The Neuronatin Gene Resides in a "Micro-Imprinted" Domain on Human Chromosome 20Q11.2

doi:10.1006/geno.2001.6612, available online at http://www.idealibrary.com on IDEAL Article

The Neuronatin Gene Resides in a “Micro-imprinted” Domain on Human Chromosome 20q11.2

Heather K. Evans,1,2 Andrew A. Wylie,1,3 Susan K. Murphy,1 and Randy L. Jirtle1,*

1Department of Radiation Oncology, and 2University Program in Genetics, Duke University Medical Center, Durham, North Carolina 27710, USA 3Safety Assessment, AstraZeneca Pharmaceuticals, Alderley Park, Macclesfield, SK104TG, UK

*To whom correspondence and reprint requests should be addressed. Fax: (919) 684 5584. E-mail: [email protected].

A small fraction of the genome contains genes that are imprinted and thus expressed exclusively from one parental allele. We report here that the human neuronatin gene (NNAT) on chromosome 20q11.2 is imprinted and transcribed specifically from the paternal allele. The region containing NNAT has multiple CpG islands, and methylation analysis showed that a 1.8-kb CpG island in its promoter region exhibits differential methylation in all tissues examined. This finding is consistent with the island acting as a component of the NNAT imprint control domain. NNAT lies within the singular 8.5-kb intron of the gene encoding bladder cancer-associated protein (BLCAP), which, as we demonstrate, is not imprinted. This study provides the first example, to our knowledge, in humans of an imprinted gene contained within the genomic structure of a nonimprinted gene. Thus, NNAT is in an imprinted “microdomain,” making this locus uniquely suited for the investigation of mechanisms of localized imprint regulation.

Key Words: human chromosome 20, genomic imprinting, neuronatin, NNAT, BLCAP

INTRODUCTION regulatory domains, it is imperative that additional imprinted loci undergo thorough characterization in many species. To facilitate interspecies comparison of imprint regula- Genomic imprinting refers to parent-of-origin-specific, tory regions, we determined the imprint status of the neu- monoallelic expression of a gene. Imprinted genes are unique ronatin gene (NNAT), the human homologue of a gene known in that although both alleles have identical sequence infor- to be imprinted in mice [4]. NNAT maps to human chromo- mation and share the same cellular environment, only one of some 20q11.2–q12 [5,6] and mouse distal chromosome 2 [4]. the alleles is expressed. Imprinted genes share several com- The NNAT amino acid sequence is 50% homologous to the mon characteristics including the presence of tandem repeats, proteolipids peroxisomal membrane protein-1 and phospho- differential methylation of CpG islands, boundary elements, lamban [5] and therefore constitutes a putative member of nontranslated RNA species, and temporal differences in DNA the proteolipid family. Proteolipids are typically small pro- replication. All of these factors seem to have a mechanistic teins consisting of one hydrophobic domain and one function in imprint regulation; however, the nature of the pri- hydrophilic domain, and they often function as the regulatory mary imprint mark established in the gametes remains elusive subunit of a membrane channel. [reviewed in 1]. NNAT is highly expressed in the central nervous system Approximately 45 imprinted genes have been identified, from mid-gestation through early postnatal development, and estimates indicate that as many as 200 may exist in the correlating with the onset and termination of brain devel- mammalian genome [2]. Only a handful of these known opment in mice and humans. Its expression seems to be lim- imprinted loci have been analyzed in any detail, most notably ited, however, to the anterior pituitary in the adult brain the insulin-like growth factor-2 (Igf2)/H19 domain on mouse [4,7–9]; tissue culture studies indicate that Nnat is also chromosome 7 (human chromosome 11p15.5) and the insulin- expressed in adult mouse pancreatic -cells [10,11]. like growth factor-2 receptor (Igf2r; also known as M6P/IGF2R) Although computer- simulated virtual northern blot analy- locus on mouse chromosome 17 (human chromosome 6q27) sis using the SAGE library demonstrated that NNAT expres- [reviewed in 1,3]. However, to better identify and understand sion in gliobastomas is equivalent to that seen in normal the function of the common genetic features found in imprint brain tissues, it is highly overexpressed in medulloblastomas

GENOMICS Vol. 77, Numbers 1–2, September 2001 99 Copyright © 2001 by Academic Press. All rights of reproduction in any form reserved. 0888-7543/01 $35.00 Article doi:10.1006/geno.2001.6612, available online at http://www.idealibrary.com on IDEAL

(http://www.ncbi. nlm.nih.gov/SAGE/SAGEcid.cgi?cid= 117546). A Mouse Nnat is expressed from the paternal allele and lies in a distinct imprinted chromosomal region 30 Mb upstream from the imprinted Gnas cluster [4,12]. Mouse embryos carry- ing either maternal (no functional copies of Nnat) or paternal (two functional copies of Nnat) duplications of distal chromosome 2 have been studied for phenotypic abnormalities [13]. Although spatial expression of Nnat in mice with a paternal duplication seemed identical to that of wild-type mice by in situ hybridization, there was a notable increase in the depth of the B cerebellar folds. Conversely, there was a reduction of cerebellar fold depth in embryos with a maternal duplication that were shown to lack Nnat expression by in situ hybridization [13]. Although mouse Nnat is known to be imprinted, the imprint status of the human homologue has not been determined. We demonstrate here that human NNAT is imprinted and expressed exclusively from the paternal allele during development. Furthermore, we show that this imprinted gene is unusually positioned in that it lies within the singular intron of the gene encoding bladder cancer-associated protein (BLCAP), a gene that is not imprinted.

RESULTS

Imprinting Status of Human NNAT The genomic structure of NNAT consists of three exons extending over approximately 2.7 kb (Fig. 1A). At least two alternatively spliced forms of NNAT exist in the human: a FIG. 1. Predicted genomic structure and expression analysis of NNAT. (A) 1.2-kb -form that incorporates all three exons and a 1.1-kb Predicted genomic structure of NNAT. Black boxes, exons; *, 28-bp region in -form consisting of only the first and third exon. Both of intron 1 containing four single-nucleotide polymorphisms (positions these isoforms are expressed in fetal brain [6]. Sequencing of 131,670–131,679; BAC RP11-425M5). (B) Parent-of-origin-dependent, monoal- these three NNAT exons in over 40 individuals failed to lelic expression of NNAT. The allelic expression of NNAT was determined using fetal brain tissue heterozygous for an A/T polymorphism (lower left demonstrate any polymorphisms, indicating a highly con- arrow) in intron 1. RNA analysis of this tissue demonstrates exclusive expres- served coding region. Indeed, NNAT mRNA remains 98% sion of the T allele (bottom right, arrowhead). The maternal genotype is A/A conserved between mouse and human. However, within (top left, arrow), demonstrating that the expressed T allele was inherited from intron 1 of NNAT we identified a polymorphic region con- the father, who must have either a TT genotype or a TA genotype. sisting of a 28-bp stretch of T residues, four of which were polymorphic for a T→A transversion (heterozygosity frequency, 67%; n = 30; bacterial artificial chromosome (BAC) the cDNA analyzed represents unspliced, nuclear RNA. RP11-425M5 positions 131,670–131,679; Figs. 1A and 1B). Therefore, both the - and -isoforms of human NNAT are To screen for expression, we used RT-PCR on total RNA monoallelically expressed from the same allele. isolated from adrenal gland, brain, gut, heart, kidney, liver, To determine the parental origin for the expressed NNAT lung, muscle, placenta, and spleen fetal tissues. These and all allele, we first obtained maternal genotypes using decidua other RT reactions included a negative control (data not tissue corresponding to the informative fetal brain samples shown) to ensure that genomic DNA did not contaminate analyzed. For two of these informative samples, the maternal subsequent PCR reactions, which would lead to the amplifi- genotypes were homozygous for the A allele (Fig. 1B). Only cation of both alleles and thus mask imprinting. Only central the T allele was expressed in these two fetal brain tissues, nervous system tissue yielded detectable cDNA products. To demonstrating that human NNAT is imprinted and expressed determine if NNAT is monoallelically expressed, we analyze specifically from the paternal allele. d RNA from 11 informative fetal brain samples by RT-PCR followed by nucleotide sequencing. Analysis of the cDNA in Imprinting Status of Human BLCAP all brain samples showed expression of only one allele (Fig. We also identified genes adjacent to NNAT, as imprinted 1B). As the polymorphism used here is in intron 1 of NNAT, genes often occur in clusters. A BLAST search [14] revealed

100 GENOMICS Vol. 77, Numbers 1–2, September 2001 Copyright © 2001 by Academic Press. All rights of reproduction in any form reserved. doi:10.1006/geno.2001.6612, available online at http://www.idealibrary.com on IDEAL Article

that a second gene, BLCAP, is on the same BAC clone con- from fetal brain, kidney, liver, and pancreas demonstrated taining NNAT (RP11-425M5; GenBank acc. no. AL109614). that the promoter CpG island for NNAT (region 3, positions Alignment of the BLCAP mRNA sequence to BAC RP11- 132,019–133,799; BAC RP11-425M5) contained a 1:1 distribu- 425M5 showed that this gene consists of two exons (Fig. 2A). tion of methylated to unmethylated alleles at all CpG dinu- Exon 1 encompasses approximately 75 bp, whereas the sec- cleotides examined, consistent with the results expected for ond exon is 1.9 kb and contains most of the 5 untranslated differential methylation (Fig. 3B). The two smaller CpG region, the entire protein coding sequence, and the 3 untrans- islands (regions 2 and 4, positions 130,973–131,529 and lated region. This 2-kb cDNA encodes a putative 87-amino- 134,849–135,251, respectively; BAC RP11-425M5) also pre- acid protein. Although the first exon lies about 6 kb down- sented a similar pattern of differential methylation in these stream of NNAT, the second exon is only 1.8 kb upstream of four tissues. In contrast, all CpG dinucleotides examined in NNAT. Thus, NNAT sits within intron 1 of BLCAP, and the the BLCAP promoter CpG island (region 1, positions two genes are transcribed in opposite directions (Fig. 2A). 125,283–126,670; BAC RP11-425M5) were unmethylated in all The close proximity of NNAT and BLCAP indicated that tissues examined (Fig. 3B). the regulatory mechanisms controlling the imprinting of NNAT could also affect BLCAP expression. To determine the imprint status of BLCAP, we identified an A/G polymorphism in the intron of BLCAP A 600 bp upstream of the NNAT transcription start site (heterozygosity frequency, 25%; n = 20; BAC RP11-425M5 position 133,286; Fig. 2A). As BLCAP is transcribed in an ori- entation opposite to that of NNAT, the BLCAP-specific RT primers should not reverse transcribe NNAT. To further ensure that NNAT transcripts would not mask the imprint status of BLCAP, we used two sep- B arate BLCAP RT primers that are approximately 1.1 kb upstream of the NNAT transcription start site. We detected BLCAP cDNA products in all fetal tissues examined and analyzed them for monoallelic expression. BLCAP demonstrated biallelic expression in the fetal brain samples (n = 4) in which NNAT was shown to be imprinted (Fig. 2B). Additionally, BLCAP was biallelically expressed in fetal adrenal gland (n = 2), heart (n = 3), kidney (n = 2), liver (n = 2), lung (n = 3), and placenta (n = 2) tissues (Fig. 2B).

Methylation Analysis of CpG Islands Imprinted genes are normally associated with CpG islands that are within or near these genes, and differential methylation of these islands is important in the mainte- nance of their imprint status [1,15]. Analysis of the NNAT/BLCAP genomic domain with the computer program Webgene (http://www.itba.mi.cnr.it/webgene/) demonstrated the presence of four CpG FIG. 2. Predicted genomic structure of the BLCAP/NNAT domain and expression analysis of BLCAP. (A) islands in this region (Fig. 3A). The pro- Predicted genomic structure of BLCAP/NNAT. Open boxes, BLCAP exons; filled boxes, NNAT exons. Arrows indicate direction of gene transcription. *Single-nucleotide polymorphism in intron 1 of BLCAP moter regions of both genes contained CpG (position 133,286; BAC RP11-425M5) used to analyze gene expression. (B) Biallelic expression of BLCAP. islands greater than 1 kb, and a CpG island The imprinting status of BLCAP was determined using fetal tissue heterozygous for an A/G polymor- less than 500 bp was detected within the phism (arrow) within intron 1 of BLCAP but upstream of the NNAT transcription start site. RNA analy- coding sequence of each gene. sis of fetal adrenal gland, brain, heart, kidney, liver, lung, and placental tissues demonstrates that both Sequencing of bisulfite-treated DNA the maternal and paternal alleles (arrowheads) are expressed in all tissues examined.

GENOMICS Vol. 77, Numbers 1–2, September 2001 101 Copyright © 2001 by Academic Press. All rights of reproduction in any form reserved. Article doi:10.1006/geno.2001.6612, available online at http://www.idealibrary.com on IDEAL

FIG. 3. Methylation analysis of NNAT/BLCAP domain CpG islands. (A) NNAT/BLCAP genomic domain. Open boxes, BLCAP; filled boxes, NNAT. Arrows indicate direction of gene transcription. Horizontal bars indicate locations of CpG islands: region 1, positions 125,283–126,670; region 2, positions 130,973–131,529; region 3, positions 132,019–133,799; region 4, positions 134,849–135,251; BAC RP11-425M5. Ovals indicate the location of nucleotides examined for CpG methylation: region 1, positions 125,439–125,598, 125,852–126,003, and 126,341–126,481; region 2, positions 131,211–131,396; region 3, positions 133,167–133,318 and 133,567–133,718; region 4, positions 134,857–135,151; BAC RP11-425M5. Open ovals, unmethylated CpG islands; gray ovals, differential methylation. (B) Methylation analysis of CpG islands in the BLCAP/NNAT genomic domain. Genomic DNA isolated from fetal brain, kidney, liver, and pancreas was treated with sodium bisulfite and amplified by PCR for methylation analysis of each of the CpG islands. Bisulfite treatment converts unmethylated cytosines to uracils. Representative sequences are shown for regions 1–4. Arrowheads, CpG dinucleotides; diamonds, positions of cytosines not present in CpG dinucleotides and therefore unmethylated. The presence of bands in both the T and C lanes indicates differential methylation of cytosine residues within the CpG dinucleotides; the presence of a band only in the C lane indicates complete methylation of cytosine residues on both alleles within the CpG dinucleotides; and the presence of a band only in the T lane indicates that cytosine residues are unmethylated within the CpG dinucleotides. All cytosines not present in CpG dinucleotides (diamonds) are unmethylated, and therefore serve as an internal control.

DISCUSSION which was not imprinted in all tissues examined. Of the four CpG islands identified within the NNAT/BLCAP genomic We have demonstrated here that NNAT is imprinted in the domain, the 1.8-kb NNAT promoter CpG island showed con- human central nervous system during development. This sistent differential methylation in all tissues investigated, gene lies within the 8.5-kb intron of a second gene, BLCAP, whereas the 1.3-kb BLCAP promoter CpG island remained

102 GENOMICS Vol. 77, Numbers 1–2, September 2001 Copyright © 2001 by Academic Press. All rights of reproduction in any form reserved. doi:10.1006/geno.2001.6612, available online at http://www.idealibrary.com on IDEAL Article

unmethylated on both parental alleles, concordant with the product in fetal brain (data not shown). This indicates that these respective imprint status of each gene. ESTs are either unspliced BLCAP RNA or genomic DNA con- This atypical genomic structure of an imprinted gene tamination in the EST database. Finally, imprinted genes tend within the intron of a nonimprinted gene has been docu- to occur in clusters, yet additional BLAST searches of overlap- mented before in mice but not in humans. The mouse ping BACs using both the nonredundant and EST database imprinted gene U2af1-rs1 sits within the 26-kb intron of the showed no other genes within a 280-kb region. biallelically expressed Murr1 on proximal chromosome 11 [16]. We have provided evidence for differential methylation of Nonetheless, there are several differences between the mouse the NNAT promoter CpG island, however, which is consistent U2af1-rs1 and human NNAT domains. U2af1-rs1 lies within an with findings for many other imprinted genes [1]. In contrast, intron three times the size of the BLCAP intron in which NNAT the BLCAP promoter CpG island is unmethylated. This princi- is situated, and whereas NNAT contains three introns, U2af1- pal difference in CpG island methylation correlates with the rs1 is intronless. Furthermore, no human homologue of mouse imprint status of these two genes and is concordant with the U2af1-rs1 has been identified. proposed mechanistic involvement of differential methylation In contrast, the NNAT/BLCAP genomic structure and in the regulation of imprinting [1]. Mouse Nnat also shows dif- imprint status seem to be conserved between humans and mice ferential methylation in the promoter region, with the maternal (GenBank acc. no. AB041829). Therefore, NNAT is the first exam- allele being fully methylated and the paternal allele being ple, to our knowledge, of an imprinted gene that lies within the unmethylated [13]. Thus, we predict that the large CpG island intron of a separate, biallelically expressed gene in both humans in the NNAT/Nnat promoter region may be crucial in regulat- and mice. Although BLCAP homologues are present in species ing the imprint status of this gene. The smaller CpG islands such as zebrafish and Drosophila melanogaster, NNAT homo- located in the coding regions of BLCAP and NNAT also seem logues are not found outside the mammalian lineage. These find- to be differentially methylated, although not at a ratio of 1:1. ings indicate that the NNAT/BLCAP structure originated either Therefore, the involvement of these CpG islands in regulating through a retrotransposition event early in mammalian evolu- NNAT imprinting is less certain. tion or through a gain-of-function event within the intron of Recent studies of IGF2 and IGF2R indicate that imprinting BLCAP that resulted in the formation of NNAT. evolved approximately 150 million years ago in an ancestor com- BLCAP was initially identified in a mRNA differential dis- mon to marsupials and eutherian mammals [22,23]. The most play between invasive and noninvasive human bladder transi- actively debated theory for why imprinting evolved concludes tional cell carcinomas [17]. The downregulation of BLCAP in that imprinted genes result as a consequence of a parental con- invasive transitional cell carcinomas indicates its possible func- flict to control the extraction of resources from the mother by her tion as a tumor suppressor. The finding that BLCAP is not offspring [24]. This hypothesis predicts that paternally expressed imprinted is intriguing, given its close proximity to NNAT. genes would promote offspring growth, whereas maternally Studies of the IGF2/H19 domain have shown that imprinting expressed genes would reduce their growth. control mechanisms are capable of exerting influence over dis- Nonetheless, embryos with maternal or paternal duplica- tances of at least 100 kb [reviewed in 1]. Therefore, mechanisms tions of Nnat show no substantial size difference from that of must exist within the NNAT/BLCAP domain that allow these their wild-type littermates; the only observed morphological two genes to be regulated differently despite their propinquity. changes are alterations in the depth of cerebellar folds [13]. This For example, differential imprint regulation of NNAT and region of the brain is involved in the coordination and control BLCAP may exist because chromatin boundary elements such of voluntary movement. Thus, mice with Nnat mutations may as those established by CTCF prevent the imprint of NNAT show abnormalities in basic behavioral activities such as walk- from spreading to nearby genes [18–21]. CTCF-binding sites ing or eating. As female mice with null mutations for the were found in both the IGF2/H19 and DLK1/GTL2 imprinted imprinted genes mesoderm-specific transcript (Mest) and Peg3 domains; however, consensus CTCF-binding sites are not pres- show severe nurturing deficiencies [25,26], it will also be impor- ent in the NNAT/BLCAP domain. The lack of CTCF-binding tant to determine if Nnat inactivation affects behavior. sites does not preclude the possibility that insulators other than Furthermore, it will be important to determine if imprinting of CTCF may lie between NNAT and BLCAP. NNAT and other genes involved in neuronal function occurred The NNAT/BLCAP domain also lacks other characteristics later in mammalian evolution than the imprinting of IGF2 and common to imprinted genes. Imprinted genes are often associ- IGF2R, genes with a considerable effect on body size during ated with tandem repeat sequences [1]; however, tandem repeats development. were not detected within this domain (tandem repeat finder, The human imprinted gene NNAT lies within a small intron http://c3.biomath.mssm.edu/trf.basic.submit.html; nucleic acid of BLCAP, a nonimprinted gene. This finding indicates that all dot plot, http://arbl.cvmbs.colostate.edu/molkit/dnadot/). of the control elements necessary for initiating and maintaining Additionally, antisense RNAs often occur near imprinted genes. the NNAT imprint could exist within the 8.5-kb intron of BLAST searches of RP11-425M5 did identify two overlapping, BLCAP. Consequently, this “micro-imprinted” domain is unspliced expressed sequence tags (ESTs) between exon 1 of uniquely suited for investigating the control elements NNAT and exon 2 of BLCAP; however, both 5 and 3 rapid required for localized regulation of genomic imprinting in amplification of cDNA ends (RACE) failed to detect a cDNA the human genome.

GENOMICS Vol. 77, Numbers 1–2, September 2001 103 Copyright © 2001 by Academic Press. All rights of reproduction in any form reserved. Article doi:10.1006/geno.2001.6612, available online at http://www.idealibrary.com on IDEAL

MATERIALS AND METHODS REFERENCES 1. Reik, W., and Walter, J. (2001). Genomic imprinting: parental influence on the genome. Nat. Rev. Genet. 2: 21–32. PCR amplification and DNA sequencing. Human fetal tissues were obtained 2. Barlow, D. P. (1995). Gametic imprinting in mammals. Science 270: 1610–1613. 3. Vu, T. H., and Hoffman, A. R. (2000). Comparative genomics sheds light on mechanisms from the National Institutes of Health-supported Birth Defects Research of genomic imprinting. Genome Res. 10: 1660–1663. Laboratory at the University of Washington. DNA was isolated from these tis- 4. Kagitani, F., et al. (1997). Peg5/Neuronatin is an imprinted gene located on sub-distal sues with the Dneasy kit (Qiagen, Valencia, CA). NNAT intronic polymor- chromosome 2 in the mouse. Nucleic Acids Res. 25: 3428–3432. phisms were amplified using the forward primer 5-TCGGAGAAAAG- 5. Dou, D., and Joseph, R. (1996). Cloning of human neuronatin gene and its localization to CACTTGG-3 and the reverse primer 5-CCTCACTTCTCGCAATGG-3 for 35 chromosome-20q11.2-12: the deduced protein is a novel “proteolipid’. Brain Res. 723: 8–22. cycles for DNA or 40 cycles on cDNA at 94°C for 30 s, 60°C for 30 s, and 72°C 6. Dou, D., and Joseph, R. (1996). Structure and organization of the human neuronatin gene. for 30 s. PCR products were separated by electrophoresis on 1% agarose gels, Genomics 33: 292–297. bands corresponding to the correct size were excised, and the DNA was eluted 7. Joseph, R., Dou, D., and Tsang, W. (1994). Molecular cloning of a novel mRNA (neuronatin) that is highly expressed in neonatal mammalian brain. Biochem. Biophys. Res. Commun. using GenElute Agarose spin columns (Sigma, St. Louis, MO). The PCR prod- 201: 1227–1234. ucts were then sequenced with the primer 5 -TTCTTCCTGCAAAGGGG-3 8. Joseph, R., Dou, D., and Tsang, W. (1995). Neuronatin mRNA: alternatively spliced forms using radiolabeled terminator cycle sequencing according to the manufacturer’s of a novel brain-specific mammalian developmental gene. Brain Res. 690: 92–98. protocol (USB Corporation, Cleveland, OH). 9. Wijnholds, J., Chowdhury, K., Wehr, R., and Gruss, P. (1995). Segment-specific expression An intronic BLCAP polymorphism was amplified using the forward primer of the neuronatin gene during early hindbrain development. Dev. Biol. 171: 73–84. 5-TCGAAACGACCTGTCCAG-3 and the reverse primer 5-GTCATCTGCC- 10. Niwa, H., Harrison, L. C., DeAizpurua, H. J., and Cram, D. S. (1997). Identification of pan- CCATAGAGC-3. Cycling conditions were identical to those used for NNAT creatic beta cell-related genes by representational difference analysis. Endocrinology 138: except that the annealing temperature was increased to 62°C. The PCR prod- 1419–1426. 11. Arava, Y., Adamsky, K., Ezerzer, C., Ablamunits, V., and Walker, M. D. (1999). Specific ucts were then sequenced as described above with the primer 5 -CTGAGCCA- gene expression in pancreatic -cells: cloning and characterization of differentially GAGCACTGCG-3 . expressed genes. Diabetes 48: 552–556. Genomic imprinting analysis. RNA Stat-60 (Tel-Test, Friendswood, TX) was 12. Williamson, C. M., et al. (1998). Localisation of the imprinted gene neuronatin, Nnat, con- used according to the manufacturer’s protocol to isolate total RNA from fetal firms and refines the location of a second imprinting region on mouse chromosome 2. tissue samples. The isolated RNA (2 g) was treated with DNaseI (Life Cytogenet. Cell Genet. 81: 73–78. 13. Kikyo, N., et al. (1997). Genetic and functional analysis of neuronatin in mice with mater- Technologies, Grand Island, NY) and then divided into two samples for reverse nal or paternal duplication of distal Chr 2. Dev. Biol. 190: 66–77. transcription in either the presence or the absence of reverse transcriptase. The 14. Altschul, S. F., et al. (1997). Gapped BLAST and PSI-BLAST: a new generation of protein NNAT RT primer was 5 -ACCTTGCCAAGTGCTCTG-3 , and the RT primer for database search programs. Nucleic Acids Res. 25: 3389–3402. BLCAP was either 5-GGCACTGTACAGAACCTGACCT-3 or 5-TGC- 15. Li, E., Beard, C., and Jaenisch, R. (1993). Role of DNA methylation in genomic imprinting. GAGGGACCCACAAGAC-3. Reverse transcription used Superscript II (Life Nature 366: 362–365. Technologies). The resulting cDNAs were amplified by PCR and sequenced as 16. Nabetani, A., Hatada, I., Morisaki, H., Oshimura, M., and Mukai, T. (1997). Mouse U2af1- described above. All DNA amplifications and imprint status determinations rs1 is a neomorphic imprinted gene. Mol. Cell. Biol. 17: 789–798. were made two or more times. 17. Gromova, I., Gromov, P., and Celis, J. E. (1999). Identification of true differentially expressed mRNAs in a pair of human bladder transitional cell carcinomas using an DNA methylation analysis using bisulfite sequencing. Genomic DNA sam- improved differential display procedure. Electrophoresis 20: 241–248. ples were treated with sodium bisulfite using the CpGenome DNA modifica- 18. Kanduri, C., et al. (2000). Functional association of CTCF with the insulator upstream of tion kit (Intergen, New York, NY). Bisulfite-treated DNA was amplified using the H19 gene is parent of origin-specific and methylation-sensitive. Curr. Biol. 10: 853–856. two rounds of nested PCR (30 cycles of 95°C for 30 s, 60°C for 30 s, and 72°C 19. Bell, A. C., and Felsenfeld, G. (2000). Methylation of a CTCF-dependent boundary con- for 30 s), purified, and sequenced as described above. Primer sets were selected trols imprinted expression of the Igf2 gene. Nature 405: 482–485. 20. Hark, A. T., et al. (2000). CTCF mediates methylation-sensitive enhancer-blocking activity to amplify DNA fragments from the four identified CpG islands present in the at the H19/Igf2 locus. Nature 405: 486–489. BLCAP/NNAT domain; these primer sequences are available on request from 21. Wylie, A. A., Murphy, S. K., Orton, T. C., and Jirtle, R. L. (2000). Novel imprinted the authors. A 300-ng amount of bisulfite-treated DNA was empirically deter- DLK1/GTL2 domain on human chromosome 14 contains motifs that mimic those impli- mined to avoid stochastic PCR amplification bias due to a limiting number of cated in IGF2/H19 regulation. Genome Res. 10: 1711–1718. starting DNA molecules, and this amount was therefore used as template in 22. Killian, J. K., et al. (2000). M6P/IGF2R imprinting evolution in mammals. Mol. Cell 5: the first round of PCR amplification. The conversion of all non-CpG dinu- 707–716. cleotide cytosines served as a positive control to ensure that the initial DNA 23. O’Neill, M. J., Ingram, R. S., Vrana, P. B., and Tilghman, S. M. (2000). Allelic expression of template underwent complete denaturation. All bisulfite-treated DNA ampli- IGF2 in marsupials and birds. Dev. Genes Evol. 210: 18–20. 24. Moore, T., and Haig, D. (1991). Genomic imprinting in mammalian development: a fication reactions were duplicated. parental tug-of-war. Trends Genet. 7: 45–49. RACE analysis. For RACE, we used the Marathon-Ready fetal brain cDNA kit 25. Lefebvre, L., et al. (1998). Abnormal maternal behaviour and growth retardation associ- according to the manufacturer’s directions (Clontech, Palo Alto, CA). For 5- ated with loss of the imprinted gene Mest. Nat. Genet. 20: 163–169. RACE we used the primer 5-TCGAAACGACCTGTCCAG-3, and for 3-RACE 26. Li, L., et al. (1999). Regulation of maternal behavior and offspring growth by paternally we used the primer 5-CAGGTCCTCTGCTTCATGAG-3. expressed Peg3. Science 284: 330–333.

ACKNOWLEDGMENTS We thank members of the Jirtle laboratory for critical reading of the manuscript. We thank the Birth Defects Research laboratory at the University of Washington for tissue samples. This study was supported by National Institutes of Health grants CA25951 and ES08823, Department of Defense grant DAMD17-98-1-8305, Sumitomo Chemical Company, and AstraZeneca Pharmaceuticals. Further informa- tion on genomic imprinting is available at http://www.geneimprint.com.

RECEIVED FOR PUBLICATION MARCH 16; ACCEPTED JUNE 28, 2001.