902

BIOLOGY OF REPRODUCTION 69, 902–914 (2003)Published online before print 14 May 2003.DOI 10.1095/biolreprod.103.017293

Disruption of Imprinted Gene Methylation and Expression in ClonedPreimplantation Stage Mouse Embryos1

Mellissa R.W. Mann,3 Young Gie Chung,4 Leisha D. Nolen,3 Raluca I. Verona,3 Keith E. Latham,2,4

and Marisa S. Bartolomei3

Howard Hughes Medical Institute and Department of Cell and Developmental Biology,3 University of PennsylvaniaSchool of Medicine, Philadelphia, Pennsylvania 19104The Fels Institute for Cancer Research and Molecular Biology and Department of Biochemistry,4 Temple UniversitySchool of Medicine, Philadelphia, Pennsylvania 19140

ABSTRACT

Cloning by somatic cell nuclear transfer requires that epi-genetic information possessed by the donor nucleus be repro-grammed to an embryonic state. Little is known, however, aboutthis remodeling process, including when it occurs, its efficiency,and how well epigenetic markings characteristic of normal de-velopment are maintained. Examining the fate of epigenetic in-formation associated with imprinted genes during clonal devel-opment offers one means of addressing these questions. We ex-amined transcript abundance, allele specificity of imprintedgene expression, and parental allele-specific DNA methylationin cloned mouse blastocysts. Striking disruptions were seen intotal transcript abundance and allele specificity of expressionfor five imprinted genes. Only 4% of clones recapitulated a blas-tocyst mode of expression for all five genes. Cloned embryosalso exhibited extensive loss of allele-specific DNA methylationat the imprinting control regions of the H19 and Snprn genes.Thus, epigenetic errors arise very early in clonal developmentin the majority of embryos, indicating that reprogramming isinefficient and that some epigenetic information may be lost.

early development, embryo, gene regulation

INTRODUCTION

The recent success of somatic cell cloning in mammalsgives promise to applications such as species preservation,livestock propagation, and cell therapy for medical treat-ment [1–7]. Importantly, this success provides a strikingdemonstration of the remarkable capacity of the egg cyto-plasm to modify nuclear function. Because embryonic genetranscription commences at the late one-cell stage in themouse, the ability to obtain live offspring suggests thatmodification of the somatic genome may be completed fair-ly quickly. Such an ability would reflect the principal func-tion of the oocyte, which is to transform the haploid ge-

1This research was supported by the National Institutes of Health(HD38381 and 5 T32 CA09214-20) and the Howard Hughes MedicalInstitute. M.R.W.M. was supported by a grant from The Lalor Foundation.R.I.V. is a Rena and Vic Damone fellow of the American Cancer Society.M.R.W.M, Y.G.C., and L.D.N. contributed equally to this work2Correspondence: Keith E. Latham, The Fels Institute for Cancer Researchand Molecular Biology and Department of Biochemistry, Temple Univer-sity School of Medicine, Philadelphia, PA 19140. FAX: 215 707 1454;e-mail: klatham@temple.edu

Received: 18 March 2003.First decision: 7 April 2003.Accepted: 8 May 2003.Q 2003 by the Society for the Study of Reproduction, Inc.ISSN: 0006-3363. http://www.biolreprod.org

nomes from two highly differentiated gametes into a single,diploid, totipotent embryonic genome.

Many questions regarding this reprogramming have yetto be addressed. When is the donor cell nucleus repro-grammed, how efficient is reprogramming, and to what de-gree is characteristic epigenetic information retained? Ad-dressing such basic questions should provide new insightregarding the molecular mechanisms that control epigeneticinformation during normal development, the underlyingability of the cloned embryo to either maintain or modifysuch information, and the potential risks that may be as-sociated with therapeutic or reproductive cloning.

Recent studies of cloned embryos during preimplanta-tion development have revealed striking defects, indicatingthat cloned embryos do not faithfully recapitulate many ofthe essential early events of normal development. Clonedembryos exhibit defects in the expression of key regulatorygenes such as Oct4, a POU domain, class 5, transcriptionfactor 1 [8]. These embryos display dramatic alterations inculture-medium preferences with a shift toward somatic cellcharacteristics, indicating a lack of nuclear reprogrammingin genes affecting basic physiology and metabolism [9].Cloned preimplantation embryos also exhibit defects in de-methylation processes, including global demethylation anddemethylation of some repetitive elements [10–13]. Finally,cloned embryos aberrantly express the somatic form of theDNA methyltransferase protein, DNMT1, and are ineffi-cient at nuclear uptake of the maternally inherited oocyteform [14]. These observations indicate that epigenetic in-formation may not be faithfully preserved during earlyclonal development.

One way to address the ability of cloned embryos toreprogram epigenetic information is to elucidate the fate ofimprinted gene modifications during clonal development.To date, most studies have been limited to the analysis ofa few, rare, surviving clones that reached fetal, neonatal, oradult stages of development [15–19]. To our knowledge, nostudy thus far has examined imprinting during the earlieststages of clonal development to determine how epigeneticinformation in cloned embryos may be affected or assessedallele specificity of imprinted gene expression in conjunc-tion with DNA methylation analyses in somatic cell clonedembryos. We therefore undertook a detailed analysis of al-lele-specific expression and DNA methylation of imprintedgenes in cloned mouse blastocysts.

For this analysis, we developed novel methods to assayimprinted gene methylation using small numbers of embry-os and to assay parental allele expression of multiple genesat the single-embryo level. By applying this combination

903EPIGENETIC ANALYSIS OF CLONED BLASTOCYSTS

of methodologies, we have achieved what we believe to bethe first detailed analysis of the effects of cloning on epi-genetic information associated with imprinted genes duringthe earliest stages of cloned embryo development. Our re-sults indicate that cloned blastocysts rarely, if ever, displaynormal expression patterns of imprinted genes, even inmorphologically high-quality cloned embryos. Defects ingene expression are accompanied by aberrant methylationpatterns of the H19 and Snrpn (for small nuclear ribonu-cleoprotein N) genes. These results suggest that at leastsome forms of epigenetic information are not faithfully re-tained during cloning and that such defects arise early dur-ing clonal development. The observed disruptions in DNAmethylation and expression of imprinted genes are mostlikely only a subset of the spectrum of defects related toreprogramming of the donor nucleus to an embryonic state.

MATERIALS AND METHODS

Generation of a Substrain of B6(CAST 7) MiceFor allele-specific expression studies, cumulus cells were obtained from

F1 hybrid females that were derived from crosses with C57BL/6J (B6) fe-males and Mus musculus castaneus (CAST) males (The Jackson Laboratory,Bar Harbor, ME) and with B6(CAST7) females and B6 males. TheB6(CAST7) substrain was generated by backcrossing [CASTXB6]F1 prog-eny to B6 mice for two generations. Progeny heterozygous for CAST andB6 on chromosome 7 were intercrossed to generate mice homozygous forCAST on chromosome 7. Mice were genotyped with MIT markers (Re-search Genetics, Carlsbad, CA) approximately every 5 cM as describedpreviously [20]. All studies adhered to procedures consistent with the Na-tional Research Council Guide for the Care and Use of Laboratory Animals.

Generation of Clones and Control EmbryosCloned embryos were generated as described previously [9]. Recipient

oocytes for all studies were from (B6D2)F1 females. Cumulus cell nucleiwere isolated from individual cells by several rounds of trituration into theinjection pipette and then injected into the oocytes, and the oocytes wereactivated as described previously [6, 9]. Tetraploid control embryos wereproduced by injecting nuclei into intact eggs. Parthenogenetic control em-bryos were obtained by activation of intact oocytes in the same manner.Cloned, parthenogenetic, and tetraploid embryos were cultured at 378C inan atmosphere of 5% CO2 in air [6]. Fertilized embryos were cultured in5% CO2, 5% O2, and 90% N2. Media included in the present study wereas described previously [9]. Embryos were cultured continuously in CZBplus glucose or initially in Whitten medium or in KSOM (for potassiumsimplex-optimized medium with increased salt concentrations) withoutamino acids, then switched at the eight-cell stage to KSOM augmentedwith amino acids.

Isolation of mRNA from Pooled Blastocysts

Pools of 12–20 blastocysts were placed in 20 ml of guanidine thiocy-anate lysis buffer (5 M guanidine thiocyanate, 0.5% sarcosyl, 25 mMsodium citrate [pH 7.0], and 20 mM dithiothreitol [DTT]), vortexed, andstored at 2808C. Before mRNA isolation, Dynabeads Oligo (dT)25 (Dynal,Lake Success, NY) was equilibrated with 100 ml of 13 dilution buffer(100 mM Tris-HCl [pH 8.0], 400 mM LiCl, and 20 mM EDTA) accordingto the manufacturer’s instructions. The volume of frozen embryo lysatewas brought up to 100 ml with guanidine thiocyanate lysis buffer andthawed quickly by vortexing. Dilution buffer (100 ml) was added, and theembryo lysate was incubated with the equilibrated Dynabead Oligo (dT)25for 5 min at room temperature with continuous shaking. This mixture waswashed twice in 200 ml of washing buffer (10 mM Tris-HCl [pH 8.0],0.15 M LiCl, and 1 mM EDTA) with 0.1% lauryl sarcosinate and threetimes in 200 ml of washing buffer. The mRNA was eluted by addition of9.4 ml of H2O and incubation at 658C for 2 min. For the reverse transcrip-tion (RT) reaction, 9.6 ml of RT mix (13 First-Strand buffer (Invitrogen,Grand Island, NY), 10 mM DTT (Invitrogen), 0.5 mM of each dNTP(Amersham Bioscience Corp, Piscatoway, NJ), 25 ng of oligo (dT)12–18(Amersham), and 20 U of RNaseOut Recombinant Ribonuclease Inhibitor[Life Technologies, Carlsbad, CA]) were added to the mRNA. The mixturewas incubated at 428C for 2 min, and then 200 U of Superscript II (Life

Technologies) were added and first-strand synthesis allowed to proceed for1 h at 428C. The reaction was heat inactivated at 958C for 10 min.

Allele-Specific Expression AssaysFor the Igf2r (insulin-like growth factor II receptor), Meg3 (maternally

expressed gene 3), and Ascl2 (achaete-scute complex homolog-like 2) RTexpression assays, polymerase chain reaction (PCR) amplification was con-ducted on two to four embryo equivalents of cDNA under conditions specificfor each primer set. To a Ready-To-Go PCR Bead (Amersham BioscienceCorp, Piscataway, NJ), 0.3 mM of each primer (Invitrogen, Grand Island, NY)and [32P]dCTP (1 mCi; Perkin Elmer, Boston, MA) were added. The Igf2rprimers, Ir1 (59-GAGACCTCACCCTCATCTATTC-39) and Ir2 (59-GCA-CACAGCAGCATCTTCAG-39), amplified a 388-base pair (bp) fragment(958C for 2 min followed by 35 cycles at 958C for 15 sec, 588C for 10 sec,and 728C for 20 sec) containing a polymorphism between B6 (A) and CAST(G) (position 1549, MMU04710) [21]. Restriction digestion with TaqI resultedin 210- and 178-bp fragments in CAST, whereas the B6 amplicon was un-cleaved. A 337-bp Meg3 fragment was amplified with Meg3 (59-CCAA-AGCCATCATCTGGAATC-39) and Meg4 (59-CAGCCCTGTGAGGTAG-GAAC-39) primers at 958C for 2 min followed by 34 cycles at 958C for 15sec, 558C for 10 sec, and 728C for 20 sec (polymorphism at position 1570,MMGT12) [22]. Restriction digestion with SfcI resulted in 250- and 88-bpfragments in B6, whereas the CAST amplicon was uncut. The Ascl2 primers,Ascl1 (59-TGAGCATCCCACCCCCCTA-39) and Ascl2 (59-CCAAACAT-CAGCGTCAGTATAG-39), amplified a 474-bp fragment (958C for 2 min fol-lowed by 35 cycles at 958C for 15 sec, 588C for 10 sec, and 728C for 20 sec).A polymorphic SfcI restriction site between B6 (T) and CAST (C) (position10828, AF139595) (L. Lefebvre, A. Nagy, and J. Mann, personal communi-cation) distinguished the parental alleles (CAST, 266- and 207-bp fragments;B6, 474 bp). Products were resolved on a 7% polyacrylamide gel. After a 16-h exposure, the relative band intensities were quantified using ImageQuant(Molecular Dynamics, Sunnyvale, CA).

The H19 and Snrpn expression LightCycler (LC) assays were conductedon cDNA using the LC Real Time PCR System (Roche Molecular Biochem-icals, Indianapolis, IN). The H19 LC assay was performed as described pre-viously [23] except that two embryo equivalents of cDNA were used, ampli-fication proceeded for 45 cycles, and the contribution of each allele was cal-culated as the peak area of the melting curve generated at the allele-specifictemperature, approximately 67.58C for B6 and 61.58C for CAST. The Snrpnprimers, Sn1 (59-CTCCACCAGGAATTAGAGGC-39) and Sn3 (59TATAGT-TAATGCAGTAAGAGG39), were used to amplify a 155-bp region of theSnrpn gene (MMSMN [21]). Fluoresence resonance energy transfer (FRET)hybridization probes were designed to the B6 amplicon. The Snrpn sensorprobe (59-GAAGCATTGTAGGGGAAGAGAA-FL-39; Idaho Technologies,Salt Lake City, UT) spans a single nucleotide polymorphism at nucleotide 915between B6 (C) and CAST (T) and was labeled with fluorescein at the 39end. The Snrpn anchor probe (59-RED640-GGCTGAGATTTATCA-ACTGTATCTTAGGGTC-P-39; Idaho Technologies) was labeled with LC-Red640 at the 59 end and phosphorylated at the 39 end. To a Ready-To-GoPCR Bead, 5.12 ml of H2O, 0.38 ml of TaqStart Antibody (BD BiosciencesClontech, Palo Alto, CA), and 1.5 ml of 25 mM MgCl2 (final concentration,3.0 mM) were added, and the reaction was incubated at room temperature for5 min. After incubation, a final concentration of 12% dimethyl sulfoxide(DMSO), 0.5 mM of each primer, and 0.3 mM of each probe was added tothe mix and the volume brought to 12.5 ml. From this reaction mix, 10 mlwere removed and added to a LC glass capillary (Roche Molecular Biochem-icals), and 10 ml of cDNA (two embryo equivalents) and H2O were addedfor a final reaction volume of 20 ml. After an initial denaturation step at 958Cfor 2 min, amplification was performed for 65 cycles at 958C for 1 sec, 508Cfor 15 sec, and 728C for 6 sec. A single fluorescence acquisition occurred atthe end of each annealing step. After amplification, a final denaturation andannealing step was conducted (958C for 3 min, 358C for 2 min) followed bya melting-curve analysis with fluorescence acquisition occurring continuouslyas the temperature was increased from 358C to 858C in 0.28C increments.After background subtraction, the contribution of each allele was calculatedas the peak height of the melting curve generated at the allele-specific tem-perature, approximately 56.58C for B6 and 51.08C for CAST (LC Data Anal-ysis Software, Roche Molecular Biochemicals).

Synthesis of a Reusable Dynabead Oligo (dT)25-cDNALibrary from Single Blastocysts

Individual blastocyst-stage embryos were placed in 20 ml of guanidinethiocyanate lysis buffer, vortexed, and stored at 2808C. Dynabeads Oligo(dT)25 were equilibrated and embryo lysates processed as described above.After removal of the last wash, 10 ml of RT mix (13 First-Strand buffer,

904 MANN ET AL.

10 mM DTT, 0.5 mM dNTPs each, 20 U of RNaseOut Recombinant Ri-bonuclease Inhibitor, and 50 U of Superscript II) were added to the Dyn-abead Oligo (dT)25, and the complex was incubated at 428C for 1 h whilerotating in a hybridization oven. The resulting Dynabead Oligo (dT)25covalently linked cDNA library was washed twice in 10 ml of TNT buffer(Tris-EDTA buffer, 0.01% IGEPAL(NP-40), and 0.01% Tween 20), andthe RNA was removed after an incubation at 958C for 1 min. After re-moval of the RNA, the Dynabead Oligo (dT)25-cDNA library was washedtwice in 100 ml of TNT buffer and stored at 48C in 100 ml of TNT buffer.

Before PCR amplification, second strand was synthesized from theDynabead Oligo (dT)25-cDNA library. To a Ready-To-Go PCR bead, 23forward primer was added in a volume of 25 ml. Ten microliters of 23forward primer-PCR mix was added to the Dynabead Oligo (dT)25-cDNAlibrary, and second strand was synthesized by one cycle in a Hybaid rapidcycler (958C for 15 sec, annealing temperature for 10 sec, and 728C for20 sec followed by denaturation at 948C for 2 min). Second-strand productwas removed quickly from the Dynabead Oligo (dT)25-cDNA library andthen centrifuged to collect any remaining condensation. The second-strandproduct was then separated a second time to ensure removal of all Dyn-abeads (any remaining Dynabeads were resuspended in TNT and addedback to the library). In preparation for the next gene of interest, the Dyn-abead Oligo (dT)25-cDNA library was washed twice in 100 ml of TNTbuffer and stored at 48C in 100 ml of TNT buffer.

Allele-Specific Expression Assays UsingSecond-Strand Product

For RT-PCR analysis, 23 reverse primer and [32P]dCTP (2 mCi) wereadded to a Ready-To-Go PCR Bead. Ten microliters of 23 reverse primer-PCR reaction mix were added to the second-strand product that contained23 forward primer, resulting in a final concentration of 0.3 mM for eachprimer. The PCR amplification was conducted as described above.

The H19 and Snrpn LC expression assays were conducted on second-strand product using the LC Real Time PCR System. For the H19 LCexpression assay, Ready-To-Go PCR Beads were preincubated withTaqStart Antibody, after which a final concentration of 4.5 mM MgCl2,0.6 mM reverse primer, and 0.3 mM of each probe was added to the mixand the volume brought to 25 ml. From this reaction mix, 10 ml wereremoved and added to a glass capillary, and then 10 ml of second-strandproduct were added (final concentration, 3.0 mM MgCl2, 0.3 mM of eachprimer, and 0.15 mM of each probe). Amplification and analysis wereperformed as described above.

For the Snrpn LC expression assay, 11.62 ml of H2O, 0.38 ml ofTaqStart Antibody, and 3 ml of 25 mM MgCl2 were added to a Ready-To-Go PCR Bead, and the reaction was incubated at room temperature for5 min. After incubation, a final concentration of 24% DMSO, 1.0 mMreverse primer, and 0.6 mM of each probe was added to the mix and thevolume brought to 25 ml. From this reaction mix, 10 ml were removedand added to a glass capillary. Then, 10 ml of second-strand product wereadded (final concentration, 3.0 mM MgCl2, 12% DMSO, 0.5 mM of eachprimer, and 0.3 mM of each probe), and amplification was performed asdescribed above.

Glyceraldehyde-3-Phosphate Dehydrogenase GeneExpression Analysis

For analysis of glyceraldehyde-3-phosphate dehydrogenase (Gapd) lev-els, GAPDHF1 (59-ATCACTGCCACCCAGAACAC-39) and GAPDHB1(59-ATCCACGACGGACACATTGG-39) primers were used to amplify a185-bp region in the Gapd gene. Second-strand synthesis was carried outas described above with 23 forward primer (0.6 mM) at an annealing stepof 588C. For the LC analysis, Ready-To-Go PCR beads were preincubatedwith TaqStart antibody, and then a final concentration of 0.6 mM reverseprimer, 6 mM MgCl2, and 23 SYBR Green (Molecular Probes, Eugene,OR) was added and the volume brought to 25 ml. Ten microliters of thisreaction mix were added to a glass capillary together with 10 ml of second-strand product (final concentration, 3.0 mM MgCl2, 0.3 mM of each primer,and 13 SYBR Green). After an initial denaturation step at 958C for 2 min,amplification was performed for 24 cycles at 958C for 0 sec, 588C for 10sec, and 728C for 9 sec, with a single fluorescence acquisition step at theend of each elongation step. Melting-curve analysis was performed after afinal denaturation step at 958C for 0 sec followed by 658C for 15 sec, afterwhich the temperature was increased to 978C in 0.18C increments with con-tinuous acquisition. After background subtraction, the total amount of prod-uct was calculated as the peak area of the melting curve generated (;928C).

Parental allele-specific expression patterns for all genes were calculatedas the percentage expression of the B6 or CAST allele relative to the total

expression of both alleles. Expression was classified as monoallelic (de-fined as #10% expression from the normally silent allele), biallelic, or noexpression. Total expression levels for each gene were classified indepen-dently as no, low (H19, crosspoint value [cp] . 38; Meg3, Igf2r, andAscl2, 102–103 cpm; Snrpn, cp . 43; and Gapd, .0.008 peak area), me-dium (H19, cp . 34; Meg3, Igf2r, and Ascl2, 104–105 cpm; Snrpn, cp .35; and Gapd, .0.1 peak area), and high (H19, cp . 30; Meg3, Igf2r,and Ascl2, 106–107 cpm; Snrpn, cp . 30; and Gapd, .1.0 peak area)expression. The cp was determined by the second-derivative maximummethod using the LC Data Analysis Software. Each value was generatedduring the log-linear phase at the threshold cycle (i.e., the point at whichthe fluorescence signal exceeds the level of background noise).

Allele-Specific DNA Methylation AnalysisThe DNA was isolated from pools of 25–30 blastocysts and subjected

to bisulfite modification, PCR amplification, subcloning, and sequencingas previously described [24] with the following modifications. The muta-genized DNA was resuspended in 5 ml of TE (10 mM Tris-Cl [pH 8.0],1 mM EDTA), and 1 ml was used for PCR amplification. All mutagenizedDNAs were subjected to multiple independent PCR amplifications to en-sure recovery of different strands of DNA. For the H19 differentiallymethylated domain (DMD) and promoter proximal region, the regionsfrom 1304 to 1726 bp and from 4397 to 4778 bp (U19619) were assayed,respectively [24]. For Snrpn, the region from 2073 to 2601 bp (AF081460)was assayed [25]. Parental alleles were distinguished by single nucleotidepolymorphisms (SNPs) in the regions of interest. The SNPs for H19 werepreviously described [26]. The SNPs for Snrpn were as follows (B6/CAST): 2181 G/A, 2191 T/A, 2251 T/G, 2260 T/A, 2268 G/A, 2281 C/T, 2292 C/T, and 2348 G/T.

RESULTS

Elucidating the fate of epigenetic information before thesevere attrition that occurs soon after implantation and dur-ing fetal life should provide insight regarding how rapidlyreprogramming of the donor nucleus occurs, the efficiencyof this reprogramming, and the fidelity of maintaining epi-genetic information characteristic of normal development.The analysis of specific genes for which epigenetic regu-lation is well-studied and which likely are representative ofsimilarly controlled genes should address these key ques-tions. In the present study, the expression, imprinted regu-lation, and methylation states of multiple imprinted geneswere examined in cloned blastocyst-stage embryos.

Expression of Imprinted Genes in Pooled Blastocysts

Two imprinted genes expressed at significant levels in blas-tocysts are H19 and Snurf-Snrpn (herein called Snrpn). TheH19 gene transcription initiates at the blastocyst stage, where-as expression of the Snrpn gene begins at the 4-cell stage [20,27]. Allele-specific expression of these genes was initially ex-amined in pools of cloned blastocysts [B6(CAST7)XB6 orB6XCAST cumulus cell nuclei injected into B6D2 oocytesfrom which spindle-chromosome complexes had been re-moved (designated as cCBBD and cBCBD, respectively)],parthenogenetic blastocysts [activated oocytes (pC/B)], tetra-ploid blastocysts [B6(CAST7)XB6 or B6XCAST cumuluscell nuclei injected into intact B6D2 oocytes (designated astCBBD and tBCBD, respectively)], in vitro-cultured blasto-cysts [B6(CAST7)XB6 (iCB)], and in vivo-derived blastocyst[B6(CAST7)XB6 (vCB)]. These middle three sets of embryosserved as controls to reveal possible effects of cloning andculturing procedures.

Diploid parthenogenetic B6(CAST7)XB6 blastocysts,which possess two maternal genomes, displayed expressionof both CAST and B6 H19 alleles and no expression of theoppositely imprinted Snrpn gene, as expected (Table 1). Tet-raploid blastocysts contain four haploid genomes, three ofwhich are maternal. Expression of H19 in pooled tCBBD

905EPIGENETIC ANALYSIS OF CLONED BLASTOCYSTS

TABLE 1. Expression of H19 and Snrpn in pooled lysates.a

Sample Embryo type

H19 C allele (%)

Expected Observed

Snrpn B allele (%)

Expected Observed

vCB1vCB2

Fertilized/in vivoFertilized/in vivo

100100

100100

100100

100100

pC/B1pC/B2pC/B3pC/B4

ParthenoteParthenoteParthenoteParthenote

50b

505050

48253536

0000

0000

tCBBD1c

tCBBD2c

tCBBD3c

tCBBD4c

TetraploidTetraploidTetraploidTetraploid

33333333

310

2633

100100100100

100NE100100

iCB2iCB3iCB4

Fertilized/culturedFertilized/culturedFertilized/cultured

100100100

NE100100

100100100

100NA100

cCBBD1c

cCBBD2c

cCBBD3c

cCBBD4c

cCBBD5c

ClonesClonesClonesClonesClones

100100100100100

100100100NENE

100100100100100

100100NE100NE

H19 B allele (%)

Expected Observed

Snrpn C allele (%)

Expected Observed

tBCBD1d Tetraploid 100 100 100 100cBCBD1d

cBCBD2d

cBCBD3d

cBCBD4d

cBCBD5d

cBCBD6d

ClonesClonesClonesClonesClonesClones

100100100100100100

NE100NENENENE

100100100100100100

NE100NENE100NE

a B, B6; C, CAST; D, D2; i, in vitro cultured; NA, not assayed; NE, noexpression detected.b Assuming an equal number of parthenotes (pC/B) in pool that possessC allele and that possess B allele.c CBBD, donor nuclei are from [B6(CAST7)XB6] F1 female cumulus cellsinjected into B6D2 oocytes.d BCBD, donor nuclei are from (B6XCAST) F1 female cumulus cells in-jected into B6D2 oocytes.

blastocysts [one CAST, two B6 or D2 maternal alleles (D2transcripts are indistinguishable from B6), and one paternalB6 allele] was observed from the maternal CAST somaticgenome and the B6 oocyte genome(s), indicating that theH19 gene of the donor cell was reactivated and expressedin blastocyst-stage embryos (Table 1). The maternal CASTsomatic Snrpn allele remained silent. Analysis of the tBCBDpool (three maternal B6 or D2 alleles and one paternal CASTallele) likewise demonstrated imprinted expression of H19and Snrpn, because the paternal CAST somatic H19 allelemaintained its silent state and the paternal CAST somaticSnrpn allele was solely expressed (Table 1). Similar to invitro-cultured and in vivo-derived blastocysts, pools ofcloned blastocysts exhibited imprinted H19 and Snrpn ex-pression with only the maternal H19 alleles and only thepaternal Snrpn alleles transcribed (Table 1).

Expression Levels of Imprinted Genesin Individual Blastocysts

Whereas a majority of parthenogenetic, tetraploid, andfertilized embryo pools yielded measurable signals for bothH19 and Snrpn mRNAs, approximately half (5 of 11) ofthe cloned embryo pools failed to yield signals for eithermRNA. Because cloned embryos have been shown previ-ously to contain reduced numbers of cells in comparison tothe other embryo types [9], the simplest explanation is thata significant number of cloned embryos within some of thepooled samples were developmentally delayed and failed

to activate expression of these genes to a detectable level.Thus, it was essential to account for individual embryoquality by assaying expression within single embryos. Todo this, we developed a method that synthesizes reusableDynabead Oligo (dT)25-cDNA libraries from individualblastocysts. This allows expression from multiple genes tobe analyzed in a single embryo. Individual cloned embryoswere classified, with respect to size and morphology, asexcellent (well-expanded and comparable in size to fertil-ized control blastocysts), good (expanded but smaller thanfertilized blastocysts, with cell debris possible), or poor(poorly expanded with a small cavity and numerous ex-cluded cells) and were assayed for expression of five im-printed genes and one nonimprinted gene (Fig. 1).

Expression of imprinted genes was examined in blasto-cysts obtained by culture in a system that we previouslydetermined supports a high rate (35%) of blastocyst for-mation: Whitten medium to the 8-cell stage and KSOMaugmented with amino acids thereafter (WK) [9]. This cul-ture system produces cloned blastocysts at an efficiencycomparable to that reported by others [6]. A significantnumber of cloned blastocysts failed to express any of theimprinted genes (Fig. 1). These embryos were generally ofgood or poor morphology. The majority of expressingclones had excellent morphologies. The compromised mor-phology exhibited by approximately 50% of the blastocystsis consistent with the observation that many cloned embry-os likely arrest during the preimplantation period [28].

The analysis of transcript abundance in the expressing em-bryos revealed a large degree of variability among clonescompared to in vivo-derived blastocysts (Fig. 1). No discern-ible pattern emerged with regard to the number or identity ofthe expressed genes in the cloned samples. All permutationswere observed from only one imprinted gene to five genesbeing expressed in a given embryo, suggesting that each generesponded independently to the cloning process.

The apparent deficiencies in imprinted gene expressioncould conceivably be attributed to an effect of culture ratherthan to an effect of cloning per se [20]. To distinguish be-tween these possibilities, we examined gene expression inadditional single cloned blastocysts in a second culture sys-tem that also supports a high rate (22%) of blastocyst for-mation: KSOM without amino acids to the 8-cell stage fol-lowed by KSOM with amino acid augmentation (KK) [29;unpublished results]. The results obtained with these KKcloned embryos were virtually identical to those obtainedwith the WK combination (Fig. 1). In contrast, the expres-sion level of imprinted genes in parthenogenetic embryoswas better in KK than in WK (Fig. 1). Parthenogeneticembryos resembled fertilized embryos, for which KSOMsupports a more in vivo-like pattern of gene expression thanWhitten medium [30], as demonstrated by Gapd expres-sion. The opposite held true for tetraploid blastocysts, be-cause transcript levels for all genes were higher in WK-cultured tetraploids.

It is noteworthy that good correlation was found betweenthe level of expression of the imprinted genes analyzed andthe expression of the control gene Gapd: Medium to highlevels of Gapd were associated with overall medium to highlevels of imprinted gene expression, whereas low or lackof expression of Gapd was correlated with similar low lev-els of expression of imprinted genes (Fig. 1). This resultindicates that effects of the cloning procedure on expressionmay be widespread and that differences in Gapd expressionamong cloned blastocysts likely reflect the general healthof the embryo.

906 MANN ET AL.

FIG. 1. Total transcript abundance in individual WK- and KK-cultured cloned (c), parthenogenetic (p), tetraploid (t), and in vivo-derived (v) blastocystsfor imprinted genes and one nonimprinted gene. Levels for each gene were classified as no (black), low (blue), medium (yellow), or high (red) expression.A significant number of cloned blastocysts failed to show any expression and were identified by good (G) to poor (P) morphology compared to embryoswith excellent (E) morphology. Embryos were cultured initially in Whitten media or in KSOM without amino acids and then switched at the 8-cell stageto KSOM augmented with amino acids. Clones, B6XCAST cumulus cell nuclei injected into B6D2 oocytes from which spindle-chromosome complexeshad been removed; parthenotes, activated B6(CAST7)XB oocytes; tetraploids, B6XCAST cumulus cell nuclei injected into intact B6D2 oocytes; in vivo-derived, B6(CAST7)XB6 blastocysts; b followed by a number, individual numbered blastocysts.

Parental Allele-Specific Expression of Imprinted Genesin Individual Blastocysts

The expression levels of imprinted genes varied amongindividual cloned blastocysts, but the question remainedwhether imprinted expression was preserved in these em-

bryos. A number of outcomes may be anticipated from theanalysis of allele specificity. First, the somatic cell nucleusmay be reprogrammed, and imprinted expression will mim-ic that of a normal fertilized embryo. Second, the somaticnucleus may not be capable of being reprogrammed andwill maintain the somatic imprint associated with the cu-

907EPIGENETIC ANALYSIS OF CLONED BLASTOCYSTS

FIG. 2. Allelic expression patterns of im-printed genes in individual WK- and KK-cultured cloned blastocysts. Data are pre-sented for embryos that expressed Gapd atmedium to high levels. Black bar heightindicates the percentage of maternal alle-lic expression; gray bar height shows thepercentage of paternal allelic expressionrelative to the total expression of both al-leles. Monoallelic expression was ob-served in cumulus cells for all genes thatare expressed. Fertilized, in vivo-derivedblastocysts display monoallelic expressionfor H19 and Snrpn and biallelic expres-sion of Ascl2 and Igf2r [20, 31–33]. De-tails are as described in Fig. 1.

mulus cell. Third, imprinting may be lost because of in-complete reprogramming or other epigenetic events. To dis-tinguish among these possibilities, embryos were tested forparental allele-specific expression of five imprinted genesthat have various allelic expression patterns in normal blas-tocysts and cumulus cells.

Control blastocysts behaved as expected. Parthenoge-netic blastocysts expressed both maternally derived allelesof the H19, Meg3, Igf2r, and Ascl2 genes (data not shown).No expression was observed for Snrpn in parthenotes. Intetraploids (tBCBD), H19 and Meg3 genes displayed im-printed expression in most cases, whereas the Igf2r andAscl2 genes exhibited a low level of expression from thenormally silent somatic CAST paternal allele (data notshown). The Snrpn gene displayed exclusively imprintedexpression from the somatic paternal CAST allele in thetBCBD embryos.

Parental allele-specific expression was examined incloned blastocysts after exclusion of samples that failed toexpress Gapd at medium to high levels (Fig. 2). Of the 17embryos that expressed H19, nearly all (15 embryos) tran-scribed only the maternal H19 allele. Thus, appropriate pa-rental allele expression was maintained for H19 in most

cloned blastocysts. Furthermore, cloned blastocysts exhib-ited exclusively maternal Meg3 and paternal Snrpn expres-sion (Fig. 2). Thus, genomic imprinting was retained withhigh fidelity for these genes in healthy clones.

Two imprinted genes, Ascl2 and Igf2r, exhibit biallelicexpression in normal blastocysts [31–33], thereby providingan informative test of reprogramming from a monoalleliccumulus pattern to a biallelic mode of expression. In 14cloned blastocysts that expressed Ascl2, 8 clones exhibitedmaternal-only expression, 4 clones displayed biallelic ex-pression, and 2 clones had paternal-only expression (Fig.2), suggesting that the imprint associated with this gene wasnot reprogrammed in a large proportion of embryos. Al-most all cloned blastocysts (91%) that were cultured in WKexhibited a biallelic Igf2r expression pattern, indicating aswitch in mode of expression. This result differed from thatobtained with KK-cultured cloned blastocysts (50% bial-lelic expression).

The allelic expression patterns for these five imprintedgenes revealed that only two cloned embryos exhibited asomewhat normal pattern of imprinted gene regulation (b23WK and b23 KK) (Fig. 2). In summary, a large number ofcloned blastocysts failed to express one or more imprinted

908 MANN ET AL.

genes, suggesting that expression from the somatic nucleuswas not properly activated. In the remaining clones, thesomatic imprint associated with the H19, Meg3, and Snrpngenes appeared to be retained and to direct monoallellicexpression. For the Ascl2 and Igf2r genes, some cloneswere not efficiently reprogrammed (maternal expression),whereas others adopted a biallelic expression pattern.

Allele-Specific Methylation of Imprinting Control Regions

To understand the effects of cloning on imprinted generegulation, it was essential to assess imprinted gene ex-pression in conjunction with DNA methylation analyses.We observed that many of the cloned embryos had lowgene transcript levels, lacked expression, or exhibited al-lelic expression patterns uncharacteristic of blastocysts.These patterns could be the result of an alteration of im-printing marks, because methylation of distinct CpG-richregions around imprinted genes plays an important role inthe regulation of monoallelic expression of these genes. Weanalyzed methylation of regions that determine the imprintto assess directly the epigenetic state of cloned blastocysts.

Methylation of two regions that are essential to the reg-ulation of H19- and Snprn-imprinted expression was as-sayed by bisulfite mutagenesis. The DMD of the H19 geneis paternally hypermethylated [26], whereas the Snrpn pro-moter-exon 1 region is maternally hypermethylated [34].These patterns are present in in vivo-derived blastocysts[26; J. Trasler and M. Toppings, personal communication]and were observed in donor cumulus cell nuclei (Figs. 3–5). In comparison, these alleles were significantly demeth-ylated in cloned blastocysts. Some strands exhibited thetypical pattern of hypermethylation, but a large proportionof strands at both imprinted loci lacked significant meth-ylation (Figs. 3–5). This heterogeneity of methylation ob-served in cloned blastocysts was not found in the cumuluscell population.

The extent of methylation loss was affected by the cultureregime employed. Clones cultured in WK exhibited moredramatic demethylation than the clones cultured in KK.Whereas 30% of DNA strands from WK-cultured clonedblastocysts maintained a hypermethylated pattern on the pa-ternal H19 allele (defined as .50% CpGs on a given strandmethylated), 79% of the paternal alleles from KK-culturedcloned blastocysts were hypermethylated (Fig. 3). A similarloss of methylation (27% hypermethylated strands) was ob-served at the Snrpn locus for cloned blastocysts cultured inWK, but KK-cultured clones showed an intermediate level(55%) of hypermethylated strands (Fig. 4).

Control embryos were examined to determine the effectof nuclear transfer, oocyte activation, and embryo cultureon the DNA methylation patterns of imprinted genes. Invitro-cultured fertilized, parthenogenetic, and tetraploidblastocysts showed less pronounced perturbations in meth-ylation than cloned embryos, except for H19 in KK-cul-tured blastocysts. Considerable methylation was maintainedat the H19 DMD and the Snprn promoter-exon 1 region,although the levels were slightly lower than those detectedin cumulus cells (Figs. 3–5). Thus, whereas culturing em-bryos affects the methylation of imprinted genes, this cul-ture effect does not fully explain the patterns of hypome-thylation seen in cloned blastocysts, nor can the nucleartransfer and activation processes fully account for the hy-pomethylation of cloned blastocysts.

A third region of differential methylation was examinedin cumulus cells and in control and cloned blastocysts.

Though not part of the region regulating imprinted expres-sion, the promoter proximal region of the H19 gene dis-plays differential methylation in midgestation embryos butnot in blastocysts [26]. Similar to conceptuses, DNA fromcumulus cells was methylated on both parental alleles, withhigher levels of methylation found on the paternal allele(Fig. 3). Thus, this region provided an additional test ofreprogramming from a differentially methylated cumuluspattern to one of hypomethylation in blastocysts. Controlblastocysts exhibited little or no methylation on the mater-nal and paternal alleles, as has been reported for in vivo-derived blastocysts [26]. Very little methylation remainedin cloned blastocysts, indicating that the somatic nucleushad lost the hypermethylation associated with this region.

DISCUSSION

Our data reveal that cloned embryos exhibit widespreaddefects in imprinted gene regulation on all three levels ex-amined: total transcript abundance, allele specificity of ex-pression, and allelic DNA methylation. Nearly all embryosexhibited some level of abnormality, even clones of highmorphological quality. Considerable heterogeneity wasseen among individual embryos. These observations indi-cate that defects in epigenetic inheritance arise very earlyduring clonal development, that a significant proportion ofclones fail to express any of the imprinted or nonimprintedgenes analyzed, and that epigenetic information associatedwith imprinted genes is not faithfully retained in the ma-jority of cloned embryos. These disruptions are most likelyonly a subset of the range of defects related to the repro-gramming of genes to an embryonic state. The proportionof embryos exhibiting a comparatively normal pattern ofimprinted gene expression at the blastocyst stage is consis-tent with the proportion of live-born cumulus cell clones asobserved by us and others using similar methods [6; un-published observations].

Defects in the Expression Level of Imprinted Genes

Successful cloning can be expected to occur throughproper recapitulation of the normal embryonic pattern ofgene expression. One defect that was observed in the anal-ysis of individual cloned blastocysts was the significantnumber (;30%) that failed to express any of the five im-printed genes, as would be expected for normal embryos.We conclude either that these clones underwent compara-tively little reprogramming of the cumulus cell genome orthat generalized transcription failure occurred, which cor-related with the clone’s compromised morphology and poorGapd expression. In the remaining clones, considerable het-erogeneity was found in the level of expression and in boththe number and identities of expressed genes. Previous ex-amination of gene expression levels in placentae of cesar-ean-delivered pups derived from embryonic stem cell nu-clei also revealed heterogeneity in the levels of expressionof several imprinted genes [35]. Misregulation of one locuswas found to be independent of abnormal expression atother imprinted loci. The similarity between the data pre-sented here and the results from term placentae of embry-onic stem cell clones suggests that differences in epigeneticstate among clones arise early during preimplantation de-velopment. A recent study of Oct4 gene expression sug-gested an early origin for defects affecting expression ofthat gene, with a large number of clones failing to reactivateexpression of Oct4 and Oct4-related genes in individualcloned blastocysts [8, 36]. Thus, our data, together with

909EPIGENETIC ANALYSIS OF CLONED BLASTOCYSTS

FIG. 3. Methylation status of individualDNA strands in the H19 upstream DMD(top) and promoter proximal region (bot-tom) as determined by bisulfite analysis.Unmethylated CpGs are represented asempty circles; methylated CpGs are de-picted as filled circles. Each line denotesan individual strand of DNA. Cumuluscells (BC) and in vitro-cultured (iCB), par-thenogenetic (pB/C), tetraploid (tBCBD),and cloned (WK-cultured clones cCBBDand cBCBD and KK-cultured clones c-BCBD) blastocysts were assayed. Partheno-tes have two maternal alleles. Only datafor the somatic paternal allele are shownfor tetraploids. Fewer strands were ob-tained for this allele, because they are ex-pected at a lower (one in four) frequency.In the DMD, the B6 and D2 alleles have apolymorphism at CpG 8; therefore, onlyCAST alleles have this position indicated.Missing circles (CpGs) are the result ofambiguous sequencing results.

data from others, indicate that most cloned embryos areunable to recapitulate a normal embryonic pattern of geneexpression.

Defects in Allele Specificity of ImprintedGene Expression

Two differing expectations exist for the control of im-printed genes during cloning. For many imprinted genes,cloning should not alter the parental imprints present in thedonor cell genome, because both blastocysts and somaticcells have identical imprints and show monoallelic expres-

sion. For some imprinted genes, however, biallelic expres-sion is expected for normal blastocysts, so recapitulation ofthe embryonic program would be accompanied by a shiftfrom monoallelic to biallelic expression. Examination ofallele-specific expression revealed appropriate imprintedexpression for Meg3 and Snrpn in all expressing embryosand appropriate imprinted expression of H19 in a majorityof embryos. Thus, for these genes, imprinting informationappeared to be retained in cloned embryos. For the Ascl2and Igf2r genes, the expected shift from a monoallelic to abiallelic mode of expression was seen in at least half theembryos, whereas other clones were not efficiently repro-

910 MANN ET AL.

FIG. 4. Methylation status of individualDNA strands in the Snrpn promoter-exon1 region as determined by bisulfite analy-sis. Cumulus cells (BC) and in vitro cul-tured (iCB), parthenogenetic (pB/C and pC/B), tetraploid (tCBBD), and cloned (WK-cultured clones cCBBD and cBCBD andKK-cultured clones cBCBD) blastocystswere assayed. The CAST allele has a poly-morphism at CpG 1; this position is indi-cated for B6 and D2 alleles only. Detailsare as described in Fig. 3.

grammed but, instead, maintained a somatic expression pat-tern. A greater fraction (91%) of embryos displayed bial-lelic Igf2r expression using the WK culture system. In the-ory, a shift to biallelic expression could reflect either truerecapitulation of the biallelic embryonic pattern of expres-sion or simply a loss of imprinting.

These data suggest that cloned embryos were partly suc-cessful at remodeling the somatic nucleus, but an assess-ment of individual clones revealed that rarely did all fivegenes behave in the same manner. In fact, many clonesappeared to lose expression or maintain a somatic patternof imprinted expression for some genes. One cloned blas-tocyst maintained a somatic expression pattern for all fivegenes (b4 KK). Even among those embryos displaying ex-

cellent morphologies, rarely did these embryos exhibit anentirely normal imprinting pattern. Of the 48 cumulus cellclones examined, 2 clones (4%) exhibited a blastocyst pat-tern of imprinted expression for the five genes. This numberapproximates the proportion of embryos that typically sup-port term development [6].

Defects in Imprinted Gene Methylation

Just as imprinted patterns of expression should be re-tained for some genes during clonal development, so tooshould allele-specific DNA methylation. Failure to do socould compromise later development or viability. We foundthat cloned embryos exhibit loss of methylation during de-

911EPIGENETIC ANALYSIS OF CLONED BLASTOCYSTS

FIG. 5. Cumulative data of the H19 upstream region (A) and the Snrpn promoter-exon 1 region (B) of cumulus cells and in WK- and KK-culturedblastocysts. Bar height indicates the fraction of strands that have a methylated CpG at each specific site (numbers correspond to positions depicted inFigs. 3 and 4). Paternal and maternal alleles are depicted by gray and black bars, respectively. Horizontal lines in the cumulus cell graphs indicate theaverage fraction of methylation observed in the in vivo-derived blastocysts.

912 MANN ET AL.

velopment to the blastocyst stage at the H19 DMD regionand at the Snrpn promoter-exon 1 region. The degree ofmethylation loss is much greater than that seen in controlembryos, indicating that the loss is not caused simply byin vitro culture or other procedures but, rather, is a specificattribute of cloned embryos. The loss of methylation at theH19 and Snrpn loci in cloned blastocysts contrasts with themonoallelic pattern of expression observed for these genes.These data indicate that even when substantial loss of DNAmethylation occurs, cloned embryos can maintain appro-priate allele-specific expression. This was a surprising re-sult, because loss of methylation typically is associated withloss of imprinted gene regulation [37]. It should be noted,however, that the precise molecular mechanisms controllingallelic expression and silencing for imprinted genes haveonly been partly illuminated for somatic tissues, with littleinformation documenting the existence or regulation ofthese mechanisms in preimplantation embryos.

One possibility to account for the apparent discrepancybetween loss of DNA methylation and maintenance of al-lele specificity of expression is that partial DNA methyla-tion may be adequate to maintain appropriate expression atthe blastocyst stage. Alternatively, some aspect of somaticcell chromatin structure may persist throughout the preim-plantation period in cloned embryos to maintain allele spec-ificity of expression. Support for the latter model comesfrom the study of H19 methylation in spermatogenic cells[38]. Parental allelic identity was retained in the absence ofdifferential methylation, implicating another epigeneticmodification in identity preservation. In either case, themodifications may not be sufficient to maintain allelic ex-pression patterns when transcription of these genes increas-es after implantation.

Another possible explanation for the apparent discor-dance in imprinted expression and methylation is that theobserved partial DNA methylation pattern reflects the em-bryo composition within the pools. For this explanation toapply, one must assume that the nonexpressing cloneswould have undergone preferential loss of methylation atthese loci whereas cloned blastocysts displaying imprintedexpression would maintain allele-specific methylation. In-terestingly, we found that the H19 promoter proximal re-gion is demethylated in all embryos. Thus, demethylationevents of unique sequences, like the H19 promoter proxi-mal region, may occur readily in all embryos, similar tobovine cloned blastocysts [39], whereas mechanisms thatprotect against inappropriate demethylation events mayonly operate in a select few clones.

If monoallelic expression is retained despite an erosionof differential methylation, this would pose an interestingproblem for cloned embryos, because biallelic expression(i.e., loss of imprinting) likely would manifest rapidly asdevelopment progressed. In this case, only the most healthycloned embryos that preserved imprinted expression andmethylation would be expected to display extended devel-opmental potential. To this point, loss of imprinting has not,to our knowledge, been observed in midgestation clonedconceptuses [40, 41].

Resolution of these alternatives will require single-em-bryo methylation analysis. At this time, however, availabletechnology does not permit comprehensive analysis ofDNA methylation to be applied with great confidence tosingle embryos, especially those embryos of good or poormorphological classes, which are handicapped by severelyreduced cell numbers.

Possible Defects in Nuclear-Cytoplasmic Interactionsin Cloned Embryos

The observed disruptions in DNA methylation and ex-pression of imprinted genes most likely are only part of aspectrum of defects related to reprogramming of epigeneticinformation to an embryonic state. These results suggestthat reprogramming of the donor cell nucleus by the eggcytoplasm may be impaired. In fact, reprogramming of thesomatic genome may be largely stochastic in nature, withthe result that only a small fraction of cloned embryos areable to regulate their genes appropriately. Earlier nucleartransfer studies revealed a striking effect of one-cell stagecytoplasm on cleavage-stage nuclei [42]. It was hypothe-sized that the one-cell stage cytoplasm has the capacity torender chromatin transcriptionally inactive so that it maybe remodeled for zygotic transcription. Such effects offer apossible explanation for the large number of clones thatfailed to initiate specific gene expression both in the presentand in earlier studies [8]. More recent studies have indi-cated that the cloned embryo lacks certain gene-regulatorycapacities that appear to be encoded uniquely by an au-thentic embryonic genome. Cloned preimplantation embry-os exhibited defects in demethylation processes, includingglobal demethylation and demethylation of some repetitiveelements [10–13]. The presence of an oocyte nucleus intetraploid embryos overcomes the inefficient demethylationof satellite regions in cloned blastocysts [43]. One expla-nation for the failure to demethylate cloned embryos maybe the ectopic presence of DNA methyltransferase. Bycounteracting the normal preimplantation demethylationprocesses, the net result would be maintenance of methyl-ation at these repetitive elements. Cloned embryos aber-rantly express the somatic form of DNMT1 [14]. Tetraploidembryos successfully prevent this aberrant expression, in-dicating that the presence of an authentic set of gamete-derived chromosomes is required for correct regulation ofDNMT1 expression. Thus, gamete-derived genomes maybe endowed with protective epigenetic properties that pre-vent inappropriate expression of genes that provide essen-tial regulatory functions. The somatic cell genomes used tomake cloned embryos would lack these important attributesand, as a result, would be expected to undergo an unpre-dictable series of epigenetic modifications. These ooplasmiceffects may reverberate through preimplantation develop-ment, because cloned embryos are also inefficient at nucle-ar uptake of the maternally inherited oocyte form ofDNMT1 at the 8-cell stage [14]. The substantial loss ofmethylation we observed in clones may be partly attribut-able to the failure of cloned embryos to regulate correctlyDNA methyltransferase at the eight-cell stage, suggestingthat the protective mechanism employed by preimplantationembryos to preserve gametic methylation of imprintedgenes failed in cloned blastocysts. Thus, most, if not all,somatic cell clones would be expected to show epigeneticdefects, even those rare clones that develop to advancedstages or birth, as has been seen [17, 35].

A spectrum of outcomes has been predicted based on thesuccess of reprogramming [44]. In the first category, repro-gramming by the egg cytoplasm has failed, resulting in pre-implantation lethality of these clones. In the present study,these embryos likely are characterized by compromisedmorphology, lack of imprinted gene expression, and littleor no Gapd expression. In the second category, the eggcytoplasm has limited success at remodeling the somaticchromatin. Initially, cloned embryos survive implantation,

913EPIGENETIC ANALYSIS OF CLONED BLASTOCYSTS

but ultimately, abnormal phenotypes and/or lethality en-sues. The stage at which these developmental defects arisemay be dependent on the extent of the partial reprogram-ming. We suggest that reprogramming errors are establishedearly during clonal development and that the majority ofclones experience fatal reprogramming errors, resulting inmajor embryonic wastage by the early postimplantationstages. Finally, in the third category, reprogramming iscompleted to an extent that the majority of genes can beengaged in a normal pattern of expression so that devel-opment and survival of cloned animals results. We predictthat all clones surviving to advanced stages or birth fallinto this grouping and that, although imprinting may appearto be normal, the eventual fate of these clones will dependon the number of nonimprinted genes that still harbor re-programming errors. Given the vast degree of embryonicwastage and the poor degree of reprogramming of both im-printed and nonimprinted genes in preimplantation embry-os, it appears that cloned embryos may be rather ineffectiveat reprogramming the donor somatic genome.

ACKNOWLEDGMENTS

We would like to thank Susan Lee and Bela Patel for technical assis-tance; Barney Crum and Brian Caplin for technical advice; Louis Lefeb-vre, Andras Nagy, and Jeff Mann for the Ascl2 polymorphism; JacquettaTrasler for help with the Snrpn methylation assay; and Jacquetta Traslerand Marc Toppings for sharing unpublished data.

REFERENCES

1. Baguisi A, Behboodi E, Melican DT, Pollock JS, Destrempes MM,Cammuso C, Williams JL, Nims SD, Porter CA, Midura P, PalaciosMJ, Ayres SL, Denniston RS, Hayes ML, Ziomek CA, Meade HM,Godke RA, Gavin WG, Overstrom EW, Echelard Y. Production ofgoats by somatic cell nuclear transfer. Nat Biotechnol 1999; 17:456–461.

2. Campbell KH, McWhir J, Ritchie WA, Wilmut I. Sheep cloned bynuclear transfer from a cultured cell line. Nature 1996; 380:64–66.

3. Chesne P, Adenot PG, Viglietta C, Baratte M, Boulanger L, RenardJP. Cloned rabbits produced by nuclear transfer from adult somaticcells. Nat Biotechnol 2002; 20:366–369.

4. Kato Y, Tani T, Sotomaru Y, Kurokawa K, Kato J, Doguchi H, YasueH, Tsunoda Y. Eight calves cloned from somatic cells of a single adult.Science 1998; 282:2095–2098.

5. Onishi A, Iwamoto M, Akita T, Mikawa S, Takeda K, Awata T, Han-ada H, Perry AC. Pig cloning by microinjection of fetal fibroblastnuclei. Science 2000; 289:1188–1190.

6. Wakayama T, Perry ACF, Zuccotti M, Johnson KR, Yanagimachi R.Full-term development of mice from enucleated oocytes injected withcumulus cell nuclei. Nature 1998; 394:369–373.

7. Shin T, Kraemer D, Pryor J, Liu L, Rugila J, Howe L, Buck S, Mur-phy K, Lyons L, Westhusin M. A cat cloned by nuclear transplanta-tion. Nature 2002; 415:859.

8. Boiani M, Eckardt S, Scholer HR, McLaughlin KJ. Oct4 distributionand level in mouse clones: consequences for pluripotency. Genes Dev2002; 16:1209–1219.

9. Chung YG, Mann MR, Bartolomei MS, Latham KE. Nuclear-cyto-plasmic ‘‘tug of war’’ during cloning: effects of somatic cell nucleion culture medium preferences of preimplantation cloned mouse em-bryos. Biol Reprod 2002; 66:1178–1184.

10. Bourc’his D, Le Bourhis D, Patin D, Niveleau A, Comizzoli P, RenardJ, Viegas-Pequignot E. Delayed and incomplete reprogramming ofchromosome methylation patterns in bovine cloned embryos. CurrBiol 2001; 11:1542–1546.

11. Dean W, Santos F, Stojkovic M, Zakhartchenko V, Walter J, Wolf E,Reik W. Conservation of methylation reprogramming in mammaliandevelopment: aberrant reprogramming in cloned embryos. Proc NatlAcad Sci U S A 2001; 98:13734–13738.

12. Kang YK, Koo DB, Park JS, Choi YH, Chung AS, Lee KK, HanYM. Aberrant methylation of donor genome in cloned bovine embry-os. Nat Genet 2001; 28:173–177.

13. Kang YK, Koo DB, Park JS, Choi YH, Kim HN, Chang WK, Lee

KK, Han YM. Typical demethylation events in cloned pig embryos.Clues on species-specific differences in epigenetic reprogramming ofa cloned donor genome. J Biol Chem 2001; 276:39980–39984.

14. Chung YG, Ratnam S, Chaillet JR, Latham K. Abnormal regulationof DNA methyltransferase expression in cloned mouse embryos. BiolReprod 2003; 69:146–153.

15. Eggan K, Akutsu H, Loring J, Jackson-Grusby L, Klemm M, RideoutWM III, Yanagimachi R, Jaenisch R. Hybrid vigor, fetal overgrowth,and viability of mice derived by nuclear cloning and tetraploid embryocomplementation. Proc Natl Acad Sci U S A 2001; 98:6209–6214.

16. Ogonuki N, Inoue K, Yamamoto Y, Noguchi Y, Tanemura K, SuzukiO, Nakayama H, Doi K, Ohtomo Y, Satoh M, Nishida A, Ogura A.Early death of mice cloned from somatic cells. Nat Genet 2002; 30:253–254.

17. Tamashiro KL, Wakayama T, Akutsu H, Yamazaki Y, Lachey JL,Wortman MD, Seeley RJ, D’Alessio DA, Woods SC, Yanagimachi R,Sakai RR. Cloned mice have an obese phenotype not transmitted totheir offspring. Nat Med 2002; 8:262–267.

18. Tanaka S, Oda M, Toyoshima Y, Wakayama T, Tanaka M, YoshidaN, Hattori N, Ohgane J, Yanagimachi R, Shiota K. Placentomegaly incloned mouse concepti caused by expansion of the spongiotrophoblastlayer. Biol Reprod 2001; 65:1813–1821.

19. Ohgane J, Wakayama T, Kogo Y, Senda S, Hattori N, Tanaka S, Yan-agimachi R, Shiota K. DNA methylation variation in cloned mice.Genesis 2001; 30:45–50.

20. Doherty AS, Mann MRW, Tremblay KD, Bartolomei MS, SchultzRM. Differential effects of culture on imprinted H19 expression inthe preimplantation mouse embryo. Biol Reprod 2000; 62:1526–1535.

21. Szabo PE, Mann JR. Biallelic expression of imprinted genes in themouse germ line: implications for erasure, establishment, and mech-anisms of genomic imprinting. Genes Dev 1995; 9:1857–1868.

22. Schmidt JV, Matteson PG, Jones BK, Guan XJ, Tilghman SM. TheDlk1 and Gtl2 genes are linked and reciprocally imprinted. Genes Dev2000; 14:1997–2002.

23. Thorvaldsen JL, Mann MR, Nwoko O, Duran KL, Bartolomei MS.Analysis of sequence upstream of the endogenous H19 gene revealselements both essential and dispensable for imprinting. Mol Cell Biol2002; 22:2450–2462.

24. Davis TL, Trasler JM, Moss SB, Yang GJ, Bartolomei MS. Acquisi-tion of the H19 methylation imprint occurs differentially on the pa-rental alleles during spermatogenesis. Genomics 1999; 58:18–28.

25. Lucifero D, Mertineit C, Clarke HJ, Bestor TH, Trasler JM. Methyl-ation dynamics of imprinted genes in mouse germ cells. Genomics2002; 79:530–538.

26. Tremblay KD, Duran KL, Bartolomei MS. A 59 2-kilobase-pair regionof the imprinted mouse H19 gene exhibits exclusive paternal meth-ylation throughout development. Mol Cell Biol 1997; 17:4322–4329.

27. Szabo PE, Mann JR. Allele-specific expression and total expressionlevels of imprinted genes during early mouse development: implica-tions for imprinting mechanisms. Genes Dev 1995; 9:3097–3108.

28. Wakayama T, Yanagimachi R. Cloning the laboratory mouse. SeminCell Dev Biol 1999; 10:253–258.

29. Heindryckx B, Rybouchkin A, Van Der Elst J, Dhont M. Effect ofculture media on in vitro development of cloned mouse embryos.Cloning 2001; 3:41–50.

30. Ho Y, Wigglesworth K, Eppig JJ, Schultz RM. Preimplantation de-velopment of mouse embryos in KSOM: augmentation by amino acidsand analysis of gene expression. Mol Reprod Dev 1995; 41:232–238.

31. Latham KE, Doherty AS, Scott CD, Schultz RM. Igf2r and Igf2 geneexpression in androgenetic, gynogenetic, and parthenogenetic preim-plantation mouse embryos: absence of regulation by genomic imprint-ing. Genes Dev 1994; 8:290–299.

32. Lerchner W, Barlow DP. Paternal repression of the imprinted mouseIgf2r locus occurs during implantation and is stable in all tissues ofthe post-implantation mouse embryo. Mech Dev 1997; 61:141–149.

33. Rossant J, Guillemot F, Tanaka M, Latham K, Gertenstein M, NagyA. Mash2 is expressed in oogenesis and preimplantation developmentbut is not required for blastocyst formation. Mech Dev 1998; 73:183–191.

34. Shemer R, Birger Y, Riggs AD, Razin A. Structure of the imprintedmouse Snrpn gene and establishment of its parental-specific methyl-ation pattern. Proc Natl Acad Sci U S A 1997; 94:10267–10272.

35. Humpherys D, Eggan K, Akutsu H, Hochedlinger K, Rideout WMIII, Biniszkiewicz D, Yanagimachi R, Jaenisch R. Epigenetic instabil-ity in ES cells and cloned mice. Science 2001; 293:95–97.

36. Bortvin A, Eggan K, Skaletsky H, Akutsu H, Berry D, YanagimachiR, Page DC, Jaenisch R. Incomplete reactivation of Oct4-related genes

914 MANN ET AL.

in mouse embryos cloned from somatic nuclei. Development 2003;130:1673–1680.

37. Li E, Beard C, Jaenisch R. Role for DNA methylation in genomicimprinting. Nature 1993; 366:362–365.

38. Davis TL, Yang GJ, McCarrey JR, Bartolomei MS. The H19 meth-ylation imprint is erased and reestablished differentially on the paren-tal alleles during male germ cell development. Hum Mol Genet 2000;9:2885–2894.

39. Kang YK, Park JS, Koo DB, Choi YH, Kim SU, Lee KK, Han YM.Limited demethylation leaves mosaic-type methylation states incloned bovine preimplantation embryos. EMBO J 2002; 21:1092–1100.

40. Humpherys D, Eggan K, Akutsu H, Friedman A, Hochedlinger K,

Yanagimachi R, Lander ES, Golub TR, Jaenisch R. Abnormal geneexpression in cloned mice derived from embryonic stem cell and cu-mulus cell nuclei. Proc Natl Acad Sci U S A 2002; 99:12889–12894.

41. Inoue K, Kohda T, Lee J, Ogonuki N, Mochida K, Noguchi Y, Ta-nemura K, Kaneko-Ishino T, Ishino F, Ogura A. Faithful expressionof imprinted genes in cloned mice. Science 2002; 295:297.

42. Latham KE, Garrels JI, Solter D. Alterations in protein synthesis fol-lowing transplantation of mouse 8-cell stage nuclei to enucleated 1-cell embryos. Dev Biol 1994; 163:341–350.

43. Kang YK, Koo DB, Park JS, Choi YH, Lee KK, Han YM. Influenceof oocyte nuclei on demethylation of donor genome in cloned bovineembryos. FEBS Lett 2001; 499:55–58.

44. Rideout WM III, Eggan K, Jaenisch R. Nuclear cloning and epigeneticreprogramming of the genome. Science 2001; 293:1093–1098.