Mice with maternal duplication of proximal Chromosome 11 (MatDp(prox11)), where Meg1/Grb10 is located, exhibit pre- and postnatal growth retardation. To elucidate the responsible imprinted gene for the growth abnormality, we examined the precise structure and regulatory mechanism of this imprinted region and generated novel model mice mimicking the pattern of imprinted gene expression observed in the MatDp(prox11) by deleting differentially methylated region of Meg1/Grb10 (Meg1-DMR). It was found that Cobl and Ddc, the neighboring genes of Meg1/Grb10, also comprise the imprinted region. We also found that the mouse-specific repeat sequence consisting of several CTCF-binding motifs in the Meg1-DMR functions as a silencer, suggesting that the Meg1/Grb10 imprinted region adopted a different regulatory mechanism from the H19/Igf2 region. Paternal deletion of the Meg1-DMR (+/ΔDMR) caused both upregulation of the maternally expressed Meg1/Grb10 Type I in the whole body and Cobl in the yolk sac and loss of paternally expressed Meg1/Grb10 Type II and Ddc in the neonatal brain and heart, respectively, demonstrating maternalization of the entire Meg1/Grb10 imprinted region. We confirmed that the +/ΔDMR mice exhibited the same growth abnormalities as the MatDp(prox11) mice. Fetal and neonatal growth was very sensitive to the expression level of Meg1/Grb10 Type I, indicating that the 2-fold increment of the Meg1/Grb10 Type I is one of the major causes of the growth retardation observed in the MatDp(prox11) and +/ΔDMR mice. This suggests that the corresponding human GRB10 Type I plays an important role in the etiology of Silver-Russell syndrome caused by partial trisomy of 7p11-p13.
The mouse Meg1/Grb10 gene was originally identified as a maternally expressed imprinted gene using a subtractive protocol between androgenetic and normal fertilized embryos (1). It exhibits maternal expression in almost all tissues and organs while the human orthologue, GRB10, exhibits a biallelic expression in the corresponding regions. However, in the brain, both human GRB10 and mouse Meg1/Grb10 have been reported to be preferentially expressed from paternal alleles (2–6). Interestingly, these genes have two corresponding promoter regions. Type I transcripts from upstream promoters display maternal and biallelic expression in mice and humans, respectively, in almost all tissues. Type II transcripts from downstream promoters display paternal expression in the brain in both species (6–9).
The DMR locating on human GRB10 and mouse Meg1/Grb10 region (Meg1-DMR, also known as Grb10 CpG island 2 (CGI 2) DMR) overlaps the brain-specific downstream promoter in both species; it is hypermethylated on the inactive maternal allele while non-methylated on the active paternal allele, suggesting that DNA methylation directly regulates the imprinted paternal expression of GRB10 and Meg1/Grb10 Type II transcripts (6–8). In addition, we have also demonstrated the existence of a mouse-specific repeat (MSR) sequence in the mouse Meg1-DMR, where an insulator CTCF protein binds in a DNA-methylation-sensitive manner. Therefore, we have proposed a molecular mechanism model in which the CTCF binding to the MSR controls the mouse-specific maternal expression of Meg1/Grb10 Type I transcript via an insulator function, as has been observed in the Igf2/H19 region (6,10,11).
Until recently, there was no evidence for the existence of other imprinted genes around Meg1/Grb10 (5). However, it has been reported that one of the promoter variants of mouse Ddc, which is located on the 3′ side of Meg1/Grb10, exhibits heart-specific paternal expression (12), suggesting that the Meg1/Grb10 region also comprises an imprinted gene cluster consisting of certain other genes where the Meg1-DMR regulates all the imprinted genes, as in the case of other imprinted domains. Thus, in order to elucidate the precise structure of this imprinted region, more extensive search of the imprinted genes around Meg1/Grb10 is required.
Meg1/Grb10 encodes an adaptor protein which binds to certain tyrosine-kinase receptors, such as insulin receptor or insulin-like growth factor 1 receptor (13–24), suggesting that the protein is implicated in growth. It is known that the mice with maternal disomy Chromosome 11 (MatDi(11)) and maternally duplicated proximal Chromosome 11 (MatDp(prox11)), where Meg1/Grb10 locates, exhibit pre- and postnatal growth retardation, while mice with paternal disomy of Chromosome 11 (PatDi(11)) or paternally duplicated proximal Chromosome 11 (PatDp(prox11)) exhibit pre- and postnatal promotion of growth (25–27), suggesting that imprinted Meg1/Grb10 is a candidate gene for the growth phenotypes. However, the precise role of the Meg1/Grb10 protein in growth remains unclear, because both positive and negative effects on the signaling pathways via those receptors have been reported in vitro (13,17,22–24,28–37). Then, the in vivo inhibitory role in growth was clearly demonstrated by Meg1/Grb10 knockout (KO) mice that exhibited embryonal and placental overgrowth when the knockout allele was transmitted from their mother, as in the case of PatDi(11) and PatDp(prox11) mice (38).
In contrast, overproduction effects of Meg1/Grb10 on fetal or neonatal growth have still not been successfully obtained. We previously generated Meg1/Grb10 transgenic (Meg1Tg) mice in which Meg1/Grb10 cDNA was transcribed from the chicken β-actin promoter and stimulated by the CMV enhancer (CAG vector) (39,40). Although an approximately 1.5-fold Meg1/Grb10 expression level was observed in the Meg1Tg embryos compared with that of controls, embryonic growth retardation was not observed. However, it is highly probable that the Meg1 Tg mice did not in fact represent the precise expression profile of Meg1/Grb10 during development, because the transgene did not have the original promoter and coupled with an external enhancer.
In this study, we analyzed the imprinted expression of the neighboring genes of Meg1/Grb10 and demonstrated that the Meg1/Grb10 imprinted region encompassed approximately 650 kb, including both of the neighboring Cobl and Ddc genes. In vitro functional assays showed that the MSR sequence in the Meg1-DMR can function as a silencer, suggesting that the Meg1/Grb10 imprinted region is regulated by a different mechanism from that of the H19/Igf2 region. Finally, we constructed a novel model mouse expressing a double dosage of the Meg1/Grb10 Type I transcript in the correct tissues and organs by deletion of the imprinting control center, the Meg1-DMR. These mice exhibited the same growth retardation phenotype as the MatDi(11) and MatDp(prox11) mice, as expected. We also observed that a subtle increment of the Meg1/Grb10 Type I expression significantly affected fetal and neonatal growth by comparing two slightly different types of mice with the Meg1-DMR deletion, those with and without Neo genes (neomycin-resistance genes) in the deletion construct. This indicates that the pre- and postnatal growth retardation observed in both the Meg1-DMR deletion mice and the MatDi(11) and MatDp(prox11) mice is mainly attributable to the double dosage of the Meg1/Grb10 Type I transcript. Based on these results, we discuss the possibility that the corresponding GRB10 Type I transcript expressed in the entire body could be responsible for the etiology of the human Silver-Russell syndrome (SRS) caused by partial duplication of 7p11–p13, even if it is not imprinted in humans.
The Meg1/Grb10 imprinted cluster contains novel imprinted genes
Meg1/Grb10 has long been known as an isolated single imprinted gene in the mouse proximal Chromosome 11 (1,5) (Fig. 1A and B), but recently it was reported that Ddc transcribed from exon 1a (hereafter referred to as Ddc-exon1a) showed complete paternal expression only in neonatal heart, by comparing the expression profiles between mice with PatDp(prox11) and MatDp(prox11) (12). We further demonstrated that an alternative promoter variant of Ddc which transcribed from exon 1 (hereafter referred to as Ddc-exon1) and Cobl also showed paternally and maternally biased expression in a tissue-specific manner, respectively.
By analyzing mice produced by reciprocal crosses between C57BL/6J (B6) and JF1/Msf (JF1) (41), we examined the imprinting status of several transcripts from neighboring genes of Meg1/Grb10, such as Cobl and Ddc locating on the 5′ and 3′ sides of Meg1/Grb10, respectively, and Fignl1 and Ikzf1, next to Ddc (Fig. 1A–I). There was a slightly biased expression from paternal alleles of Ddc-exon1 in yolk sac and liver, but not in embryos (Fig. 1C–E). Interestingly, another transcript, Ddc-exon1a, did exhibit a paternally biased expression in embryos. In addition, we reconfirmed the results that Ddc-exon1a exhibited heart-specific paternal expression in neonates while it was biallelically expressed in neonatal brain (Fig. 1C, D and F). Cobl exhibited biallelic expression in all tissues examined except yolk sac, where the expression pattern was slightly maternally biased (Fig. 1C and G). These biased expression patterns of Ddc and Cobl were confirmed by allele-specific quantitative PCR (Supplementary Material, Fig. S1). Fignl1 and Ikzf1 exhibited biallelic expression and no clearly biased expression in any of the tissues examined by direct sequence analysis and restriction fragment length polymorphism (RFLP), respectively (Fig. 1C, H and I). These results demonstrate that at least two neighboring genes of Meg1/Grb10, Cobl and Ddc, are imprinted, and that the Meg1/Grb10 imprinted region is comprised of approximately 650 kb in the mouse proximal Chromosome 11.
As previously shown, the Meg1-DMR corresponding to CpG island 2 (CGI 2) of Meg1/Grb10 is a primary DMR in this region (Fig. 1A) (6,7). Next, we examined the DNA methylation status of the other CpG islands (CGIs) in the promoter regions of Cobl, Fignl1 and Ikzf1. As shown in Figure 1A and J–L, we confirmed that all the CGIs examined were hypomethylated as well as CGI 1 of Meg1/Grb10, which locates on the promoter region of Meg1/Grb10 Type I transcript (6,7). No CGIs and no differentially methylated CpG sites are reported in and around the Ddc-exon1a promoter region (12). The same was true for the Ddc-exon1 promoter (data not shown). Thus, the Meg1-DMR, the CGI 2 DMR of Meg1/Grb10, is the only primary DMR in this imprinted region which possibly regulates all of these imprinted genes.
The MSR sequence can function as a silencer
Next, we addressed the mechanism of imprinted regulation by in vitro functional assays using several vector constructs, especially the role of the MSR sequence (shown in Fig. 2A) containing several CTCF binding sites within the Meg1-DMR. Unexpectedly, we found that the MSR, by itself, could function as a silencer.
First, the insulator action was examined using two different assays: transient assay using luciferase vectors (data not shown) and stable assay using neomycin resistance gene (Neo) vectors (Fig. 2B) (42). In both vectors, the MSR or H19 DMR insulator was positioned between the promoter and the enhancer. The MSR inserted in either direction significantly reduced the number of Neo-resistant colony as in the case of the control H19 DMR, although that in the forward direction was weaker than the reverse direction (Fig. 2B). These results indicate that the MSR within the Meg1-DMR had some capacity to reduce the Neo expression. However, using these kinds of assays, it is essentially impossible to discriminate insulator activity from silencer activity.
We then tested whether the MSR had silencer activity by using vector constructs without an enhancer. In these vectors, the luciferase gene was driven by the Meg1-Type I promoter and the MSR or the H19 DMR was placed downstream. This experiment clearly demonstrated that the MSR in either the forward or reverse orientation could down-regulate the luciferase activity while no such effect was observed with the H19 DMR (Fig. 2C). Importantly, its silencer function was DNA methylation sensitive because the luciferase expression increased 2–3-fold when fully methylated fragments were inserted (Fig. 2D). The human sequence corresponding to the MSR (h-MSR, shown in Fig. 2A) has no CTCF binding sites (43) and showed no silencer function (Fig. 2C), and DNA methylation did not affect the result (Fig. 2D). These results indicate that the MSR containing several CTCF-binding sites within the Meg1-DMR, by itself, can play a silencer role in a DNA-methylation-dependent manner. Although it is not easy to integrate the silencer function of the MSR with the regulatory mechanism of the Meg1/Grb10 imprinted region, it is probable that the Meg1-DMR plays a different role from the H19 DMR.
Paternal inheritance of the Meg1ΔDMR allele results in maternalization of the Meg1/Grb10 imprinted region by disruption of imprinted regulation
We generated mice with the Meg1-DMR deletion (Meg1ΔDMR) to produce mice exhibiting biallelic expression of Meg1/Grb10 Type I transcript during development in order to directly see the growth inhibitory effect (Fig. 3). We confirmed that mice with paternal transmission of the Meg1ΔDMR induced expression of the Meg1/Grb10 Type I transcript from the paternal allele, and moreover, disrupted imprinted expression patterns of Ddc and Cobl were observed in these mice.
Parent-of-origin-specific expression of the imprinted genes was examined using two reciprocal F1 mice, one from a hetero mutant dam and JF1 sire (ΔDMR/JF1) and the other from a JF1 dam and hetero mutant sire (JF1/ΔDMR). RT–PCR analysis demonstrated that almost all of the Meg1/Grb10 expression was due to the Type I transcript (data not shown) and exhibited maternal expression in all of the tissues examined except neonatal brain in the wild-type (JF1/+ and +/JF1) mice, as previously reported. However, paternal transmission of ΔDMR (JF1/ΔDMR mice) activated expression of the paternally silenced allele, resulting in biallelic expression of the Meg1/Grb10 Type I from the upstream promoter (Fig. 4A). This biallelic expression is different from that observed in normal neonatal brain, in which Meg1/Grb10 exhibits biallelic expression due to a combined expression of both the maternally expressed Type I from exon 1a and the paternally expressed Type II from the brain-specific exon 1b promoter (Fig. 4A and B).
In the case of Ddc, RFLP analysis indicated that the expression levels of paternal Ddc-exon1 in yolk sac and liver, as well as those of Ddc-exon1a in embryos, were reduced relative to those of JF1/+ (Fig. 4C and D). A 50 to 60% reduction was confirmed in each case by allele-specific quantitative PCR analysis (Fig. 4E–G). The paternal expression of the Ddc-exon1a in neonatal heart was completely repressed in JF1/ΔDMR (Fig. 4D). In contrast, paternal Cobl expression was increased in the JF1/ΔDMR yolk sac compared with that of JF1/+. Although we failed to detect any obvious alteration in expression pattern between JF1/+ and JF1/ΔDMR by RFLP analysis (Fig. 4H), allele-specific quantitative PCR analysis revealed a 40% increment in paternal Cobl expression (Fig. 4I). These results demonstrate that the Meg1-DMR regulates the entire Meg1/Grb10 imprinted region consisting of Cobl, Meg1/Grb10 and Ddc. In all of the transcripts examined, maternal Meg1ΔDMR transmission did not have any effect on imprinted expression in the region (Fig. 4A–D and H).
The Meg1-DMR hypermethylated in the maternal allele has both a promotional role for the maternally expressed Meg1/Grb10 Type I and maternally biased Cobl and an inhibitory role for the paternally expressed Meg1/Grb10 Type II and the paternally expressed/paternally biased Ddc, respectively, while the Meg1-DMR hypomethylated in the paternal allele acts vice versa (Fig. 5). Paternal deletion of the Meg1-DMR leads to the activation of the upstream maternally expressed genes and repression of downstream paternally expressed genes, and therefore, is comparable to the hypermethylated Meg1-DMR in the maternal allele (Fig. 5). Thus, it is concluded that the paternal deletion of the Meg1-DMR causes maternalization of the Meg1/Grb10 imprinted region on paternal Chromosome 11.
Biallelic expression of Meg1/Grb10 Type I in the entire body results in pre- and postnatal growth retardation
The mice with maternal deletion of the Meg1-DMR did not display any clear phenotypes (data not shown). However, the mice with paternal deletion of the Meg1-DMR (+/ΔDMR) exhibited growth retardation from 12.5 dpc (days post coitus) and an ∼30% reduction of the fetal weight was evident at term. Interestingly, placental growth retardation was slightly delayed and exhibited from 13.5 dpc and its weight exhibited a 20% reduction at term, suggesting Meg1/Grb10 functions primarily in embryos and secondarily in placentas (Fig. 6A and B). After birth, the growth retardation, which was ∼60% of the normal weight, was maintained throughout the postnatal period to the adult stage (Fig. 6C and D), suggesting that biallelic expression of the Meg1/Grb10 Type I in the entire body led to growth retardation in both the pre- and postnatal periods.
Negative correlation between the Meg1/Grb10 expression level and pre- and postnatal growth was confirmed by comparing Meg1ΔDMR mice and Meg1ΔDMRneo mice. The Meg1ΔDMRneo mice retained the neomycin resistance gene (Neo) replaced by the Meg1-DMR (Fig. 3). +/ΔDMRneo mice carrying the Neo allele displayed mild growth retardation (74% of normal at 20 days after birth) compared to the +/ΔDMR mice without the Neo allele (57% of the normal). Importantly, a significant difference in the expression level of Meg1/Grb10 Type I was confirmed in the 12.5 dpc embryos between +/ΔDMR and +/ΔDMRneo mice: the expression level of Meg1/Grb10 Type I was higher in the former fetuses (1.93-fold compared with normal), which displayed severe growth retardation compared with the latter (1.65-fold) (Table 1), while the expression levels of Cobl and Ddc did not display any significant differences between these two mice. These results indicate that fetal growth was extremely sensitive to the level of Meg1/Grb10 Type I, and that its elevated expression is the primary cause of the pre- and postnatal growth retardation observed in +/ΔDMR mice.
|Genotype||Weight (+/+=1)||mRNA level (12.5 dpc, +/+=1)|
|12.5 dpc||18.5 dpc||20 D (male)||Meg1/Grb10||Cobl||Ddc|
|Genotype||Weight (+/+=1)||mRNA level (12.5 dpc, +/+=1)|
|12.5 dpc||18.5 dpc||20 D (male)||Meg1/Grb10||Cobl||Ddc|
D, days after birth.
*P < 0.05, **P < 0.01.
This study has demonstrated that the Meg1/Grb10 imprinted region on the mouse proximal Chromosome 11 consists of at least five imprinted transcripts of the three genes, Cobl, Meg1/Grb10 Types I and II, and Ddc-exon1 and -exon1a and that they are, respectively, regulated by the Meg1-DMR in a tissue-specific manner. It has also demonstrated that the deletion of the Meg1-DMR is comparable to the fully methylated status of the Meg1-DMR in the maternal alleles in terms of genomic imprinting regulation. Therefore, its paternal deletion maternalizes the entire Meg1/Grb10 imprinted region.
The DMR coupled with the CTCF-binding sites is very effective for the regional control of multiple imprinted genes. The human homologous DMR region has no such CTCF-binding sequences and the GRB10 Type I transcript exhibits biallelic expression in almost all tissues, and there have been no obvious data demonstrating the imprinting expression of COBL and DDC, although GRB10 Type II transcript does exhibit paternal expression in the brain, as is also the case with mice (5,6,12). Therefore, the function of the CTCF-binding sequences is of considerable interest.
Previously, we reported that CTCF insulator protein bound to the mouse-specific CTCF-binding repeat sequence in the Meg1-DMR in a methylation-sensitive manner (6). There have not been any CTCF binding sites reported in the corresponding human sequence (h-MSR) (43). We then proposed that the insulator function of the CTCF is essential for the regulation of the maternally expressed Meg1/Grb10 Type I transcript in mice as in the case of the Igf2/H19 region (6,10,11). In this insulator model, the paternal expression of the Meg1/Grb10 Type II transcript is explained by the absence of DNA methylation of its promoter region within the Meg1-DMR, and the maternal expression of Meg1/Grb10 Type I transcript is explained by the enhancer blocking activity of the CTCF bound to the MSR sequence within the Meg1-DMR (6). The maternal expression of the Cobl located upstream of the Meg1-DMR fits this model, but paternal expression of the Ddc that locates downstream of the Meg1-DMR does not. Because there are no DMRs around Ddc-exon1 and -exon1a (12), its maternal repression cannot thus be explained. Therefore, we investigated the mechanism of the imprinting regulation by Meg1-DMR using several reporter vector constructs transfected into cultured cells, and demonstrated that the MSR sequence within Meg1-DMR, but not its corresponding human h-MSR sequence, functioned as a silencer, although an insulator function could not be excluded. The Meg1-DMR seems to function as a boundary between the maternally expressed upstream and paternally expressed downstream regions (Fig. 5). As the DNA methylation-sensitive silencer is not sufficient to explain the mechanism of this imprinted region as well as the insulator model, as discussed above, the involvement of other factors is necessary. As a genome-wide interaction between cohesin and CTCF was recently demonstrated (44–47), we may account for the three-dimensional structure change induced by the cohesin-CTCF complex in the regulation of this imprinted region. Therefore, further experiments will be required to elucidate the precise regulation mechanism of the Meg1/Grb10 imprinted region.
The reason for the biased expression of Ddc-exon1 in yolk sac and neonatal liver, Ddc-exon1a in embryo and Cobl in yolk sac remains unknown. It is highly possible that transcripts from some promoters have parent-of-origin-specific expressions, but other transcripts from other promoters have non-imprinted expression. Then, we may simply recognize them as biased when detecting both at the same time. However, so far, we have not detected any novel promoters of Cobl and Ddc (data not shown). Alternatively, the co-existence of several different cell types in the tissues examined also may contribute to the biased expression patterns to some extent. As described above, all of the biased expressions of these transcripts were observed to be highly tissue-specific, such as yolk sac, embryos and neonatal heart, suggesting that the cell-type-specific monoallelic expression, such as the brain-specific paternal expression of Meg1/Grb10 Type II, may occur in a minority of cells in these tissues using cell-type-specific transcription factors.
The +/ΔDMR mice exhibit biallelic expression of the Meg1/Grb10 Type I transcripts, resulting in severe embryonic and placental growth retardation. It is assumed that the +/ΔDMR mice have the same expression pattern of imprinted genes in the Meg1/Grb10 imprinted region as those of the MatDi(11) and MatDp(prox11) mice, because they have one maternal and maternalized Meg1/Grb10 imprinted region. An essentially identical phenotype was observed between these mice, indicating that the most probable cause of the growth inhibition is the double dosage of Meg1/Grb10 Type I (Table 2).
|Body weight (+/+=1)||Placenta weight (+/+=1)|
|Neonate||0.608 (2D)||0.603 (at birth)||0.642 (5D)||N.D.||N.D.|
|Body weight (+/+=1)||Placenta weight (+/+=1)|
|Neonate||0.608 (2D)||0.603 (at birth)||0.642 (5D)||N.D.||N.D.|
D, days after birth. N.D., no data.
As shown in this study, the +/ΔDMR mice have an additional loss of the paternal expression of Meg1/Grb10 Type II in brain and also Ddc-exon1a in neonatal heart. Therefore, logically, it is not possible to conclude that the double dosage of Meg1/Grb10 Type I expression is the direct cause of such growth inhibitory effect in this and the MatDi(11) and MatDp(prox11) experiments. It is possible that a loss of the paternal expression of Meg1/Grb10 Type II in the brain and/or Ddc-exon1a in the heart also contributes to or is a major cause of the phenotype. One piece of supportive evidence of Meg1/Grb10 Type I being a major contributor comes from the experimental finding that fetal and postnatal growth was very sensitive to the expression level of the Meg1/Grb10 Type I, which was obtained by a comparison with two mice lines, one with and one without the neomycin construct (Table 1). Our analysis of the +/ΔDMRneo mice with Neo revealed that an artificial transcript, possibly from the Neo promoter, was expressed in the anti-sense direction to Meg1/Grb10 by accident, suggesting that the Meg1/Grb10 Type I expression was reduced by the influence of this anti-sense RNA to some degree. Therefore, final expression level of Meg1/Grb10 Type I only reached 1.6–1.7-fold compared to the expected value of 2-fold observed in the +/ΔDMR mice (without Neo). The +/ΔDMRneo mice also lack the paternal expression of Meg1/Grb10 Type II in the brain and Ddc-exon1a in heart (data not shown), and exhibit the same level of Cobl and Ddc in whole embryo. Therefore, we can conclude that the growth inhibitory effect observed in these mice is dependent on the expression level of Meg1/Grb10 Type I and that a slight change (1.6–1.7- to 2.0-fold increment) of its expression level has a great impact on embryonic and postnatal growth (0.74 to 0.57-fold weight reduction of neonates).
Another piece of supportive evidence comes from the previous work on Meg1/Grb10 knockout mice (Meg1/Grb10 KO) (38). This report showed that maternal transmission of loss of the Meg1/Grb10 Type I in the entire body resulted in fetal growth promotion with disproportionate liver overgrowth, demonstrating that the loss of maternally expressed Meg1/Grb10 Type I in the entire body was responsible for the growth promotion effects observed in the PatDi(11) and PatDp(prox11) mice. However, there were not any notable growth effects observed upon its paternal transmission in the pre- and postnatal stages, so it is clear that the loss of paternal expression of Meg1/Grb10 Type II in the brain itself does not have any growth effect. Together with their results, it is concluded that Meg1/Grb10 Type I plays an essential role in growth regulation via its strong growth inhibitory effect in vivo. However, placental growth retardation in the +/ΔDMR seems milder than that in MatDi(11) (Table 2), suggesting that there is/are other imprinted gene(s) involved to placental growth in proximal Chromosome 11 other than the Meg1/Grb10 imprinted region.
In humans, SRS is characterized by pre- and postnatal growth failure (48). In approximately 30–65% patients, hypomethylation of H19 DMR associated with a reduced expression of IGF2 has been observed (48–54). Maternal disomy of Chromosome 7 has also been demonstrated in ∼10% of the cases (48). This suggests that at least one imprinted gene on Chromosome 7 is involved in the pathogenesis of SRS, such as the paternally expressed PEG1/MEST and PEG10 that have growth promotional effects and/or putative maternally expressed genes that have inhibitory function (55,56). Peg1/Mest knockout mice display significant pre- and postnatal growth retardation, although no PEG1/MEST mutations have been identified as yet (57–59). It has recently been reported that a patient with paternal deletion of PEG10 and the neighboring imprinted gene SGCE showed pre- and postnatal growth retardation due to severe placental defects like the Peg10 knockout mice, although the mice died at an early stage of development (60,61).
GRB10 has been one of the strong candidate genes implicated in the growth retardation of SRS because its mouse ortholog Meg1/Grb10 is maternally expressed (1). However, it is widely accepted that human GRB10 Type I is biallelically expressed in the majority of tissues and organs, while only paternal expression of GRB10 Type II in brain has been confirmed (2–4). Therefore, it is not likely that GRB10 is the gene responsible in the SRS case of maternal disomy of Chromosome 7, although we cannot rule out the possibility that GRB10 does exhibit imprinted maternal expression in some growth affecting tissue(s). Recently, cases of SRS having duplicated 7p11–p13 including the GRB10 gene (partial trisomy of 7p11–p13) were reported (62–64). In these cases, it is highly probable that a 50% increase in the expression of GRB10 located on 7p11–p13 causes SRS, because we have demonstrated that subtle changes of the expression of the growth inhibitory factor Meg1/Grb10 has significant influence on pre-and postnatal growth (Table 1). It was also reported recently that the patients with a duplication of 7p11.2–p12 not including GRB10 showed no association with futures of SRS, including the growth retardation effect (65). Therefore, it is strongly suggested that an increased level of GRB10 is at least one of the major causes of the growth retardation seen in SRS patients with a duplication of 7p11–p13.
MATERIALS AND METHODS
Generation of Meg1ΔDMR mice
To generate the Meg1ΔDMR targeting vector, a DNA fragment including the Meg1-DMR region was screened from the 129SvJ lambda genomic library (Stratagene). A 9.7 kb fragment (SacII-EheI) and a 1.0 kb fragment (HindIII-EcoRI) were ligated with loxP—neomycin resistance gene—loxP fragment. The targeting vector was linearized and electroporated into ES cells (CCE) of 129/Sv/Ev mouse origin. The cells were incubated under G418 selection for 1 week and 1382 colonies were obtained. Of them, 120 colonies were selected, and the genomic DNA were prepared and digested with NdeI and SpeI for Southern blot analysis to identify the correctly targeted cells. A 5′ probe (probe 1) and 3′ probe (probe 2) were used to detect a 16.7 and 4.7 kb Meg1ΔDMRneo and a 16.7 and 3.5 kb Meg1ΔDMR fragment, respectively. The Meg1ΔDMRneo targeted ES cells that resulted from homologous recombination were used to generate chimeric mice by blastocyst injection. Male chimeras were bred to C57BL/6J females and their agouti progeny were genotyped by PCR amplification of tail DNA samples using the Meg1ΔDMRneo allele and endogenous Meg1/Grb10 forward primers along with a common reverse primer. To generate the Meg1ΔDMR allele, the neomycin-resistance gene was eliminated by injection of Cre recombinase expression vector into fertilized eggs derived from hetero Meg1ΔDMRneo sire and C57BL/6J dam. Tail DNA samples were used for Southern blot (described above) and PCR analysis using the Meg1ΔDMR allele and endogenous Meg1/Grb10 forward primers along with a common reverse primer to identify the mice with Meg1ΔDMR allele. The primers used are listed below. Meg1ΔDMRneo allele forward primer: 5′-GGACGTAAACTCCTCTTCAGACC-3′, Meg1ΔDMR allele forward primer: 5′-GAACCTCTTCGAGGGACCTAAT-3′, endogenous Meg1/Grb10 forward primer: 5′-CTAAGGTTGGCCTGGTCATC-3′ and common reverse primer: 5′-GCTCTGTCCAGGCCTCTTATC-3′.
Polymerase chain reaction
For PCR reactions, 20 ng of genomic DNA or 5 ng of cDNA in a 50 µl reaction mixture containing 1 x ExTaq buffer (TaKaRa), 2.5 mm dNTP mixture, 0.8 µM primers and 1.25 U of ExTaq (TaKaRa) were subjected to 30–35 PCR cycles. PCR was carried out under the following conditions: 96°C for 15 s, 55–65°C for 30 s and 72°C for 30–90 s. The PCR primer sequences used in each PCR are listed in each section.
Bisulfite methylation assay
Genomic DNA from the BJ 13.5 dpc embryo, placenta and yolk sac were used and modified with sodium bisulfite as described previously. The treated DNA was amplified by PCR using the primer pairs listed below. Cobl: 5′-AAATATATATAAGGGTTTTTATAGTAGTAT-3′ and 5′-TAAATTCTAAAACAAAAACTAAAAC-3′, Fignl1: 5′-GGATAGAAGAAAATAAAATTAAAGTATA-3′ and 5′-AACTAAATCTCCCTCATAAAAC-3′, Ikzf1: 5′-GAGGGTGTTTTTTTTTGATT-3′ and 5′-TTATTAAAATACCCTACTAAATTAACTAC-3′. The PCR products were subcloned into pGEM T-Easy vector (Promega) and sequenced. CpG methylation status was analyzed by the QUMA program (http://quma.cdb.riken.jp/top/quma_main_j.html) (66). In Cobl and Ikzf1 analysis, the first 19 and 25 CpGs were used, respectively.
The reporter plasmids pHNE and pHNIE (top and second column of Fig. 2B, respectively) were obtained from Dr K. Ishihara (Kumamoto University, Japan) (42). The pHNE vector consists of a neomycin-resistance (Neo) gene driven by the H19 promoter, a 2.5 kb NsiI-BglII fragment containing the enhancers and 1.8 kb AatII-HindIII fragment containing the H19 DMR insulator. This insulator prevents influence from the adjacent regions. The pHNIE vector is composed of pHNE containing an additional H19 DMR insulator between Neo and the enhancers. To make test vectors, the MSR sequence within Meg1-DMR fragments was generated by PCR and inserted into the XhoI site of pHNE between the Neo gene and the enhancers. For the PCR reaction, 20 ng of genomic DNA in a 50 µl reaction mixture containing 1× buffer for KOD Dash (TOYOBO), 2.0 mm dNTP mixture, 0.8 µM primers and 1.0 U of KOD Dash (TOYOBO) were subjected to 35 PCR cycles. PCR was carried out under the following conditions: 96°C for 30 s, 68°C for 2 s and 74°C for 30 s. The primer pairs were designed with XhoI site on the 5′-end of each primer. The PCR product was digested with XhoI (TaKaRa) and ligated with the pHNE vector. The following primer sequences were used: 5′-GATGGCATTCGGGAGGCTGTGTTGC-3′ and 5′-GCTCTGGAGCCTAGAGGAGCGCGG-3′. The direction of the fragments inserted into pHNE was confirmed by sequencing. The reporter constructs were linearized with MluI (TOYOBO), and 0.17 pmol of each construct was transfected into L cell by electroporation using Nucleofector I (amaxa biosystems) with a pGL3-control vector (Promega) encoding the firefly luciferase gene. After 48 h, cells were replated and selected by G418 (400 µg/ml). Colonies were counted after 2 weeks of selection. The number of G418-resistant colonies was corrected for transfection efficiency based on the luciferase activity and normalized to that obtained with pHNE.
Test reporter constructs were co-transfected into L cells with a pMLuc2 renilla luciferase vector (Novagen) as an internal control using Lipofectamine 2000 reagent (Invitrogen) according to the manufacturer's instructions. After 24 h, cells were analyzed for luciferase activity using the Dual-Glo Luciferase Assay System (Promega) according to the manufacturer's instructions. The assays were performed in triplicate and more than three times. Luciferase activity of each sample was measured by a GENE LIGHT 55A (Microtec) and normalized to a control (renilla luciferase activity).
pGL3Mp vector was generated by inserting the Meg1/Grb10 Type I promoter region into the BglII-HindIII site of the pGL3-Basic vector (Promega). The Meg1/Grb10 promoter region was PCR-amplified using the forward primer 5′-ACAATGCACACAGCACAGC-3′ with BglII site on the 5′-end and the reverse primer 5′-ACTCCTACCTGACGTGCAGC-3′ with HindIII site on the 5′-end. The PCR product was digested with BglII and HindIII (TaKaRa) and ligated with pGL3-Basic vector. A KOD Dash (TOYOBO) was used for the PCR. Twenty nanogram of genomic DNA in a 50 µl reaction mixture containing 1× buffer for KOD Dash (TOYOBO), 2.0 mm dNTP mixture, 0.8 µM primers, 2.5% DMSO and 1.0 U of KOD Dash (TOYOBO) were subjected to 35 PCR cycles. PCR was carried out under the following conditions: 96°C for 30 s, 64°C for 2 s and 74°C for 30 s. To generate reporter constructs, test DNA fragments (MSR, h-MSR, H19 DMR, control for Meg1-DMR, control for h-MSR and control for H19 DMR) were generated by PCR using primers designed with BglII and SalI sites at their 5′-end. For MSR, the PCR conditions and primer sequence were described above. For others, PCR was performed using ExTaq (TaKaRa), and the PCR primers were listed below. h-MSR: 5′-CGGAGGCTGAGTATTGCAG-3′ and 5′-GGCGCAGAAAACCGAC-3′, H19 DMR: 5′-CTTGGACGTCTGCTGAATCAG-3′ and 5′-CGTCTGCCGAGCAATATGTAGTA-3′, control for MSR: 5′-ACACTCTAGCCCAGGAGCAA-3′ and 5′-GAGGGCAATACACACTGCCT-3′, control for h-MSR: 5′-TTTGGTGGGCTAACGATGAT-3′ and 5′-ATGGACTCTCCAGCATGGAG-3′, control for H19 DMR: 5′-GAAGGCTTAGTCCCTGGGAG-3′ and 5′-CTCCTTTAAATTCGTGACGACA-3′. The PCR products were digested with BglII and SalI (TaKaRa) and ligated with the pGL3Mp vector. The reporter constructs were transfected into L cells, and the luciferase activity of each construct was measured as described above.
Methylated cassette approach
pGL3Mp vectors with insertion of MSR-F/R and control for MSR were digested with HpaI and SalI (TaKaRa), and the pGL3Mp vector fragment and the insert fragments of each construct (MSR-F/R, h-MSR or control for MSR) were purified by gel extraction using the RECO CHIP (TaKaRa). The insert fragments were methylated with SssI metyltransferase (New England Biolabs), and the methylation status of the insert fragments was analyzed by digestion with HpaI for MSR and h-MSR, and with HpyCH4IV for control. The mock-methylated fragments were treated in the same way without the addition of SssI. The methylated and mock-methylated insert fragments were ligated within the pGL3Mp vector. Each ligation mix was phenol/chloroform extracted and used directly for transfection. The luciferase activity of each construct was measured as described above.
Allelelic expression analysis
The allelic expression of the Meg1/Grb10, Meg1/Grb10 Type I, Meg1/Grb10 Type II, Cobl, Ddc, Ikzf1 and Fignl1 genes between JF1 and C57BL/6J (or ΔDMR (=129/Sv/Ev)) was detected by PCR-RFLP and direct sequencing analysis. PCR were performed under the conditions described above. The following primers were used for DNA amplifications. Meg1/Grb10: 5′-CTTGATACCACCCAGAAAGTCTG-3′ and 5′-AACCCAAAGCATTTGGCAG-3′, Meg1/Grb10 Type I: 5′-CACGAAGTTTCCGCGCA-3′ and 5′-AGTATCAGTATCAGACTGCATGTTG-3′, Meg1/Grb10 Type II: 5′-GCGATCATTCGTCTCTGAGC-3′ and 5′-AGTATCAGTATCAGACTGCATGTTG-3′, Cobl: 5′-AAGTGAATGAGGACGGCG-3′ and 5′-GGTGAGAAGGATTCAGGTGG-3′, Ddc-exon1: 5′-TTCGCAGAGCTGGACAATC-3′ and 5′-TGCAAGCATAGCTGGGTATG-3′, Ddc-exon1a: 5′-CGAATAGAGAGGAGGCGAT-3′ and 5′-TGCAAGCATAGCTGGGTATG-3′, Ikzf1: 5′-CTTTCGGGATCCCTTTGAGT-3′ and 5′-CCTTCAGCACATTGCACAAC-3′, Fignl1: 5′-TTGTGTTCCTTCTGGCTGTG-3′ and 5′-CAGCTTCATCAATCTCTTGGG-3′. For RFLP analysis of Meg1/Grb10, Cobl, Ddc and Ikzf1, the PCR products were digested with MspI (TaKaRa), BanI (New England Biolabs), MspI and MspI, and subjected to agarose gel electrophoresis. For direct sequencing analysis of Fignl1, the PCR products were sequenced directly.
Real-time quantitative PCR
The assays were performed in triplicate and the copy number of genes examined were calculated with a LightCycler 480 (Roche Diagnostics) by using Power SYBR Green PCR Master Mix (Applied Biosystems) for the analysis of total expression and TaqMan Gene Expression Master Mix with TaqMan MGB probes (Applied Byosystems) for the analysis of allele-specific expression. PCR was carried out under the following conditions: 96°C for 15 s, 65°C for 30 s and 72°C for 30 s for the total expression analysis, and 96°C for 15 s, 60°C for 30 s for the allele-specific analysis. For the total expression analysis, the data for each gene were normalized to an internal standard (β-actin), and for the B6 or Meg1ΔDMR allele-specific analysis, the expression levels were normalized to the JF1 allele expression level. The sequences of the primers for the TaqMan probes used are listed below. For total expression analysis: for Meg1/Grb10, 5′-GTTTCTGAGAATTCTCTGTGGC-3′ and 5′-CTGTGAGACTCCTCGCGG-3′; for Cobl, 5′-TCTGTGAAAGTGCCAGCATC-3′ and 5′-TGTGGACAGCAGCAGGATAG-3′; for Ddc, 5′-TAGAATGTACGGAGTCAAGGGG-3′ and 5′-AGCAGCTCTGCTTTCATTCTTT-3′; for β-actin, 5′-AAGTGTGACGTTGACATCCG-3′ and 5′-GATCCACATCTGCTGGAAGG-3′. For allele-specific analysis: for Cobl, 5′-AAGGCTATGACATGCATCAGGTT-3′ and 5′-TGGTGAAATCTCAGGCTCCAT-3′ with TaqMan probe 5′-CTGTGAAGGAGCCTT-3′ (B6 or Meg1ΔDMR) or 5′-CTGTGAAAGGGCCTT-3′ (JF1); for Ddc-exon1, 5′-CAATGCCATCCAGATAGTCAGCTA-3′ and 5′-AGTGGACCTGTGAAGAATCCAAA-3′ with TaqMan probe 5′- TCTCCTCCGGAATT-3′ (B6 or Meg1ΔDMR) or 5′-CTCTCCTTCGGAATT-3′ (JF1); for Ddc-exon1a, 5′-CAATGCCATCCAGATAGTCAGCTA-3′ and 5′-CCAGCTGCCTTTTTCAACATG-3′ with the same TaqMan probes as Ddc-exon1.
Results are presented as the means±standard deviation. Statistical analyses were performed using Mann–Whitney U-test.
This work was supported by grants from Creative Science Research, the research program of Japan Society for the Promotion of Science (JSPS), the Japanese Ministry of Education, Global Center of Excellence (GCOE) Program, “Internaional Research Center for Molecular Science in Tooth and Bone Diseases”, the Mitsubishi Foundation and the Ministry of Health, Labor and Welfare for Child Health and Development (17C-2) and a Grant-in-Aid for Scientific Research on Priority Areas form the Ministry of Education, Culture, Sports, Science and Technology of Japan (1508023) to F.I., and the Asahi Glass Foundation and JSPS, Grants-in Aid for Scientific Research to T.K.-I.
We thank S. Aizawa of Center for Developmental Biology, RIKEN for providing the DT-A vector that was used for making Meg1ΔDMR construct, E. Robertson of University of Oxford for the CCE ES cells, Y. Nakahara and M. Takabe of the Mitsubishi Kagaku Institute of Life Sciences for animal breeding and H. Hasegawa, N. Kawabe of the Tokai University for histological analysis. The JF1/Msf mouse strain (RBRC00639) was provided by RIKEN BRC, which is participating in the National Bio-Resource Project of the MEXT, Japan. Pacific Edit reviewed the manuscript prior to submission.
Conflict of Interest statement. None declared.