Abstract
Monoallelic gene expression at the Igf2/H19 locus is controlled by paternal allele-specific DNA methylation of the imprinting control region (H19 ICR) that is established during spermatogenesis. We demonstrated that the H19 ICR fragment in transgenic mice acquires allele-specific methylation only after fertilization, which is essential for maintaining its allelic methylation during early embryogenesis. We identified a DNA element required for establishing postfertilization methylation within a 118 bp (m118) region. A previously generated knock-in mouse whose endogenous H19 ICR was substituted with the human H19 ICR (hIC1; 4.8 kb) sequence revealed that the hIC1 sequence was partially methylated in sperm, although this methylation was lost by the blastocyst stage, which we assume is due to a lack of an m118-equivalent sequence in the hIC1 transgene.
To identify a cis sequence involved in postfertilization methylation within the hIC1 region, we generated three transgenic mouse lines (TgM): one carrying an 8.8 kb hIC1 sequence joined to m118 (hIC1+m118), one with the 8.8 kb hIC1 and one with the 5.8 kb hIC1 sequence joined to m118 (hIC1–3′+m118). We found that the hIC1–3′ region was resistant to de novo DNA methylation throughout development. In contrast, the 5′ portion of the hIC1 (hIC1–5′) in both hIC1+m118 and hIC1 TgM were preferentially methylated on the paternal allele only during preimplantation. As DNA methylation levels were higher in hIC1+m118, the m118 sequence could also induce imprinted methylation of the human sequence. Most importantly, the hIC1–5′ sequence appears to possess an activity equivalent to that of m118.
Introduction
Genomic imprinting is an epigenetic phenomenon in mammals that causes the transcription of genes to differ substantially depending on their parental origin. This mono-allelic gene expression pattern is regulated by imprinting control regions (ICRs), which are cis-regulatory DNA sequences frequently found either within or surrounding clustered imprinted gene loci that have different DNA methylation patterns depending on their parental origin. ICR sequences that acquire germline-specific DNA methylation during gametogenesis are called gametic differentially methylated regions (gDMRs). While most gamete-derived DNA methylation is lost after fertilization as part of chromatin reprogramming, allelic methylation within ICRs is faithfully maintained to control imprinted gene expression. Therefore, a specific mechanism must exist that maintains allelic methylation at restricted sequences in spite of genome-wide epigenetic reprogramming during preimplantation development.
The DNA methylation-dependent function of murine H19 ICR on distal chromosome 7, a DMR in the insulin-like growth factor 2 (Igf2)/H19 gene locus, has been intensively studied. The H19 ICR, as well as its human ortholog, Imprinting Center 1 (IC1, Fig. 1A), do not acquire DNA methylation on the maternal allele in either oocytes or after fertilization. Remaining unmethylated permits the interaction of these alleles with CCCTC binding factor (CTCF) to bring about insulator activity in somatic cells. Consequently, activation of the Igf2 promoter by a shared downstream enhancer is blocked, whereas the maternal allele of the H19 non-coding RNA gene is activated. In addition, CTCF, as well as Sox/Oct binding, is essential for protecting the unmethylated maternal copy of H19 ICR from de novo methylation during the postimplantation period (1–4). In contrast, the H19 ICR sequence in pro-spermatogonia acquires DNA methylation that is maintained after fertilization to inhibit its binding to CTCF, allowing for preferential Igf2 promoter activation by the shared enhancer on the paternal allele. The failure to establish, maintain or erase DNA methylation and consequent abnormal expression of imprinting genes leads to developmental collapse and human imprinting disorders, such as Beckwith-Wiedemann overgrowth syndrome (BWS; OMIM130650) and Silver-Russell growth retardation syndrome (SRS; OMIM180860).
Generation and structural analyses of YAC-TgM. (A) (Top) Genomic structure of the human IGF2/H19 locus. Expression of both genes (solid boxes) depends on the shared 3′ enhancer (double ovals). The solid and open rectangles indicate the positions of Sox/Oct and CTCF binding sites, respectively (some are presumptive). The extent of the human Imprinting Center 1(IC1) sequence that was employed in the study by Hur et al. (12) and in this study is shown below the map by thick horizontal lines. (Middle) Schematic representation of the YAC transgenes. The positions of the β-like globin genes (solid boxes; transcribed in the right to left direction) are shown relative to the LCR (gray box). SfiI and PstI (only those in the YAC arms) enzyme sites are shown by vertical lines. Probes (solid rectangles) used for Southern blot analyses in panels B and C and expected restriction enzyme fragments (SfiI) with their sizes (open bars) are shown. The enlarged map shows the detailed structure of the ‘hIC1+m118’ fragment that was inserted between the LCR and the ε-globin gene. The positions of loxP5171 and loxP2272 are indicated as solid and open triangles, respectively. (Bottom) Segments of the transgenes integrated in each TgM line (139 and 171) are indicated by thick horizontal lines. (B) End-fragment analysis of the transgenes. Somatic (tail tip) DNA of the YAC-TgM was digested with PstI and Southern blots were hybridized separately to the probes. (C) Long-range structural analyses of the transgenes. DNA from thymus cells was digested with SfiI in agarose plugs and separated by pulsed-field gel electrophoresis, and Southern blots were hybridized separately to probe. Thin vertical lines in panels B and C indicate that lanes were run on the same gel but were noncontiguous. (D) In vivo Cre-loxP recombination in the parental ‘hIC1+m118’ transgene generates either ‘hIC1 (=∆m118)’ or ‘hIC1–3′+m118 (=∆5′)’ daughter transgenes. Positions of AvaII sites and expected fragment sizes are shown. Positions of primers (arrows A~C) and HS1–3′ probe (hatched rectangle) used for PCR (in panel E) and Southern blot (in panel F) analyses, respectively are shown. (E) Tail DNA from each YAC-TgM sublines was subjected to PCR analyses by using primer sets shown above each panel. (F) Tail DNA was digested with AvaII, separated on agarose gel, and Southern blot was hybridized to the HS1–3′ probe.
We previously generated a yeast artificial chromosome (YAC) TgM, in which a 2.9 kb mouse H19 ICR fragment was inserted into the non-imprinted human β-globin locus YAC (150 kb, (5)). Although the transgenic H19 ICR was not methylated in sperm, it starts to acquire paternal allele-specific DNA methylation soon after fertilization (6). We therefore assumed that a putative epigenetic mark other than DNA methylation is deposited within the H19 ICR during gametogenesis to distinguish its parental origin for postfertilization imprinted DNA methylation (5,6). We went on to show that an H19 ICR fragment with its 5′-segment truncated (765 bp) could not be methylated after fertilization in the TgM context. Importantly, the same mutation at the endogenous H19 ICR impaired maintenance of paternal allele hypermethylation after fertilization and caused SRS-like growth retardation without affecting its methylation status in sperm (6). We therefore concluded that specific sequences within the 5′-segment of the H19 ICR is probably involved in its de novo methylation and helps maintain the methylation of the paternal H19 ICR during preimplantation development. We have narrowed down the responsible sequence to a 118 bp region (m118) and successfully recapitulated allele-specific DNA methylation of the lambda DNA sequence by incorporating m118, as well as CTCF binding sites and Sox/Oct motifs, at both transgenic and endogenous loci in mouse (7,8).
Although both the mouse and human H19 ICRs carry CTCF and Sox/Oct binding sites, their overall sequence homology is low (9,10). Therefore, we could not identify an orthologous m118 sequence in hIC1, and imprinted methylation of mouse and human H19 ICRs may not be established or maintained by the same molecular mechanisms.
Jones et al. (11) generated a TgM carrying the 100 kb hIC1 fragment, yet failed to recapitulate the expected DNA methylation pattern; they observed methylation in neither male germ cells nor on the paternal allele in somatic cells of single-copy TgM. Recently, Hur et al. (12) replaced the endogenous 3.8 kb murine H19 ICR sequence with a 4.8 kb hIC1 sequence to test hIC1 functions at the orthologous mouse locus. Even in this knock-in mouse, the paternally inherited hIC1 in somatic cells did not exhibit significant DNA methylation at any developmental stages studied. They found that the hIC1 sequence was only partially methylated in sperm of mutant mice, the level of which, however, became lower in two-cell embryos and was eventually lost by the blastocyst stage. We assumed that the loss of methylation in their experimental setup was due to the lack of a postfertilization, imprinted methylation mechanism because m118 was removed from the locus by knock-in replacement (12). In addition, the putative orthologous m118 sequence of hIC1 may reside outside the 4.8 kb hIC1sequence, because this 4.8 kb sequence corresponds roughly to the allele-specifically methylated portion of the human Igf2/H19 locus, while m118 is located within the 5′ non-DMR portion of the mouse locus (8).
In this study, we extended the range of the hIC1 sequence to ~9 kb and asked if it acquires gametic and/or postfertilization DNA methylation in the TgM context. In addition, we tested if m118 can confer postfertilization methylation imprinting even on the hIC1 sequence. We observed no methylation of the 9 kb hIC1 sequence in the sperm, demonstrating that either a longer sequence was necessary or that hIC1 was somehow resistant to gametic methylation in mouse. After fertilization, paternal allele-specific methylation was established in the hIC1 sequence under the control of m118 in a region-specific manner; however, this methylation disappears as development proceeds. Most importantly, the hIC1 sequence alone (i.e. in the absence of m118) can acquire paternal allele-specific, postfertilization methylation, demonstrating that the human sequence carries the orthologous activity for postfertilization methylation imprinting.
Results
Generation of YAC transgenic mice harboring hIC1
Hur et al. (12) generated a humanized Igf2/H19 knock-in locus in mouse by using the 4.8 kb hIC1 sequence (Fig. 1A, top) and found that it is partially methylated in the mouse sperm. The observed methylation is likely under the control of mouse sequences surrounding the hIC1 knock-in site. To facilitate the detection of potential postfertilization imprinting involving DNA methylation in our test fragments, we employed a TgM strategy, as gametic methylation was not anticipated (5,13). The m118 sequence was joined to the 8.8 kb hIC1 (3.0 kb hIC1–5′ + 5.8 kb hIC1–3′) in order to determine if it affects the methylation of the human DNA sequence. By adopting a transgene co-placement strategy (14), either of the upstream sequences, m118 or hIC1–5′, could be removed later. The floxed, chimeric fragment (hIC1+m118) was inserted into a human β-globin YAC and used to generate TgM (Fig. 1A, middle). Two independent single-copy TgM lines (lines 139 and 171) were established and the structure of integrated YAC transgenes was analyzed by end-fragment (Fig. 1B) as well as long-range Southern blot analyses (Fig. 1C). Both lines contain sequences stretching at least from LCR-5′ to the β-globin regions (Fig. 1A, bottom). Each TgM line was subsequently crossed with cre-TgM to allow excision of either m118 or the hIC1–5′ sequences (Fig. 1D). Offspring were screened for successful homologous recombination events by polymerase chain reaction (PCR) (Fig. 1E). After eliminating the Cre transgene by mating, the fixed genotypes of hIC1+m118, hIC1 and hIC1–3′+m118 mice were confirmed by Southern blotting (Fig. 1F).
DNA methylation analysis of hIC1+m118 TgM
The methylation status of somatic (tail tip) cell DNA of the hIC1+m118 TgM was analyzed by bisulfite sequencing (Fig. 2). The hIC1–5′ and 5′ end of the hIC1–3′ sequences were highly methylated after both paternal and maternal transmission (regions I, II and IV), except for the partially methylated region III, which contains the Sox/Oct and CTCF motifs. In contrast, both 5′ (regions V and VII) and 3′ (region VI) portions of the hIC1–3′ sequence were hypomethylated irrespective of their parental origin, which is consistent with that by Hur et al. (12). These results were also confirmed by methylation sensitive Southern blotting (Supplementary Material, Fig. S1). Thus, the methylation status of the transgenic hIC1 sequence in mouse somatic cells differed significantly from that in human cells.
DNA methylation status of the hIC1+m118 fragment in somatic cells of YAC-TgM. (A) Map of the hIC1+m118 transgene fragment. The m118, hIC1–5′ and hIC1–3′ sequences are denoted by solid, gray and open boxes, respectively. The regions analyzed by bisulfite sequencing in B are denoted by thick horizontal lines (I–VII). (B) Tail genomic DNAs of TgM inheriting the transgene either paternally or maternally were subjected to bisulfite sequencing, the methylation levels of which were also determined by Southern blotting in Supplementary Material, Fig. S1. Each horizontal row represents a single DNA template molecule. Methylated and unmethylated CpG motifs are shown as filled and open circles, respectively. Positions of Sox/Oct and CTCF binding sites are shown by solid and open rectangles, respectively. Calculated methylation levels (%) are shown for each cluster.
To determine whether the lack of differential methylation at the hIC1 sequence in tail somatic cells is due to loss of its establishment or a failure to maintain it after it has been established, we conducted bisulfite sequencing analysis at an earlier stage of development (Fig. 3). In the blastocyst stage embryo, the 5′ portion of the paternally inherited transgene (regions I and II of the hIC1–5′ in Fig. 3A) was moderately methylated while the maternal allele was almost devoid of methylation (Fig. 3B). Therefore, allele-specific methylation of the 5′ portion of the transgene was previously established, but was subsequently lost, presumably due to allele-nonspecific de novo DNA methylation during the postimplantation period. On the other hand, regions IV and VII of the hIC1–3′ sequence were hypomethylated regardless of its parental origin (Fig. 3B), indicating that DNA methylation was not acquired even after paternal transmission at this region.
DNA methylation status of the hIC1+m118 fragment in blastocysts of YAC-TgM. (A) Map of the hIC1+m118 transgene fragment. The regions analyzed by bisulfite sequencing in B are denoted by thick horizontal lines (I, II, IV and VII). (B) Blastocysts that inherited the transgenes either paternally or maternally were pooled, embedded in agarose beads and treated with sodium bisulfite. The beads were used to amplify the regions shown in A by PCR. PCR products were individually subcloned and sequenced. The results from single beads are presented together in a cluster.
Next, to determine if the paternal allele-specific methylation detected in blastocyst stage embryos was established during spermatogenesis, we conducted bisulfite sequencing analyses of the transgene (Fig. 4A) in sperm (Fig. 4B), as well as in oocytes (Fig. 4C). DNA methylation at region I was observed in neither of these cell types, demonstrating that differential DNA methylation of the 5′ segment of the hIC1–5′ sequence was established after fertilization. This result was quite different from that of Hur et al. (12), because they observed partial methylation of the hIC1 sequence corresponding to the hIC1–3′ region (Fig. 4A) in sperm of the humanized knock-in mouse. We therefore assume that the partial methylation they observed is governed by the mouse endogenous sequence around the knock-in site of the Igf2/H19 locus.
DNA methylation status of the hIC1+m118 fragment in gametic cells of YAC-TgM. (A) Map of the hIC1+m118 transgene fragment. The regions analyzed by bisulfite sequencing in B are denoted by thick horizontal lines (I, III, V and VI). (B) Sperm DNA was treated with sodium bisulfite and used to amplify the regions shown in A by PCR. PCR products were individually subcloned and sequenced. (C) Pooled oocytes embedded in agarose beads were analyzed by bisulfite sequencing as described in the legend to Figure 3B.
DNA methylation analysis of hIC1 TgM
We previously demonstrated that the m118 sequence of the mouse H19 ICR is essential for maintaining its differential DNA methylation status after fertilization, and that this sequence conferred postfertilization, allele-specific methylation of the lambda DNA sequence (8). To determine if imprinted methylation found in the hIC1+m118 YAC TgM was governed by the accompanying m118 sequence, we removed m118 and analyzed the methylation status of the remaining hIC1 sequence (Fig. 5A).
DNA methylation status of the hIC1 fragment in somatic cells of YAC-TgM. (A) Map of the hIC1 transgene fragment. The hIC1–5′ and hIC1–3′ sequences are denoted by gray and open boxes, respectively. The regions analyzed by bisulfite sequencing in B are denoted by thick horizontal lines (I–VII). (B) DNA methylation status of the transgene was determined by bisulfite sequencing analysis as described in the legend to Figure 2. Positions of Sox/Oct and CTCF binding sites are shown by solid and open rectangles, respectively. Calculated methylation levels (%) are shown for each cluster.
To determine their DNA methylation statuses in somatic cells, the paternally and maternally inherited hIC1 Tg sequences in the tail tip samples were subjected to bisulfite sequencing (Fig. 5B). Except for the hypomethylated region III, the hIC1–5′ sequences were hypermethylated, regardless of their parental origin (regions I and II). On the contrary, the hIC1–3′ sequence (regions V–VII), except for the region IV was hypomethylated after both paternal and maternal transmission, suggesting that the sequence is highly resistant to postimplantation, de novo methylation activity (Fig. 5B). This methylation pattern, which was also confirmed by methylation sensitive Southern blotting (Supplementary Material, Fig. S2), resembles that of the hIC1+m118 Tg (Fig. 2).
In the blastocyst stage embryos (Fig. 6), the 5′ portion of the hIC1–5′ sequence (region I) acquired paternal allele-specific DNA methylation (Fig. 6B), although the level of methylation tended to be lower than that in the hIC1+m118 transgene (P = 0.129 and 0.016 for lines 171 and 139, respectively, Fig. 3B), suggesting that the m118 sequence carried preferential DNA methylation activity over the hIC1 sequence. More importantly, the result also demonstrated that the hIC1–5′ sequence itself harbored allele-specific DNA methylation activity. The 5′ end of the hIC1–3′ sequence (regions IV and VII) was hypomethylated regardless of its parental origin (Fig. 6B), which was also the case for the hIC1+m118 transgene (Fig. 3B). As regions I, II and IV were highly methylated in tail somatic cells (Fig. 5B), it was suggested that the sequence lacked protection-against-de novo DNA methylation activity during the postimplantation period.
DNA methylation status of the hIC1 fragment in blastocysts of YAC-TgM. (A) Map of the hIC1 transgene fragment. The regions analyzed by bisulfite sequencing in B are denoted by thick horizontal lines (I, II, IV and VII). (B) Blastocysts that inherited the transgenes either paternally or maternally were pooled and embedded in agarose beads and treated with sodium bisulfite. The beads were analyzed as described in the legend to Figure 3B.
To determine the developmental timing of DNA methylation acquisition at the hIC1–5′ region (Fig. 7A), we performed bisulfite sequencing analysis of two-cell stage embryos. As shown in Figure 7B, the paternally inherited transgene at region I was slightly methylated, though the level of methylation was lower than that in the blastocyst stage embryos (Fig. 6B). Since the same sequence was almost devoid of methylation in sperm (Fig. 7C), this suggests that DNA methylation at the paternally inherited transgenic hIC1–5′ sequence commenced soon after fertilization.
DNA methylation status of the hIC1 fragment in 2 cells and sperm of YAC-TgM. (A) Map of the hIC1 transgene fragment. The region I (thick horizontal line) was analyzed by bisulfite sequencing in B and C. (B) Two cell stage embryos that inherited the transgene paternally were pooled and embedded in agarose beads. The beads were analyzed as described in the legend to Figure 3B. (C) Sperm DNA was treated with sodium bisulfite and used to amplify the region I by PCR. PCR products were individually subcloned and sequenced.
DNA methylation analysis of hIC1–3′+m118 TgM
Thus far, we have shown that hIC1–3′ sequence is likely resistant to DNA methylation at all the developmental stages we studied. However, it is possible that DNA methylation activity of the m118 sequence is insufficient to reach the hIC1–3′ region in the hIC1+m118 TgM, because the intervening hIC1–5′ region (3 kb) is larger than the mouse H19 ICR. If so, m118 may introduce allele-specific DNA methylation into the hIC1–3′ sequence when it is directly joined to the target sequence.
To test this hypothesis, we removed the hIC1–5′ sequence from the transgene, and placed m118 adjacent to the hIC1–3′ sequence (Fig. 8A). Bisulfite sequencing analysis of the hIC1–3′+m118 transgene in tail somatic cells indicated that the 5′ end of the transgene alone (region IV in Fig. 8A) was highly methylated with no significant bias based on parental origin (Fig. 8B). Southern blot analysis confirmed our bisulfite sequencing results (Supplementary Material, Fig. S3). In addition, the analysis revealed that the middle part of the hIC1–3’ was moderately methylated in a manner independent of parental origin (Supplementary Material, Fig. 3B and C).
DNA methylation status of the hIC1–3′+m118 fragment in somatic cells of YAC-TgM. (A) Map of the hIC1–3′+m118 transgene fragment. The m118 and hIC1–3′ sequences are denoted by solid and open boxes, respectively. The regions analyzed by bisulfite sequencing in B are denoted by thick horizontal lines (IV–VII). (B) DNA methylation status of the transgene was determined by bisulfite sequencing analysis as described in the legend to Figure 2. Positions of Sox/Oct and CTCF binding sites are shown by solid and open rectangles, respectively. Calculated methylation levels (%) are shown for each cluster.
We went on to conduct bisulfite sequencing analysis of the hIC1–3′+m118 transgene in the blastocyst stage embryos (Fig. 9). Allele-preferential methylation was observed at regions IV, VII and V of the paternally inherited transgene, demonstrating that the hIC1–3′ sequence was properly subjected to methylation imprinting under the control of m118. In addition, m118’s activity appeared to be distance-dependent because the region VII did not exhibit significant allele-specific methylation within the hIC1+m118 transgene (Fig. 3).
DNA methylation status of the hIC1–3′+m118 fragment in blastocysts of YAC-TgM. (A) Map of the hIC1–3′+m118 transgene fragment. The regions analyzed by bisulfite sequencing in B are denoted by thick horizontal lines (IV, V and VII). (B) Blastocysts that inherited the transgenes either paternally or maternally were pooled and embedded in agarose beads and treated with sodium bisulfite. The beads were analyzed as described in the legend to Figure 3B.
Finally, we analyzed the methylation status of the hIC1–3′+m118 sequence in germ cells (Fig. 10). The sequences were not methylated in either sperm or oocytes, which suggests that differential methylation of the hIC1–3′+m118 Tg in blastocyst stage embryo (Fig. 9) is established after fertilization.
DNA methylation status of the hIC1–3′+m118 fragment in gametic cells of YAC-TgM. (A) Map of the hIC1–3′+m118 transgene fragment. The region VII (thick horizontal line) was analyzed by bisulfite sequencing in B and C. (B) Sperm DNA was treated with sodium bisulfite and used to amplify the region VII by PCR. PCR products were individually subcloned and sequenced. (C) Pooled oocytes embedded in agarose beads were analyzed by bisulfite sequencing as described in the legend to Figure 3B.
Discussion
We determined the DNA methylation dynamics of transgenic hIC1 fragments in mice, which are summarized in Figure 11. We did not detect any allelic differences in hIC1 DNA methylation status in tail somatic cells, i.e. hIC1–5′ region was bi-allelically methylated, while both copies of the hIC1–3′ region were unmethylated. In the early stages of development, methylation imprinting was observed in blastocysts, which was, however, lost at some point after the implantation period.
Summary of DNA methylation dynamics of the hIC1 sequences in YAC-TgM. (A) The hIC1+m118 sequence did not acquire methylation in germ cells. After fertilization, only the paternal hIC1–5′ region acquired DNA methylation. However, such asymmetric methylation was lost as both alleles acquired de novo DNA methylation after implantation period. Even under these circumstances, the hIC1–3′ region on both alleles was not methylated throughout development. Positions of regions analyzed by bisulfite sequencing (thick horizontal lines) and Zfp57 binding motifs (vertical lines) are shown at the bottom. (B) The hIC1 sequence did not acquire methylation in the sperm. After fertilization, only the paternal hIC1–5′ region acquired DNA methylation, although to a lesser extent than in the presence of m118 sequence. This asymmetric methylation was lost as both alleles acquired de novo DNA methylation after implantation period. Again, the hIC1–3′ region on both alleles was not methylated throughout development. (C) The hIC1–3′+m118 sequence did not acquire methylation in germ cells. After fertilization, only the 5′ portion of the paternal hIC1–3′ sequence acquired DNA methylation, presumably under the influence of m118 sequence. This asymmetric methylation was lost as 5′ end of the hIC1–3′ sequence were methylated, while the other region became hypomethylated, on both alleles after implantation period.
Recently, Hur et al. (12) found that the 4.8 kb hIC1 sequence knocked-in at the mouse endogenous Igf2/H19 gene locus acquired DNA methylation in sperm. We believe that this phenotype is not attributable to the intrinsic activity of the 4.8 kb hIC1 sequence because gametic methylation of the mouse H19 ICR fragment did not happen in our TgM model (5,13) and the endogenous sequences surrounding the Igf2/H19 locus appear to be responsible for its gametic methylation (8). Consistent with this idea, we found that an 8.8 kb hIC1 sequence, including the whole 4.8 kb sequence, did not acquire DNA methylation in the TgM sperm (Fig. 11A and B). Therefore, we did not successfully identify the hIC1 sequence responsible for regulating autonomous DNA methylation in sperm in this study.
On the other hand, an unmethylated 8.8 kb hIC1 sequence acquired DNA methylation after fertilization in an allele-specific fashion, although only on its 5′ portion (Fig. 11B). This result indicates that the ‘postfertilization methylation imprinting activity’ resides within the 8.8 kb hIC1 sequence (Fig. 11B). Furthermore, when the hIC1 sequence was supplemented with m118 (hIC1+m18), the differential methylation status of the sequence became even more obvious (Fig. 11A), demonstrating that m118 could also act on hIC1, as was also the case for lambda DNA (8).
Paternal allele-specific methylation of the hIC1 Tg was pronounced in its 5′ portion and not detectable in the hIC1–3′ region in blastocyst stage embryos (Fig. 11A and B). This was due, at least in part, to the distance dependency of the methylation imprinting activity of m118 and presumably, the hIC1–5′ as well. Consistent with this notion, the 5’portion of the paternal hIC1–3′ sequence became methylated when it was placed proximal to the m118 (Fig. 11C). Because the m118 sequence of the mouse H19 ICR is located outside of the 5′ border of its DMR extent, the human sequence orthologous to m118 may be located in the 5′ portion of the hIC1–5′ sequence.
Unlike the mouse H19 ICR transgene, where differential DNA methylation was established during preimplantation and maintained faithfully thereafter (5), the methylation imprinting established at the hIC1 transgenes in blastocyst stage embryos was eventually lost because of gaining DNA methylation on the maternal allele (Fig. 11A–C). Maintenance of the hypomethylated state of the maternal H19 ICR has been shown to depend on the presence of CTCF and Sox/Oct binding motifs (1–4). A small number of consensus binding motifs for these factors in the hIC1–5′ region may cause de novo DNA methylation in this region after the implantation period. In addition, while the Oct consensus binding motif is well conserved in the hIC1–5′ region, the Sox consensus motif is poorly conserved (15). This may destabilize binding of Sox/Oct proteins to the hIC1–5′ region and allow its de novo DNA methylation after the postimplantation period, even on the maternal allele.
In contrast, both maternal and paternal copies of the hIC1–3′ region in the TgM were hypomethylated even after the implantation period when genome-wide de novo DNA methylation activity becomes dominant. As mentioned above, CTCF and Sox/Oct motifs are essential for maintaining the hypomethylated state of the maternal H19 ICR/hIC1; there are seven CTCF and two Sox/Oct motifs in the hIC1–3′ region. Hori et al. (16) reported that Sox/Oct binding motifs within the mouse H19 ICR are essential for its DNA demethylation activity in mouse embryonal carcinoma P19 cells. They also proposed that CpG methylation of the Sox binding motifs in the mouse H19 ICR interferes with its binding to Sox/Oct proteins, thereby inhibiting active DNA demethylation, and therefore maintaining the hypermethylated state of the paternal H19 ICR (15). In the hIC1 sequence, however, CpG motifs are not found in the Sox binding motifs within the hIC1–3′ region. Therefore, the lack of methylation in the hIC1–3′ region on both paternal and maternal alleles during gametogenesis and the preimplantation period may enable its continuous binding to Sox/Oct and CTCF proteins, thereby maintaining its hypomethylated state by antagonizing de novo DNA methylation after implantation.
While maternally inherited deletions within the IC1 are frequently associated with its hypermethylated state, likely due to a reduced number of CTCF/Oct/Sox binding motifs, the same mutation is rarely associated with IC1 hypomethylation and reduced IGF2 gene expression when paternally inherited. Abi Habib et al. (17) suggested that deletions that are more proximal to the IGF2 gene might cause hypomethylation of the IC1 and SRS features after their paternal transmission. Based on these and other observations, Sparago et al. (18) proposed that the DNA methylation status and clinical phenotype may depend on the specific cis elements disturbed by the deletions. The ZFP57 sites may be candidates for such cis elements because germline mutations in ZFP57 were shown to be associated with decreased DNA methylation of particular imprinted gene loci (19), but none has reported their link to SRS (20). Our current results suggest the presence of a cis DNA sequence in the hIC1–5′ region that is potentially involved in the maintenance of hIC1 methylation on the paternal allele. This region of the sequence is located outside the hIC1 DMR and carries only one ZFP57 binding motif (while hIC1–3′ region carries 12; Fig. 11A, bottom). Therefore, a search for mutations in this region may help to understand the molecular mechanism for pathogenesis of imprinting disorders, such as SRS. In addition, differential methylation status of the hIC1–5′ region was once established yet lost after the implantation period in the TgM. Thus, a future challenge will be to determine how methylation imprinting is maintained at the hIC1 locus in human cells.
In summary, postfertilization, paternal allele-specific methylation activity of m118, which we proposed is an essential component of the mechanism to maintain paternal H19 ICR methylation during the preimplantation period, appears to also regulate hIC1 methylation in mouse. Alternatively, an unknown epigenetic modification set on the hIC1 sequence during gametogenesis and used to recognize its parental origin after fertilization is shared by human and mouse sequences in the mouse environment. Identifying the hIC1–5′ sequence equivalent to m118 should provide insights concerning the regulation of imprinting at the human Igf2/H19 locus.
Materials and Methods
Retrieving of human H19 ICR sequences
Human H19 ICR (hIC1) sequence was subcloned from the human PAC clone (RP5-998 N23; AC123789.6; GenBank) by using two retrieving vectors.
To generate pHS1/flox118(−)/Ret-hIC1(−) retrieving vector, following two sets of PCR primers and human genomic DNA were used to amplify the Ret-hIC1–5′ (nucleotides 78 662–78 953 in RP5-998 N23) and Ret-hIC1–3′ (nucleotides 84 071-84 525) sequences:
Ret-hIC1–5’-5S: 5’-ggtgaatTC/AGATCTCTTTCATTGTCAT-3′ (EcoRI/BglII).
Ret-hIC1–5′-3A: 5’-GGAGGCCGAGacGcGTGGATCACCTGAG-3′ (MluI).
Ret-hIC1–3’-5S: 5’-TGCCCAaCGcgTGTGGGGACTCTGTCCT-3′ (MluI).
Ret-hIC1–3′-3A: 5’-GCACACGTCTCTCTCACCCAGCACCCAT-3′.
The mouse H19 ICR sequence (m118) was PCR-amplified by the following primer set, digested with BamHI and inserted at BamHI site of the pHS1/loxP(w+) (21), to generate pHS1/flox118(−).
5’del_fr-3A8G+B: 5’-ctagagatctggatccAAGCTTTCCTGCTCACTG-3′ (BamHI).
ICRcore-118-3A2:5’-cacttaggatcCACCATGGCCCTTTAGC-3′ (BamHI).
The Ret-hIC1–5′- and 3′-homology fragments, digested with EcoRI/MluI and MluI/EcoRI, respectively, were introduced together into the EcoRI-digested pHS1/flox118(−). Resultant plasmid, pHS1/flox118(−)/Ret-hIC1(−), was linearlized by digestion with MscI (at 78 899) and BstEII (at 84 110) and used for retrieving reaction.
To generate pRet-hIC1XL(−) retrieving vector, following PCR primer set and human genomic DNA were used to amplify the Ret-hIC1XL-5′ (nucleotides 75 704–75 962 in RP5-998 N23) sequence:
Ret-hIC1XL-5’-5S4: 5’-GGAGGCTGGAaTtC/A-GATCtGGGCTGCTGC-3′ (EcoRI/BglII).
Ret-hIC1XL-5′-3A3: 5’-TGGGCTCACGCgTGGTGGCTGAGT-3′ (MluI).
The Ret-hIC1XL-5′- and Ret-hIC1–3′-homology fragments, digested with EcoRI/MluI and MluI/EcoRI, respectively, were introduced together into the EcoRI-digested pBluescriptII KS(+). Resultant plasmid, pRet-hIC1XL(−), was linearlized by digestion with BstEII (at 75 922/84 110) and used for retrieving reaction.
Linearized retrieving vectors were used to transform Escherichia coli cells (strain EL250; a gift from N.A. Jenkins, National Cancer Institute, Frederick, Maryland, USA) harboring the RP5-998 N23 PAC. After selection on the basis of ampicillin resistance, transformants that underwent accurate recombination were identified by restriction enzyme digestion and DNA sequencing. Two plasmids, pHS1/flox118(−)/hIC1(−) and phIC1XL(−), each carrying 5.8 kb (78,664 [BglII]-84,493 [EcoRI]) and 8.8 kb (75,704–84,493 [EcoRI]) sequences, respectively, were successfully obtained.
Yeast targeting vectors and homologous recombination in yeast
The co-placement target backbone vector, pHS1/loxP-5171-B-2272-5171-G-2272 (pCop5B25G2), for introducing floxed sequences into the human β-globin HS1 region (nucleotides 13 299–14 250 in HUMHBB; GenBank) was described elsewhere (4). The fragment harboring 5′ segment of the human β-globin HS1+loxP5171 sequences was recovered from this plasmid by digestion with KpnI and inserted into KpnI site of the pBluescriptII KS(+), in which HindIII site in the MCS was prospectively disrupted, to generate pBSII∆H/K–K. The linker oligonucleotide (5’-AGCTGGAATTCC-3′) was annealed and inserted into HindIII site of the HS1 sequence (at nucleotide 13 769 in HUMHBB) to generate EcoRI site (underlined in the oligo) in this position of the plasmid (pBSII∆H/E/K–K). The KpnI fragment was recovered from this plasmid and inserted back into equivalent portion of pCop5B25G2 to generate pHS1E/loxP-5171-B-2272-5171-G-2272 (pCopE5B25G2 in short).
The mouse H19 ICR 118 bp fragment, digested with BamHI (see above), was inserted into BglII site of pCopE5B25G2 to generate pHS1E/loxP-5171-B-2272-5171-[118]-2272 (pCopE5B25[118]2). Then, 5′ segment of the hIC1 sequence (2961 bp [75704-78 664]) was recovered from phIC1XL(−) by BglII digestion and inserted into BamHI site of pCopE5B25[118]2 to generate pHS1E/loxP-5171-[5’hIC1]-2272–5171-[118]-2272 (pCopE5[5′]25[118]2). Finally, 3′ segment of the hIC1 sequence (5824 bp [78665-84 488]) was recovered from pHS1/flox118(−)/hIC1(−) by EcoRI digestion and inserted into EcoRI site of pCopE5[5′]25[118]2 to generate pHS1[3’hIC1]/loxP-5171-[5’hIC1]-2272–5171-[118]-2272 (pCop[3′]5[5′]25[118]2). In each cloning step, the correctness of DNA construction was confirmed by DNA sequencing.
The targeting vector was linearized by digestion with SpeI [at nucleotide 13 670 in HUMHBB] and used to mutagenize the human β-globin YAC (A201F4.3) (22). 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 mouse eggs from C57BL/6 J (Charles River) mice. Tail DNA from founder offspring was screened first by PCR, followed by Southern blotting. Structural analysis of the YAC transgene was performed as described elsewhere (23). The Zp3-Cre TgM (Jackson Laboratory) (24) was mated with parental YAC-TgM lines to derive sublines carrying either mouse H19 ICR (118 bp) or human H19 ICR (3 kb) sequences (co-placement strategy, (14)). Successful cre-loxP recombination was confirmed by Southern blot, as well as PCR analyses.
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 (MEXT), Japan.
DNA methylation analysis by Southern blotting
Genomic DNA extracted from tail somatic cells was first digested by DraI or BglII and then subjected to the methylation-sensitive enzymes HhaI or BstUI, respectively. Following size separation in agarose gels, Southern blots were hybridized with α-32P-labeled probes and subjected to X-ray film autoradiography.
DNA methylation analysis by bisulfite sequencing
Preimplantation embryos were embedded in agarose beads and treated with sodium bisulfite as described previously (13). Genomic DNA extracted from adult male sperm or the tail tips of ~1-week-old animals was treated with sodium bisulfite using the EZ DNA Methylation Kit (Zymo Research). Sperm and tail tip DNA was digested with DraI prior to the treatment. Subregions (I–VII) of the transgenic hIC1 were amplified by PCR. The PCR products were subcloned into the pGEM-T Easy vector (Promega) for sequencing analyses. PCR primers are listed in Table 1.
Primer sets for bisulfite sequencing analysis
| Regions | 5′-primer | Sequence |
|---|---|---|
| 3′-primer | ||
| I | hIC1-MA-5S-4 | 5′-TATTTGGGTTTTGTTAGTTTTTTG-3’ |
| hIC1-MA-3A-1 | 5’-CTCCTTCCATCTCACTACTCTAAA-3’ | |
| II | hIC1-MA-5S6 | 5′-TTTTTAAGTAGGTGGGATTATAGG-3’ |
| hIC1-MA-3A6 | 5’-ACACACAATACTCATATATCTATC-3’ | |
| III | hIC1-MA-5S3 | 5′-AAGATGGTTTTGATTTTTTATGAT-3’ |
| hIC1-MA-3A2 | 5’-ACTAAATCCCCTATACTATACTAC-3’ | |
| IV | hIC1-MA-5S5 | 5′-ATTAGGTTTTTAGTTATGTATGGG-3’ |
| hIC1-MA-3A5 | 5’-AAAATTCAAATCTCTTTCATTATCA-3’ | |
| V | hIC1-MA-5S2 | 5′-TAAGGAATTGGGATATTTTATTGT-3’ |
| hIC1-MA-3A3 | 5’-AAACCTCCAACAAAAACAAAAATC-3’ | |
| VI | hIC1-MA-5S1 | 5′-AAGTTAGGAGATTAAGGTAGATTTTT-3’ |
| hIC1-MA-3A4 | 5’-CATTAACATTCTCATTCAATACAAAT-3’ | |
| VII | hIC1-MA-5S7 | 5′-GGGTTGTGATGTGTGAGTTTGTAT-3’ |
| hIC1-MA-3A7 | 5’-ATTTCACCAAAAAAACCAAACATT-3’ |
| Regions | 5′-primer | Sequence |
|---|---|---|
| 3′-primer | ||
| I | hIC1-MA-5S-4 | 5′-TATTTGGGTTTTGTTAGTTTTTTG-3’ |
| hIC1-MA-3A-1 | 5’-CTCCTTCCATCTCACTACTCTAAA-3’ | |
| II | hIC1-MA-5S6 | 5′-TTTTTAAGTAGGTGGGATTATAGG-3’ |
| hIC1-MA-3A6 | 5’-ACACACAATACTCATATATCTATC-3’ | |
| III | hIC1-MA-5S3 | 5′-AAGATGGTTTTGATTTTTTATGAT-3’ |
| hIC1-MA-3A2 | 5’-ACTAAATCCCCTATACTATACTAC-3’ | |
| IV | hIC1-MA-5S5 | 5′-ATTAGGTTTTTAGTTATGTATGGG-3’ |
| hIC1-MA-3A5 | 5’-AAAATTCAAATCTCTTTCATTATCA-3’ | |
| V | hIC1-MA-5S2 | 5′-TAAGGAATTGGGATATTTTATTGT-3’ |
| hIC1-MA-3A3 | 5’-AAACCTCCAACAAAAACAAAAATC-3’ | |
| VI | hIC1-MA-5S1 | 5′-AAGTTAGGAGATTAAGGTAGATTTTT-3’ |
| hIC1-MA-3A4 | 5’-CATTAACATTCTCATTCAATACAAAT-3’ | |
| VII | hIC1-MA-5S7 | 5′-GGGTTGTGATGTGTGAGTTTGTAT-3’ |
| hIC1-MA-3A7 | 5’-ATTTCACCAAAAAAACCAAACATT-3’ |
Primer sets for bisulfite sequencing analysis
| Regions | 5′-primer | Sequence |
|---|---|---|
| 3′-primer | ||
| I | hIC1-MA-5S-4 | 5′-TATTTGGGTTTTGTTAGTTTTTTG-3’ |
| hIC1-MA-3A-1 | 5’-CTCCTTCCATCTCACTACTCTAAA-3’ | |
| II | hIC1-MA-5S6 | 5′-TTTTTAAGTAGGTGGGATTATAGG-3’ |
| hIC1-MA-3A6 | 5’-ACACACAATACTCATATATCTATC-3’ | |
| III | hIC1-MA-5S3 | 5′-AAGATGGTTTTGATTTTTTATGAT-3’ |
| hIC1-MA-3A2 | 5’-ACTAAATCCCCTATACTATACTAC-3’ | |
| IV | hIC1-MA-5S5 | 5′-ATTAGGTTTTTAGTTATGTATGGG-3’ |
| hIC1-MA-3A5 | 5’-AAAATTCAAATCTCTTTCATTATCA-3’ | |
| V | hIC1-MA-5S2 | 5′-TAAGGAATTGGGATATTTTATTGT-3’ |
| hIC1-MA-3A3 | 5’-AAACCTCCAACAAAAACAAAAATC-3’ | |
| VI | hIC1-MA-5S1 | 5′-AAGTTAGGAGATTAAGGTAGATTTTT-3’ |
| hIC1-MA-3A4 | 5’-CATTAACATTCTCATTCAATACAAAT-3’ | |
| VII | hIC1-MA-5S7 | 5′-GGGTTGTGATGTGTGAGTTTGTAT-3’ |
| hIC1-MA-3A7 | 5’-ATTTCACCAAAAAAACCAAACATT-3’ |
| Regions | 5′-primer | Sequence |
|---|---|---|
| 3′-primer | ||
| I | hIC1-MA-5S-4 | 5′-TATTTGGGTTTTGTTAGTTTTTTG-3’ |
| hIC1-MA-3A-1 | 5’-CTCCTTCCATCTCACTACTCTAAA-3’ | |
| II | hIC1-MA-5S6 | 5′-TTTTTAAGTAGGTGGGATTATAGG-3’ |
| hIC1-MA-3A6 | 5’-ACACACAATACTCATATATCTATC-3’ | |
| III | hIC1-MA-5S3 | 5′-AAGATGGTTTTGATTTTTTATGAT-3’ |
| hIC1-MA-3A2 | 5’-ACTAAATCCCCTATACTATACTAC-3’ | |
| IV | hIC1-MA-5S5 | 5′-ATTAGGTTTTTAGTTATGTATGGG-3’ |
| hIC1-MA-3A5 | 5’-AAAATTCAAATCTCTTTCATTATCA-3’ | |
| V | hIC1-MA-5S2 | 5′-TAAGGAATTGGGATATTTTATTGT-3’ |
| hIC1-MA-3A3 | 5’-AAACCTCCAACAAAAACAAAAATC-3’ | |
| VI | hIC1-MA-5S1 | 5′-AAGTTAGGAGATTAAGGTAGATTTTT-3’ |
| hIC1-MA-3A4 | 5’-CATTAACATTCTCATTCAATACAAAT-3’ | |
| VII | hIC1-MA-5S7 | 5′-GGGTTGTGATGTGTGAGTTTGTAT-3’ |
| hIC1-MA-3A7 | 5’-ATTTCACCAAAAAAACCAAACATT-3’ |
Statistical analysis
The differences in the means of methylation rate between the paternal region I of the hIC1+m118 transgene and that of the hIC1 transgene were tested using an unpaired t-test. Differences between results at a P value of <0.05 were considered statistically significant.
Acknowledgements
We thank Dr Akiyoshi Fukamizu (University of Tsukuba) for his continuous support.
Conflict of Interest statement. The authors declare that they have no conflict of interest.
Funding
This work was supported in parts by a research grant from the Takeda Science Foundation (to K.T.), Astellas Foundation for Research on Metabolic Disorders (to H.M.), a Grant-in-Aid for Scientific Research (B; KAKENHI Grant number 19H03134 to K.T. and C; KAKENHI Grant number 20 K06481 to H.M.) from JSPS (Japan Society for the Promotion of Science) and Grant-in-Aid for Scientific Research on Innovative Areas (KAKENHI Grant number 20H05379 to H.M.) from the MEXT (Ministry of Education, Culture, Sports, Science and Technology of Japan).
References
Author notes
Katsuhiko Hirakawa and Hitomi Matsuzaki contributed equally to this work.
