Skip Navigation

This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (44)
Right arrowRequest Permissions
Google Scholar
Right arrow Articles by Arnaud, P.
Right arrow Articles by Kelsey, G.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Arnaud, P.
Right arrow Articles by Kelsey, G.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?

Human Molecular Genetics, 2003, Vol. 12, No. 9 1005-1019
DOI: 10.1093/hmg/ddg110
© 2003 Oxford University Press

Conserved methylation imprints in the human and mouse GRB10 genes with divergent allelic expression suggests differential reading of the same mark

Philippe Arnaud1, David Monk2, Megan Hitchins2,{dagger}, Emma Gordon1, Wendy Dean1, Colin V. Beechey3, Jo Peters3, William Craigen4, Michael Preece2, Philip Stanier2, Gudrun E. Moore2 and Gavin Kelsey1,*

1Developmental Genetics Programme, The Babraham Institute, Cambridge CB2 4AT, UK, 2Department of Fetal and Maternal Medicine, Institute of Reproductive and Developmental Biology, Faculty of Medicine, Imperial College, Du Cane Road, London W12 0NN, UK, 3MRC Mammalian Genetics Unit, Harwell, Didcot, Oxfordshire OX11 0RD, UK and 4Departments of Molecular and Human Genetics, and Pediatrics, Baylor College of Medicine, Houston, TX, USA

Received December 17, 2002; Accepted February 21, 2003


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 RESULTS
 DISCUSSION
 MATERIALS AND METHODS
 REFERENCES
 
Grb10/GRB10 encodes a cytoplasmic adapter protein which modulates coupling of a number of cell surface receptor tyrosine kinases with specific signalling pathways. Mouse Grb10 is an imprinted gene with maternal-specific expression. In contrast, human GRB10 is expressed biallelically in most tissues, except for maternal-specific expression of one isoform in muscle and paternal expression in fetal brain. Owing to its location in 7p11.2–p12, GRB10 has been considered a candidate gene for the imprinted growth disorder, the Silver–Russell syndrome (SRS), but its predominantly biallelic expression argues against involvement in the syndrome. To investigate the discrepant imprinting between mouse and human, we compared the sequence organization of their upstream regions, and examined their allelic methylation patterns and the splice variant organization of the mouse locus. Contrary to expectation, we detected both maternal and paternal expression of mouse Grb10. Expression of the paternal allele arises from a different promoter region than the maternal and, as in human, is restricted to the brain. The upstream regions are well conserved, especially the presence of two CpG islands. Surprisingly, both genes have a similar imprinted methylation pattern, the second CpG island is a differentially methylated region (DMR) with maternal methylation in both species. Analysis of 24 SRS patients did not reveal methylation anomalies in the DMR. In the mouse this DMR is a gametic methylation mark. Our results suggest that the difference in imprinted expression in mouse and human is not due to acquisition of an imprint mark but in differences in the reading of this mark.


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
RESULTS
 DISCUSSION
 MATERIALS AND METHODS
 REFERENCES
 
Genomic imprinting is a form of epigenetic inheritance in mammals in which a small subset of autosomal genes show parental-allele-specific expression (1,2). Many of these genes are implicated in the control of fetal growth (3). As a result, aberrant imprinting or inappropriate expression of imprinted genes gives rise to a variety of disease syndromes (4). To accomplish monoallelic expression, imprinted genes are subject to germline specific epigenetic modifications (imprint marks), of which CpG methylation is a major component (5). Most imprinted genes examined so far show differentially methylated regions (DMRs), where the methylation state of the parental alleles differs. These can be considered as either germline (primary) DMRs, as methylation is inherited from the male or female gamete and subsequently maintained throughout development, or somatic (secondary) DMRs, at which parental-allele-specific methylation is established after fertilization (6). Several observations clearly indicate that germline DMRs play a crucial role in the establishment of imprinting, and are associated with imprinting control regions (ICRs) (7–10). Ultimately, the epigenetic features at ICRs are read in differing ways to ensure proper parental-allele-specific expression of the genes or imprinted clusters (6,11). Despite the fact that methylation at a germline DMR is maintained from the mature gamete to adulthood, at some imprinted genes monoallelic expression may be limited to selected tissues or specific developmental stages (12). In some cases, such restricted imprinted expression can be correlated with the presence of secondary, tissue-specific DMRs (13–15).

How genes are differentially marked in the germlines remains to be fully elucidated. This issue could be addressed by comparative analysis of the few genes whose imprinting is known to differ between species. For example, the Rasgrf1 gene is imprinted with paternal-specific expression in Mus musculus and Rattus, but not in Peromyscus (16). Imprinted expression is correlated with the presence of a block of direct repeats found next to the paternally methylated DMR of Rasgrf1 (16). This repeat region has recently been shown to be required for establishment of methylation at the Rasgrf1 DMR and proper imprinted expression (17). The imprinting of the mouse Impact gene but the absence of imprinting of the human orthologue also suggests a role for direct repeats (18). Species differences in imprinting may also result from differences in the reading of imprint marks. Thus, although the mouse Igf2r and human IGF2R genes both have a maternally methylated intronic CpG island containing numerous direct repeats, only the mouse gene is monoallelically expressed (19,20). In this case, monoallelic expression of mouse Igf2r seems to be governed by an imprinted antisense transcript emanating from the intronic CpG island, which is absent in humans. Grb10/GRB10 falls into this class of genes with a species-specific pattern of imprinting. Grb10 (also known as Meg1), which maps to mouse proximal chromosome (Chr) 11, is imprinted with maternal-specific expression (21). In contrast, human GRB10 is expressed almost exclusively from the paternal allele in the fetal brain, biallelic in all other tissues, except for maternal-specific expression of one isoform ({gamma}1) in skeletal muscle (22).

Grb10/GRB10 encodes the growth factor receptor bound protein 10, which belongs to a small family of cytoplasmic adapter proteins that mediate the coupling of a number of cell surface receptor tyrosine kinases and signalling molecules with specific signalling pathways. GRB10 could play a role in mitogenesis through its involvement in the signal transduction pathway for insulin and/or insulin-like growth factors. However, whether its effect on insulin receptor signalling is inhibitory or stimulatory remains to be resolved (23–26).

Human GRB10 is of particular interest as it maps to the region p11.2–p12 on Chr 7, within one candidate interval for the Silver–Russell syndrome (SRS) growth disorder (27,28). SRS is characterized by severe pre- and post-natal growth restriction and a spectrum of additional dysmorphic features. SRS is a genetically heterogeneous disorder, with maternal uniparental disomy (UPD) for Chr 7 found in up to 10% of cases (29). However, the finding of biallelic expression in most tissues, including tissues important for linear growth (22,30), and the failure to identify mutations in GRB10 in 139 SRS patients (29), argue against an involvement of GRB10 in SRS. Therefore, although the properties and localization of the gene make it a plausible candidate, the contribution, if any, of GRB10 to SRS requires further investigation.

To investigate the discrepant imprinting of the mouse and human homologues and to determine if an alteration in epigenotype of GRB10 could contribute to SRS, we have compared the genomic organization of the upstream regions of these genes and examined their allelic methylation patterns.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
DISCUSSION
 MATERIALS AND METHODS
 REFERENCES
 
Conserved organization of the upstream regions of mouse Grb10 and human GRB10
Grb10 and GRB10 do not apparently map to clusters of imprinted genes. In fact, the genes flanking Grb10/GRB10 and other genes centromeric to GRB10 have all been shown to be expressed biallelically (31–33). Therefore, elements controlling their imprinting are expected to reside within the genes or in close proximity. To identify candidate ICRs, we compared the genomic organisation of human GRB10 and mouse Grb10, with particular emphasis on a 20 kb region surrounding the putative transcription start sites. Comparison was made using the human GRB10 (AC004920) and mouse Grb10 (AL663087) genomic sequences. The graphical output of pairwise sequence alignments made using VISTA and schematic representation of other sequence features are shown in Figure 1. Within this upstream region, there are two domains of marked sequence similarity, domain 1 and 2, separated by a less conserved region. Domains 1 and 2 each contain CpG islands (CGIs; identified using standard criteria: >200 bp, CpG:GpC obs/exp >0.6, C+G frequency >0.5, window size 100 bp) in both human and mouse sequences. CGI1 is 1384 bp in human and 1349 bp in mouse; CGI2 is shorter, being 764 bp in human and 861 bp in mouse. A third ‘weaker’ CpG island denoted mCGI3 (201 bp in length, CpG obs/exp=0.6, C+G frequency=0.51) was detected specifically in the mouse sequence in a region of lower similarity (<65%). Interestingly, whereas there is good overall sequence alignment in domain 1 particularly in CGI1, the alignment in domain 2 is interrupted. The disruption is due mainly to blocks of direct repeats present specifically in the mouse sequence (Fig. 1B). Four sets of CG-rich direct repeats (10mers to 21mers), all of which share a consensus which includes CACGCGCC, are present in mouse CGI2 from position 65 749 to 66 243 (Table 1). These repeats account for nearly 35% of the CpG island in length and 45% (47/104) in CpG content. A second block of a C-rich tandem repeat (28 bp repeated 9.2 times) is present downstream of CGI2 in mouse. In the human sequence, in contrast, we detected only a single direct repeat unit, a 7 bp sequence repeated 4.6 times, at the 3' end of CGI2 (Table 1). The finding of species-specific tandem repeats within or in the vicinity of a CpG island may be relevant to cis-acting elements involved in the species-specific imprinted expression of Grb10/GRB10. There are no other CpG islands within the human and mouse genes, or in their close vicinity, indicating that these identified CpG islands are likely locations for ICRs.



View larger version (44K):
[in this window]
[in a new window]
  		Figure 1. Conserved organization of the upstream region in mouse Grb10 and human GRB10. Schematic representation of two-way sequence alignment of the human and mouse upstream regions using VISTA, using the human as reference sequence in (A), and the mouse as reference sequence in (B). Regions of nucleotide similarity of >75% in a window of 100 bp are shown in grey. The human sequence is from AC004920, the mouse is from AL663087, the positions in each sequence are given on the horizontal axes. Note that both sequences are reversed to display them with respect to the transcription orientation of GRB10 and Grb10. The main features of the regions analysed are shown schematically. Interspersed repetitive elements were determined using RepeatMasker, and are identified according to the key at the bottom of the figure. The directions of the arrows indicate transcriptional orientation of the respective repetitive sequences. Two CpG islands (CGI 1 and 2, grey boxes) were identified by CpGplot, a third CpG island (mCGI3) was identified in the mouse sequence using Webgene CpG island prediction program. Direct repeat blocks in the mouse are indicated by hatched boxes (see Table 1 and text for details). When present in the regions analysed, exons for the human (22) and mouse (see text) are shown.

 

View this table:
[in this window]
[in a new window]
  		Table 1. Sequence and positions of direct repeats in upstream region of mouse Grb10 and human GRB10 genes. Tandem repeats identified using Tandem Repeat Finder. The consensus units are given by the upper italicized lines, bases diverging from the consensus are in bold. Nucleotide positions are given according to AC04920 and AL663087 for human and mouse, respectively. Note that sequences are displayed in a reversed orientation with respect to the transcription orientiation of GRB10 and Grb10
 
We determined the repetitive sequence content in the human and mouse upstream sequences using RepeatMasker. Few regions in either species match the consensus sequences of interspersed repeat families and, for most of these matches, homology is limited to a short part of the consensus. The mobile element content (retroelements plus DNA transposons) found in the region spanning CGI1 and CGI2 is 4.5% in mouse and 21.5% in human. This difference is due mainly to a nearly full-length Hal1 Line element (non-autonomous L1 element lacking ORF2) present in five fragments in the human sequence (between positions 47 200 and 44 100). A region downstream of CGI1, sharing homology with the core and the 3' part of a Mir retroposon is conserved between human and mouse. A full-length DNA transposon MER type 2 inserted downstream of the mCGI3 is only found in the mouse (Fig. 1).

Genomic organization, splice variants, and isoform- and tissue-specific imprinting of mouse Grb10
Five splice variants of human GRB10 (hGr10ß, {gamma}, {varepsilon}, {zeta}, {sigma}), and a putative testis-specific splice variant (accession no. AJ271366), have been described (22,34). To date, mouse Grb10 is known to be expressed only as two alternatively spliced transcripts, mGrb10{alpha} (35), and mGrb10{delta} (36). We established the likely genomic organization of mouse Grb10 and structure of its splice variants by compiling information from the Ensembl mouse genome server and splice variants described in the literature and/or available in EST databases. In total, Grb10 contains at least 19 exons and encompasses nearly 110 kb of genomic DNA. The three first exons are untranslated (Un1, Un2, Un3) and among the 16 remaining exons two coding exons (exons 3 and 4) are not present in all predicted isoforms (Fig. 2A and B). Human GRB10 contains 22 exons spanning ~300 kb, five being untranslated (22).



View larger version (52K):
[in this window]
[in a new window]
  		Figure 2. Organization, splice variants, and imprinted expression of mouse Grb10. (A) The exon–intron structure of mouse Grb10 as presented at the Ensembl mouse genome server and inferred from known splice variants and available database entries is shown. Exons translated in all isoforms are represented as filled boxes; hatched boxes are alternatively spliced coding exons; untranslated exons and the 3'UTR are shown as open boxes. The positions of the CpG islands described in Figure 1 are shown above. The locations of primers (A to G) used for RT–PCR are indicated by arrows. Not drawn to scale. (B) Mouse Grb10 splice variants. The previously described {alpha} and {delta} isoforms correspond to accession numbers U18996 and AF022072. Isoform ß1 is represented by the Ensembl identifier ENSMUST00000020393; ß2 by ENSMUST00000047154. For all isoforms, the 3'UTR is assumed from information given in Ensembl. Vertical arrows indicate the predicted translation start of each isoform. (C) RT–PCR analysises of expression of Grb10 isoforms {alpha} and {delta} in brain, liver and kidney of neonatal mice with PatDp(prox11), MatDp(prox11) and one normal littermate, using amplification primers identified in (A). (D) RT–PCR analysises of expression of Grb10 isoforms ß1 and ß2 in the same samples. (E) RT–PCR analysis of expression of all Grb10 isoforms in the same samples. All experiments were performed in three independent RT samples. As controls (not shown), the paternally expressed U2af1-rs1 RNA was not detected in the MatDp(prox11) samples, Hprt RNA was amplified in all samples to demonstrate RNA/cDNA integrity, and samples on which RT–PCR was performed without reverse transcriptase gave no product for any primer set.

 
The two characterized splice variants in mouse, {alpha} and {delta}, both initiate at Un1 in CGI1, nearly 70 kb upstream of the first translated exon and lack the untranslated exon Un2 (Figs 1 and 2). mGrb10{delta} is 75 bp shorter, due to the lack of the third translated exon, and appears to be the predominant isoform in mouse tissues (Fig. 2C) (36). In addition, two further isoforms are predicted, represented as the Ensembl transcript identifiers ENSMUST00000020393 and ENSMUST00000047154, which we propose to designate mGrb10ß1 and mGrb10ß2, respectively (Fig. 2B). mGrb10ß1 and mGrb10ß2 differ in the alternative splicing of exons 3 and 4. Interestingly, the predicted first exon of these two isoforms is the untranslated exon Un2, 57.4 kb upstream of the translation start site, which overlaps the CpG island mCGI3 (Figs 1 and 2).

So far, all analyses of mouse Grb10 have consistently shown expression specifically of the maternal allele (21,37). Recently, however, biallelic expression in brain has been reported (33). All these analyses were made by RT–PCR or northern blotting using primers or probes in the 3' part of the transcript, which could not distinguish between the various isoforms described in Figure 2B and, in particular, could not differentiate transcripts initiating in Un1 and Un2. Here, we have investigated isoform-specific imprinting of mouse Grb10 by RT–PCR using RNAs from tissues (brain, liver and kidney) from neonatal mice with maternal or paternal duplication for proximal Chr 11 [designated MatDp(prox11) and PatDp(prox11)]. To test imprinted expression of mGrb10{alpha} and mGrb10{delta}, RT–PCRs were done with a primer in Un1 (primer A) in combination with primers in exons Un3 or 4 (primers C or D, respectively). RT–PCR products were detected for primer combinations A+C and A+D in the RNAs from normal and MatDp(prox11) tissues, but not from PatDp(prox11) tissues, clearly indicating maternal expression for the mGrb10{alpha} and/or mGrb10{delta} isoforms in the three tissues tested, including brain (Fig. 2C). Furthermore, we could discriminate between mGrb10{alpha} and mGrb10{delta} in the A+D RT amplicons. The presence of two RT–PCR products in amplicons from normal and MatDp(prox11) tissues with a weaker product corresponding to mGrb10{alpha} confirms that both transcripts are maternally expressed and that mGrb10{delta} is the predominant isoform. In striking contrast, use of primers specific for the predicted ß1 and ß2 isoforms (primer B in Un 2 in combination with primers C, D or E) detected expression only in brain and in PatDp(prox11) RNA but not in MatDp(prox11) RNA (Fig. 2D). The presence of two products of similar intensity in the B+E RT amplicon obtained from normal and PatDp(prox11) brain indicate that the ß1 and ß2 alternatively spliced isoforms are paternally expressed in brain to a similar extent. The use of primers designed to amplify all splice-variants (F+G, situated in the 3' part of the transcript) resulted in a compilation of the patterns obtained with the splice-variant specific primers, with apparent biallelic expression of Grb10 in brain and only maternal expression in others tissues (Fig. 2E). These results were confirmed by performing similar RT analysis in brain and liver from mice with maternal or paternal disomy for the entire Chr 11 (data not shown). We failed to detect any PCR product containing both Un1 and Un2 exons (using primers A+B'), suggesting that mCGI3 is the likely origin for the ß1 and ß2 transcripts (not shown).

Altogether, our results suggest that the mGrb10{alpha} and mGrb10{delta} isoforms initiate in CGI1 and are maternally expressed in all tissues, whereas the ß1 and ß2 isoforms initiating in mCGI3 are specifically expressed in brain from the paternal allele. The finding of paternally expressed isoforms in brain has parallels with human GRB10 and suggests a more conserved imprinting profile in this tissue than previously thought.

CGI2 is a maternally methylated DMR in human GRB10 and mouse Grb10
DNA methylation is associated with the differential transcriptional activity of imprinted genes. Therefore, we wished to determine if imprinted methylation is present at the mouse Grb10 and/or human GRB10 5' UTR, and whether it could be associated with the isoform-specific and species-specific patterns of monoallelic/biallelic expression. To address this question in human, we analysed blood DNAs from three normal individuals and three patients with maternal disomy for the entire Chr 7 (mUPD7), and lymphoblastoid cell line DNA from one patient with paternal disomy for the entire Chr 7 (pUDP7). The mouse Grb10 5'UTR methylation pattern was determined by comparing 13.5 dpc whole embryo DNAs from two MatDp(prox11), one PatDp(prox11) and one normal littermate. Using a bisulfite based approach, we analysed eight regions in the human sequence (H1–H8) and five (M0–M4) in mouse (Fig. 3). These regions are in or near CGI1 and 2 in mouse and human, or in regions with a suitable amount of CpG sites for a bisulfite analysis, between the two CpG islands in human (Fig. 3).



View larger version (50K):
[in this window]
[in a new window]
  		Figure 3. Bisulfite sequence analysis of methylation in human GRB10 and mouse Grb10. (A) For the human GRB10 upstream region, bisulfite-modified DNA was analysed with eight PCR products (H1–H8; shown as grey bars below the scheme of the GRB10 upstream region). Methylation was examined in DNA from lymphocytes from controls (n=3) and mUDP7 (n=3), and from a pUDP7 lymphoblastoid cell line. Details of methylation for products H1, H2, H6 and H7 are shown. Each row of dots represents the series of CpGs in an individual sequence molecule, in which methylated CpG is shown as a solid circle and unmethylated CpG as an open circle. The properties of the region are displayed at the top of the figure according to the scheme in Figure 1. The vertical arrow above the H2 product marks the HpaII site described as differentially methylated (22). (B) The equivalent analysis of the mouse Grb10 upstream region. Methylation was assayed in DNAs (13.5 dpc embryo) from PatDp(prox11) (n=1), MatDp(prox11) (n=2) and one normal littermate. Analysis was made with five PCR products (M0–M4); the results are shown for M1–M4. N/D, not done.

 
In human, the 3' part of CGI1 appears to be mainly unmethylated (Fig. 3, product H1). This is in agreement with previous Southern analysis demonstrating that CGI1 was unmethylated on both alleles (22). A region directly downstream of the CpG island was shown to be a paternally methylated region in the same Southern analysis (22). However, our bisulfite sequencing results do not show this region as being a DMR. Sequences from PCR product H2 revealed that partially methylated and unmethylated molecules are indeed found in normal DNA, and that 47% of the HpaII sites analysed by Southern were methylated (corresponding CpG indicated by an arrow in Fig. 3A). However, both mUPD7 and pUPD7 DNA also contained partially methylated and unmethylated molecules. This may imply that this region is a transition between the unmethylated CGI1 and the downstream methylated region, and has a stochastic methylation profile: regions sequenced between the two CpG islands were highly methylated on both alleles (Fig. 3A, PCR products H3, H4 and H5; data not shown).

In contrast to CGI1, CGI2 was found to constitute a DMR, as methylated and unmethylated molecules were obtained in equal proportions in normal DNA. Moreover, sequences from mUPD7 DNA were very highly methylated whilst sequences from pUDP7 DNA were unmethylated (Fig. 3A, product H7). Sequences from PCR product H8 confirmed that the whole CGI2 is likely to be a DMR (data not shown). The upstream border of the DMR is immediately 5' of CGI2, as the region 750 bp upstream is methylated on both alleles (Fig. 3A, product H6).

A similar pattern of methylation was found for mouse Grb10. Methylation at CGI1 was examined with PCR products at the 5' and 3' parts of the CpG island (Fig. 3B, products M0 and M1). Sequences obtained for M1 from both MatDp(prox11) and PatDp(prox11) DNA contained essentially no methylation (Fig. 3B); products for the M0 amplicon were digested with informative enzymes and also found to be free of methylation (data not shown). CGI2, however, is included in a maternally methylated DMR, as sequences immediately upstream of the CpG island (product M3) and at the 3' end (product M4) showed high levels of methylation in MatDp(prox11) DNA, very little methylation in PatDp(prox11) DNA, and a mixture of highly methylated and unmethylated sequences in normal DNA (Fig. 3B). (Because of the extensive direct repeats within CGI2, it was not possible to obtain bisulfite sequences across the whole CpG island.) The different methylation states of mouse CGI1 and CGI2 observed by bisulfite sequencing were confirmed by Southern blot analysis (data not shown) with the same DNA samples. Methylation analysis at mCGI3 is reported below.

In conclusion, although human GRB10 is mainly biallelically expressed, except for paternal expression in brain, and mouse Grb10 is maternally expressed, except for isoform-specific paternal expression in brain, both genes possess a similar maternally methylated DMR in CGI2.

Mouse CGI1 is maintained unmethylated in adult tissues
The presence of an equivalent DMR but no conservation of imprinted expression is observed for the mouse Igf2r and human IGF2R genes in a similar configuration to that which we observe for GRB10/Grb10. In mouse, maternal-specific expression of Igf2r correlates with the establishment of a secondary paternal DMR at the CpG island promoter region, whereas the promoter region remains unmethylated in the biallelically expressed human IGF2R gene. As a first step to see whether further parallels may exist between GRB10/Grb10 and IGF2R/Igf2r in mechanisms of imprinting, we wished to determine if a secondary paternal DMR is acquired in mouse CGI1 at later stages of development. Southern blots of HindIII–MspI and HindIII–HpaII digested genomic DNA from adult tissues (brain, liver and muscle) of MatDp(prox11), PatDp(prox11) and normal mice were hybridized with probes for CGI1 (probe A, data not shown). The HindIII fragment analysed was digested to the same extent with MspI and HpaII in all DNA samples (data not shown), indicating that no methylation is subsequently established at CGI1.

Tissue-specific hypomethylation of mCGI3
Hybridization of Southern blots of MatDp(prox11) and PatDp(prox11) embryo DNAs with a probe (Fig. 4A; probe B) lying between the two CpG islands confirmed that CGI2 is a DMR with maternal methylation, but the HpaII digestion pattern also suggested that hypomethylation or partial methylation of the paternal allele extended upstream of CGI2 and may cover mCGI3. As mCGI3 may contain the initiation site for the brain-expressed ß1 and ß2 transcripts, we were prompted to examine mCGI3 methylation in tissues in more detail. Results of bisulfite sequence analysis (Fig. 4B) showed that, whereas the maternal allele is mainly methylated in the three adult tissues examined, methylation of the paternal allele is more variable. In brain, in particular, this region was found to have a low level of methylation in PatDp(prox11) DNA, whereas in kidney there was no apparent difference in parent allele methylation. Therefore, mCGI3 appears to be a tissue-specific DMR, where the variable methylation pattern found on the paternal allele in brain could correspond to a mixture of tissues expressing and not expressing Grb10. Alternatively, these patterns suggest that mCGI3 is partially protected from de novo and/or spreading of methylation from surrounding regions on the paternal allele, and this protection varies between and within tissues.



View larger version (39K):
[in this window]
[in a new window]
  		Figure 4. Analysis of mCGI3 methylation in embryo and adult mouse tissues. (A) Southern analysis of DNAs (13.5 dpc embryo) from PatDp(prox11), MatDp(prox11) and one normal littermate. At the top, a representation of the region analysed: the fragment produced by HindIII (Hind) digestion and position of MspI/HpaII sites (vertical bars) are shown; the probe used (probe B) is depicted as black bar. DNAs were digested with HindIII alone (—), HindIII+MspI (M) and HindIII+HpaII (H) and hybridized with probe B. (B) Methylation at mCGI3 analysed in adult tissue DNAs by bisulfite modification and sequencing of PCR product M5, representative results are shown. Explanation for symbols as in legend to Figure 3.

 
Mouse CGI2 is a germline DMR
To determine whether any of the mouse Grb10 CpG islands constitute a germline DMR and therefore a likely ICR, methylation in gametes and pre-implantation embryos was analysed by bisulfite sequencing. The region immediately upstream of CGI2 (Fig. 5, product M3) was found to be methylated in oocyte, unmethylated in sperm and contained a similar ratio of methylated and unmethylated sequences in morula DNA. This pattern, together with the parental-allele-specific methylation at later stages, strongly indicates the presence of a germline DMR with methylation of maternally derived alleles maintained throughout embryonic development. A similar profile was found for the 3' part of CGI2 (Fig. 5, product M4), although some unmethylated sequences were obtained from oocyte DNA (we consider it unlikely that these sequences represent contaminating maternal material, as the same batch of oocytes was used). The germline DMR does not extend far upstream of CGI2, as sequences obtained for mCGI3 showed high levels of methylation in both germlines, with erasure during pre-implantation stages (Fig. 5, product M5). Methylation at the tissue-specific DMR mCGI3 must be acquired after implantation. Finally, we analysed the region downstream of CGI1 (Fig. 5, product M2). Although highly methylated in oocytes and mainly unmethylated in sperm, there was very little methylation in morulae, indicating demethylation of oocyte-derived methylation in pre-implantation embryos. The methylation of this region in oocyte could suggest that the entire region between CGI1 and CGI2 is methylated in oocyte.



View larger version (41K):
[in this window]
[in a new window]
  		Figure 5. Germline and early embryo methylation profile of mouse Grb10. Methylation analysed in DNA from oocytes, sperm and morulae by bisulfite sequencing. Analysis of sperm and morula DNAs was repeated in independent experiments, the results of one experiment are shown. Explanation for symbols as in legend to Figure 3.

 
The CGI2 DMR is conserved in SRS patients
SRS is a congenital disease characterized by pre- and postnatal growth deficiency and other growth abnormalities for which the molecular aetiology is unknown. The properties and localization of human GRB10 make it a plausible candidate, although its predominant biallelic expression is less easy to reconcile. Having identified that CGI2 of human GRB10 is a maternally methylated DMR, and that the homologous mouse CGI2 is a germline DMR and candidate control element for Grb10 imprinting, we examined whether anomalies in CGI2 methylation could be found in SRS patients. The methylation of CGI2 in blood DNA from 24 SRS non-mUPD7 patients was tested by bisulfite treatment and PCR followed by digestion with informative restriction enzymes. Sequencing was used to confirm a limited number of samples. For all 24 SRS samples a normal imprinted methylation pattern of CGI2 was observed: all PCR products contained unmethylated and methylated molecules (Fig. 6). In parallel and by the same approach, we checked CGI2 methylation in normal fetuses in tissues with monoallelic and biallelic expression, and in sperm. GRB10 is expressed from the paternal allele in fetal brain, expression is biallelic in liver, and the {gamma}1 isoform is maternally expressed in muscle. CGI2 was found to be unmethylated in sperm, and all three tissues tested contained a similar ratio of methylated and unmethylated molecules, indicating that the CGI2 DMR is maintained in tissues (Fig. 6). This result indicates that the tissue-specific and isoform-specific imprinted expression of human GRB10 is not directly attributable to methylation changes at the CGI2 DMR.



View larger version (22K):
[in this window]
[in a new window]
  		Figure 6. Human CpG island 2 methylation in SRS patients and in tissues with biallelic or monoallelic expression of GRB10. PCR products obtained from bisulfite-treated DNAs of fetal tissues and of three SRS patients digested with TaiI. Following bisulfite treatment three TaiI sites (ACGT) are created in methylated molecules but not in unmethylated ones (ATGT). Digestion by TaiI into fragments of 330, 70, 65 and 20 bp indicates the DNA was methylated for the three sites; undigested PCR products represent molecules unmethylated at each of these sites. As a control, digested PCR products of the same region obtained from normal and mUPD7 lymphocyte, and pUPD7 lymphoblastoid DNAs are shown. The methylation state of these products has been determined by sequencing.

 

   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
 DISCUSSION
MATERIALS AND METHODS
 REFERENCES
 
To begin to account for the differences in imprinted expression of mouse Grb10 and human GRB10 we compared the genomic organization of the upstream regions of these genes and examined their allelic methylation patterns. Both genes have two conserved CpG islands. CGI1 gives rise to the maternally expressed transcripts in mouse, although it is unmethylated on both parental alleles (Fig. 7). CGI1 is also unmethylated in humans and most isoforms predicted to initiate from this region are expressed biallelically (22). Two additional mouse isoforms were shown to be expressed specifically in the brain and from the paternal allele (Fig. 7). These are associated with mCGI3, which shows tissue-specific variation in methylation, with lower methylation content on the paternal allele in brain than other tissues. CGI2 is a DMR conserved in both species, with a gametic imprint demonstrated in mouse (Fig. 7). This was unexpected, thus, the conserved methylation is not translated into similar patterns of monoallelic expression.



View larger version (12K):
[in this window]
[in a new window]
  		Figure 7. Summary of genetic and epigenetic features found in mouse Grb10 upstream region. Proposed start sites of the maternal (Mat: {alpha} and {delta}) and paternal (Pat: ß1 and ß2) transcripts are given. Methylation of the region is a compilation of bisulfite and Southern analysis: black symbols, methylated; grey symbols, partially methylated; white symbols, unmethylated. Hatched boxes represent direct repeat blocks.

 
Brain-specific imprint switching of mouse Grb10
An unexpected finding of this study was the paternal-specific expression in the mouse of the predicted alternative transcripts ß1 and ß2. The previously characterized maternally expressed variants {alpha} and {delta} initiate at Un1 in CGI1, whereas the predicted first exon of the ß1 and ß2 transcripts is Un2, contained in weak CpG island mCGI3. Recently, we have confirmed by 5'-RACE (unpublished results) that brain-specific paternal transcripts initiate in mCGI3, but also in the germline DMR (immediately upstream of and at the 3' end of CGI2). Therefore, paternal expression arises from different promoters than maternal expression. The unmethylated states of the paternal allele in the germline DMR and the fact that mCGI3 shows differential methylation with more prominence in brain are consistent with location of promoters activities. Promoter specific imprinting has been described for other imprinted genes, such as human IGF2 (38,39) and MEST (40). In the Gnas cluster, the maternally derived NESP55, paternally derived XL{alpha}S and predominantly biallelically derived GS{alpha} proteins are produced as a result of different promoter usage and alternative splicing of a single transcription unit (41,42). In human GRB10, monoallelic (brain and muscle) and biallelic transcripts are likely to originate from the same promoter, situated in CGI1, which is controlled in a tissue-specific manner (22). However, to our knowledge, mouse Grb10 is the first example of the use of oppositely imprinted promoter activities resulting in production of the same protein, albeit different isoforms.

In mouse brain, therefore, there is expression of the {alpha} and {delta} isoforms from the maternal allele and of the ß1 and ß2 transcripts from the paternal allele. Whether this occurs in the same cells or distinct cell populations still needs to be demonstrated. It will also be important to determine whether all isoforms have similar properties. The role of GRB10 in non-neural tissues is best surmised from the imprinted phenotypes associated with uniparental duplications of Chr 11 in the mouse: maternal duplication leads to prenatal growth restriction and paternal duplication enhanced growth. Grb10 is the most plausible candidate for these effects, an attractive explanation being that GRB10 acts as a negative regulator of the insulin and/or insulin-like growth factor type 1 receptors, although there are conflicting data on whether the adapter protein suppresses or promotes mitogenesis in cell culture assays in response to the respective ligands (23–26). The significance of the paternal-specific expression of Grb10 in brain—which occurs in both mouse and human—is not obvious, if maternally and paternally expressed forms couple to the same receptors. Alternatively, GRB10 mediates a distinct role in brain by coupling with a different class of receptor tyrosine kinases. Interestingly, mouse GRB10 forms a constitutive complex in vivo with the ubiquitin protein ligase (E3) Nedd4 (43). The Angelman syndrome gene UBE3A, which is monoallelically expressed in distinct brain regions, encodes the ubiquitin ligase (E6) (44,45).

CGI structure and direct repeats
CGI2 of mouse Grb10, the germline DMR, contains a block of CG-rich direct repeats. Many, but not all, imprinted genes contain or are closely linked to such direct repeats. These are generally found in or close to CpG islands and their presence may be conserved between mice and human, although there need be no sequence similarity. It is suggested that such repeats attract de novo methylation or help maintain methylation patterns in a sequence-independent manner (46). Such a role for direct repeats is supported by comparative analysis of genes whose imprinting differs between species (16,18). Deletion of the repeat block at the Rasgrf1 locus eliminates imprinted methylation and expression (17). However, deletion of direct repeats at the H19 and U2af1-rs1 loci did not prevent the establishment of normal methylation patterns (47–49). In Grb10, the presence of a direct repeat in CGI2 per se does not seem obviously necessary for the acquisition of imprinted methylation, as CGI2 is a DMR in both mouse and human, although the human CGI2 lacks such extensive direct repeat blocks. An alternative view of the presence of direct repeats in imprinted gene CpG islands is that they may be an evolutionary consequence of the inherent instability of methylated CpGs, owing to deamination of methyl-cytosine (50). Thus, the methylation of critical control elements in the germline presents a paradox. Direct repeats with CG-rich content might reflect expansion events that help to maintain a particular density of CpGs of the CGI. This might be the case if there is selective pressure for a threshold of methylation for the function of the CGI. In the case of Grb10, the selection may be to maintain the activity of the maternal allele to prevent overgrowth of the fetus.

Models for monoallelic expression of Grb10
Considering the different patterns of monoallelic/biallelic expression, the finding of a similar pattern of imprinted methylation in mouse Grb10 and human GRB10 was unexpected. Other than CGI1, CGI2 and mCGI3, there are no other CGIs in the region in either human or mouse. In addition, the identification of CGI2 as a germline DMR in mouse strongly suggests that it marks the ICR. The conserved methylation implies that the difference in imprinted expression is not based on acquisition of the imprint mark, but in differences in the reading of this mark in mouse and human. A variety of models could be proposed for monoallelic expression of Grb10, but any model must ultimately also be able to explain the species difference.

Methylation of CGI2 is permissive for expression of Grb10. This is confirmed by the analysis of cloned embryos produced from primordial germ cells which have erased imprints of the parental alleles (51). Grb10 falls into the category of imprinted genes for which lack of expression is the default state, such that acquisition of methylation in the female germline is necessary for expression of the maternal allele. Methylation must influence the Grb10 promoter by an indirect mechanism and, currently, there are two mechanisms for such monoallelic expression: antisense transcripts and the enhancer/boundary model.

The paradigm for the antisense model is provided by Air and Igf2r (52). We do not favour this model for Grb10, because we have not been able to demonstrate antisense transcripts (in EST database searches and RT–PCR experiments with primers designed to putative antisense transcripts revealed by NIX analysis), and also because the promoter-associated CGI1 remains unmethylated at all stages of development (methylation of the repressed allele seems to ensue in cases where antisense transcripts are implicated) (52). Instead, we favour a model in which maternal Grb10 expression is controlled by an enhancer/boundary mechanism, as exemplified by the reciprocal imprinting of Igf2 and H19 (7). The prediction of this model is that CGI2 in its unmethylated form recruits chromatin boundary proteins that interfere with productive interactions between the promoter at CGI1 and enhancers lying downstream of CGI2. The absence of monoallelic expression in most tissues in human could be attributed to the failure to recruit such a protein, whatever the methylation of CGI2. Alternatively, expression of human GRB10 could be controlled by newly recruited enhancers which lie upstream of CGI2 and are immune to the presumptive boundary. In the case of Igf2 and H19, the boundary is affected by the multifunctional DNA binding factor CTCF, which binds to the H19 DMR in a methylation-sensitive manner (53–56). Using the consensus described in (57) and (58) we found a number of putative CTCF binding sites in both mouse and human CGI1. At CGI2, three sites fitting a relaxed consensus were detected in the human and one in the mouse CGI2 (data not shown), and an additional site upstream of CGI2 in the human sequence. The in vivo function of these sites will need to be investigated to substantiate an enhancer/boundary model. A boundary could be created by unrecognized factors. Since the mouse and human CGI2s differ most obviously in the presence of the direct repeats, we should also consider the possibility that the repeat block could comprise a boundary. A second difference is the Hal1 LINE element found between CGI1 and CGI2 specifically in human GRB10. This element could also interfere with communication between a control element at CGI2 and the promoter at CGI1 (by an unknown mechanism), but such a role for a LINE has not yet been described at an imprinted locus. Finally, any model will need to be compatible with paternal expression of the ß1 and ß2 transcripts in the brain. mCGI3, which contains a promoter for these isoforms, is a DMR with maternal methylation which is established after implantation and is probably tissue-specific. This promoter, therefore, may be controlled directly by methylation in concert with brain-specific factors. If the enhancer/boundary model is correct, the enhancer(s) for this brain-specific promoter may need to be upstream of CGI2.

Methylation analysis in Silver–Russell syndrome patients
The genomic region surrounding GRB10 has been strongly associated with SRS owing to the previous reports of several cytogenetic abnormalities involving 7p11.2–p13 (59). Several duplications involving GRB10 have been reported, suggesting that overexpression of this gene maybe involved in the growth restriction associated with the phenotype (27,28,59). Having identified that CGI2 of human GRB10 is a maternally methylated DMR, and that the homologous mouse CGI2 is a germline DMR and candidate control element for Grb10 imprinting, we examined whether anomalies in CGI2 methylation could be found in non-mUPD7 SRS patients. Such imprinting mutations have been identified in patients with the Beckwith–Wiedemann, Prader–Willi and Angelman syndromes (60–62). In all 24 SRS patients analysed the methylation patterns were identical to normal control DNA and different from that of mUPD7, suggesting that there is no abberant imprinting. Collectively, the absence of epimutations and the previous reports of no coding mutations in 139 SRS patients strongly suggests GRB10 does not play a key role in the aetiology of this syndrome.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
 DISCUSSION
 MATERIALS AND METHODS
REFERENCES
 
Sequence analysis
The human GRB10 sequence was AC004920, being the full sequence of PAC RP5-898O18; the mouse Grb10 sequence was from AL663087 (BAC RP23-431M5). Two-way sequence alignment of the human and mouse Grb10/GRB10 5'UTR regions was performed using VISTA at www-gsd.lbl.gov/vista/. The structure of Grb10 was taken from the entry at Ensembl mouse genome server (www.ensembl.org/Mus musculus/), with additional information as noted in the text. Further sequence analyses were performed with the following programmes. CpG islands were identified using CpGplot at http://bioweb.pasteur.fr/seqanal/interfaces/cpgplot.html, and the Webgene CpG island prediction program at http://l25.itba.mi.cnr.it/genebin/wwwcpg.pl. Interpsersed repeat families were identified using RepeatMasker at http://repeatmasker.genome.washington.edu/, and direct repeats with Tandem Repeat Finder at http://tandem.biomath.mssm.edu/trf/trf.html (63).

Collection of human material
Fetal tissues were obtained from terminated pregnancies at Queen Charlotte's Hospital. Samples were washed in sterile PBS and snap frozen in liquid nitrogen. Local ethics approval for obtaining fetal samples was granted by the Research Ethics Committee of Hammersmith, Queen Charlotte's and Chelsea and Acton Hospitals Research Ethics Committee (2001/6028). Twenty-four SRS patients presenting with characteristic phenotype were included in an epimutation screen. These patients were a subset of 53 previously described (64). Major structural abnormalities, trisomy mosaicism and mUPD7 for Chr 7 had been ruled out. The mUPD7 and control DNA samples were derived from uncultured lymphocytes. The pUPD7 DNA sample was derived from cultured lymphoblast cells. The patients had undergone microsatellite repeat marker analysis to confirm uniparental disomy for entire Chr 7 (64,65).

Collection of mouse material
Mouse fetuses and adults with maternal and paternal duplication for proximal Chr 11, referred to as MatDp(prox11) and PatDp(prox11), respectively, were generated by the standard protocol of inter-crossing mice heterozygous for a reciprocal translocation T(7;11)40Ad and identified using appropriate genetic markers (66). The mice had inherited two maternal or two paternal copies of Chr 11 proximal to the breakpoint of T(7;11)40Ad in G-band 11D. Neonatal mice with maternal or paternal disomy for the entire Chr 11, MatDi(11) and PatDi(11) were generated by inter-crossing mice heterozygous for the Robertsonian translocation Rb(11.13)4Bnr and identified using the recessive visible marker vestigial tail (vt) (67). Snap-frozen whole 13.5 dpc embryos and tissue samples from neonatal mice for both maternal and paternal lines were used in methylation and expression studies.

Mouse oocytes were obtained from F1 (C57BL/6JxCBA/Ca)xF1 crosses, and morulae from a C57BL/6JxCBA/Ca cross. Oocytes were collected from superovulated immature females as described in Hogan et al. (68). Mature spermatozoa were isolated from caudal epididymis of adult F1 mice.

RNA isolation and RT–PCR analysis
Total RNA was isolated from homogenized tissues using the guanidinium isothiocyanate extraction technique. RNA was treated with amplification-grade DNaseI (Invitrogen) to degrade any genomic DNA present in the sample. First-strand cDNAs were synthesized from 1 µg denatured total RNA with Superscript RNaseH-reverse transcriptase (Invitrogen) using random hexamer primers. After a prior step at 25°C for 10 min and 42°C for 2 min for primer annealing, cDNA synthesis was performed at 42°C for 50 min. The reaction was terminated at 70°C for 15 min. Duplicate sets of samples were produced with RT omitted to detect amplification from contaminating DNA.

PCR reactions were performed as previously described. For cDNA templates PCR was performed within the log–linear phase of the reaction. Depending on the amplicon analysed the log–linear phase has been estimated to be between 32 and 35 cycles. The sequences of primers depicted in Figure 2 are as follows: A, 5'-GAGCACGAAGGTTCCGCGCA-3'; B, 5'-ATCGCCATCTACAGTTTCTG-3'; B', 5'-CAAGGTACA-GAGCTAGGACG-3'; C, 5'-CTGGTTGGCTTCTTTGTTGT-GG-3'; D, 5'-GGTGGACACTGGTTCTTAGGT-3'; E, 5'-CAC-AGTGACTTTTGTAAACCA-3'; F, 5'-TCTTTGTGAAGTCC-AATAAC-3'; G, 5'-ACAGGATCATCAAGCAACAA-3'. PCR amplification of mouse Hprt using forward primer 5'-AAGGACCTCTCGAAGTGTTG-3' and reverse primer 5'-GACGCAGCAACTGACATTTC-3' was performed for 32 cycles to check the integrity of the cDNAs. To check parental origin of RNA samples used, U2af1-rs1 was amplified using previously designed primers, with annealing at 60°C for 35 cycles (33).

DNA extraction
For fetal tissues (mouse and human), sperm, lymphocyte and lymphoblastoid cell line, DNA was extracted according to standard techniques. Samples of 500 ng to 1 µg DNA were used per bisulfite treatment. Oocytes (700) or morulae (15–40) were resuspended in 32.5 µl of a solution containing 10 µg glycogen, 1 mM SDS, and 280 µg/ml proteinase K, incubated for 90 min at 37°C, and then for 15 min at 95°C in a thermocycler. This lysate was used directly for bisulfite treatment.

Bisulfite treatment
DNA in a volume of 32.5 µl was denaturated by the addition of 1.1 µl NaOH 10 N for 15 min at 50°C. For bisulfite treatment, 200 µl of approximatively 4 M sodium bisulfite, pH 5.0 (Sigma; final concentration 3.5 M), 1.5 µl 75 mM hydroquinone (Sigma, final concentration 0.5 mM), 5 µg glycogen was added and the mixture incubated for 4 h at 55°C. Desalting was carried out using the QIAquick PCR purification kit (Qiagen), and the eluted DNA (in 50 µl Tris–Cl, pH 7.5) was desulfonated by treatment with 1.6 µl NaOH 10 N. DNA was ethanol-precipitated and resuspended in H2O.

PCR, digestion, cloning and sequencing
A nested primer strategy was used to amplify bisulfite-treated oocyte and morula DNAs. For this, 5 µl of resuspended DNA (~100 cells starting material equivalent) were used per PCR. PCR primer sequences are available on request. Amplification, cloning and sequencing were performed as previously described (69). Prior to cloning, the methylation pattern of the native DNA was first assessed by restriction digestion of the PCR products. Briefly, following bisulfite treatment new restriction sites are created, whereby some sites appear only if their recognition site corresponds to a methylated cytosine in the native DNA (therefore unconverted in the PCR product). For example, ACGC in the native sequence will become ATGT if the first cytosine is unmethylated or ACGT, if it is methylated. The latter results in the creation of a TaiI site. Such sites can be found frequently in the PCR products. Thus, digestion indicates the ‘methylation pattern’ of the population of molecules in PCR products, information that can be used to ensure that the population of sequences subsequently obtained is not substantially biased towards methylated or unmethylated molecules. Following sequencing, we estimated the overall conversion frequency at 98.6% (ranging from 94 to 100% according to the PCR product) on the basis of the number of unconverted non-CpG sites. A random distribution of residual unconverted non-CpG cytosines allowed us to conclude that most sequences with similar CpG methylation patterns were not clonal: for those with complete conversion of non-CpG cytosines, we had no evidence for clonality because of the diversity of methylation patterns in the given PCR products. For all DNAs except from oocytes, results were obtained from two independent experiments. In addition, extra bisulfite treatments were performed and methylation patterns assessed by restriction digestion (as described above). No discrepancies were found with the patterns obtained from previously sequenced products. For oocyte DNA the results shown were obtained from a single treatment performed on a batch of 700 oocytes. PCRs for products M4 and M5 were repeated twice from this batch.

Southern blot hybridization
Fifteen micrograms of genomic DNA were predigested with HindIII, then digested with methylation sensitive HpaII or its isoschizomer MspI before electrophoresis and Southern transfer to Hybond N+ membrane. Membranes were probed with either a 3.2 kb XmnI–HindIII (74 507–77 761 of AL663087) fragment containing Grb10 CpG island 1 (probe A), a 1.8 kb PstI–HindIII (64 851–66 661 of AL663087) fragment containing CpG island 2, or a 4.5 kb PCR product (67 661–72 240 of AL663087) from between the two CpG islands (probe B: primers 5'-GTAGACCTCCGATTCCGAAG-3' and 5'-GCTATAGCCCTAGATGTTAACC-3'). Both restriction fragment probes were derived from BAC 431M5 and cloned into pGEM3Z vector (Promega). Filters were prehybridized (0.2 M NaPO4 pH 7.2, 1 mM EDTA, 1% BSA, 7% SDS, 15% deionized formamide) for a minimum of 2 h/65°C, and hybridized with radiolabelled probe overnight. Filters were washed three times (40 mM NaPO4, 1 mM EDTA, 1% SDS) for 10 min/65°C. Blots were exposed to X-ray film with intensifying screens at -70°C.


   ACKNOWLEDGEMENTS
 
All animal studies were carried out under the guidance issued by the Medical Research Council (MRC) in The Use of Animals for Medical Research (July 1993) and Home Office Project Licence No. 30/1518.

We gratefully acknowledge Dr J. Richard Chaillet (Rangos Research Center, Pittsburgh, PA, USA) for helpful advice in the bisulfite technique for germ cells. We thank Rachel Smith, Antonius Plagge and Wolf Reik for valuable comments on the manuscript. This work was supported by the MRC and the Dunhill Medical Trust (support to M.H.). P.A. holds a Marie Curie Individual Fellowship from the European Community Programme in Human Potential (under contract number HPMF-CT-2001-01122), and G.K. is a Senior Fellow of the MRC.

This paper is dedicated to the memory of Maurice Arnaud.


   FOOTNOTES
 
* To whom correspondence should be addressed. Tel: +44 1223496332; Fax: +44 1223496022; Email: gavin.kelsey{at}bbsrc.ac.uk Back

{dagger} Present address: Molecular Embryology Unit, Institute of Child Health, University College London, 30 Guilford Street, London WC1N 1EH, UK. Back


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
 DISCUSSION
 MATERIALS AND METHODS
 REFERENCES
 

  1. Surani, M.A. (1998) Imprinting and the initiation of gene silencing in the germ line. Cell, 93, 309–312.[CrossRef][ISI][Medline]

  2. Tilghman, S.M. (1999) The sins of the fathers and mothers: genomic imprinting in mammalian development. Cell, 96, 185–193.[CrossRef][ISI][Medline]

  3. Tycko, B. and Morison, I.M. (2002) Physiological functions of imprinted genes. J. Cell Physiol., 192, 245–258.[CrossRef][ISI][Medline]

  4. Morison, I.M. and Reeve, A.E. (1998) A catalogue of imprinted genes and parent-of-origin effects in humans and animals. Hum. Mol. Genet., 7, 1599–1609.[Abstract/Free Full Text]

  5. Li, E., Beard, C. and Jaenisch, R. (1993) Role for DNA methylation in genomic imprinting. Nature, 366, 362–365.[CrossRef][Medline]

  6. Constancia, M., Pickard, B., Kelsey, G. and Reik, W. (1998) Imprinting mechanisms. Genome Res., 8, 881–900.[Abstract/Free Full Text]

  7. Thorvaldsen, J.L., Duran, K.L. and Bartolomei, M.S. (1998) Deletion of the H19 differentially methylated domain results in loss of imprinted expression of H19 and Igf2. Genes Dev., 12, 3693–3702.[Abstract/Free Full Text]

  8. Wutz, A., Theussl, H.C., Dausman, J., Jaenisch, R., Barlow, D.P. and Wagner, E.F. (2001) Non-imprinted Igf2r expression decreases growth and rescues the Tme mutation in mice. Development, 128, 1881–1887.[Abstract]

  9. Bourc'his, D., Xu, G.L., Lin, C.S., Bollman, B. and Bestor, T.H. (2001) DNMT3L and the establishment of maternal genomic imprints. Science, 294, 2536–2539.[Abstract/Free Full Text]

  10. Hata, K., Okano, M., Lei, H. and Li, E. (2002) DNMT3L cooperates with the DNMT3 family of de novo DNA methyltransferases to establish maternal imprints in mice. Development, 129, 1983–1993.

  11. Reik, W. and Walter, J. (2001) Genomic imprinting: parental influence on the genome. Nat. Rev. Genet., 2, 21–32.[ISI][Medline]

  12. Liu, J., Yu, S., Litman, D., Chen, W. and Weinstein, L.S. (2000) Identification of a methylation imprint mark within the mouse Gnas locus. Mol. Cell Biol., 20, 5808–5817.[Abstract/Free Full Text]

  13. Feil, R., Walter, J., Allen, N.D. and Reik, W. (1994) Developmental control of allelic methylation in the imprinted mouse Igf2 and H19genes. Development, 120, 2933–2943.[Abstract]

  14. Constancia, M., Dean, W., Lopes, S., Moore, T., Kelsey, G. and Reik, W. (2000) Deletion of a silencer element in Igf2 results in loss of imprinting independent of H19. Nat. Genet., 26, 203–206.[CrossRef][ISI][Medline]

  15. Takada, S., Paulsen, M., Tevendale, M., Tsai, C.E., Kelsey, G., Cattanach, B.M. and Ferguson-Smith, A.C. (2002) Epigenetic analysis of the Dlk1-Gtl2 imprinted domain on mouse chromosome 12: implications for imprinting control from comparison with Igf2-H19. Hum. Mol. Genet., 11, 77–86.[Abstract/Free Full Text]

  16. Pearsall, R.S., Plass, C., Romano, M.A., Garrick, M.D., Shibata, H., Hayashizaki, Y. and Held, W.A. (1999) A direct repeat sequence at the Rasgrf1 locus and imprinted expression. Genomics, 55, 194–201.[CrossRef][ISI][Medline]

  17. Yoon, B.J., Herman, H., Sikora, A., Smith, L.T., Plass, C. and Soloway, P.D. (2002) Regulation of DNA methylation of Rasgrf1. Nat. Genet., 30, 92–96.[CrossRef][ISI][Medline]

  18. Okamura, K., Hagiwara-Takeuchi, Y., Li, T., Vu, T.H., Hirai, M., Hattori, M., Sakaki, Y., Hoffman, A.R. and Ito, T. (2000) Comparative genome analysis of the mouse imprinted gene Impact and its nonimprinted human homolog IMPACT: toward the structural basis for species-specific imprinting. Genome Res., 10, 1878–1889.[Abstract/Free Full Text]

  19. Smrzka, O.W., Fae, I., Stoger, R., Kurzbauer, R., Fischer, G.F., Henn, T., Weith, A. and Barlow, D.P. (1995) Conservation of a maternal-specific methylation signal at the human IGF2R locus. Hum. Mol. Genet., 4, 1945–1952.[Abstract/Free Full Text]

  20. Killian, J.K., Nolan, C.M., Wylie, A.A., Li, T., Vu, T.H., Hoffman, A.R. and Jirtle, R.L. (2001) Divergent evolution in M6P/IGF2R imprinting from the Jurassic to the Quaternary. Hum. Mol. Genet., 10, 1721–1728.[Abstract/Free Full Text]

  21. Miyoshi, N., Kuroiwa, Y., Kohda, T., Shitara, H., Yonekawa, H., Kawabe, T., Hasegawa, H., Barton, S.C., Surani, M.A., Kaneko-Ishino, T. and Ishino, F. (1998) Identification of the Meg1/Grb10 imprinted gene on mouse proximal chromosome 11, a candidate for the Silver–Russell syndrome gene. Proc. Natl Acad. Sci. USA, 95, 1102–1107.[Abstract/Free Full Text]

  22. Blagitko, N., Mergenthaler, S., Schulz, U., Wollmann, H.A., Craigen, W., Eggermann, T., Ropers, H.H. and Kalscheuer, V.M. (2000) Human GRB10 is imprinted and expressed from the paternal and maternal allele in a highly tissue- and isoform-specific fashion. Hum. Mol. Genet., 9, 1587–1595.[Abstract/Free Full Text]

  23. Morrione, A., Valentinis, B., Resnicoff, M., Xu, S. and Baserga, R. (1997) The role of mGrb10 alpha in insulin-like growth factor I-mediated growth. J. Biol. Chem., 272, 26382–26387.[Abstract/Free Full Text]

  24. Stein, E.G., Gustafson, T.A. and Hubbard, S.R. (2001) The BPS domain of GRB10 inhibits the catalytic activity of the insulin and IGF1receptors. FEBS Lett., 493, 106–111.[CrossRef][ISI][Medline]

  25. Wang, J., Dai, H., Yousaf, N., Moussaif, M., Deng, Y., Boufelliga, A., Swamy, O.R., Leone, M.E. and Riedel, H. (1999) GRB10, a positive, stimulatory signaling adapter in platelet-derived growth factor BB-, insulin-like growth factor I-, and insulin-mediated mitogenesis. Mol. Cell Biol., 19, 6217–6228.[Abstract/Free Full Text]

  26. O'Neill, T.J., Rose, D.W., Pillay, T.S., Hotta, K., Olefsky, J.M. and Gustafson, T.A. (1996) Interaction of a GRB-IR splice variant (a human GRB10 homolog) with the insulin and insulin-like growth factor I receptors. Evidence for a role in mitogenic signaling. J. Biol. Chem., 271, 22506–22513.[Abstract/Free Full Text]

  27. Monk, D., Wakeling, E.L., Proud, V., Hitchins, M., Abu-Amero, S.N., Stanier, P., Preece, M.A. and Moore, G.E. (2000) Duplication of 7p11.2–p13, including GRB10, in Silver–Russell syndrome. Am. J. Hum. Genet., 66, 36–46.[CrossRef][ISI][Medline]

  28. Joyce, C.A., Sharp, A., Walker, J.M., Bullman, H. and Temple, I.K. (1999) Duplication of 7p12.1–p13, including GRB10 and IGFBP1, in a mother and daughter with features of Silver–Russell syndrome. Hum. Genet., 105, 273–280.[CrossRef][ISI][Medline]

  29. Hitchins, M.P., Stanier, P., Preece, M.A. and Moore, G.E. (2001) Silver–Russell syndrome: a dissection of the genetic aetiology and candidate chromosomal regions. J. Med. Genet., 38, 810–819.[Abstract/Free Full Text]

  30. McCann, J.A., Zheng, H., Islam, A., Goodyer, C.G., and Polychronakos, C. (2001) Evidence against GRB10 as the gene responsible for Silver–Russell syndrome. Biochem. Biophys. Res. Commun., 286, 943–948.[CrossRef][ISI][Medline]

  31. Wakeling, E.L., Abu-Amero, S.N., Stanier, P., Preece, M.A. and Moore, G.E. (1998) Human EGFR, a candidate gene for the Silver–Russell syndrome, is biallelically expressed in a wide range of fetal tissues. Eur. J. Hum. Genet., 6, 158–164.[CrossRef][ISI][Medline]

  32. Wakeling, E.L., Hitchins, M.P., Abu-Amero, S.N., Stanier, P., Moore, G.E. and Preece, M.A. (2000) Biallelic expression of IGFBP1 and IGFBP3, two candidate genes for the Silver–Russell syndrome. J. Med. Genet., 37, 65–67.[Free Full Text]

  33. Hitchins, M.P., Bentley, L., Monk, D., Beechey, C., Peters, J., Kelsey, G., Ishino, F., Preece, M.A., Stanier, P. and Moore, G.E. (2002) DDC and COBL, flanking the imprinted GRB10 gene on 7p12, are biallelically expressed. Mamm. Genome, 13, 686–691.[CrossRef][ISI][Medline]

  34. Daly, R.J. (1998) The GRB7 family of signalling proteins. Cell Signal., 10, 613–618.[CrossRef][ISI]