hiaorporating Mouse Genome
Mamma1ian
Genome
Allele-specific methylation at the promoter-associated CpG island of mouse Copg2 Jihye Yun, Chang Won Park, Young Jae Lee, Jae Hoon Chung Department of Biological Sciences, Korea Advanced Institute of Science and Technology, Taejon 305-701, South Korea Received: 15 November 2002 / Accepted: 14 February 2003
Abstract
We previously reported that the mouse Copg2 (Coatomer Protein Subunit Gamma 2) gene was imprinted in the intraspecific F l hybrid mice between C57BL/6 and M. m. molossinus. In this study, methylation status at the promoter-associated CpG island and allele-specific expression pattern of Copg2 were investigated in the interspecific F 1 hybrid (C57BL/6 x M. spretus). We found that Copg2 was biallelically expressed in the interspecific F 1 hybrid, despite the presence of a differentially methylated region (DMR) in the CpG island. Therefore, paternalspecific methylation at the CpG island of Copg2 is not accompanied by allele-specific expression of Copg2 in the F 1 hybrid (C57BL /6 x M. spretus). The results suggest that the imprinted pattern of Copg2 expression is subject to variable epigenetic control systems of different mouse species. In addition, we identified a novel antisense transcript, Copg2AS2, at the Copg2 promoter region. The antisense transcript was expressed monoallelically, indicating that the differential expression of the antisense transcript is correlated with the differential methylation at the CpG island of Copg2.
Genomic imprinting is an epigenetic modification on chromosomal regions, which leads to the preferential expression of a specific parental allele. Though the precise mechanism of imprinting remains elusive, the biased expression of a parental allele usually accompanies the differential methylation at CpG
The nucleotide sequence data reported in this paper have been submitted to GenBank and have been assigned the accession number AF402597. Correspondence to: J.H. Chung ; Email:
[email protected] 376
islands with differential chromation conformation (Hark and Tilghman 1998; Szabo et al. 2002). Tandem repeat elements (Szebenyi and Rotwein 1994), antisense transcripts (Wutz et al. 1997), trans-acting DNA binding proteins (Hark et al. 2000 ; Bell and Felsenfeild 2000), and regional clustering (Ainscough et al. 1998; Feinberg 1999) also constitute prominent features common to the imprinted genes. The characteristic features prevalent among imprinted genes could be exploited to trace novel imprinted genes. By using the clustering behavior of imprinted genes, two mouse genes, Mitt/Lb9 and Copg2, linked to Peg1/hest on mouse Chromosome (Chr) 6, were identified to be imprinted maternally and paternally, respectively, in the intraspecific F 1 hybrid between C57BL /6 and M. m. molossinus (Lee et al. 2000). In contrast, human COPG2 on Chr 7q32 was reported to exhibit biallelic expression pattern in the examined tissues (Yamasaki et al. 2000). The allele-specific expression of imprinted genes is often associated with allele-specific methylation at the CpG island (Li et al. 1993). In this study, we determined a differentially methylated region (DMR) within the CpG island located at the promoter region of Copg2, using the interspecific F, hybrid (C57BL/6 x M. spretus) rather than the intraspecific F 1 hybrid (C57BL/6 x M. m. molossinus) used in the previous study. The pattern of methylation at the promoter-associated CpG island of Copg2 was analyzed by using a sodium bisulfite sequencing method. We further examined the allelic expression of Copg2 in the interspecific F l hybrid. In addition, we found a novel antisense transcript expressed monoallelically at the promoter region of Copg2.
Materials and methods
The sequence of the Copg2 promoter has been submitted to the GenBank database under the accession number AF402597.
DOI: 10.1007/s00335-002-2252-X • Volume 14, 376-382 (2003) • O Springer-Verlag New York, Inc. 2003
J. YUN ET AL: ALLELE-SPECIFIC METHYLATION OF MOUSE Copg2
Mice. F I hybrid mice (C57BL/6 female x M. spretus male) were purchased from The Jackson Laboratory (Bar Harbor, Me.). Primer extension analysis. was performed with total RNAs of mouse brain and testis. Primer Cg-R31 (5'-CAGACTCCTCGTCCTTCTTGTCG-3') was 5'labeled with [y- 32 P]ATP (3000 Ci/mmol) by T4 polynucleotide kinase (Sambrook and Russell 2001). The labeled primer was added to 5 µg of total RNA, heated at 90°C for 10 min, and chilled on ice. The primer was annealed in 30 µL of hybridization buffer (12 mM Tris-HC1 pH 7.0, 0.56 M NaCl, 80% formamide) at 30°C for 12 h. The RNA/primer hybrid was ethanol-precipitated and dissolved in 25 µL of reverse transcription buffer containing 30 units of RNasin (Promega, Madison, Wi.) and 200 units of Superscript II Reverse Transcriptase (Gibco BRL, Gaithersburg, Md.). Primer extension was carried out at 42°C for 90 min. RNA was degraded by incubation with RNaseA (40 gg/mL) for 30 min. The extension products were then purified by phenolchloroform extraction prior to electrophoresis on an 8% sequencing polyacrylamide gel. Bisulfite treatment. Bisulfite treatment was performed as previously described (Clark et al. 1994) with several modifications. Five microgram of genomic DNA isolated from brains of F 1 hybrid (C57BL/6 female x M. spretus male) was denatured with 2 µL of 3 M NaOH and mixed with 200 µL of freshly prepared 3.5 M NaHSO 3 and 10 mivt hydroquinone (pH 5.0). Reaction mixtures were incubated at 55°C for 16 h in darkness. The modified DNA was recovered by PCR purification columns (Qiagen) and treated with NaOH (0.3 M final) at 40°C for 30 min. For the complete conversion of C to U, bisulfite treatments were repeated twice with the same procedure.
377
C57BL/6 allele from the M. spretus allele by creating a restriction fragment length polymorphism (RFLP) with mismatched PCR (Jeong et al. 2000), exploiting a G/C polymorphism at the CpG island. Plasmids containing the bisulfite PCR products were sequenced by the Bigdye Terminator Cycle Sequencing Kit (Perkin-Elmer, ABI) and an automatic sequencer (Model 373A, Perkin-Elmer ABI). Allelic expression of Copg2 sense and antisense RNAs. A single nucleotide polymorphism located
in exon 19 of mouse Copg2 was found between C57BL/6 and M. spretus, creating an Alul site (AGCT) only in the C57BL/6 allele. Reverse transcription (RT) was performed with Superscript II Reverse Transcriptase (Gibco BRL), using strandspecific primer G2-R18 (5'-CTTCATCTGGA ACTCCAGTATTAG-3'). The RT product of the Copg2 transcript was amplified by PCR with primers, G2-F5 (5'-AAATTACACTTGTGACTCCT-3') and G2-R5 (5'-GCACAGGCTCAGAAGACTTAA AC-3'). PCR amplification was performed for 33 cycles (94°C for 30 s, 61°C for 30 s, and 72°C for 30 s). The PCR products were digested with AluI and analyzed on a 5% polyacrylamide gel. For the determination of the allelic origin of Copg2 antisense RNA, AvaII RFLP between C57BL/6 and M. spretus was used. RT was performed with strand-specific primer CuR2 (5'-TGACATTTTCAAGCTCCTAGAAG-3') located at 490 bp upstream of Copg2 translation start codon. The RT product of Copg2 antisense transcript was amplified by PCR with primers Cg-F36 (5'-CCCACTCCTCACTCTCCA AG-3') and Cg-R31 (5'-CAGACTCCTCGTCCTTC TTGTCG-3'). PCR amplification was performed for 33 cycles (94°C for 30 s, 58°C for 30 s, and 72°C for 30 s). The PCR products were digested with Avail and electrophoresed on a 5% polyacrylamide gel.
Methylation analysis by bisulfite sequencing.
Bisulfite-treated DNAs were amplified by hot-start PCR. Primer sequences for the amplification of a 645-bp fragment at the CpG island of Copg2 were 5'GGTTTGTAGAGTTTGTGAAG-3' (BS-F 1 , forward) and 5'-CCCTAAAACCTAACCC- CCC-3' (BS-R4, reverse). The primer set encompassed all 78 CpG dinucleotides within the CpG island. PCR amplification was carried out in 50 pL of 10 mivz Tris-HC1 (pH 8.3), 40 mlvi KCl, 1.5 mivt MgC1 2, 1 maul DTT, 0.5 µg/mL acetylated BSA, and 200 µM of each dNTP with 50 pmol of each primer and 2.5 units of Taq DNA polymerase for 40 cycles consisting of 30 s at 94°C, 30 s at 60°C, and 30 s at 72°C. PCR products were cloned into the pGEM-T Easy Vector (Promega). Before sequencing, we differentiated the
Results
Determination of the Copg2 transcription start site. To identify the transcription start site of mouse
Copg2, we performed the primer extension analysis. An antisense primer, Cg-R31 (see Materials and methods), was derived from nucleotides +15 to +39 relative to the translation start site of mouse Copg2. The analysis revealed that the transcription start site is located 61 bp upstream of the start codon (Fig. 1). Three minor bands were also visible, indicating putative alternative transcription start sites. The region neighboring —61 position of the major transcription start site shows sequence similarity to the putative 5'-end of the human COPG2 cDNA (Blagitko et al.
378
J. YUN ET AL: ALLELE-SPECIFIC METHYLATION OF MOUSE
Copg2
samples, which would represent no amplification bias. Each of the 14 individual clones was analyzed by automated sequencing. The results are summarized in Fig. 2B. With minor variations of the methylation pattern in the region encompassing 300 bp of the CpG island, the methylation level was observed to be higher in the paternal M. spretus allele than in the maternal C57BL/6 allele. The rest of the CpG dinucleotides, completely unmethylated in both parental alleles, were not shown by circles. Therefore, it was concluded that the parental alleles of Copg2 are differentially methylated at the promoter-associated CpG island of Copg2 in the interspecific F l hybrid (C57BL/6 x M. spretus).
Fig. 1. Identification of the transcription start sites of mouse Copg2. The 5'-end of Copg2 mRNA was determined by primer extension analysis with primer Cg-R31. The lanes marked as G, A, T, and C show the dideoxy sequencing ladders produced with the same primer. Translation starts at the nucleotide position of +1. A major band and three minor bands are indicated by an arrow and asterisks, respectively. Nucleotide sequences around the position of the bands are shown on the left. Boxed residue represents the major transcription start site.
1999). The Copg2 promoter sequence was analyzed in order to understand the structure of Copg2 promoter (GenBank Accession No. AF402597). Three GC-box sequences and a CREB binding site were found in the 310-bp region immediately upstream of the translation start codon. The promoter contains neither TATA box nor CAAT box (data not shown). Methylation analysis of the CpG island at the mouse Copg2 promoter region. A CpG island is located downstream of Vmel gene (Jeong et al.
2000) adjacent to Copg2 (Fig. 2A). The CpG island encompasses the promoter, exon 1, and a part of intron 1 of Copg2. We determined methylation status of the CpG island in each parental allele, to investigate a correlation between the CpG methylation and the allelic expression (Li et at 1993). Bisulfite sequencing was performed with DNA isolated from brains of the F 1 hybrids (C57BL/ 6 x M. spretus). The sequence polymorphism at position —138 relative to the transcription start site provided a physical marker by which we could distinguish the C57BL/6 allele from the M. spretus allele. Specifically, the G in C57BL/6 allele was substituted with the C in the M. spretus allele. All of the 78 CpG dinucleotides at the CpG island were examined with three independent bisulfite-treated
Allelic expression of Copg2 in interspecific hybrids. To examine the allelic expression of
mouse Copg2 in the interspecific F I hybrid (C57BL/ 6 x M. spretus), we exploited an AluI RFLP in exon 19 between C57BL/6 (AGCT) and M. spretus (GGCT). The AIul site is present only in the C57BL/ 6 allele but not in the M. spretus allele (Fig. 3A). Strand-specific RT products were amplified by PCR with primers G2-F5 and G2-R5 in exons 18 and 19, respectively. These primers could eliminate any possible amplification of contaminating genomic DNA because the genomic product would include intron 18. After digestion with AluI, PCR products derived from C57BL/6 allele resulted in 71- and 58bp fragments, while PCR products from the M. spretus allele remained intact as an 129-bp fragment. As shown in Fig. 3B, both parental alleles were observed to be expressed in brain, liver, and kidney. The hiallelic expression of Copg2 in the brain of interspecific F I hybrid (C57BL/6 x M. spretus) contrasts with the imprinted pattern of Copg2 in the brain of intraspecific F 1 hybrid mice between C57BL/6 and M. m. molossinus (Lee et al. 2000). This discrepancy suggests that paternal epigenetic control system of the M. spretus abolish the imprinted pattern of Copg2 observed among M. musculus mice. A novel antisense transcript encompassing the CpG island of Copg2 and its allelic expression. To
search for possible antisense transcript(s) in the vicinity of the DMR, we used a strand-specific RT primer (CuR2) for reverse transcription (Fig. 4A). Subsequent PCR was carried out with Cg-F36 and Cg-R31 primers, with the RT products as templates. PCR products were observed in all of the tested tissues, indicating that an antisense RNA was ubiquitously transcribed from the CpG island of mouse Copg2 (Fig. 4B). To exclude the possibility of genomic contamination, we performed independent experiments several times with an RT-negative control
J. YUN ET AL: ALLELE-SPECIFIC METHYLATION OF MOUSE
Copg2
379
Fig. 2. Methylation profile of all CpG dinucleotides in a 620-bp region of the CpG island. A. The CpG dinucleotide at position —197 relative to the transcription start site of Copg2 was regarded as a start site for the CpG island. Promoter region, exon 1 of mouse Copg2, and poly (A) signal of Vmel are indicated. Rectangular arrow represents the transcription start site of Copg2. A vertical bar with G/C represents the position of polymorphism between C57BL/6 and M. spretus. A dotted line indicates the CpG island that includes a total of 78 CpGs. B. The methylation status of the CpG island of Copg2 in brain tissues of F I hybrid (C57BL/6 x M. spretus) was analyzed by sequencing of PCR-amplified products from bisulfite-modified genomic DNA. Each row of circles represents a separate clone. Methylated CpGs are designated by filled circles, and unmethylated CpGs are designated by open circles. A rectangular arrow indicates the transcription start site. The rest of the CpGs not shown by circles were completely unmethylated in both parental alleles. The data are derived from pools of three independent experiments in which different samples were used to prepare DNA for bisulfite treatment and amplification of the region.
(data not shown). We designated this novel antisense transcript as Copg2AS2 to discriminate it from the Copg2AS reported previously (Lee et al. 2000). An AvaII RFLP between the C57BL/6 allele and the
M. spretus allele was found and used to determine the allelic expression of Copg2AS2 in the F I hybrid. As shown in Fig. 4B, Copg2AS2 was transcribed
exclusively from the maternal allele (C57BL/6). This
Fig. 3. Location of specific primers and biallelic expression of Copg2 transcript. A. Strand-specific reverse transcription (RT) was performed with primer G2-R18. The subsequent RT product of Copg2 transcript was amplified by PCR with G2-F5 and G2-R5 primers and digested with Alul. The polymorphic Alul site was marked by an asterisk. Expected DNA fragments, after AluI digestion, were depicted below the map. B. Strand-specific RT-PCR and Alul digestion were carried out with RNA derived from brain, liver, and kidney of 2-month-old F I hybrids (C57BL/6 x M. spretus, C x S). The identical analysis was applied to the RNAs isolated from inbred M. spretus and C57BL/6 mice for the control. The allelic origin of the digested products is indicated on the right. M represents DNA markers.
380
J. YUN ET AL: ALLELE-SPECIFIC METHYLATION OF MOUSE
Copg2
Fig. 4. A novel antisense transcript encompassing the CpG island of Copg2 and its monoallelic expression in mouse tissues of F I hybrids (C57BL /6 x M. spretus). A. Positions of specific primers that were used for the investigation of allelic expression at the Copg2 locus. Strand-specific RT was performed by using the primer CuR2. The RT products from each tissue were amplified by PCR with primers Cg-F36 and Cg-R31 and digested with AvaII. The polymorphic AvaII site was marked by an asterisk (G/C). The CpG island and differentially methylated region (DMR) of Copg2 were indicated by a dotted line and a bar, respectively. Expected DNA fragments, after AvaII digestion, were represented below the map. B. All antisense transcripts from brain, liver, and kidney showed monoallelic expression of Copg2AS2 in the F I hybrid mice (C57BL /6 x M. spretus, C x S) after digestion with AvaII. Genomic DNAs (gDNA) from the F I hybrid were amplified and digested with AvaII as positive controls. The allelic origin of the digested DNA fragments was indicated on the right (C, C57BL /6 allele S, M. spretus allele). ;
result suggests that the observed paternal-specific methylation of the CpG island correlates with the exclusive maternal expression of the Copg2AS2 from this region.
Discussion
Our previous study with intraspecific F I hybrid mice between C57BL/6 and M. m. molossinus as well as the presence of a CpG island at the promoter region of Copg2 suggested a candidate region for DMR, which controls the allele-specific expression of Copg2. However, we could not determine the DMR in the intraspecific F I hybrid because there was no informative polymorphism at the CpG island between the two strains. In this study, we demonstrated that the DMR is located within the CpG island, including the exon 1 of Copg2, with the interspecific F I hybrid (C57BL/6 x M. spretus). The availability of an informative polymorphism, by which we could physically discriminate between parental alleles of the interspecific F I hybrid enabled us to define the DMR through bisulfite sequencing. We showed the biallelic expression of Copg2 in the interspecific F I hybrid, although our previous
work demonstrated that Copg2 is paternally imprinted in the intraspecific F I hybrid between C57BL/6 and M. m. molossinus (Lee et al. 2000). Using bisulfite sequencing with DNA isolated from brains of the interspecific F 1 hybrid, we found that the promoter-associated CpG island of Copg2 exhibits preferential paternal methylation. Furthermore, we identified a novel antisense transcript, Copg2AS2, at the Copg2 promoter region of the interspecific F I hybrid. The antisense transcript was expressed exclusively from the maternal allele. This suggests that the preferential paternal methylation at the CpG island may correlate with the maternal expression of the Copg2AS2 from this region. The distinct expression patterns of Copg2 allele in two different hybrids may be attributable to modified epigenetic environments dictated by different genetic backgrounds. An analogous phenomenon is exemplified by human COPG2, which was recently discovered to be expressed biallelically throughout the fetal and extraembryonic tissues (Yamasaki et al. 2000). It is speculative that the allele-specific expression pattern of Copg2, which is found in interspecific hybrid mice between C57BL/6 and M. spretus, resembles that of human COPG2
J. YUN ET AL: ALLELE-SPECIFIC METHYLATION OF MOUSE
Copg2
more closely than that of Copg2 observed in the intraspecific hybrid mice between C57BL/6 and M. m. molossinus. If this is the case, the void epigenetic compensation between two parental species could provide a clue for elucidating the functional components responsible for the difference between their epigenetic systems. There seems to be a tendency for the epigenetic control of COPG2 to evolve toward a more relaxed state as the functional complexity of an organism increases. An accidental failure of maintaining the imprinted pattern could happen in crosses between animals with distant epigenetic backgrounds. Epigenetic controls of Copg2 in varying situations imply that the imprinting of Copg2 is dispensable at least for the survival of the progeny. Putative beneficial effects of relaxed imprinting in Copg2 on the organism's survival or on its reproductive success could have promoted a transition to the biallelic expression of the gene with dispensable epigenetic constraints, which is required for a high degree of functional complexity. In Northern hybridization analysis with a riboprobe to detect the Copg2AS2 mRNA, we detected a large diffused signal in every tissue examined (data not shown). The diffused signal is likely to represent either unusually large RNAs degraded during a preparation procedure or a collection of RNAs having heterogeneous sizes caused by varying 5'- and/or 3'-ends. Our previous report described that Copg2AS spans 3'-UTR of Peg1/Mest and 3'-UTR of Copg2 Lee et al. 2000). However, Copg2AS does not seem to be the intrinsic antisense transcript of Copg2 since it might represent additional 3'-UTR of Peg1/ Mest like human PEG1/MEST encoding an additional 2.7 kb of 3'-UTR overlapping with at least four exons and introns of COPG2 (Yamasaki et al. 2000). It is notable that a novel Copg2AS2 is independent of Copg2AS since allelic origin of each transcript is opposite to the other. Copg2AS is subject to maternal imprinting (paternal expression), while Copg2AS2 is maternally expressed. In addition, we could not detect any antisense transcript in the intron 1 region that is approximately 1 kb apart from the exon 1. The presence of antisense transcripts is common to most imprinted regions (Li et al. 2000). It has been proposed that there is an "expression competition" between antisense and sense transcripts, probably via simple transcription exclusion or competition for transcription factors (Reik and Constancia 1997). Copg2 sense RNA is expressed from both parental alleles, while Copg2AS2 is expressed exclusively from the maternal allele in our allelic expression study. Therefore, the genomic imprinting pattern of
381
Copg2 sense and antisense RNAs does not support the expression-competition model. Alternatively, this novel antisense transcript, Copg2AS2, might be related to the Vmel located upstream of Copg2. Vmel exhibited a variegated monoallelic expression with different patterns among individuals, suggesting that a complex epigenetic mechanism operates in this locus (Jeong et al. 2000). In this case, monoallelic expression of Copg2AS2 could be related to the monoallelic expression pattern of Vmel via an "expression competition" or other mechanisms. It would be interesting to see whether the expression pattern of Copg2AS2 is correlated with that of Vmel among individuals, tissues, or at various developmental stages in mouse. Acknowledgments
This work was supported by grant No. R01-199900131 from the Basic Research Grant Program of the Korea Science and Engineering Foundation. References 1. Ainscough JF, John RM, Surani MA (1998) Mechanism of imprinting on mouse distal chromosome 7. Genet Res 72, 237-245 2. Bell AC, Felsenfeild G (2000) Methylation of a CTCFdependent boundary controls imprinted expression of the Igf2 gene. Nature 405, 482-485 3. Blagitko N, Schulz U, Schinzel AA, Ropers H, Kalscheuer NM (1999) y2-COP, a novel imprinted gene on chromosome 7q32, defines a new imprinting cluster in the human genome. Hum Mol Genet 8, 23872396 4. Clark SJ, Harrison J, Paul CL, Frommer M (1994) High sensitivity mapping of methylated cytosines. Nucleic Acids Res 22, 2990-2997 5. Feinberg AP (1999) Imprinting of a genomic domain of llpl5 and loss of imprinting in cancer: an introduction. Cancer Res 59, 1743-1746 6. Hark AT, Tilghman SM (1998) Chromatin conformation of the H19 epigenetic mark. Hum Mol Genet 7, 1979-1985 7. Hark AT, Schoenherr CJ, Katz DJ, Ingram RS, Levorse JM et al. (2000) CTCF mediated methylation-sensitive enhancer-blocking activity at the H19/Igf2 locus. Nature 405, 486-489 8. Jeong S, Lee YJ, Jang JS, Park CW, Chung JH et al. (2000) A novel epigenetic control operating on Vme1+ locus leads to variegated monoallelic expression. Biochem Biophys Res Commun 279, 884-890 9. Lee YJ, Park CW, Hahn Y, Park J, Lee J et al. (2000) Mitt /Lb9 and Copg2, new members of mouse imprinted genes closely linked to Peg1/Melt. FEBS Lett 472, 230-234
382
10. Li E, Beard C, Jaenisch R (1993) Role for DNA methylation in genomic imprinting. Nature 366, 362-365 11. Li T, Vu TH, Zeng ZL, Nguyen BT, Hayward BE et al. (2000) Tissue-specific expression of antisense and sense transcripts at the imprinted gene locus. Genomics 69, 295-304 12. Reik W, Constancia M (1997) Genomic imprinting. Making sense or antisense? Nature 389, 669-671 13. Sambrook J, Russell DW (2001) Molecular Cloning: A Laboratory Manual, third ed. (Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press) 14. Szabo PE, Tang SH, Reed MR, Silva FJ, Tsark WM et al. (2002) The chicken beta-globin insulator element conveys chromatin boundary activity but not imprinting
J. YUN ET AL: ALLELE-SPECIFIC METHYLATION OF MOUSE
Copg2
at the mouse Igf2/H19 domain. Development 129, 897-904 15. Szebenyi G, Rotwein P (1994) The mouse insulin-like growth factor II/cation-independent mannose 6-phosphate (IGF-II/MPR) receptor gene: molecular cloning and genomic organization. Genomics 19, 120-129 16. Wutz A, Smrzka OW, Schweifer N, Schellander K, Wagner EF et al. (1997) Imprinted expression of the Igf2r gene depends on an intronic CpG island. Nature 389, 745-749 17. Yamasaki K, Hayashida S, Miura K, Masuzaki H, Ishimaru T et al. (2000) The novel gene, y2-COP (COPG2), in the 7q32 imprinted domain escapes genomic imprinting. Genomics 68, 330-335