the pHS1/loxPw+/ICR(−)(4). In each cloning step, the correctness of DNA sequences was confirmed by DNA sequencing. The ΔSOM fragment (2340 bp, from 1126 to 3503 nt; AF049091, GenBank) was released by BglII digestion (31).

LCb (λ + CTCF + b) fragment

For embedding the ‘b’ sequence into the λ + CTCF fragment (DDBJ accession no. AB691538) (12), three fragments were PCR-amplified by using the following primer sets and bacteriophage λ or murine H19 ICR DNA as templates: L + CTCF-LS6: 5′-TGTCAGTCCGGAACTTTTCATTTA-3′ (BspEI) and L + C + b-3A1: 5′-ATTGTTTTGAATTCTGGTGATTGC-3′(EcoRI), L + C + b-5S1: 5′-TGATGTAGAATTCCTCAGC TGCCA-3′ (EcoRI) and L + C + b-3A2: 5′-CCCAATTAATCACTTAAGGAACCG-3′ (AseI), or L + C + b-5S2: 5′-CAATACATTAATATTGGACTCAAG-3′ (AseI) and L + CTCF-LA3: 5′-GTATGTACAACAATTGCATGTCCAGAG-3′ (BsrGI). After treating with appropriate restriction enzymes (in parentheses), three fragments were linked together and replaced with the corresponding portion of the pλ + CTCF (12). The LCb fragment was released by BamHI digestion.

Targeting vector and homologous recombination in yeast

The ΔSOM and the LCb fragments were introduced into BglII and BamHI sites of pHS1/loxP-5171-B-2272-5171-G-2272, respectively (Fig. 1B). The resultant targeting vector was linearized with SpeI (at nt position 13670 in HUMHBB) and used for mutagenizing the human β-globin YAC (A201F4.3). Successful homologous recombination in yeast was confirmed by Southern blot analyses with several combinations of restriction enzymes and probes.

Generation of YAC-TgM

Purified YAC DNA was microinjected into fertilized eggs from CD1 mice (ICR; Charles River Laboratories) and TgM in the founder offspring were screened by PCR and Southern blotting. The structural analysis of the YAC transgene was performed as described elsewhere (34), and revealed that both lines (No. 1 and 112) carried intact, single-copy transgene. The ICR’ TgM carries a pseudo-WT ICR fragment (2.4 kb) embedded in the human β-globin YAC (GenBank accession no. AB775805; detailed information on this YAC-TgM is available on request).

Animal experiments were performed in a humane manner and approved by the Institutional Animal Experiment Committee of the University of Tsukuba. Experiments were conducted in accordance with the Regulation of Animal Experiments of the University of Tsukuba and the Fundamental Guidelines for Proper Conduct of Animal Experiments and Related Activities in Academic Research Institutions under the jurisdiction of the Ministry of Education, Culture, Sports, Science and Technology of Japan.

Methylation analysis by Southern blotting

Genomic DNA was prepared from tail-tip cells (1- to 2-weeks-old) of TgM using standard procedures. Restriction enzyme-digested DNA was separated on an agarose gel, and transferred to a nylon membrane. The membrane was hybridized with the α-³²P-labeled probes and subjected to X-ray film autoradiography.

Methylation analysis by bisulfite sequencing

Genomic DNA was extracted from tail-tip (2-weeks-old; pool of 2–6), blastocyst cells, or MII oocytes of TgM. The DNA was treated with sodium bisulfite using the EZ DNA Methylation Kit (Zymo Research). Transgenic ICR-specific nested PCR, PCR product cloning and sequence analysis were performed as described previously (4,5). PCR primers used are summarized in the Supplementary Material, Tables S1 and S2.

SUPPLEMENTARY MATERIAL

Supplementary Material is available at HMG online.

FUNDING

This work was supported in parts by a research grant from the Mochida Memorial Foundation for Medical and Pharmaceutical Research and a Grant-in-Aid for Young Scientists (S) from the Ministry of Education, Culture, Sports, Science and Technology of Japan (KAKENHI Grant number 20678002) to K.T.

ACKNOWLEDGEMENTS

We thank Drs Doug Engel (University of Michigan) and Jörg Bungert (University of Florida) for critically reading the manuscript.

Conflict of Interest statement. None declared.

Author notes

†

These two authors contributed equally to this work.