Parental allele specific II Beckwith-Wiedemannjmg.bmj.com/content/jmedgenet/30/5/353.full.pdfof a complete form of Beckwith-Wiedemann ... 120 units ofrestriction endonuclease, accord-

J Med Genet 1993; 30: 353-362

ORIGINAL ARTICLES

Parental allele specific methylation of the humaninsulin-like growth factor II gene andBeckwith-Wiedemann syndrome

Helene Schneid, Danielle Seurin, Marie-Paule Vazquez, Micheline Gourmelen,Sylvie Cabrol, Yves Le Bouc

AbstractIn an attempt to elucidate the role ofmethylation in parental imprinting atthe IGF-II gene locus, for which imprint-ing has already been described in themouse; we undertook an allele specificmethylation study of the human IGF-IIgene (mapped to 1lpl5.5) in a controlpopulation and in patients with Beck-with-Wiedemann syndrome.

In control leucocyte DNA (16 unrelatedadults and eight families), the maternalallele of the IGF-II gene was specificallyhypomethylated, whereas no such allelespecific methylation was found for eitherthe insulin or the calcitonin genes whichare located in llpl5.5 and lpl5.1, re-spectively. Furthermore, the IGF-II genespecific hypomethylation was localisedon the 5' portion of exon 9.

In the patients with Beckwith-Wiede-mann syndrome in which the IGF-II geneis thought to be involved and wherepaternal isodisomy has been described,hypomethylation of the maternal allelewas conserved in leucocyte DNA, but ab-normal methylation was detected in mal-formed tissues where the paternal allelewas also demethylated. Some specificmechanism linked to methylation there-fore seems to be involved in the patho-genesis of Beckwith-Wiedemann syn-drome.(J7 Med Genet 1993;30:353-62)

Various family studies have indicated thatthere is a link between Beckwith-Wiedemannsyndrome and the llpl5 region of chromo-some 1112 which also carries the insulin-likegrowth factor II (IGF-II) gene.3 This syn-drome causes childhood malformation and ischaracterised by a number of congenital dis-orders (neonatal gigantism, omphalocele, mac-roglossia, visceromegaly, hemihypertrophy,and neonatal hypoglycaemia) and is frequentlyassociated with tumours (nephroblastoma,adrenocortical carcinoma, neuroblastoma, andhepatoblastoma).45 Expression of IGF mRNAhas been seen to increase in a variety of tu-mours '5 and IGF-II is known to play a pivo-tal role in cell proliferation'6 and embryonicdevelopment.'7 In addition, DeChiara et al'8

have recently used homologous recombinationto show that the murine IGF-II gene is subjectto differential parental imprinting. Themechanisms involved in the regulation of thisimprinting remain largely unknown, but someobservations indicate a relationship betweenparental imprinting and methylation of thealleles.'1820We have therefore analysed the genomes of a

control population and of children with Beck-with-Wiedemann syndrome and their familiesin order to determine whether DNA methyla-tion might be linked to differential parentalimprinting of the human IGF-II gene.

PatientsTwenty-seven unrelated subjects and thebrothers (P3C2 and P16C2) of two patients(P3C1 and P16C1) were analysed (15 boys and14 girls). Their clinical findings are listed intable 1 which shows the major anomalies asso-ciated with Beckwith-Wiedemann syndrome.In 16 cases (patients P1 to P15, including thesibs P3C1 and P3C2) the symptoms were thoseof a complete form of Beckwith-Wiedemannsyndrome. For the patient P3C2, a fetal serumsample was obtained for a variety of echo-graphic abnormalities and for karyotype analy-sis. Four patients (P16 to P19) had a smallernumber of anomalies and were diagnosed ashaving an incomplete form of the syndrome.Patient P16C1 had a brother, P16C2, whohad a neuroblastoma, but no symptoms ofBeckwith-Wiedemann syndrome. Four otherpatients (P20 to P23) were not diagnosed ashaving the syndrome, but nevertheless pre-sented isolated major clinical symptoms typ-ical of it and were therefore classified as caseswith isolated signs of the Beckwith-Wiede-mann syndrome. Two patients exhibited mac-roglossia (P20 and P21), one hemihypertrophy(P22), and one an omphalocele (P23). Amongthe 25 patients with signs of Beckwith-Wiede-mann syndrome, seven developed tumours:P1, a neuroblastoma and a nephroblastoma;P2, a nephroblastoma; P3Cl, a ganglioneur-oma; P3C2, bilateral nephroblastoma; P6, anephroblastoma; P16C1, an adrenocorticalcarcinoma; and P16C2, a neuroblastoma. Inaddition to the above, four children were

Institut National de laSante et de laRecherche Medicale,Unite 142 H6pitalSaint-Antoine, Paris,France.H SchneidD Seurin

Service de ChirurgieMaxillo-faciale,H6pital Trousseau,Paris, France.M-P Vazquez

Service d'ExplorationsFonctionnellesEndocriniennes,Hopital Trousseau,26 Avenue ArnoldNetter, 75012 Paris,France.M GourmelenS CabrolY Le Bouc

Correspondence toDr Le Bouc.Received 15 September 1992.Revised version accepted20 November 1992.

353

on 11 July 2018 by guest. Protected by copyright.

http://jmg.bm

j.com/

J Med G

enet: first published as 10.1136/jmg.30.5.353 on 1 M

ay 1993. Dow

nloaded from

http://jmg.bmj.com/

Schneid, Seurin, Vazquez, Gourmelen, Cabrol, Le Bouc

co cq

00

00

*0 v

U

-5

.t; U o

*W3 E=i C

a4 U

0r t

v 0

U Cl0 E

U/ 0 0

- 0

C)

0X N

¶5S. c

*3 0.2eSH:

ClC

0W.C 0

05 30,_C _

_0.-O

0.55_UUU N

0.

_0~0.

0.

o0.

EC,

0.

0

+

0tr

v ++ +

00O

SZ++ °o ++ ++ + +

It r- + ++++ + +t

S + ° ++ +

++ +0

uL + + +0 0t

0 ou +> in+ + ++

+0

e +

L + ++ ±+ ++

+o

z + ob + +++

OLf

enc

;Z+ +e + + +m

m£ ++ + + + ++

.T + o7+ + + + +

0

+z+ + + + +

0Sz++ -+ + ++ + + +

0(3Z++ ++++ + + + ± +

0

t ++S+no tn + ++ + + +

;g + + 0 C> + + + + + + +

tr)dUEs+++0+ + + + + +

(10XOO +++ +++ + + +++

354

0

.0

.00

0

.00

000

.0

0

0

0

0

0

0

0

-00

E0

11z

0.

+


http://jmg.bm

j.com/

J Med G


ay 1993. Dow

nloaded from

http://jmg.bmj.com/

Parental allele specific methylation of the human insulin-like growth factor II gene

investigated with no signs of Beckwith-Wiede-mann syndrome, but with embryonic tumoursoften associated with the syndrome, such asadrenocortical carcinoma (P24), neuroblas-toma (patients P25 and P26), and nephroblas-toma (P27).

Family analysis was possible in 14 cases,both parents of P1, P3, P5, P6, P7, P9, PIO,P13, P14, P16, P19, P20, and only the mothersof patients P2 and P24.

ControlsThe control population has been describedpreviously'5: 16 unrelated subjects and eightfamilies (including 10 children and 16 parents)were studied.

the IGF-II cDNA was used as probe. Num-bering of exons was based on the nomenclaturefor the IGF genes as published.24 An exon 7specific probe (a 159 bp fragment) wasobtained by EcoRI/AvaII digestion, and anexon 9 probe (a 356 bp fragment) by SalIlEcoRI digestion of the 663 bp cDNA insert.The human insulin 2-7 kb BglII-XhoI frag-

ment was kindly provided by P Holthuizen25and the 827 bp human calcitonin cDNA probeby A Julienne.26DNA probes were labelled with a32P-dATP

by random priming (kit, Amersham, England)to a specific activity of 2 x 109 cpm/ g DNA, asindicated by the manufacturer.

Materials and methodsSAMPLESBlood samples from controls and patients werecollected in anticoagulant (EDTA) and storedat - 80°C until use for DNA extraction. Tis-sues, either from tumours (P1, P24) or frompartial glossectomy (P1, P6 to PlO, P13, P14,P20), were removed from surgical patients,frozen immediately in liquid nitrogen, andstored at - 80°C until use.

ISOLATION OF DNA AND RNA FROM BLOOD ANDTISSUESLeucocyte DNA was isolated as previouslydescribed.2' Total DNA and RNA wereextracted simultaneously from frozen tissue(for the nine samples mentioned above exceptfor P13 for which only DNA analysis wasavailable) according to Chirgwin et aP2 withsome modifications as previously described.'5Spectrometry at 260 and 280 nm was used forDNA and RNA quantification. Nucleic acidintegrity was routinely checked by ethidiumbromide staining.

PROBESThe human placental IGF-II cDNA insertP35 (vector kgt 10)23 was subcloned in plasmidPGEM4 (pTG 3907). The 663 bp EcoRI frag-ment containing 15 bp of the 5' untranslatedregion (exon 6), the coding region, and 99 bpof the 3' untranslated region (exons 7, 8, 9) of

8 9

| H H | H*H*A A AAA

RFLP

1 *6 kb10kb

0-5kb 1-1kb

A

04 kb 1-1 kb , Re:

fra l

0 9 kb len

Figure I Map of the relationship between AvaII and HpaII sites on the IGFgene around exons 7, 8, and 9; hatched areas correspond to the cDNA probe us,A = AvaII site: CCA (orT) GG. A* = specific AvaII site, methylation sensitivedepending on tissue of origin.'5 H=HpaII site: CCGG. H*=specific HpaII sitsensitive to parental allele specific methylation. The fragments detected by Soutblotting after AvaII digestion are shown at the bottom.

SOUTHERN BLOT ANALYSISTen micrograms of DNA were digested with120 units of restriction endonuclease, accord-ing to the manufacturer's specifications(Appligene, France or New England BiolabsInc, USA). DNA fragments were separated byelectrophoresis on 0-7% to 1-2% agarose gels,then denatured and transferred to a nylonmembrane (Gene Screen Plus, New EnglandNuclear, USA) using the alkaline method.Finally they were hybridised to 32P-labelledprobes as previously described.2'Enzymes were selected for Southern blot

analysis on the basis of the polymorphic pat-terns yielded: AvaII and SacI for the IGF-IIgene,'5 Sacl and TaqI for the insulin gene,27and TaqI for the calcitonin gene.28Double digests were used to study allele

methylation (fig 1). The initial enzyme wasused to detect polymorphism and to distin-guish the alleles from each other. The seconddigestion was done with a methylation sensit-ive enzyme, HpaII, through which methyl-ation of the CCGG site could be determinedby comparison with the fragments obtainedwith its methylation insensitive isoschizomerMspI.A specific profile was detected with AvaII

for the IGF-II gene in some tissues whereDNA was demethylated and the 1 6 kb frag-ment was cleaved to a 05 kb fragment, aspreviously described'5 (fig 1).

DOT BL-OT ANDT NORTHTF.RM RT XT ANAT VY.TV.APi "J.%F fl'JLJ A V1DJLu n1w1MX313

For dot blot analysis, RNA samples were dot-ted onto a Hybond-N membrane (Amersham,England) using a Hybri-dot (BRL, USA)apparatus as previously described.'5 ForNorthern blot analysis, denatured total RNAwas loaded onto 1-2% agarose-formaldehydegels for electrophoresis, then transferred toHybond-N membranes (Amersham, Eng-land), as previously reported.'5

striction The blots were hybridised according to thegment manufacturer's instructions with 3 x 10 cpm/gth ml of 32p labelled probe, then washed twice'-II at room temperature and once at 65°C, each

ed. for 20 minutes, in 0-1 x SSPE (1 x SSPEt 001 mol/l NaH2PO4, pH 68, 0001 mol/l

'hern EDTA-Na2, 0 15 mol/l NaCl) and 0 1% SDS,before being exposed to x ray film (Curix RP1-

Exons 7

5'

A A*

355


http://jmg.bm

j.com/

J Med G


ay 1993. Dow

nloaded from

http://jmg.bmj.com/

356Schneid, Seurin, Vazquez, Gourmelen, Cabrol, Le Bouc

Agfa) at - 80°C with intensifying screens(Dupont Cronex Hiplus).

ResultsALLELE SPECIFIC METHYLATION OF THE IGF-IIGENE IN CONTROL LEUCOCYTESIn order to determine whether DNA methyla-tion is linked to differential imprinting of theIGF-II gene, Southern blot analyses weredone, using a system of double digestion toidentify the methylation state of each of thealleles of the IGF-II gene, as described above.Sixteen unrelated adults were investigated, allof whom were heterozygous with AvaII forthe IGF-1I gene. As can be seen from acomparison of the AvaIl digest and the AvaII/MspI double digest shown in fig 2, all AvaIIfragments (1 6, 1-1/0 9, 0 4kb) were cleavedby MspI, indicating that all contained someCCGG sequence. The AvaII/HpaII doubledigest, by contrast, showed a conserved AvaIIprofile, but with major differences in the poly-morphic 1 1/09 kb fragments. Both alleleswere capable of cleavage by HpaII, but in all16 samples examined, only one allele wascleaved, whether it was the 1 1 kb (nine sam-ples, 56%) or the 0 9 kb allele (seven samples,

_ _

44%). This indicates specific hypomethylationof only one allele.The same analyses were done in eight

informative families (10 children and 16 par-ents), and in all cases it was the maternal allelethat was hypomethylated. Two examples areshown in fig 3.

In order to localise the AvaII fragmentsdetected by Southern blotting, hybridisationexperiments were done with IGF-II exon 7and exon 9 specific probes. As shown in fig 4,the 16 kb AvaII fragment was specificallydetected by the exon 7 probe, whereas thepolymorphic 1 1/09 kb fragments weredetected by the exon 9 probe represented in fig1. The maternal allele specific hypomethyla-tion site was therefore localised at exon 9 of theIGF-II gene.

ALLELE METHYLATION OF THE INSULIN ANDCALCITONIN GENES IN CONTROL LEUCOCYTESThe same methodology was used to investigatemethylation of two other genes localised on theshort arm of chromosome 11, insulin (11 p1 5.5)and calcitonin (1 Ipi 5.1). For the insulin gene,both alleles were partially demethylated, indic-ating an absence of allele specific demethyla-

_.e;;

Figure 2 Southern blot analysis of control leucocyte DNA digested with AvaII, AvaII/MspI, or AvaII/HpaII,using IGF-II cDNA as probe. (1) Homozygote for the 0 9 kb fragment. (2) Homozygote for the 1 1 kb fragment.(3, 4, 5, 6, 7, 8) Unrelated heterozygotes for the 0 9/11 kb fragments.

.s4w OMW to to *_ 4mm__

SW_

7 X;ALI:

Figure 3 Family study: Southern blot analysis of leucocyte DNA digested with AvaII and AvaII/HpaII, usingIGF-II cDNA as probe. Two control families (A and B) were examined. F, father; M, mother; C, child.


http://jmg.bm

j.com/

J Med G


ay 1993. Dow

nloaded from

http://jmg.bmj.com/


A B4 a'I A v

kb 3 3

16 - 4

J.1-!- _

0 §-9-

2 3

4I_

em

oil 4mp~-2A___f_.*-gm _W..

cDNA exor. 7 -DNA exorn q

Figure 4 Southern blot analysis of control leucocyte DNA digested with AvaII, using the IGF-II cDNA probe(exons 7, 8, 9) and exon 7 (A) or exon 9 (B) specific probes. (1) Homozygote for the 0 9 kb fragment,(2) homozygote for the 1 1 kb fragment, (3) heterozygote for the 0 9/11 kb fragments.

tion, at least in the regionfragments (data not showrcalcitonin gene, double dHpaII yielded two polywhich were shorter thanTaqI alone, indicating totboth alleles, without anythe polymorphic fragment

ANALYSIS OF THE INSULIN/:PATIENTS WITH BECKWITH-SYNDROMEThe genome of children i

demann syndrome was a

-7-

2.8~~~~~~~~~~~~#

72_3 "

28 - . *a

2-3 - 4| s

Figure 5 Southern blot analysiDNA from a control family, usi;probe with TaqI or TaqI/HpaI]C, child.

of the polymorphic determine whether abnormalities weren). In the case of the implicated at the insulin/IGF-II locus. Leuco-ligestion with TaqIl cyte DNA was analysed in the 29 patientsmorphic fragments listed in table 1, 20 of whom were diagnosed asthose obtained after having Beckwith-Wiedemann syndrome (Pl-tal demethylation of P19, including P3CI and P3C2). In addition,allele specificity for samples of tissue DNA (following partialts studied (fig 5). glossectomy or tumorectomy) were examined

when they were available. AvaIl, SacI, andTaqI were used to analyse the restriction frag-

IGF-II LOCUS IN ment length polymorphisms (RFLP) at the-WIEDEMANN insulin/IGF-II locus as mentioned above. For

the calcitonin gene, a control gene distal to thewith Beckwith-Wie- insulin/IGF-II locus, TaqI was used. Wherenalysed in order to possible, the study was extended to family

analyses.Among the 29 patients examined, seven

F e. e ia were homozygous for the insulin/IGF-II locusand therefore non-informative (P4, P6, P10,P20, P22, P26, and P27) (data not shown). Out

* of the 22 informative cases, five exhibitedprofiles with abnormalities (P1, P3C1, P3C2,

_e am 3 P9, and P24, as described in detail below andin table 2) and the remaining 17 had normalheterozygote profiles for either the IGF-IIgene (P2, Pll, P13, P15, P16CI, P16C2, P18,P19, P21, P23) or the insulin gene (P5, P7, P8,P12, P14, P16C1, P17, P18, P19, P23). Allelefrequencies of AvaII and Sacl RFLPs at theIGF-II gene were the same as those of thecontrol population, both in the group of 15unrelated children with Beckwith-Wiedemann

i syndrome and in the group of 19 cases con-sidered to have complete or incomplete Beck-with-Wiedemann syndrome (table 3). Fur-thermore, if the four contiguous loci, AvaII,Sacl for the IGF-II gene, and TaqI, SacI forthe insulin gene, were taken together, thenumbers of homozygotes and heterozygoteswere no different from the controls (table 4).

In four out of the 16 cases with Beckwith-Wiedemann syndrome (P1, P3CI, P3C2, andP9) abnormal heterozygote profiles weredetected for the insulin/IGF-II locus. These

s of control leucocyte are represented in table 2 and shown in fig 6ng the calcitonin cDNAr* F, father; M, mother; where disproportionate allele intensities were

detected, suggesting uniparental disomy for

357


http://jmg.bm

j.com/

J Med G


ay 1993. Dow

nloaded from

http://jmg.bmj.com/


Table 2 Genotypes of some patients with either Beckwith-Wiedemann syndrome or isolated tumours. Southern blotanalysis of the IGF-II gene (AvaII and SacI), insulin gene (SacI and TaqI), and calcitonin gene (TaqI).

IGF-II Insulin Calcitonin

Restriction endonuclease AvaII SacI Sacl TaqI TaqI(alleles (kb)) A: 1-1 A: 10 0 A: 7-5 A: 5 6 A: 8-0

B: 0 9 B: 7-6 B: 6 0 B: 4 5 B: 6 5

Patient Associatedtumour

Patients with P1 F A/A BR'B A B A,B AAcomplete forms of M A/B A/B BRB B B BBBeckwith-Wiedemann C Alb a,'B A b A/b A/Bsyndrome To A,b a'B A/b ANb A B

Tu A/A B!B A A A A A B NephroblastomaNk A,/b b/B A!/b A/b A B

P3 F AA B B B1B B/B A,BM A/B A/B A, B A iB B,BCl A/b a,,B a B alB B B Ganglioneuroma*

C2f A,b a'B a B a'B BB'B BilateralC2 A,tb a,B a B a 'B B,'B nephroblastoma*

P9 F A,'B AB B,B B'B A AM A:B A B A'B B BC A,B A/B A)B A B AIBTo a/B A/b a,B a B A'B

Patient with P24 M B/B B B B B B/Bisolated tumour C A/'B A/B B, B B/B B/B Adrenocortical

Tu A/b a/B B,'B B/B B:B carcinoma

F= father. M = mother. C = affected child. To = tongue tissue after partial glossectomy. '1'u = tumour after tumorectomy.Nk = normal kidney.Cl and C2 were two affected children from the same family (see table 1), C2f being a fetal sample from the child C2.Alleles are represented as A and B.A/b or a/B: disproportionate profile, the less abundant allele being printed in lower case.* Tissue not available.

the insulin/IGF-II locus. In the nephroblas-toma of P1, the maternal allele (09kb) wastotally missing (fig 6). Allele loss was alsodetectable in leucocyte DNA in P1, but aheterozygote profile with disproportionateintensities was visible in tongue (fig 6) andhealthy kidney tissue (data not shown). InP3C1 and her brother, P3C2 (in fetal P3C2fand newborn P3C2 leucocyte DNA), hetero-zygote profiles were detectable, but alleleintensities were also disproportionate (fig 6).P9 had a normal heterozygote leucocyte DNAprofile, but a disproportionate profile for

Table 3 Allele frequencies of AvaII and SacI polymorphismis at the IGF-II locus(1Ip15.5) in Beckwith-Wiedemann syndrome patients.

AvaIl Sacl

Alleles A: 1 1 kb B: 0 9kb A: 10-0kb B: 7 6kb

Complete forms ofBeckwith-Wiedemann 70 0% 3000% 36-7% 63-3%(n= 15)Complete and incomplete forms ofBeckwith-Wiedemann 68-4% 31-6%/6 36-8% 63 2%(n= 19)Controls* 75.0% 25 0% 35 0% 6500%(n= 37)

* Control population previously reported.'5

Table 4 Homozygote and heterozygote frequencies offour RFLPs associated with Beckwith- Wiedemannsyndrome, insulin (TaqI,SacI) and IGF-II(SacI,AvaII) loci taken together.

Frequency of Frequency ofhomozygosity heterozygosity

Complete forms ofBeckwith-Wiedemann 20% 80%(n = 15)Complete and incompleteforms ofBeckwith-Wiedemann 16% 84%(n = 19)Controls* 18% 82%(n = 34)* Control population previously reported for the four contigu-ous loci (insulin/RsaI, insulin/HindlIl, IGF-II ISacI, IGF-II/Aval 1).36

tongue tissue (fig 6). In addition, a 0 5 kbfragment was detected in tissue samples (ton-gue and tumour) after AvaII digestion, aspreviously detected in a number of other tissuesamples.'" This was attributable to a specificAvaII site being methylation sensitive,depending on the tissue concerned (fig 1).Among the four children with isolated tu-

mours (P24-P27, table 1), tumoral DNA wasavailable in only one case, P24, and an abnor-mal profile with disproportionate intensity wasdetected for the insulin/IGF-II locus (table 2).For this particular patient, the similar abnor-mal profile was detected in the adrenocorticalcarcinoma, but not in leucocyte, DNA (fig 6).

For these five patients (P1, P3C1, P3C2, P9,and P24, table 2), family analyses alwaysshowed a maternal origin of the partially ortotally missing allele. The disproportionateprofiles were located only at the insulin andIGF-II loci and did not extend to the calcit-onin gene which, in the informative cases,exhibited a normal profile (patients P1 and P9,table 2).

IGF-II MRNA EXPRESSION IN TISSUE MATERIALFROM PATIENTS WITH BECKWITH-WIEDEMANNSYNDROMEIGF-II mRNA expression was analysed in thetissues available to determine whether anydisregulation was involved. On average, IGF-II mRNA expression, measured by dot blotanalysis in the tongue tissue of the eightpatients having undergone partial glossectomy(P1, P6, P7, P8, P9, PlO, P14, P20), wassimilar to that in normal adult liver used as areference (fig 7A and B). In addition, there wasno significant difference between the two casesof macroglossia where uniparental disomy hadbeen detected (P1 and P9) and the othersanalysed. Moreover, the usual 6, 4 8, and

358


http://jmg.bm

j.com/

J Med G


ay 1993. Dow

nloaded from

http://jmg.bmj.com/


P1

kb Tu To C

1 6 -.- * IF_

1F1 -- flf09-

P3

Cl C2f C2

U, v.-,

05

0.4-

.I aw

Figure 6 Southern blot analysis ofgenomic DNA digested with AvaII using IGF-II cDNA as probe for patientsPl (Tu: tumour, To: tongue, C: leucocytes), P3 (Cl: leucocytes, C2f: fetal leucocytes, C2: newborn leucocytes),P9 (C: leucocytes, To: tongue), and P24 (Tu: tumour, C: leucocytes).

RNA 'gV 8 4 2 1 05

AP14 D

P6 * *

Pio **.

L @9

B

P7 * *

P8 * I* 't

P20 * 0 a

P1.* s

L O* is

P24 f

L

Figure 7 Dot blot analysis of IGF-II mRNA expression in tongue tissue (Pl, P6-PlO, P14, P20) from patientswith macroglossia and in an adrenocortical carcinoma (P24). In each experiment normal adult liver (L) was usedas internal control. Exposure time: A = six days, two intensifying screens; B =four days, one intensifying screen;C= 16 hours, two intensifying screens.

2-2 kb IGF-II mRNAs were detectable byNorthern blot analysis (data not shown).

In the two samples of tumour tissue avail-able, IGF-II mRNA expression could not bemeasured in the nephroblastoma of P1, whowas undergoing chemotherapy at the time ofsurgery (data not shown), but in the adreno-cortical carcinoma of P24, who was not underchemotherapy, extremely high levels of IGF-II mRNA expression were found (fig 7C).

ALLELE SPECIFIC METHYLATION IN PATIENTSWITH BECKWITH-WIEDEMANN SYNDROMEIn order to determine whether allele specificmethylation of leucocyte DNA may be linked

to the Beckwith-Wiedemann syndrome,double digestion experiments were done andcomparisons made with the control popula-tion. This type of analysis was possible onlyfor subjects with heterozygous profiles forAvaII for the IGF-II gene. In 11 of thesepatients (P1, P3C1, P9, P11, P13, P15, P16C1,P18, P19, P21, P23), the patterns were similarto those for the control population, indicatingthat only one allele was hypomethylated (datanot shown). In only one case (P2) did bothalleles remain methylated (fig 8). In the fiveinformative families of Beckwith-Wiedemannsyndrome patients (P1, P2, P3, P13, and P19)specific hypomethylation was always found forthe maternal allele (data not shown), as had

P9

%- 0

0-%

:..

P24

Tu C-

; * :-

__ 9

map,i.X -

C

359


http://jmg.bm

j.com/

J Med G


ay 1993. Dow

nloaded from

http://jmg.bmj.com/


leucocyte DNA still exhibited allele specifichypomethylation, only the 1 1 kb allele in P9or the 0 9 kb allele in P13 being demethylated(fig 9). For P24, without Beckwith-Wiede-

"WI~ mann syndrome, both alleles of the adrenocor-tical carcinoma DNA were hypomethylated(fig 9). These preliminary findings for DNAmethylation in the tissues of four informativesubjects showed a difference between leuco-cyte and tissue DNA in patients with Beck-with-Wiedemann syndrome and the patientwith an embryonic tumour, whereas allele spe-cific demethylation in control tissues (placenta,peritumoral adult kidney tissue) was similarto that in control leucocyte DNA (data notshown).

Figure 8 Southern blot analysis of patient P2 leucocyteDNA digested with AvaII or AvaII/HpaII, usingIGF-II cDNA as probe. M, mother; C, affected child.

previously been observed for the control popu-lation.None of the patients or families investigated

for the insulin and calcitonin genes showedany allele specific demethylation at least in theregion of the polymorphic fragments. Here,too, the patterns were the same as those for thecontrol population: partial demethylation ofboth alleles for the insulin gene and totaldemethylation of both alleles for the calcitoningene (data not shown).Where possible, tissue samples from

informative subjects were also analysed (ton-gue samples from P1, P9, P13 with macroglos-sia, and tumour samples from P1 and P24), asshown in fig 6. For P1, IGF-II DNAdemethylation was evident in tongue and tu-mour DNA (as indicated by the presence of the0 8, 0 7, and 0 3 kb fragments, fig 9). However,the parental origin of the demethylated allelescould not be confirmed because of the extens-ive disproportion of the alleles (fig 6). For thispatient, P1, it was possible to show withAvaII/HpaII digestion that the 09 kb frag-ment (which was maternal, table 2) was stillpresent in tongue tissue, indicating that theshorter fragments were in part paternal (fig 9).In P9 and P13, both alleles (1 1 and 0 9 kb) intongue tissue were demethylated, whereas

_;ANAMW*_04af_

DiscussionAllele specificity of human IGF-II genemethylation as related to parental origin hasbeen investigated in both a control populationand patients with Beckwith-Wiedemann syn-drome where the IGF-II gene is thought toplay a role.' 229 Differential parental imprintingof the IGF-II gene has been shown in the fetalmouse,'8 but the molecular mechanism re-mains an enigma, despite transgenic studiessuggesting that DNA methylation may beinvolved.30Our analyses of control leucocyte DNA

methylation in the IGF-II gene have shownthat allele specificity does indeed exist, sinceonly one allele was demethylated in all cases.Family analyses showed that the hypomethyl-ated allele was always maternal, which meansthat allele specific hypomethylation at theIGF-II gene is linked to parental origin.Moreover, allele specificity of methylation inthe 1ipl5 region of the chromosome waslimited to the IGF-II locus (llpl5.5). Thephenomenon was not observed either at theinsulin gene3' (this study) or at the calcitoningene (this study), both alleles being demethyl-ated simultaneously, either partially (insulin)or totally (calcitonin) in the area studied.Finally, we found the sites involved in thisallele specific methylation of the IGF-II geneto be situated in the region of exon 9.

In their homologous recombination studiesof transgenic mice which were heterozygous

aad

Figure 9 Southern blot analysis of genomic DNA digested with AvaII/HpaII, using IGF-II cDNA as probe, forpatients PI, P9, P13, and P24. C= leucocytes from affected child, To= tongue, Tu= tumour (Pl: neuroblastoma,P24: adrenocortical carcinoma). The AvaII patterns for the same patients are shown in fig 6.

360

4%10 41101-A6.


http://jmg.bm

j.com/

J Med G


ay 1993. Dow

nloaded from

http://jmg.bmj.com/


for the mutant gene, DeChiara et al"8 showedthat only the paternal allele of the IGF-II genewas expressed embryonically, whereas thematernal allele remained silent. Our observa-tions indicate that in man, allele specificmethylation of the IGF-II gene varies withdifferential parental imprinting, the maternalallele being hypomethylated and the paternalallele methylated. The allele specific IGF-IIgene methylation may be linked to mRNAexpression, demethylation of a gene frequentlybeing associated with expression.32 Neverthe-less, our results suggest that the paternal alleleis methylated at IGF-II exon 9 and is associ-ated with expression of the paternal allele.This, as DeChiara et all8 proposed, may reflectnegative imprinting, expression of a specifictrans acting repressor inhibiting expression ofthe maternal, and thereby allowing expressionof the paternal, allele.We also examined the possibility of genetic

abnormalities accounting for the Beckwith-Wiedemann syndrome. Southern blotting ex-periments using enzymes which show RFLPsshowed that in these patients (table 1) no majorchromosomal anomalies (like deletion oramplification) occurred in the lp 15 region.

Loss of heterozygosity has been described intumours where the short arm of chromosome11 (1 lpl3-ll pl 5) is involved,33 and particu-larly in Wilms' tumour.34 35 Henry et al 36reported increased leucocyte homozygosity inpatients with the Beckwith-Wiedemann syn-drome as compared to controls, but our find-ings do not support this, as we found the samefrequency of homozygosity in the two groups37(tables 3 and 4). This discrepancy may beexplained by the fact that some apparent allelelosses in fact reflect mosaics with grossly dis-proportionate alleles, which simulate homozy-gosity in some patients (as in patient 1, fig 6).Among the 'complete' Beckwith-Wiedemannsyndrome patients examined in this study,three (P1, P3C1, P9) out of 15 unrelated sub-jects (20%) (and four out of 16 when includingP3C2) exhibited total allele loss or markeddisproportion, and in all of them the maternalallele was lost (or very diminished) makingduplication of the paternal allele possible.Uniparental disomy is probably only one ofthe mechanisms that accounts for Beckwith-Wiedemann syndrome. Among the cases ofpaternal isodisomy (P1, P3C1, P9), two out ofthree unrelated children developed tumours(66%). A malignant tumour rate of 7 5% hasbeen reported for Beckwith-Wiedemann syn-drome.5 The high rate found in cases ofuniparental disomy (with a disproportionatepattern) could be important in genetic coun-selling as regards the risk of tumour in Beck-with-Wiedemann patients. Nevertheless,uniparental disomy has also been seen in non-tumoral tissue (enlarged tongue and leuco-cytes), which means that allele loss is notalways accompanied by tumour formation inthese particular tissues37 (this study). For twopatients (P1 and P9), we were able to show thatthe mosaic disomy was limited to the distal 1 Ipregion (calcitonin not included). The presenceof disproportionate allele intensities was

probably the result of mosaicism for two typesof cell, the normal (with normal heterozygoteDNA) and the paternal isodisomy cells. Thistherefore concerns a post-zygotic event, withmitotic recombination, and indicates a lowrecurrence risk for a second child in a Beck-with-Wiedemann syndrome family. It alsothrows some light on the genetic mechan-ism involved. The recurrence of somatic mo-saicism in the two sibs (P3C1 and P3C2)was highly improbable and suggests that anabnormal paternal 1 ipi5 chromosome wastransmitted.As in the control population, allele specifi-

city of IGF-II gene methylation was observedin leucocyte DNA of patients with Beckwith-Wiedemann syndrome. Hypomethylation ofthe maternal allele and methylation of thepaternal allele were seen in both groups. Inonly one case, P2, were both constitutive leu-cocyte alleles abnormally methylated, whereasinsulin and calcitonin gene methylation resem-bled that of normal controls. Here, both IGF-II alleles could have been submitted to similarpaternal imprinting, suggesting a mechanismdifferent from loss of the maternal allele withpaternal duplication.

Tissue (tongue or tumour) DNA methyla-tion in patients with Beckwith-Wiedemannsyndrome proved to be different from leuco-cyte DNA methylation. In these patients thepaternal allele of the IGF-II gene wasdemethylated, whether or not loss of heterozy-gosity had occurred. The abnormal paternalallele methylation may therefore be specific tochildren with Beckwith-Wiedemann syn-drome and embryonic tumours.Our study indicates that there is a link at the

level of the adult human IGF-II gene betweenmethylation and differential parental imprint-ing. It may be that methylation is involved inuniparental disomy and hence in the patho-genesis of Beckwith-Wiedemann syndrome.Finally, the exon 9 region of the IGF-II gene,where parental allele specific methylation isfound, may play a role in regulating IGF-IIgene expression.

This work was supported by the InstitutNational de la Sante et de la Recherche Medi-cale (INSERM), the Contrat de RechercheClinique de l'Assistance Publique (CRCAP89-57, CRCAP 923107), the Fondation pour laRecherche Medicale-92, and the UniversiteParis VI. H Schneid is a recipient of a fellow-ship of the Fondation pour la RechercheMedicale. We are grateful to Dr M Binoux forhis support and to Dr M Mannens for helpfuldiscussion. We thank Drs J J Baudon, A Bens-man, P Brun, I Grapini, M Gruner, G Lasfar-gues, M Maes, A Masson, and R Rappaportfor donating samples for analysis.

1 Koufos A, Grundy P, Morgan K, et al. Familial Wiede-mann-Beckwith syndrome and a second Wilms' tumorlocus both nap to 1 lpl5.5. Am .T Hum Genet1989;44:71 1-9.

2 Ping AJ, Reeve AE, Law DJ, Young MR, Boehnke M,Feinberg AP. Genetic linkage of Beckwith-Wiedemannsyndrome to I Ip 15. Am J Hum Genet 1989;44:720-3.

3 Tricoli JV, Rall LB, Scott J, Bell GI, Shows TB. Localiza-tion of insulin-like growth factor genes to human chromo-somes 11 and 12. Nature 1984;310:784-6.

4 Sotelo-Avila C, Gonzalez-Crussi F, Fowler JW. Completeand incomplete forms of Beckwith-Wiedemann syn-

361


http://jmg.bm

j.com/

J Med G


ay 1993. Dow

nloaded from

http://jmg.bmj.com/


drome: their oncogenic potential. 7 Pediatr 1980;96:47-50.

5 Pettenati MJ, Haines JL, Higgins RR, Wappner RS,Palmer CG, Weaver DD. Wiedemann-Beckwith syn-drome: presentation of clinical and cytogenetic data on 22new cases and review of the literature. Hum Genet1986;74: 143-54.

6 Blatt J, White C, Dienes S, Friedman H, Foley TJR.Production of an insulin-like growth factor by osteosar-coma. Biochem Biophys Res Commun 1984;123:373-6.

7 Reeve AE, Eccles MR, Wilkins RJ, Bell GI, Millow LJ.Expression of insulin-like growth factor-II transcripts inWilms' tumour. Nature 1985;317:258-60.

8 Scott J, Cowell J, Robertson ME, et al. Insulin-like growthfactor-II gene expression in Wilms' tumour and embryo-nic tissues. Nature 1985;317:260-2.

9 Haselbacher GK, Irminger JC, Zapf J, Ziegler WH, Hum-bel RE. Insulin-like growth factor II in human adrenalpheochromocytomas and Wilms' tumors: expression atthe mRNA and protein level. Proc Natl Acad Sci USA1987;84:1 104-6.

10 Tricoli JV, Rall LB, Karakousis CP, et al. Enhanced levelsof insulin-like growth factor messenger RNA in humancolon carcinomas and liposarcomas. Cancer Res1986;46:6169-73.

11 Huff KK, Kaufman D, Gabbay KH, Spencer EM, Lipp-man ME, Dickson RB. Secretion of an insulin-likegrowth factor-I-related protein by human breast cancercells. Cancer Res 1986;46:4613-9.

12 Yee D, Cullen KJ, Paik S, et al. Insulin-like growth factorII mRNA expression in human breast cancer. Cancer Res1 988,48:669 1-6.

13 Gray A, Tam AW, Dull TJ, et al. Tissue-specific anddevelopmentally regulated transcription of the insulin-like growth factor 2 gene. DNA 1987;6:283-95.

14 Gloudemans T, Prinsen I, Van Unnik JA, Lips CJ, DenOtter W, Sussenbach JS. Insulin-like growth factor geneexpression in human smooth muscle tumors. Cancer Res1 990;50:6689-95.

15 Schneid H, Seurin D, Noguiez P, Le Bouc Y. Abnormalit-ies of insulin-like growth factor (IGF-I and IGF-II)genes in human tumor tissue. Growth Regulation1992;2:45-54.

16 Daughaday WH, Rotwein P. Insulin-like growth factors Iand II. Peptide, messenger ribonucleic acid and genestructures, serum, and tissue concentrations. EndocrinolRev 1989;10:68-91.

17 DeChiara TM, Efstratiadis A, Robertson EJ. A growth-deficiency phenotype in heterozygous mice carrying aninsulin-like growth factor II gene disrupted by targeting.Nature 1990;345:78-80.

18 DeChiara TM, Robertson EJ, Efstratiadis A. Parentalimprinting of the mouse insulin-like growth factor IIgene. Cell 1991;64:849-59.

19 Reik W, Collick A, Norris ML, Barton SC, Surani MA.Genomic imprinting determines methylation of parentalalleles in transgenic mice. Nature 1987;328:248-51.

20 Sapienza C, Peterson AC, Rossant J, Balling R. Degree of

methylation of transgenes is dependent on gamete oforigin. Nature 1987;328:251-4.

21 Schneid H, Le Bouc Y, Seurin D, et al. Insulin-like growthfactor-I gene analysis in subjects with constitutionallyvariant stature. Pediatr Res 1990;27:488-91.

22 Chirgwin JM, Przybyla AE, MacDonald RJ, Rutter WJ.Isolation of biologically active ribonucleic acid fromsources enriched in ribonuclease. Biochemistry1979;18:5294-9.

23 Le Bouc Y, Noguiez P, Sondermeijer P, Dreyer D, GirardF, Binoux M. A new 5'-non-coding region for humanplacental insulin-like growth factor II mRNA expression.FEBS Lett 1987;222:181-5.

24 Holthuizen PE, LeRoith D, Lund PK, et al. In: SpencerEM, ed. Revised nomenclature for the insulin-like growthfactor genes and transcripts. Modern concepts of insulin-likegrowth factors. New York: Elsevier Science PublishingCo, 1991:733-6.

25 Bell GI, Pictet RL, Rutter WJ, Cordell B, Tischer E,Goodman HM. Sequence of the human insulin gene.Nature 1980;284:26-32.

26 Le Moullec JM, Jullienne A, Chenais J, et al. The completesequence of human preprocalcitonin. FEBS Lett1984;167:93-7.

27 Bell GI, Horita S, Karam JH. A polymorphic locus near thehuman insulin gene is associated with insulin-dependentdiabetes mellitus. Diabetes 1984;33:176-83.

28 Hoppener JW, Steenbergh PH, Zandberg J, et al. Localiza-tion of the polymorphic human calcitonin gene on chro-mosome 11. Hum Genet 1984;66:309-12.

29 Little M, Van Heyningen V, Hastie N. Dads and disomyand disease. Nature 1991;351:609-10.

30 Surani MA, Reik W, Allen ND. Transgenes as molecularprobes for genomic imprinting. Trends Genet 1988;4:59-62.

31 Ghazi H, Magewu AN, Gonzales F, Jones PA. Changes inthe allelic methylation patterns of c-H-ras-1, insulin andretinoblastoma genes in human development. Develop-ment 1990;suppl: 1 15-23.

32 Cedar H. DNA methylation and gene activity. Cell1988;53:3-4.

33 Ponder B. Cancer. Gene losses in human tumours. Nature1988;335:400-2.

34 Mannens M, Slater RM, Heyting C, et al. Molecular natureof genetic changes resulting in loss of heterozygosity ofchromosome 11 in Wilms' tumours. Hum Genet1988;81:41-8.

35 Mannens M, Devilee P, Bliek J, et al. Loss of heterozygosityin Wilms' tumors, studied for six putative tumor suppres-sor regions, is limited to chromosome 11. Cancer Res1 990;50:3279-83.

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

37 Schneid H, Vasquez MP, Seurin D, Le Bouc Y. Loss ofheterozygosity in non-tumoral tissue in two children withBeckwith-Wiedemann syndrome. Growth Regulation1991;1:168-70.

362


http://jmg.bm

j.com/

J Med G


ay 1993. Dow

nloaded from

http://jmg.bmj.com/

Documents

Parental allele specific II Beckwith-Wiedemannjmg.bmj.com/content/jmedgenet/30/5/353.full.pdfof a complete form of Beckwith-Wiedemann ... 120 units ofrestriction endonuclease, accord-