JMed Genet 1997;34:353-359

Original articles

Epigenetic modification and uniparentalinheritance ofH 19 in Beckwith-Wiedemannsyndrome

Daniel Catchpoole, Wayne W K Lam, Debbie Valler, I Karen Temple, Johanna A Joyce,Wolf Reik, Paul N Schofield, Eamonn R Maher

Cambridge UniversityDepartment ofPathology, TennisCourt Road,Cambridge, UKD CatchpooleWW K LamE R Maher

Division ofMedicalGenetics, Departmentof Paediatrics andChild Health,University ofBirmingham,Birmingham Women'sHospital, BirminghamB15 2TF, UKW W K LamE R Maher

Molecular GeneticsLaboratory,Addenbrooke's NHSTrust, Cambridge, UKD Valler

Wessex RegionalGenetics Service,SouthamptonUniversity HospitalsTrust, Southampton,UKI K Temple

Laboratory of StemCell Biology,Department ofAnatomy, DowningStreet, CambridgeUniversity,Cambridge, UKJ A JoyceP N Schofield

Laboratory ofDevelopmentalGenetics andImprinting, TheBabraham Institute,Cambridge, UKW Reik

Correspondence to:Professor Maher,Birmingham.

Received 30 August 1996Revised version accepted forpublication 8 January 1997

AbstractBeckwith-Wiedemann syndrome (BWS)is a congenital overgrowth syndrome asso-ciated with a characteristic pattern of vis-ceromegaly and predisposition tochildhood tumours. BWS is a geneticallyheterogeneous disorder; most cases aresporadic but approximately 15% are fa-milial and a small number of BWSpatients have cytogenetic abnormalitiesinvolving chromosome lIpIS. Genomicimprinting effects have been implicated infamilial and non-familial BWS. We haveinvestigated the molecular pathology of106 sporadic BWS cases; 17% (14/83) ofinformative cases had uniparental disomy(UPD) for chromosome llpl5.5. In eachcase UPD appeared to result from apostzygotic event resulting in mosaicismfor segmental paternal isodisomy. Thecritical region for isodisomy was refinedto a 25 cM interval between DllS86I andDl1S2071 which contained the IGF2, H19,and p57KIP2 genes. In three cases isodisomyfor llq markers was detected but this didnot extend further than lq13-q21 sug-gesting that complete chromosome 11 dis-omy may not produce a BWS phenotype.The allele specific methylation status ofthe H19 gene was investigated in 80sporadic BWS cases. All 13 cases withUPD tested displayed hypermethylationconsistent with an excess of paternal H19alleles. In addition, five of 63 (8%) caseswith normal biparental inheritance hadH19 hypermethylation consistent with an"imprinting centre" mutation (ICM) or"imprinting error" (IE) lesion. The phe-notype of patients with putative ICM/IEmutations was variable and overlappedwith that ofnon-UPD sporadic BWS caseswith normal H19 methylation. However,exomphalos was significantly (p<0.05)more common in the latter group. Thesefindings may indicate differential effectson the expression of imprinted genes inchromosome 1lplS according to the pre-cise molecular pathology. Analysis of H19

methylation is useful for the diagnosis ofboth UPD or altered imprinting in BWSand shows that a variety of molecularmechanisms may cause relaxation ofIGF2imprinting in BWS.(JMed Genet 1997;34:353-359)

Keywords: Beckwith-Wiedemann syndrome; Hl9; uni-parental inheritance

Genomic imprinting is a recently discoveredmechanism by which the activity of genes maybe altered epigenetically in a manner depend-ent on the parent of origin of each of the twoalleles. Further work has implicated genes sub-ject to imprinting in the control of embryonicgrowth and, of the genes now shown to beimprinted, eight have been shown to affectsome aspect of cellular or organismal growth.'Imprinted genes tend to be clustered in specificchromosomal locations, and currently there isevidence that such imprinted loci may be con-trolled by cis acting "imprinting centres" whichare involved either in the maintenance orinitiation of the imprint.'The observation that the penetrance of some

human genetic diseases was subject to a non-Xlinked parent of origin effect led to the realisa-tion that dysregulation of the normal pattern ofimprinting in certain key genes may lead tohyperplastic/trophic or frank neoplasticdisease.3 Three human diseases, Angelman,Prader-Willi, and Beckwith-Wiedemann(BWS) syndromes have been extensively stud-ied as paradigms of disordered genomicimprinting.BWS is a congenital overgrowth syndrome

characterised by the association of gigantism,macroglossia, and visceromegaly with a varietyof developmental anomalies such as exompha-los, umbilical hernias, hemihypertrophy, geni-tourinary abnormalities, and a predispositionto embryonal tumours.' The genetics of BWSare complex. Most cases are sporadic, butapproximately 15% are familial and a smallnumber have chromosomal aberrations ofchromosome lIp. Clinical and moleculargenetic studies suggest that the BWS gene, or

353

Catchpoole et al

genes, are imprinted as, in affected families,penetrance is usually more complete withmaternal inheritance.6 Genomic imprintingeffects have also been implicated in sporadicforms of the disease with a variety ofmechanisms being implicated,8-'6 including (1)uniparental disomy for 1 Ip, (2) paternallyinherited duplications of l 1pl 5, (3) maternallyinherited balanced rearrangements of chromo-some I 1, and (4) putative "imprinting centremutations" (ICM) or imprinting errors (IE), inwhich biallelic IGF2 expression is associatedwith hypermethylation of the IGF2 and H19loci. IGF2 is paternally expressed in humansand mice, and while there is no evidence thatcirculating IGF-II is normally rate limiting onfetal growth,'7 situations in which it is deletedfrom the genome result in marked fetal growthreduction.'8 Analysis of tissue targeted expres-sion of IGF-II suggests that local expression isa key factor in its biological effects.'9 A recipro-cal phenotype is seen with the equivalentmanipulations ofthe maternally expressed H 19gene which maps -100 kb telomeric of IGF2.Regulation of IGF2 and HI 9 expression maybe closely and reciprocally linked. Maternalinheritance of a deletion of the HI 9 gene andflanking sequences causes expression from thematernal IGF2 allele (which is normallysilenced).20

In humans and mice, Hi9 and IGF2 showallele specific differences in DNAmethylation," 21-25 and in previous studies we

have shown an increase in DNA methylation atthe IGF2/H 19 loci in some sporadic BWS

26patients. Initially hypermethylation was onlydetected in disomic cases, but recently a smallsubgroup of non-disomic BWS patients withIGF2/H 19 methylation changes wereidentified.'6 These patients had the combina-tion of a paternal methylation pattern on thematernal chromosome and biallelic IGF2expression. It was proposed that these patientshad mutations in a putative "imprintingcentre" or errors in imprinting mechanisms. Inaddition, Morison et af27 have reported threepatients with somatic overgrowth and neph-romegaly or Wilms' tumour who had biallelicIGF2 expression but did not satisfy thediagnostic criteria for BWS. All three cases hadpartial hypermethylation at HI 9.

In the light of increasing evidence for theheterogeneity of BWS, we have further investi-gated the origin, frequency, significance, andassociated phenotype of H 19 methylationabnormalities in 106 sporadic BWS patients,the largest sample reported to date. Thesestudies confirmed the usefulness ofHI 9 meth-ylation analysis for the molecular diagnosis ofBWS, identified a further two cases with anICM/IE, and refined the critical region for dis-omy to a 25 cM interval between D 1 1 S861 andDlIS2071.

MethodsPATIENTSA total of 106 subjects (51 male, 55 female)with sporadic BWS were investigated. BWSwas diagnosed according to previously definedcriteria: (1) three major features (anterior

abdominal wall defects, macroglossia andpre/postnatal growth >90th centile), or (2) twomajor features plus three or more of: character-istic ear signs (ear lobe creases or posteriorhelical ear pits), facial naevus flammeus,hypoglycaemia, nephromegaly, and hemi-hypertrophy.' All cases were sporadic with nofamily history of BWS, and only one patienthad a cytogenetic abnormality, a male with apaternally derived duplication of i ipi 5.5.Peripheral blood for DNA analysis was ob-tained from each affected child and theirparents. Of these patients, 49 were included ina previous study of uniparental disomy (UPD)in BWS'2 with 34 of these being part of aninvestigation into H 1 9/IGF2 methylation.26

HI 9 METHYLATION ANALYSIS

High molecular weight DNA was isolated fromperipheral blood by standard methods.28 Foranalysis of Hi9 methylation, DNA samples(8-10 ,ug) were digested with a large excess ofPstI and SmaI as described previously.26 AfterSouthern analysis and hybridisation with aHi 9 cDNA probe, the Hi 9 methylation status,as designated by methylation indices (MI), wasassessed by densitometry of autoradiographsand comparing the ratio of the 1.8 kb (uncutmethylated) and 1.0 kb (cut unmethylated)fragments for deviation from a 1:1 ratio. MIvalues of less than 0.6 had previously beenconsidered to be the upper limit for a normalpopulation.26

MOLECULAR GENETIC INVESTIGATIONAll patients were screened for evidence ofdisomy at the tetranucleotide microsatellitepolymorphism in tyrosine hydroxylase (TH)and ApaI RFLP within exon 9 of IGF228 29 inchromosome lip15.5. TH was amplified inPCR reactions as reported previously." To testfor the ApaI RFLP site in the IGF2 gene DNA(100 ng) was amplified using 50 pmol primers(IGF2-F 5'-CTTGGACTTTGAGTCAAA-TTGG-3'; IGF2-R 5'-CCTCCTTTGGT-CTTACTGGG-3') in 25 gl mixtures contain-ing 10 mmol/l Tris/HCl, pH 9.0, 50 mmol/lKC1, 1.5 mmol/l MgCl,2 0.01 w/v% gelatin, 0. 1v/v% triton X-100, 250 jimol/l each dNTP, and0.5 U Taq polymerase. Thermocycling condi-tions involved 30 cycles of 60 seconds dena-turation at 92°C, 60 seconds annealing at55°C, and two minutes for elongation at 72°C,concluding with a final extension period of fiveminutes.To increase the informativeness for UPD

detection, the extent of disomy was investi-gated using a series of microsatellite and RFLPmarkers from chromosome 1 i p and 11 q. TheRsaI polymorphic fragment in exon 5 of H19was amplified using 50 pmol of each primer(p3104F 5'-AGATTCAAAGCCTCCACGA-3'; p3461R 5'-AGTGTTTATTGATGAT-GAGTCCAG-3') in the same reaction mixtureas detailed above with the addition of: 10% v/vDMSO, 10% v/v glycerol, and 0.05% w/v poly-oxyethylene ether (W-1) (Sigma; Poole, Dor-set). In this case, thermocycling involved aninitial denaturation period of three minutes at94°C followed by 30 cycles of 96°C for 30 sec-

354

355Inheritance ofHl 9 in Beckwith-Wiedemann syndrome

1 5.515.415.315.215.1

14

DI 1 S2071H19IGF2

- TH

D1 S861

D11S92913 WT1121 1.211.12 D11S1344

11.11 Dl1S34

1 1

12in13.1 D11S95613.213.313.4135-14.114.214.3

21 D1 1S89822.122.222.323.123.223.3

2425

'94 115 131 9 51* 0@000

* 0* @ 0

0 00 0

0

153 85 101 277 98 126@0ID~ ~

@ 00 0 @0

* 0

@ 0

in *k * A

V

00

* Isodison'icUninifornmativeNormal inheritance

I 1

Figure1 Map showing UPD in BWS patients as indicated by analysis of various chromosome 11 microsatellite (DllS-numbers) and RFLP polymorphic markers. Patient identification numbers are indicated at the top of each column. TheHl 9, IGF2, and TH loci were within the disomic region in all cases and disomy extended onto 1 1q in three patients (194,115, and 277).

onds, 52°C for 60 seconds, and 72°C for twomintes, concluding with a 10 minute extensionat 72°C. Following the PCR of either IGF2 or

HI 9 fragments, 10.5 ml product was digestedwith 30 U ApaI (10 U/gl) or RsaI (10 U/jl)respectively in the appropriate reaction buffersaccording the supplier's conditions. Bothdigested and non-digested samples were elec-trophoresed through 2% agarose in TAE withDNA bands being visualised by ethidium bro-mide staining.

Amplification of the CA repeat within theWT1 gene (lp1 3) was performed essentiallyas previously outlined.30 The dinucleotiderepeat polymorphisms D11S861, D11S956,and DI 1S2017 were amplified under condi-tions described by Hudson et al,3' Smith et al,32and Browne et al,3' respectively. For analysis ofthe microsatellite polymorphisms at D 11 S929,D11S1344, and D11S898 the followingprimer sets were used: S929F 5'-AGGCCCTTCCAAGATCAG-3', S929R 5'-CCCAGTTGCCGAACTACC-3'; S1344F 5'-CCCTGAACTTCTGCATTCAC-3', S1344R 5'-

GCGCCTGGCTTGTACATATA-3'; S898F5'- AGCACCATTTGCTGAGACTG-3', S898R 5'-TGTAlYIGTA- TCGATTAACCAACTT-3'.3 DNA (50 ng) was amplified in 13 ,dreactions containing PCR buffer (10 mmol/lTris/HCl, pH 9.0, 50 mmol/l KCI, 1.5 mmol/lMgC12, 0.01% w/v gelatin, 0.1% v/v tritonX-100), 10 pmol of each primer, 125 gmol/l ofeach dNTP, and 0.25 U Taq polymerase. Thesamples were subjected to 30 PCR cycles ofdenaturation at 94°C for 45 seconds, annealingat 60'C for 30 seconds, and extension at 72°C

for 30 seconds. Upon completion, a further fiveminutes 72°C extension was performed.PCR products for each microsatellite marker

were electrophoresed through 7.5 or 10%polyacrylamide (acrylamide:bis ratio 40:1, 1 x

TBE) depending on the size of separationrequired. Gels were cast in a 20 cm x 20 cm x

1 mm vertical gel system and run in 1 x TBE.Samples (10 gl) were mixed with 3 gl (6x) gelloading dye (15% w/v ficoll, 5 mmol/l EDTA,0.1% w/v SDS, 0.25% w/v bromophenol blue,0.25% w/v xylene cyanol) and electrophoresedat 100-120 volts for approximately 16 hours.Microsatellite bands were visualised by silverstaining.

STATISTICAL ANALYSIS

Intergroup differences were compared usingthe chi-squared test with Yates' correction.Statistical significance was taken at the 5%level.

ResultsUNIPARENTAL DISOMY IN BECKWITH-WIEDEMANN

SYNDROME

A total of 69 of 83 (83%) cases informative atTH or IGF2 had normal biparental inheritancewith no evidence of paternal UPD. No cases ofheterodisomy were identified but 14 (17%)cases showed paternal isodisomy (figs 1 and 2).To investigate the origin and extent of paternaldisomy a series of chromosome 11 polymor-phic markers were analysed in these 14 cases.

All showed mosaicism for partial paternalisodisomy suggesting that all UPD cases hadarisen as a postzygotic event. The TH and

140 12 205

0* 0

* 0

0 0

Catchpoole et al

-. fkVI i:

._.V w"

Wl

) S 9 i3

Figure 2 Molecular genetic analysis using chromcmicrosatellite markers and RFLPs. Inheritance ofpalleles in BWS patient 51, showing mosaicism. Tyrhydroxylase (TH) (I1pl5.5) and Wilms' tumour(WTI) (1lp13) arefrom ethidium bromide staineagarose gels, while D11S1344 (lIpl ) and DlJS!(11q13) arefrom PAGE gels stained with silver. Mmother; C, affected child; F1 father.

w - -

Figure 3 Increased Hl9 methylation in BWSpaBW88 and BW39 are non-UPD patients with normethylation patterns (M1=0. 60 and 0.49 respectivBW51 and BW205 were identified as UPD (fig 1Ishow hypermethylation at the H19 locus. BW243 zBW15 represent BWS patients with normal inherichromosome 1 lpl5 alleles but have increased Hl 9methylation (MI=0. 87 and 0. 93 respectively) indian ICM or IE lesion.

IGF2 loci were included in the miudisomic region in all cases and the sregion for disomy was a 25 cM region bD11S861 and D11S2071 (BW131) (Disomy extended onto the long achromosome 11 in three instances (B

* *

*E E

o o

NIPNo~~~~~~~

*~~~~~~ NonUP* UPD

nA1pduplicatc

Non-UPD UPD

Figure 4 H19 methylation indices ofBWS patients. MI were derived as describetMaterials and methods. The patient represented by the triangle was shown to have

paternally derived duplication of I 1pl5.5 (MI=0. 67).

ssome 11)arentalosine1d'956

BW1 15, BW277), but did not extend as far as1 1q21 (D1 lS898) (fig 1).

H19 METHYLATION ANALYSIS

Allele specific methylation of the H19 pro-moter region was examined in the 80 BWSpatient samples in which sufficient DNA wasavailable for Southern analysis. Fig 3 illustratestypical examples; hypermethylation of thematernal allele of H19 is indicated by a reduc-tion in intensity of the 1.0 kb digestionproduct. Methylation indices (MI) were de-rived from 63 patients informative for UPDanalysis. In the 13 patients with paternal UPDanalysed, the MI ranged between 0.65 and0.84 and 40 patients, while most patients inwhom paternal UPD had been excluded had aMI <0.64 (fig 4). A previously undescribedpatient with paternally derived duplication ofllpl5.5 had an MI of 0.67, consistent with a2:1 ratio of paternal (methylated) to maternalH 19 (unmethylated) alleles. Five patients withnormal biparental inheritance, including threereported previously,"6 had a MI greater thanthose found in the UPD cases (>0.84).

- KU CLINICAL PHENOTYPE OF BWS SUBGROUPSIn a previous study we had found thathemihypertrophy was significantly more com-

-1 ki. mon and exomphalos less frequent in UPDcases than in non-UPD patients." The clinicalphenotype of non-UPD patients with(MI>0.84, n=5) and without (MI<0.6, n=19)

1- kl putative ICM/IE were compared. Postnatal

tients. growth >90th centile was similar in bothtmal groups (4/5 and 16/19 respectively); birthxely). weight tended to be higher in cases with H19) and hypermethylation (4/5 and 5/19 respectivelytanceof had a birth weight >90th centile; X2 = 2.85,

o0.05<p<O. 1). In addition, exomphalos was lessccative of common in patients with ICM/IE (0/5 versus

13/19; X2 = 4.96, p<O.05). Wilms' tumouroccurred in 1/5 non-UPD patients with a

nimally MI>0.84, 0/19 non-UPD patients with,mallest MI<0.6, and 1/12 UPD patients.setween(fig 1). Discussionirm of UNIPARENTAL DISOMY IN BECKWITH-WIEDEMANN

3W1945 SYNDROMEUniparental disomy (UPD) in humans hasbeen thought to be caused primarily by meioticnon-disjunction events followed by trisomy ormonosomy "rescue". UPD has been reportedfor 16 human chromosomes and has indicatedthe likelihood of imprinting effects in certaingenome regions.4 35 The proportion of UPDidentified (17%) in our informative BWS casesis comparable with that derived from smallerstudies.9 10 12 To date all BWS patients withUPD have had partial (segmental) and somati-cally mosaic paternal isodisomy. This is in con-trast to paternally derived UPD in Angelmansyndrome, which is usually complete andresults from errors during meiosis. The ab-

Jion sence of complete chromosome 11 isodisomyor heterodisomy could reflect the infrequencyof trisomy 1 1 compared to trisomy 15, or more

d in likely, that the presence of imprinted genes ondan chromosome 1 1 q results in a non-BWS

phenotype in patients with complete isodis-

1.0

0.9

0.8

0.7

0.6

0.5 F

0.4

0.3

0.2

0.1

356

Inheritance ofHi 9 in Beckwith- Wiedemann syndrome

omy. Hence, Webb et alr6 have reported a caseof complete paternal chromosome 11 disomyas a result oftrisomy 11 rescue which presentedwith prenatal growth retardation. Two loci forfamilial non-chromaffin paraganglioma (PGL1and PGL2) have been mapped to chromosome1 q.3 38 Familial paraganglioma only manifestswhen it is inherited from the father, suggestingthat it is caused by mutations in an imprintedpaternally expressed tumour suppressor gene.PGL2 maps to a 5 cM region of 11 qi 3 betweenD1 1S956 and PYGM but three of our patientswere disomic at DI 1S956 (fig 1). PGLI, how-ever, maps to chromosome 1 1 q2 1 which wasbeyond the limits of disomy in our BWSpatients. No other imprinted genes have beenmapped to 11 q, but imprinted genes arefrequently clustered and other genes may liewithin this interval. The presence of imprintedgenes distal to 11 qi 3 may preclude the findingof complete paternal chromosome 11 disomyin people with a BWS phenotype.

Molecular genetic analysis of a BWS associ-ated paternally derived duplication by Weks-berg et al'4 showed that the duplicated regionincluded genes distal to the calcitonin-A(CALCA) locus in llpl5.4. Analysis of thebreakpoint regions of balanced translocationsand pericentric inversions associated withBWS have indicated two breakpoint clusterregions, the most distal of which (BWSCR2)maps centromeric to a cluster of genes alreadyknown to be imprinted in the mouse, IGF2,H19, p57MPg, Ins, all of which are associatedwith embryonic growth. While IGF2 has beensuggested as a candidate for the BWS locus,one familial study has reported exclusion oflinkage to IGF239 and both biallelic and normalmonoallelic IGF2 expression has beenreported.40 Bischoff et al" mapped the minimaldisomic region in one BWS patient to anapproximately 38 cM region betweenD1lS922 (1lpl5.5) and D1IS904 (1lpl4-p13). Our results refine the critical region forBWS still further to a 25 cM interval betweenD1lS861 (1lpl5.1) and the lip telomericmicrosatellite marker DlI S207 1. This regioncontains both IGF2 and HI 9 and theBWSCR2 region defined by Redeker et al.42

Hl9 METHYLATION AND IGF2 EXPRESSION.Biallelic IGF 2 expression has been found intissues from BWS patients,'6 40 and is similarlydysregulated in Wilms' tumours. Inappropriateexpression of IGF2 from the maternal allelehas been shown to correlate with methylationof the promoter region of the neighbouringH 19 gene in Wilms' tumour43 44 and in BWS.16Hypermethylation of HI 9 is associated withtranscriptional inactivity and hypermethylationof a region of exon 9 in the human IGF2 geneis associated with active transcription.'4 Re-moval of methylation by germline deletion ofthe DNA methylase enzyme results in IGF2inactivity and transcription of the normallysilent paternal HI 9 gene,45 and deletion of theH19 gene results in IGF2 overexpression.'0 Inaddition, a small number of BWS patients(5-10%) have normal biallelic inheritance andH 19 hypermethylation which exceeds that

found in mosaic UPD patients. By analogywith similar findings in Angelman/Prader-Willipatients," it has been proposed that thesechanges may result from alteration of animprinting control region or imprintingerrors.'6 In this context all five BWS patientsidentified so far have been sporadic cases, sothere is, as yet, no definitive evidence as towhether possible ICM/IE lesions are genetic orepigenetic.To date all three BWS patients with an

ICM/IE who are informative for IGF2 poly-morphic marker studies have shown biallelicIGF2 expression'6 (unpublished observations).However, the clinical phenotype of thesepatients is not uniform and although four hadclassical BWS features, one patient displayedan incomplete phenotype with mild macroglos-sia and an umbilical hernia but normal prena-tal growth. Recently Morison et al'7 havedescribed non-classical BWS overgrowth pa-tients with biallelic IGF2 expression and H19hypermethylation. However, the degree ofhypermethylation was less than that seen in ourICM/IE patients. In addition to the fivepatients with markedly increased H 19 hyper-methylation (MI>0.84), several non-disomicpatients had methylation indices slightly higherthan the previously described upper limit ofnormal (0.6). Morison et al'7 suggested thatpatients with partial hypermethylation, such asthese, may be mosaic for imprinting mutations.In addition to putative ICM or IEs causingaberrent Hi 9/IGF2 methylation patterns andIGF2 overexpression, we have also detectedbiallelic IGF2 expression in some non-UPDBWS patients with normal H 19 methylation(Joyce et al, submitted) and Brown et al'7 havedescribed similar findings with normal H 19expression in a BWS family with a chromo-some II inversion. These findings indicate thata variety of Hi 9 dependent and independentmechanisms may lead to biallelic IGF2 expres-sion in BWS.Although biallelic IGF2 expression has been

reported in complete and incomplete forms ofsporadic BWS, normal imprinting of the IGF2gene has also been reported,40 perhaps indicat-ing that loss ofIGF2 imprinting is not the onlymechanism from which a BWS phenotype canresult. In addition to IGF2 and H 19, otherimprinted genes such as p57KIP2 map to theminimally duplicated/disomic region.48 Thephenotype of BWS is variable and so theparticular features may depend on the extent ofdisomy or the precise pathogenetic mech-anism, if this resulted in differential changes inthe expression of other imprinted genes.

PHENOTYPE/GENOTYPE CORRELATIONThe identification of clear genotype/phenotypecorrelations would enhance clinical manage-ment and provide evidence to support thehypothesis that the precise BWS phenotypemay be determined by the specific molecularpathology ofBWS. Two studies have reported ahigh incidence of Wilms' tumour in BWSpatients with UPD (40% and 50%respectively)'°" and our results continue tosuggest that although Wilms' tumour may be

357

Catchpoole et al

more common in children with BWS, theactual risk (-8%) is lower than previousestimates, which may have reflected an ascer-tainment bias. In addition, the occurrence of aWilms' tumour in a non-UPD BWS patientshows that the tumour risk is not confined toBWS UPD patients. We have previouslyobserved that hemihypertrophy was signifi-cantly more common and exomphalos was lesscommon in the disomic group than thenon-UPD group. These findings were attrib-uted to the the mosaic nature ofUPD, so it wassurprising that there was evidence to suggestthat exomphalos was less common in non-UPD cases with H19 hypermethylation than inthose with normal methylation. This may indi-cate that exomphalos is not simply a conse-quence of increased intra-abdominal pressureowing to hypertrophy, but may involve the fail-ure of specific processes such as cell adhesionor migration which could be affected by altera-tions in the activity of other imprinted genes atthe locus. Further studies are needed toconfirm these findings, to determine the risk ofneoplasia in patients with ICMIIE, and to elu-cidate the mechanisms for these genotype-phenotype correlations. Knowledge of theimprinting status of genes such as Hi9 andp57KIP2 in patients with UPD and ICM/IEcompared to non-UPD without H19 methyla-tion might provide insight into the cause ofthese associations. It is interesting to note thatH 19 expression is low or absent in the class ofpatients with a low incidence of exomphalos(that is, UPD and ICM/IE patients).

In summary, we have shown that H 19 meth-ylation analysis is a reliable method for detect-ing UPD in BWS patients and also detectssmall numbers of patients with "imprintingcentre mutations" (5-10%). We studied meth-ylation status in blood DNA and it is possiblethat H19 methylation may vary betweentissues. However, to date, the results of H19methylation analysis in cultured fibroblasts(n=3) from non-UPD patients is consistentwith those obtained from blood DNA. Thepostzygotic origin of paternal UPD in BWSmay indicate that complete isodisomy or

heterodisomy of chromosome 11 has a lowviability or non-BWS phenotype. Carefulcorrelation of the clinical phenotype with themolecular pathology may provide an insightinto the aetiology of the variable expression ofBWS and determine what screening should beoffered to BWS children. Definition of thecritical region of chromosome 1 p for UPD inBWS will provide a basis for understanding theaetiology of BWS, and the detailed comparisonof the clinical features of BWS patients withvariable lengths of segmental isodisomy mayidentify target regions for other imprintedgenes on chromosome 11. Similarly, furtherstudies of the phenotype and heterogeneity ofmolecular mechanisms in non-BWS UPD mayprovide critical insights into the role of IGF2,H19, p57KJP2, and other imprinted genes invarious facets of the BWS phenotype.

We thank the many colleagues who referred patients and the

Wellcome Trust and East Anglian Regional Health Authority

for financial support. WR acknowledges the support of ActionResearch.

1 Barlow DO. Gametic imprinting in mammals. Science1995;270:1610-13.

2 Efstradiatis A. Parental imprinting of autosomal mamma-lian genes Curr Opin Genet Dev 1994;4:265-80.

3 Reik W. Genomic imprinting and genetic disorders in man.Trends Genet 1989;5:331-6.

4 Tycko B. Genomic imprinting; mechanism and role inhuman pathology. Am J Pathol 1994;144:431-43.

5 Elliott M, Maher ER. Beckwith-Wiedemann syndrome. JMed Genet 1994;31:560-4.

6 Moutou C, Junien C, Henry I, Bonaiti-Pellie C. BeckwithWiedemann syndrome: a demonstration of the mecha-nisms responsible for the excess of transmitting females. JMed Genet 1992;29:217-20.

7 Viljoen D, Ramesar R. Evidence for paternal imprinting infamilial Beckwith-Wiedemann syndrome. J Med Genet1992;29:221-5.

8 Henry I, Bonaiti-Pellie C, Chehensse V, et al. Uniparentalpaternal disomy in a genetic cancer-predisposing syn-drome. Nature 1991;351:665-7.

9 Nystrom A, Cheetham JE, Engstr6m W, Schofield PN.Molecular analysis of patients with Wiedemann-Beckwithsyndrome II: paternally derived disomies of chromosome11. EurJrPediatr 1992;151:511-14.

10 Henry I, Puesch A, Riesewijk A, et al. Somatic mosaicism forpartial paternal isodisomy in Wiedemann-Beckwithsyndrome: a post fertilisation event. Eur J Hum Genet1993;1: 19-29.

11 Schneid H, Seurin D, Vazquez MP, Gourmelen M, CabrolS, Le Bouc Y. Parental allele specific methylation of thehuman insulin-like growth factor II gene and Beckwith-Wiedemann syndrome. J Med Genet 1993;30:353-62.

12 Slatter RE, Elliott M, Welham K, et al. Mosaic uniparentaldisomy in Beckwith-Wiedemann syndrome. J Med Genet1994;31:749-53.

13 Brown KW, Gardner A, Williams JC, Mott MG, McDer-mott A, Maitland NJ. Paternal origin of I1 p 15 duplicationsin the Beckwith-Wiedemann syndrome. A new case andreview of the literature. Cancer Genet Cytogenet 1992;58:66-70.

14 Weksberg R, Teshima I, Williams B, et al. Molecular charac-terisation of cytogenetic alterations associated with theBeckwith-Wiedemann syndrome phenotype refines thelocalisation and suggests the gene for BWS is imprinted.Hum Mol Genet 1993;2:549-56.

15 Norman AM, Read AP, Clayton-Smith J, Andrews T, Don-nai D. Recurrent Wiedemann-Beckwith syndrome withinversion of chromosome (I1)(pll.2p15.5). Am J MedGenet 1992;42:638-41.

16 Reik W, Brown K, Schneid H, Le Bouc Y, Bickmore W,Maher ER. Imprinting mutations in the Beckwith-Wiedemann syndrome suggested by an altered imprintingpattern in the IGF2-H19 domain. Hum Mol Genet 1995;4:2379-85.

17 Gluckman PD, Ambler GR What is the function of circulat-ing IGF-II in postnatal life? Mol Cell Endocrinol 1993;92:C1-3.

18 de Chiara TM, Efstradiatis A, Robertson E. Parentalimprinting of the mouse insulin-like growth factor II gene.Cell 1991;64:849-59.

19 Ward A, Bates P, Fisher R, Richardson L, Graham C.Disproportionate growth in mice with igf-2 transgenes.Proc Nad Acad Sci USA 1994;91:10365-9.

20 Leighton PA, Ingram RS, Eggenschwiler J, Efstratiadis A,Tilghman SM. Disruption of imprinting caused by deletionof the H19 gene region in mice. Nature 1995;375:34-9.

21 Sasaki H, Jones PA, Chillet JR, et al. Paternal imprinting:potentiallly active chromatin of the repressed allele of themouse igf2 gene. Genes Dev 1992;6:1843-56.

22 Bartolomei, MS, Webber AL, Brunkow ME, Tilghman SM.Epigenetic mechanisms underlying the imprinting of themouse H19 gene Genes Dev 1993;7:1663-73.

23 Ferguson-Smith AC, Sasaki H, Cattanach BM, Surani A. Aparental-origin-specific epigenetic modification of themouse H19 gene. Nature 1993;362:751-5.

24 Feil R, Walter J, Allen ND, Reik W. Developmental controlof allelic methylation in the imprinted mouse IGF2 andH19 genes. Development 1994;120:21933-43.

25 Brandeis M, Kafri T, Ariel M, et al. The ontogeny of allelespecific methylation associated with imprinted genes in themouse. EMBO_ 1993;12:3669-77.

26 Reik W, Brown KW, Slatter RE, Sartori P, Elliott M, MaherER. Allelic methylation of H 19 and IGF2 in the Beckwith-Wiedemann syndrome. Hum Mol Genet 1994;3:1297-301.

27 Morison IM, Becroft D, Taniguchi T, Woods CG, Reeve A.Somatic overgrowth associated with overexpression ofinsulin-like growth factor II. Nat Med 1996;2:311-16.

28 Sambrook J, Fritsch EF, Maniatis T. Molecular cloning - a

laboratory manual. New York: Cold Spring Harbor LaborT-tory Press, 1989:31-9.

29 Todokoro K, Fujii H, Inoue T, Yamada M. Polymerasechain reaction (PCR) for detection of Apal polymorphismat the insulin like growth factor II gene (IGF2). NucleicAcids Res 1991;19:6967.

30 Haber DA, Buckler AJ, Glaser T, et al. An internal deletionwithin an lp 13 zinc finger gene contributes to the dvelop-ment of Wilms' tumour. Cell 1990;61:1257-69.

358

Inheritance ofHl 9 in Beckwith- Wiedemann syndrome

31 Hudson TJ, Engelstein M, Lee MK, et al. Isolation andchromosomal assignment of 100 highly informative humansimple sequence repeat polymorphisms. Genomics 1992;13:622-7.

32 Smith MW, Clark SP, Hutchinson JS, et al. A sequence-tagged site map of human chromosome 11. Genomics 1995;17:699-725.

33 Browne DL, Smith EA, Dietz-Band J, Reithmann HC,Phromchotikul T, Litt M. Dinucleotide repeat polymor-phism at the human chromosome 1 Ip teleomere(D 11 S2071). Genomics 1995;25:600- 1.

34 Coullin P, Guern EL, Vignal A, et al. Assignment of 112microsatellite markers to 23 chromosome 11 subregionsdelineated by somatic hybrids: comparison with the geneticmap. Genomics 1994;21:379-87.

35 Ledbetter DH, Engel E. Uniparental disomy in humans:development of an imprinting map and its implications forprenatal diagnosis. Hum Mol Genet 1995;4:1757-64.

36 Webb A, Beard J, Wright C, Robson S, Wolstenholme JIGoodship J. A case of paternal uniparental disomy forchromosome 11. Prenat Diagn 1995;15:773-7.

37 Heutink P, van der Mey AGL, Sandkuijl LA, et al. A genesubject to genomic imprinting and responsible for paragan-gliomas maps to chromosome 1 lq23-qter. Hum Mol Genet1992;1:7-10

38 Mariman ECM, van Beersum SEC, Cremmers CWRJ,Struycken PM, Ropers HH. Fine mapping of a putativelyimprinted gene for familial non-chromaffin paragangliomasto chromosome 11 ql 3.1: evidence for genetic heterogene-ity. Hum Genet 1995;95:56-62.

39 Nystrom A, Hedborg F, Ohlsson R. Insulin-like growth fac-tor 2 cannot be linked to a familial form of Beckwith-Wiedemann syndrome. Eur7 Pediatr 1994;153:574-80.

40 Weksberg R, Shen DR, Fei YL, Song QL, Squire J. Disrup-tion of insulin-like growth factor 2 imprinting in BeckwithWiedemann syndrome. Nat Genet 1993;5:143-50.

41 Bischoff FZ, Feldman GL, McCaskill C, Subramanian S,Hughes MR, Shaffer LG. Single cell analysis demonstrat-ing somatic masaicism involving llp in a patient withpaternal isodisomy and Beckwith-Wiedemann syndrome.Hum Mol Genet 1995;4:395-9.

42 Redeker E, Alders M, Hoovers JMN, Richard CW III,Westerveld A, Mannens M. Physical mapping of 3candidate tumour suppressor genes relative to BWS associ-ated chromosomal breakpoints at lp 15.3. Cytogenet CellGenet 1995;68:22-5.

43 Moulton T, Crenshaw T, Hao Y, et al. Epigenetic lesions atthe HI 9 locus in Wilms' tumour patients. Nat Genet 1994;7:440-7.

44 Steenman MJC, Rainier S, Dobry CJ, Grundy P, Horon I,Feinberg, AP. Loss of imprinting of IGF2 is linked toreduced expression and abnormal methylation of H19 inWilms tumour. Nat Genet 1994;7:433-9.

45 Li E, Beard C, Jaenisch R. Role for DNA methylation ingenomic imprinting. Nature 1993;366:362-5.

46 Buiting K, Saitoh S, Gross S, et al. Inherited microdeletionsin the Angelman and Prader-Willi syndromes define animprinting centre on human chromosome 15. Nat Genet1995;9:3985-401.

47 Brown KW, Villar AJ, Bickmore W, et al. Imprintingmutation in the Beckwith-Wiedemann syndrome leads tobiallelic IGF2 expression through a H 19-independentpathway. Hum Mol Genet 1996;5:2027-32.

48 Matsouka et al. Proc NatlAcad Sci USA 1996;93:3026-30.

359