Unique expression patterns of H19 in human testicular cancers of different etiology

Unique expression patterns of H19 in human testicular cancers of di�erentetiology

Annemieke JMH Verkerk1, Ilana Ariel2,3, Marjolein C Dekker1, Tamar Schneider3,Ruud JHLM van Gurp1, Nathan de Groot3, Ad JM Gillis1, J Wolter Oosterhuis1,Abraham A Hochberg3 and Leendert HJ Looijenga1

1Laboratory of Experimental Patho-Oncology, Dr Daniel den Hoed Cancer Center, University Hospital, Rotterdam, TheNetherlands; 2Department of Pathology, Hadassah University Hospital, Mount Scopus; 3Department of Biological Chemistry,Institute of Life Science, The Hebrew University, Jerusalem, Israel

The expression pattern of the imprinted human H19 genewas investigated in testicular cancers of di�erentetiology, as well as in normal testicular parenchyma,parenchyma without germ cells, and adjacent totesticular germ cell tumors of adolescents and adults(TGCTs), using RNase protection analysis, mRNA insitu hybridization and reverse-transcription polymerasechain reaction. While di�erent total expression levelswere detected in spermatocytic seminomas, lymphomas,a Sertoli cell tumor and Leydig cell tumors, none showeda disturbance of monoallelic expression. Strikingly, themajority of invasive TGCTs revealed expression of bothparental alleles. The total level of expression highlycorrelated with di�erentiation lineage and stage ofmaturation, similar to that as reported during earlynormal embryogenesis. Biallelic expression could also bedetermined speci®cally in testis parenchyma containingthe preinvasive lesion of this cancer. We thereforeconclude that within the adult testis, biallelic H19expression is speci®c for TGCTs, and that the level ofexpression is dependent on di�erentiation lineage andmaturation stage. This is in agreement with the proposedprimordial germ cell-origin of this cancer, and might berelated to retention of embryonic characteristics inTGCTs. In addition, our data argue against H19 beinga tumor suppressor gene.

Keywords: testis; germ cell tumors; non-germ celltumors; H19 expression; genomic imprinting; germcell development

Introduction

Genomic imprinting, de®ned as the functional differ-ence between homologous mammalian chromosomalregions due to their parental origin (Solter, 1988;Ferguson-Smith et al., 1991; De Groot and Hochberg,1993), has become a major ®eld of interest. Multiplestudies have indicated large impact of this phenomenonon embryonic development and pathological, bothnon-neoplastic and neoplastic, conditions (Hall, 1990;Solter, 1992; Feinberg, 1993; Tycko, 1994). It wasshown that normal embryogenesis depends on a properbalance between the paternal and maternal copy ofcertain chromosomal regions (Barton et al., 1991;

Ferguson-Smith et al., 1991; Fundele and Surani,1994). Relative gain of maternal (parts of) chromo-somes results in preferential growth of somaticelements, while overrepresentation of paternal (partsof) chromosomes favors trophoblast formation. Inevolutionary terms, this was interpreted as a strifebetween the interests of the individual and the species(Moore and Haig, 1991). In addition, existence ofgenomic imprinting results in non-viability of digynic(consisting of two haploid sets of maternally derivedchromosomes) and diandric (consisting of two haploidsets of paternally derived chromosomes) embryo's andrenders sexual reproduction a necessity, which isbene®cial to the species.A number of human genes are subject to genomic

imprinting (Reed and Le�, 1994; Latham, 1995), ie.,showing a uniparental pattern of expression. One ofthese genes, H19, expressed preferentially from thematernal allele, shows a de®ned pattern of expressionduring embryonal development (Rachmilewitz et al.,1992a; Mutter et al., 1993; Lustig et al., 1994; Ohlssonet al., 1994; Svensson et al., 1995; Walsh et al., 1995).In mice, overexpression of H19 leads to intra-uterinedeath (Brunkow and Tilghman, 1991), while thecomplete absence of expression seems to have nodeleterious e�ect (Leighton et al., 1995a). H19 mostprobably acts as an RNA (Brannan et al., 1990) andmultiple functions have been suggested, includingregulation of the closely apposed and reciprocallyimprinted insulin-like growth factor 2 gene (Bartolo-mei et al., 1993; Leighton et al., 1995a,b), therebypossibly linked to its supposed action as a tumorsuppressor (Hao et al., 1993; Moulton et al., 1994;Steenman et al., 1994; Taniguchi et al., 1995), and as aso-called oncofetal gene (Ariel et al., 1995; Lustig et al.,1994; Rachmilewitz et al., 1995). Biallelic expression ofH19 has been found in a certain percentage of humancancers (Zhang et al., 1993; Moulton et al., 1994; VanGurp et al., 1994; Steenman et al., 1994; Elkin et al.,1995; Kondo et al., 1995; Taniguchi et al., 1995; Walshet al., 1995; Douc-Rasy et al., 1996; Hibi et al., 1996),referred to as relaxation or loss of imprinting (LOI),which may be accompanied by up- and downregulationof the total level of expression.Genomic imprinting is determined by the sex of the

individual, and the paternal (during spermatogenesis)or maternal (during oogenesis) pattern has to be fullypresent in mature germ cells. Therefore, erasement andthe subsequent formation of a uniparental pattern ofgenomic imprinting has to be established duringdevelopment of a primordial germ cell to a mature

Correspondence: LHJ LooijengaReceived 30 May 1996; revised 12 August 1996; accepted 3September 1996

Oncogene (1997) 14, 95 ± 107 1997 Stockton Press All rights reserved 0950 ± 9232/97 $12.00

gamete (Rossant, 1993). Somewhere in this process, theprecursor of all testicular germ cell tumors ofadolescents and adults (TGCTs), i.e., carcinoma insitu, is initiated (Skakkebaek et al., 1987; Jùrgensen etal., 1995). Histologically, TGCTs mimick earlyembryogenesis, and can be divided into seminomas,tumors showing similarities with early (primordial?)germ cells, just like carcinoma in situ and nonsemino-matous-TGCTs, which can be composed of somatic(embryonal carcinoma and teratoma) as well as extra-embryonic (yolk sac tumor and choriocarcinoma)elements (Mosto® et al., 1987). Embryonal carcinomacells are the stem cells for the more di�erentiatednonseminomatous-TGCT-components, i.e., teratoma,yolk sac tumor, and choriocarcinoma (Goodfellowand Andrews, 1982). So-called embryoid bodies,structures analogous to a developmental stage of 10days post coitum, can be found in nonseminomatous-TGCTs (Nakashima et al., 1988). Besides TGCTs,spermatocytic seminomas also originate from a cellbelonging to the germ cell lineage, most probablyspermatogonia-B (Romanenko and Persidskii, 1983;Talerman et al., 1984; Dekker et al., 1992; Burke andMosto®, 1993; Cummings et al., 1994; Eble, 1994). Inaddition, non-germ cell-derived tumors of di�erenthistogenesis can develop in the male gonad, includinglymphomas, Leydig cell and Sertoli cell tumors.Analysis of H19 expression (mono- or biallelic/total

level/histological heterogeneity) in the germ cell- andnon-germ cell-derived cancers might reveal new insightsin the possible role of genomic imprinting in general,and of H19 in particular, in the development of thesecancers of the testis.

Results

RNase protection analysis

H19 expression levels (with the expression of g-actinused as a reference, see Methodology) within the seriesof testicular non-germ cell tumors analysed are shownin Figure 1a. No expression was detected in three outof four B-cell lymphomas and in two out of threespermatocytic seminomas. One B-cell lymphoma(Lym4B) and one spermatocytic seminoma (SS3)showed a very low expression. A low to moderateexpression was present in the T-cell lymphoma andboth Leydig cell tumors. The only Sertoli cell tumorstudied showed a high expression. A low expressionwas present in normal testis parenchyma. `Low,moderate and high' expression are de®ned asdescribed in the legend of Table 1.Of the TGCTs, low H19 expression was detected in

all 24 seminomas tested (Figure 1b), within the rangeof 0.03 and 0.34 times the intensity of the actin signal,with a mean of 0.16 (standard deviation 0.09). We didnot observe di�erences between seminomas withoutand with hCG (human Chorionic Gonadotropin)-positive mononuclear cells (depicted by a (+) inFigure 1b). Concerning H19 expression in nonsemino-matous-TGCTs, three di�erent groups can be identi®ed(Figure 1c and Table 1). A low expression was presentin the ®rst four cases. Three were diagnosed as pureembryonal carcinoma (cases 1, 2 and 3), and one asembryonal carcinoma/yolk sac tumor (case 4). Tissue

slides used for veri®cation of the histological composi-tion of the samples used for RNA isolation,demonstrated that also case 4 contained mainlyembryonal carcinoma (not shown). These data suggestthat both seminomas and embryonal carcinoma show acomparable and consistent low expression of H19.A moderate expression was identi®ed in cases 5 ± 10.

The lowest expression was found in case 6, whichcontained only mature teratoma. Moderately lowexpression was found in case 8, containing mainlyembryonal carcinoma. All other samples consisted ofthe di�erent histological elements in approximatelysimilar amounts. High expression was present in thelast three cases (cases 11 ± 13). The ®rst containedmainly immature teratoma, the second both matureteratoma and choriocarcinoma, and the last mainlyimmature teratoma. These results suggest that H19 ispreferentially expressed in immature teratoma andchoriocarcinoma, and less in mature teratoma. Tostudy this in more detail, mRNA in situ hybridizationwas performed.

mRNA in situ hybridization

No signals were found by mRNA in situ hybridizationon routinely included controls in every experiment, i.e.,sense probe and RNase treated tissue sections using theantisense-probe (not shown). Expression of H19mRNA was found in the Leydig cell tumors and theSertoli cell tumor (Figure 2a and b). No expression ofH19 was present in their non-malignant counterpartsor other cell types in normal testis and in testicularparenchyma adjacent to a TGCT (not shown). The(low) expression found in normal testes using RNaseprotection analysis (and RT±PCR) can therefore notspeci®cally be attributed to one of the cell typespresent. No expression was detected in B- and T-celllymphomas, two spermatocytic seminomas (SS3 wasnot studied with mRNA in situ hybridization), andseminomas with or without hCG-positive mononuclearcells (Figure 2c and d). Also no expression wasdetected using mRNA in situ hybridization in

Table 1 Level of H19 expression relative to g--actine in non-

seminomatous testicular germ cell tumors of adolescents and adults

of different histological composition

Level of H19

expression

Case Histological composition relative to g--actine

1

2

3

4

5

6

7

8

9

10

11

12

13

pure EC

pure EC

pure EC

mainly EC, minor YS

mainly EC, minor TE(M)+YS

pure TE(M)

moderate TE(M), minor TE(I)+EC

mainly EC, minor TE(M)

50% EC, 50% TE(M)

mainly EC, minor YS, Minor TE(M+I)

mainly TE(I), minor TE(M)+EC+YS

TE(M)+CH

mainly TE(I), minor TE(M)+YS+EC

0.10

0.62

0.69

0.35

1.98

0.96

3.60

1.43

2.12

2.33

10.79

11.87

6.57

low

low

low

low

moderate

moderate

moderate

moderate

moderate

moderate

high

high

high

Cases are identically numbered as NS in Figure 1c. H19 expression

is considered low between 0.10 and 0.69 times, moderate betwen

0.96 and 3.60 times and, high when more than six times the intensity

of the actin signal; EC=embryonal carcinoma; TE(I)=immature

teratoma; TE(M)=mature teratoma; YS=yolk sac tumor;

CH=choriocarcinoma

H19 expression in testicular cancerAJMH Verkerk et al

96

intratubular seminoma (one case) and in carcinoma insitu (not shown).All di�erent histological nonseminomatous-TGCT-

elements could be analysed in independent cases andshowed no di�erences whether pure or mixed withother components. Twenty-one di�erent embryonal

carcinoma components (including all pure cases) couldbe scored, showing no H19 expression in the tumorcells, while expression was detected in the stromalelements (Figure 2e and f). One pure teratoma and 15teratoma-components of mixed nonseminomatous-TGCTs could be analysed. The immature glandular,squamous and respiratory epithelium and cartilageshowed strong expression of H19 (Figure 2g and h),while smooth muscle cells and neural elements werenegative. Also no expression was found in embryoidbodies (Figure 2i and j). While the majority of matureteratoma elements lacked H19 expression, three cases(one mixed with embryonal carcinoma, one withembryonal carcinoma/yolk sac tumor and one withyolk sac tumor) showed expression of this gene in someregions (not shown). H19 expression was also found inthree residual masses diagnosed as containing matureand no immature teratoma (one mixed with embryonalcarcinoma, one with choriocarcinoma and one withyolk sac tumor). Prominent expression of H19 waspresent in choriocarcinoma and in all trophoblasticgiant cells (not shown), as well as in the reticular areasof yolk sac tumor (Figure 2k and l). No expression waspresent in Schiller Duval bodies (Figure 2m and n).

Reverse-transcription polymerase chain reaction (RT ±PCR)

In contrast to RNase protection analysis and mRNA insitu hybridization, RT ±PCR allowed us to detect H19expression in three B-cell lymphomas out of four casestested (Figure 3). One (case 2) remained negative, justas the bilateral spermatocytic seminomas (SS1 and SS2).The expression found by RNase protection in one B-cell lymphoma and one spermatocytic seminoma (notdetectable using mRNA in situ hybridization) wascon®rmed by RT±PCR (Figure 3a, Lym4

B and SS3).RT ±PCR also con®rmed H19 expression in all othertumors showing a low, moderate or high expressionusing RNase protection analysis. In addition, RT ±PCR showed H19 expression in four samples of normaltesticular parenchyma (Figure 4a), as well as in seventesticular parenchyma samples without germ cells (notshown). By making use of a polymorphic RsaI site inexon 5 of the H19 gene (Zhang and Tycko, 1992), itwas possible to analyse whether H19 was expressedmono- or biallelically in one normal testis, two testeswithout germ cells, one spermatocytic seminoma (SS3),one B-cell lymphoma (Lym1B) and two Leydig celltumors. Expression was found to be monoallelic in allcases (Figure 3b and c, Figure 4b and c). The origin ofthe expressed allele could not be determined becauseparental blood was not available for analysis. Threenormal semen samples showed no H19 expression,while HPRT (Hypoxanthine-Phospho-Ribosyl-Trans-ferase)-transcripts were present, illustrated in Figure 5.Within the series of seminomas and nonseminoma-

tous-TGCTs studied by RNase protection analysis, 13seminomas and seven nonseminomatous-TGCTs wereinformative for the RsaI site (representative examplesare illustrated in Figure 6a, upper panel). Allseminomas showed biallelic expression, as well asmost nonseminomatous-TGCTs (Figure 6a, lowerpanel). In addition to this series, 30 nonseminoma-tous-TGCTs with di�erent histological compositionswere studied. Of these, seven out of nine informative

a

b

c

Figure 1 Expression levels of the H19 gene relative to g-actin(=1.0) as assessed by RNase protection assay in (a) normal testisparenchyma and non-germ cell tumors of the testis (NT=normaltesticular parenchyma; SS=spermatocytic seminoma (SS1 and SS2are contralateral cases of one patient); LymB=B-cell lymphoma;LymT=T-cell lymphoma; Leyd=Leydig cell tumor; Sert=Sertolicell tumor); (b) seminomas, the tumors with chorionic-gonado-tropin expressing mononuclear cells are indicated by a `+', and(c) nonseminomatous testicular germ cell tumors of adolescentsand adults


97

e f

g

mi j

k l

h

a b

c d

n

Figure 2 In situ hybridization analysis of H19 mRNA expression in testicular cancers of di�erent histogenesis. Bright ®eldillustrations of the hematoxylin-eosin-light green counterstained para�n embedded tissue sections are shown on the left (a, c, e, g, i, k)and dark ®eld on the right (b, d, f, h, j, l). (a and b) Sertoli cell tumor (3006); (c and d) seminoma (1856) (no signal abovebackground); (e and f) embryonal carcinoma (1856) with strong expression in stromal element (/); (g and h) immature teratoma


98

cases (n=9) showed biallelic expression. As TGCTsshow an aneuploid DNA content, and the ratiobetween maternal and paternal (parts of) chromo-somes is unknown, we measured the ratio of the twoparental H19 alleles of most of the informativeTGCTs by a quantitative PCR approach. In addition,the level of transcription from each allele was analysedand the quantity of transcript per allele was calculated(see Methodology). As found in normal tissue samplesused as control (Figure 6b, lane 1 and Figure 7b, lane1), 12 out of 13 seminomas showed a parental allele-

ratio of approximately 1 : 1. Expression per allelevaried between 1 : 0.7 and 1 : 1.4. One seminoma (SE5

in Figure 1b) showed a DNA ratio of 1 : 0.5 and a sixtimes higher expression of the underrepresented allele(Figure 6b). Out of nine informative primary non-seminomatous-TGCTs tested quantitatively, sevenshowed a near 1 : 1 ratio of alleles at the DNA level.The calculated expression per allele of six of thesevaried between 1 : 1 and 1 : 2. One nonseminomatous-TGCT (case 1 in Figure 1c) showed strong preferentialexpression from one allele (Figure 7a). Two other

(1856) with expression mainly in stromal elements (/1) and in respiratory epithelium (/2); (i and j) embryoid body (3006) withabsence of expression in the embryoid body itself (/1) and strong expression in the reticular stroma (/2); (k and l) yolk sac tumor(1856) with prominent expression in the reticular areas (arrows); (m and n) characteristic Schiller-Duval body (1856) in uppperhalf of the picture

M Leyd

1

Leyd

2

Leyd

3

Lym

B 1

Lym

B 2

Lym

B 3

Lym

B 4

Lym

T

SS

1

SS

2

SS

3

Ser

t

neg

. co

ntr

ol

501,489 bp—404 bp—320 bp—

501,489 bp—404 bp—320 bp—

—387 bp

—371 bp H19

HPRT

a

MM

un

dig

este

d D

NA

Leyd

2

Leyd

3

LYM

B 1

SS

3 un

dig

este

d c

DN

A

Leyd

2

Leyd

3

LYM

B 1

SS

3—949 bp—714 bp

—788 bp

—553 bp

—235 bp —235 bp

1114 bp—900 bp—692 bp—

1114 bp—900 bp—692 bp—

501,489 bp—404 bp—320 bp—242 bp—190 bp—

501,489 bp—404 bp—320 bp—242 bp—190 bp—

b c

Figure 3 (a) RT±PCR of H19 on non-germ cell tumors of the testis using primerset HN7 and HN8 (upper panel) and HPRT(lower panel), (b) PCR products from DNA of informative cases using primerset HN9 and HN10, digested with RsaI. (c) RT±PCRproducts using primerset HN9 and HN10. Digestion with RsaI shows monoallelic expression of H19 in all samples shown.(NT=normal testicular parenchyma; SS=spermatocytic seminoma; LymB=B-cell lymphoma; LymT=T-cell lymphoma;Leyd=Leydig cell tumor; Sert=Sertoli cell tumor; M=DNA molecular weight marker VIII (Boehringer)


99

nonseminomatous-TGCTs (cases 5 and 7) showedmonoallelic expression, and a strong overrepresenta-tion of the expressed allele at DNA level (Figure 7b).Analysis of DNA isolated from peripheral blood fromthe patient of case 5 and both his parents showed thatthe expressed, and overrepresented, allele is of paternalorigin. A metastasis found in one of the retroperitoneal

lymph nodes after intensive chemotherapeutic treat-ment, containing only mature teratoma, showed a 1 : 1ratio of the H19 alleles at the DNA level. Interestingly,in contrast to the primary tumor, only the maternalallele was found to be expressed. These results areillustrated in Figure 7c.To study H19 expression in the preinvasive stage of

TGCTs, RNA was isolated from testis parenchymasamples containing various amounts of carcinoma insitu cells. All testicular parenchyma samples adjacent toa TGCT without carcinoma in situ cells (n=6) showedmonoallelic expression, in agreement with normal testisparenchyma samples. Of 14 samples containingvariable amounts of carcinoma in situ (8 ± 100% ofthe seminiferous tubules), nine contained detectablebiallelic expression, with one allele clearly over-expressed, in spite of a 1 : 1 ratio of the alleles atDNA level (Figure 8). This indicates that the bulk ofH19-transcripts present is expressed from the hosttissue, which shows monoallelic expression. Theremaining cases showed monoallelic expression ofH19. Parental blood was not available of thesesamples to test parental origin of the expressedallele(s).The results obtained by RNase protection analysis,

mRNA in situ hybridization and/or RT ±PCR aresummarized in Table 2.

Discussion

To gain more insight into the role of genomicimprinting in the development of testicular cancers ofdi�erent etiology, we performed an extensive screeningfor H19 expression using RNase protection analysis,mRNA in situ hybridization and RT±PCR. The resultsof these techniques were found to be in agreement andcomplementary to each other and allowed a detailedanalysis of total level, mono- or biallelic pattern, aswell as tissue distribution of H19 expression.Low or no expression of H19 was found in testicular

lymphomas. In this context, it is noteworthy that weand others (Steenman et al., 1994; Reik et al., 1994)were not able to detect H19 expression in peripheralblood. Therefore, it would be interesting to analysewhether H19 is expressed in hematopoietic stem cells.Malignant Sertoli and Leydig cells show a moderate tohigh expression. It remains to be studied whether theexpression found in normal and germ cell-devoid testisparenchyma and parenchyma adjacent to a TGCT isdue to H19 transcription in Sertoli and/or Leydig cells,or is (also) due to other cell types present. The ®ndingof H19 expression in testicular parenchyma devoid ofgerm cells indicates that expression of this gene is atleast not restricted to the germ cell compartment.Because all informative cases showed monoallelicexpression of H19, LOI of this gene is not involvedin the pathogenesis of these cancers.Both carcinoma in situ and seminomas show

characteristics of early (primordial?) germ cells andare thought to be derived from these cells (Skakkebaeket al., 1987; Jùrgensen et al., 1993, 1995; Rajpert-DeMeyts and Skakkebaek, 1994). Because of the ®ndingof biallelic expression of imprinted genes, includingH19, in murine primordial germ cells (Szabo andMann, 1995), the biallelic expression as found in

501,489 bp—404 bp—320 bp— —371 bp

M

M

NT

1

NT

2

NT

3

NT

4

neg

. co

ntr

ol

un

dig

este

d D

NA

NT

1

NT

2

NT

3

NT

4

x Rsa I

M un

dig

este

d c

DN

A

NT

1x

Rsa

I

dig

est

con

tro

l

1114 bp—900 bp—692 bp—

501,489 bp—404 bp—320 bp—242 bp—190 bp—

1114 bp—900 bp—692 bp—

501,489 bp—404 bp—320 bp—242 bp—190 bp—

—949 bp

—788 bp—553 bp

—714 bp

—235 bp

—235 bp

a

b

c

Figure 4 PCR products of four normal testicular parenchymasamples (NT1± 4) are shown. (a) RT±PCR of all four cases showH19 expression using primers HN7 and HN8, (b) DNA: one case(NT1) is heterozygous for the RsaI site in H19 exon 5, (c) theinformative case (NT1) shows monoallelic expression (HN9/HN10RT±PCR product digested with RsaI). M=DNA molecularweight marker VIII (Boehringer)


100

900 bp692 bp

501,489 bp

spermM1 tu

mo

r

sperm

neg

. co

ntr

ol

M1 tum

or

neg

. co

ntr

ol

448 bp—339 bp—

—788 bp

—387 bp

a b

Figure 5 RT±PCR on three semen samples using (a) H19 and (b) HPRT speci®c primers. Lane 1 shows PCR product from onetumor sample used as positive control. M1=DNA molecular weight marker VIII (Boehringer), M2=Lambda6PstI

Table 2 Summary of results for H19 expression in testicular tumors

H19 expression as expressed by:

RNase mRNA in situprotection RT±PCR hybridisation

Normal testis parenchym

Testis parenchym without germ cells

Testis parenchym with CIS cells

Semen

low

nd

nd

nd

+ monoallelic

+ monoallelic

+ biallelic

±

±

nd

±

nd

Testicular non germ cell tumors

Spermatocytic seminomas

SS1, SS2

SS3

B-cell lymphomas

Lym1

Lym2

Lym3

Lym4

T-cell lymphoma

LymT

Leydig cell tumors

Sertoli cell tumor

±

low

±

±

±

low

low/moderate

low/moderate

high

±

+ monoallelic

+ monoallelic

±

+ ni

+ ni

+ monoallelic

+ monoallelic

+ monoallelic

±

nd

±

±

±

±

±

+

+

Testicular germ cell tumors

Seminomas

13 without hCG pos. cells, SE 1-13

SE 1,3,4,5,6,7,9,10,12,13

SE 2,8,11

11 with hCG pos. cells, SE 14-24

SE 16,19,20

SE 14,15,17,18,21,22,23,24

low

low

+ biallelic

+ ni

+ biallelic

+ ni

±

±

±

Non-seminomatous testicular germ cell

tumors

16 informative cases

NS 1,2,3,4,8

NS 5 and 7

9 other NS

+ biallelic

+ monoallelic

+ biallelic

variable

depending

on the

histology,

specifically:Somatic elements

embryonal carcinoma=EC± in tumor cells

+ in stroma

immature teratoma=TE(I)

stromal elements

immature glandular, squamous and re-

spiratory epithelium and cartilage

smooth muscle + neural elements

embryoid bodies

mature teratoma+TE(M)

Extra-embryonic elements

Choriocarcinoma

Yolk sac tumor

Schiller Duval bodies

variable

depending

on the

histology

+

high

±

±

±

high

high in reticular areas

±

ni=not informative for polymorphic site; nd=not done,+=present; ±= absent


101

carcinoma in situ-containing testicular parenchymaadjacent to an invasive TGCT, as well as inseminomas, is therefore not unexpected. To investigatewhether lack of biallelic expression of H19 in theminority of carcinoma in situ-containing parenchymasamples is due to the limited sensitivity of the methodused, will be veri®ed using isolated populations ofcarcinoma in situ. The expression of H19 alleles in a

1 : 1 ratio in the majority of nonseminomatous-TGCTis disconcordant with the ®nding of leakage ofimprinted genes in murine parthenogenetic andandrogenetic embryonic stem cell lines (Szabo andMann, 1994), and seems to be in contradiction to the®nding of monoallelic expression of H19 during earlymurine development (Sasaki et al., 1995). This is mostprobably due to the fact that at the developmental

un

dig

este

d

M SE

NS

SE

SE

NS

SE

NS

NS

SE

SE

DNA

cDNA

— 949 bp— 714 bp

— 235 bp

— 788 bp

— 553 bp

— 235 bp

M con

tro

l DN

A

tum

or

DN

A

tum

or

cDN

A

1114 bp—

404 bp—320 bp—

242 bp—

190 bp—

900 bp—692 bp—

501,489 bp—

—949 bp—————788 bp—714 bp—553 bp

—235 bp

a

b

Figure 6 (a) Bi-allelic expression of H19 in di�erent seminomas (SE) and nonseminomatous testicular germ cell tumors ofadolescents and adults (NS). PCR products from DNA (upper panel) and cDNA (lower panel) are generated with primerset HN9and HN10 and digested with RsaI. (b) SE5 shows a 66 higher expression of the underrepresented allele at the DNA level. Lane 1shows a blood control sample. M=DNA molecular weight marker VIII (Boehringer)


102

stages included in the latter study, primordial germcells (and thus biallelic expression of imprinted genes)are hardly present. Most interestingly, however, the®nding of monoallelic expression of the maternal alleleof H19 in a residual mature teratoma, in contrast topreferential expression of the overrepresented andpaternal allele in the primary tumor, suggests a similarphenomenon as reported in di�erentiating parthenoge-netic and androgenetic stem cells (Feil et al., 1994),showing ®nal functional establishment of genomicimprinting upon di�erentiation. Speci®cally, thisindicates that the parental alleles of H19 in TGCTscan still be distinguished from each other, in agreementwith the ®nding of preferential loss of the paternalcopy of the short arm of chromosome 11 in some

TGCTs (Lothe et al., 1993), not compatable assumingcomplete erasement of genomic imprinting. Overall,these data suggest that retention of biallelic expressionof H19 in TGCTs is due to an intrinsic characteristic ofthe cell of origin, i.e., carcinoma in situ, being themalignant counterpart of an early (primordial?) germcell. In addition, the lack of establishment ofmonoallelic expression of H19 might be related to theretention of the embryonic characteristics of TGCTs.In spite of a consistent biallelic expression of H19 in

TGCTs, RNase protection analysis and mRNA in situhybridization revealed a striking correlation betweentotal level of H19 expression, di�erentiation lineageand maturation stage, highly similar to the pattern asreported during normal embryogenesis (Rachmilewitz

M un

dig

este

d D

NA

con

tro

l sam

ple

NS

1 D

NA

NS

1 cD

NA

1114 bp—900 bp—692 bp—

501,489 bp—404 bp—320 bp—

242 bp—190 bp—

1114 bp—900 bp—692 bp—

501,489 bp—404 bp—320 bp—

242 bp—190 bp—

M

M con

tro

l sam

ple

NS

5D

NA

NS

5cD

NA

NS

7D

NA

NS

7cD

NA

949 bp788 bp714 bp553 bp

235 bp

235 bp

1114 bp900 bp692 bp

501,489 bp404 bp320 bp242 bp190 bp

un

dig

este

d D

NA

DN

A f

ath

er

DN

A m

oth

er

DN

A p

atie

nt

tum

or

DN

A

met

asta

sis

DN

A

tum

or

cDN

A

met

asta

sis

cDN

A

un

dig

este

d c

DN

A

x Rsa I

949 bp788 bp714 bp

a b

c

Figure 7 H19 PCR products (HN9/10) digested with RsaI. (a) One nonseminomatous testicular germ cell tumor of adolescents andadults (NS1) shows strong preferential expression from the allele with the restriction site. (b) Two NS (cases 5 and 7) contained animbalance at the DNA level and show monoallelic expression of the overrepresented H19-allele. (c) Summary of the results ofparental origin analysis of H19 expression regarding case 5, showing data on DNA level of both parents, the patient, the primaryand the metastatic tumor, as well as the results on RNA level


103

et al., 1992a,b; Mutter et al., 1993; Lustig et al., 1994;Ohlsson et al., 1994; Ariel et al., 1995; Svensson et al.,1995; Walsh et al., 1995). While strong expression wasdetected in some immature somatic elements ofnonseminomatous-TGCTs, as well as in choriocarcino-ma and in reticular areas of yolk sac tumor, lowexpression was present in seminomas and embryonalcarcinoma, similar to primordial germ cells (Szabo andMann, 1995) and embryonal stem cells (Lustig et al.,1994; Szabo and Mann, 1994). These data imply, that

H19 might be a putative marker to identify immatureteratomatous as well as early di�erentiation of yolk sactumor-elements in TGCTs. The additional ®nding ofH19 expression in morphologically mature teratomaelements could indicate the presence of immatureelements and thus might have clinical implications,which serves further investigation. That H19 expressioncan indeed be of clinical importance was recentlyreported by us for bladder cancer, where a strongerand biallelic expression was found in cancers with amore malignant behavior (Ariel et al., 1995; Elkin etal., 1995; Cooper et al., 1996). The ®nding of H19expression in stromal elements as part of embryonalcarcinoma and yolk sac tumor, as well as in stromalcomponents surrounding the embryoid body is ofspecial interest. Similar ®ndings have been reported inbreast cancer (Dugimont et al., 1995). A regulatoryrole for H19 in proliferation and di�erentiation of bothnormal and malignant cells might therefore besuggested, as found for stromal elements in general(Donjacour and Cunha, 1991; Bosman et al., 1993;Howlett and Bissell, 1993). In conclusion, the patternof H19 as found in the testicular cancers of di�erentetiology included in this survey, as well as in cell lines(Rachmilewitz et al., 1995) and normal developmentcontradicts the supposed tumor suppressor function ofthis gene (Hao et al., 1993), but, in contrast, suggests adi�erentiation-lineage and maturation stage-speci®cpattern of expression. This is in agreement with theproposed oncofetal properties of this gene (Rachmile-witz et al., 1992b; Biran et al., 1994; Lustig et al.,1994).The process of erasement of genomic imprinting,

which occurs during germ cell maturation, and the ®nalestablishment of a new imprint, is not well understood.It is still a matter of debate whether both the maternaland paternal imprint are erased, or only the imprint ofthe opposite sex (Rossant, 1993). In this context thespermatocytic seminomas are of particular interest.Monoallelic expression of H19 found in the onlyinformative tumor indicates that at this developmentalstage of spermatogenesis the maternal and paternalallele can still be distinguished from each other, inagreement with data from mice (Szabo and Mann,1995). This again supports the hypothesis that in earlygerm cells (and malignant counterparts and derivatives)biallelic expression of imprinted genes is not due toerasement, but a result of functional ignorance of thenon-equivalence of the parental alleles. The lack ofH19-transcripts in semen (this paper) might be due tothe ®nal establishment of a uniparental pattern ofgenomic imprinting in these mature germ cells, or maybe due to cell type-speci®c inactivity. However, the lackof H19 expression in two spermatocytic seminomas(contralateral cases of one patient) and low expressionin the third suggests that inactivation of the H19 genemight indeed be established around this stage of germcell maturation. Our hypothesis regarding the status ofgenomic imprinting in germ cells at di�erent stages oftheir maturation, as well as their malignant counter-parts and derivatives, is schematically illustrated inFigure 9.In conclusion, we illustrated that LOI of H19 is not

a general event in the pathogenesis of testicularcancers. However, biallelic expression was detected inthe majority of TGCTs, with an expression pattern

Mun

dig

este

d

0 40 35 80 85 100

100

% tubules with CIS cells

DNA

cDNA

—949 bp—714 bp

—788 bp—553 bp

—235 bp

—235 bp

Figure 8 H19 PCR products (HN9/HN10) from CIS containingtesticular parenchyma samples adjacent to an invasive testiculargerm cell tumor of adolescents and adults. The upper panelillustrates heterozygosity for the RsaI site, and the lower panelshows the allelic pattern of expression. The percentage of CIS-containing seminiferous tubules are indicated

Figure 9 Schematic representation of the model of H19expression at di�erent stages of male germ cell development andtheir possible malignant counterparts and derivatives


104

resembling normal embryogenesis. We hypothesize thatthis is due to functional ignorance of the non-equivalence of the parental alleles of H19 inprimordial germ cells, which is retained in theirmalignant counterparts (carcinoma in situ) andderivatives (seminomas and nonseminomatous-TGCTs).

Materials and methods

Tissue samples

The testicular tumors were collected at the Laboratory ofExperimental Patho-Oncology in close collaboration withpathologists and urologists in the South-Western part ofthe Netherlands. Representative parts of the tumor and theadjacent parenchyma (when available) were snap frozen inliquid nitrogen (for DNA and RNA isolation) as well as®xed in 4% bu�ered formalin overnight at roomtemperature, and subsequently embedded in para�n. TheTGCTs were classi®ed according to the World HealthOrganization recommendations (Mosto® et al., 1987),while the Working Formulation was used for lymphomas(Rosenberg et al., 1982).

H19 probe and RNase protection analysis

A cDNA fragment of the human H19 gene (exon 5:position 3030 ± 3375) (Brannan et al., 1990) including thepolymorphic RsaI site (position 3238) was cloned intoSacI/SmaI digested PGEM-3Z plasmid (Promega). Togenerate the antisense probe, in vitro transcription of1 mg of plasmid DNA, in the presence of [a-32P]CTP wasperformed using Sp6 RNA polymerase after linearizationof the plasmid with EcoRI. As control, a sense probe wasgenerated similarly after linearization of the plasmid withHindIII and transcription using T7 RNA polymerase. Asreference for the amount of RNA used for the analysis, ag-actin antisense probe was constructed as follows. A129 bp HinfI/HindIII fragment (Enoch et al., 1986) wassubcloned into the SmaI/HindIII site of the PGEM4Zplasmid (Promega). The antisense probe was generated bylinearization of the plasmid with EcoRI and transcriptionwith T7 RNA polymerase in the presence of [a-32P]CTP.Subsequently, template DNA was removed by adding2.5 U of RQ1 DNase (Promega) for 20 min at 378C. Thelabeled probes were separated from the unincorporatednucleotides using the Quick Spin Columns Sephadex G50(Boehringer Mannheim).

From each tumor total RNA was isolated fromapproximately 10 frozen tissue sections of 50 mm thicknesseach, using RNA STAT-60 (TEL-TEST). Of each sample,two 5 mm sections (the ®rst and the last in the series) were cutand stained with hematoxylin and eosin for microscopicanalysis of the histological elements present. Five mg of RNAwas used for the analysis using the Ribonuclease ProtectionAssay Kit, RPA II (Ambion) according to the manufacturer'sdescription. After hybridization, the samples were treatedwith RNase (1 : 100) for 1 h at 378C. The samples wereloaded onto a 6% polyacrylamide/8M urea gel, andelectrophoresed for 3 h at a Voltage of 1800, after whichthe gel was vacuum-dried. Subsequently, exposure wasperformed to CEA RP ®lms (medical X-ray screen ®lm bluesensitive, Cea Corps) for various lengths of time at 7808C.

Interpretation of the autoradiographs was establishedusing a videodensitometer (2600, Biorad) with appropriatesoftware applications as recommended by the supplier.Within each lane, the relative intensity of the H19 signalwas determined compared to that of the g-actin signal, aftercompensating for background signal. Due to the RNase

conditions as used by us (optimalized for H19), the protectedfragment of the actin signal is composed of three bands. Thesmallest band was used as reference in the calculations, as theintensity of this band always fell within the linear range andcould be measured by the videodensitometer.

This analysis was performed on two normal testes, four B-cell lymphomas, one T-cell lymphoma (being a metastasis ofa nodal lymphoma), one Sertoli cell tumor, two Leydig celltumors, and three spermatocytic seminomas (two werecontralateral sequential tumors of one patient). In addition,the following TGCTs were studied: 24 seminomas (11contained hCG-positive mononuclear trophoblastic cells)and 13 nonseminomatous-TGCTs (these were diagnosed asthree pure EC, one as EC/YS, one as TE (mature), one TE(mature)/CH, two EC/TE (mature), one EC/TE (mature andimmature), one EC/TE (mature)/YS, and three EC/TE(mature and immature)/YS. Trophoblastic giant cells werefound in two cases, one containing EC/YS/mature TE, andone EC/TE (mature and immature)/YS.

H19 probe and mRNA in situ hybridization

Part of the human H19 gene with a length of 800 base pairs(bp) (Lustig et al., 1994) was subcloned at the EcoRIrestriction site into the Bluescript II KS plasmid(Stratagene, GMBH). After linearization of the plasmid,in vitro transcription was performed using T7 to producethe antisense H19 probe (plasmid linearized with HindIII),while T3 was used to produce the sense probe as control(plasmid linearized with EcoRI).

The in vitro generated transcripts were labeled with 35S(107 c.p.m./mg) using a commercially available kit (RPN2006, Amersham) with polymerases of Boehringer accordingto the manufacturer's recommendations. The labeled frag-ments were separated from unincorporated nucleotides byethanol precipitation.

For mRNA in situ hybridization, 5 mm thick para�ntissue sections were cut and mounted onto 3-aminopropyl-triethoxylane (Tespa) (Sigma) coated microscope slides anddried overnight at 378C. RNase treated tissue sections as wellas the sense probe were included as negative controls. Tissuesections were depara�nized with xylene and post-®xed with4% paraformaldehyde. Subsequently, the tissue sections weretreated with proteinase K (Sigma), after which they wereacetylated and dehydrated to reduce non-speci®c binding.Hybridization was performed as described (Ariel et al., 1994;Lustig et al., 1994). The tissue sections were exposed at roomtemperature for 10 days, after which they were counterstainedwith hematoxylin-eosin-light green. Interpretation was doneby microscopic analysis under bright and dark ®eldillumination by two pathologists individually.

The mRNA in situ hybridization was applied on 21 pureseminomas (®ve also studied by RNase protection analysis),11 pure nonseminomatous-TGCTs (six EC, one TE, threeYS, 25 mixed NS (one EC/TE, one TE/CH, one TE/YS/SE,one EC/SE, one EC/TE/SE, three EC/YS, three EC/TE/YS/SE (one containing intratubular SE), three TE/YS, and 11EC/TE/YS) (seven were also studied by RNase protectionanalysis). In addition, one Sertoli cell tumor, three Leydigcell tumors, two spermatocytic seminomas, ®ve centroblasticB-cell lymphomas and one T-cell lymphoma were investi-gated (one Leydig cell tumor and one B-cell lymphoma werenot studied by RNase protection analysis). The T-celllymphoma is most likely a metastasis from a nodal primarycancer. Embryoid bodies were detected in three NS,trophoblastic giant cells also in three, and Schiller Duvalbodies in three YS. In addition, testis parenchyma wasstudied from eight adults (two normal, one adjacent to aLeydig cell tumor, two adjacent to a seminoma and ®veadjacent to a NS). Immunohistochemistry con®rmed thepresence of CIS in the adjacent parenchyma of the TGCTs(not shown). Six residual extragonadal masses (three TE,


105

one EC/TE, one TE/CH and one TE/YS) after polyche-motherapy because of a testicular NS were included in thissurvey.

First strand cDNA synthesis and PCR

For detection of H19 expression, 5 mg of total RNA,isolated as described above, was reverse transcribed. cDNAwas generated at 378C for 2 h in a total volume of 40 mlcontaining 1 mM each dNTP (Pharmacia), 1 mM dithio-threitol, 1.2 mg random hexamer primers (d[N]6) (Phar-macia), 1.2 mg oligo d(T) primer d[T]12-18 (Pharmacia),4.5 U of RNasin (Pharmacia), 50 mM Tris-HCl (pH 8.3),75 mM KCl, 3 mM MgCl2 and 1 ml of Superscript RNAseH7 RT (BRL; 200 U/ml). For the subsequent PCR 2 ml wasused.

Ampli®cations were performed using 2 ml of the RTreaction in a total volume of 50 ml containing 16TaqDNA polymerase bu�er with 1.5 mM MgCl2, 100 pM ofeach primer, 250 mM each dNTP and 1 U of Taq DNApolymerase (Promega). Two primer sets were used forampli®cation of H19, HN7 and HN8, spanning intron 2and 3 (DNA fragment 542 bp, cDNA fragment 371 bp), andHN9 and HN10, spanning intron 3 and 4 (DNA fragment949 bp, cDNA fragment 788 bp). Primer positions are:forward primer HN7 (5'-CCAGGTCTCCAGCTGGGG-TG-3') and reverse primer HN8 (5'-CTTCCAGAGCC-GATTCCTCAGT-3') derived from positions 2253 ± 2272and 2773 ± 2794; and forward primer HN9 (5'-ACTTCCTC-CAGGGAGTCGGCA-3') and reverse primer HN10 (5'-TGATGATGAGTCCAGGGCTC-CT-3') derived from posi-tions 2526 ± 2546 and 3453 ± 3474 of the published H19sequence (which was renumbered by us, starting at one atthe beginning of the published sequence (Brannan et al.,1990)). After an initial denaturation of 4 min at 948C, everyampli®cation cycle consisted (between 30 and 35 cycles) of1 min at 948C, 2 min at 64 (HN7/8) or 668C (HN9/10), and2 min at 728C. Hypoxanthine Phospho-Ribosyl-Transferase(HPRT) primers used were 243 and 244 (Gibbs et al., 1989).PCR products were visualized on 2.5 ± 3% (50% regular and50% NuSieve GTG (FMC) agarose gels stained withethidium bromide.

The samples showing H19 expression were studied forinformativity of the polymorphic RsaI restriction site in exon5 (Zhang and Tycko, 1992), contained within primerset HN9and HN10. Therefore high molecular weight DNA wasisolated from peripheral blood of the patient (whenavailable), normal testicular parenchyma, or the tumorsample itself when no control tissue was available, usingproteinase K-sodium dodecyl sulfate treatment followed byphenol-chloroform extraction and ethanol precipitation(Maniatis et al., 1982). Again tissue sections were used toverify histological composition. Ampli®cation products (5 ±10 ml were digested to completion with 40 U of RsaIendonuclease (Pharmacia). Heterozygous samples showedan uncleaved band of 949 bp, and cleaved bands of 714 bpand 235 bp. The matched tumor RNAs were judged as

biallelic when the cDNA ampli®cation products afterdigestion showed the uncleaved 788 bp band as well as thecleaved 553 bp and 235 bp bands. Completion of digestionswas tested by always including a sample homozygous for theallele with the RsaI site. In addition at least threeindependent digests were done for each sample. Thisprotocol was applied to all testicular non-germ cell tumors,seminomas and nonseminomatous-TGCTs studied by RNaseprotection analysis. In addition, six seminomas and fournonseminomatous-TGCTs, two normal testis samples, seventesticular parenchyma without germ cells (after intensivehormonal treatment for sex-reversal), and three normalsemen samples and 20 testicular parenchyma samples withvariable numbers of carcinoma in situ containing seminifer-ous tubules were analysed.

For the quanti®cation of alleles by (RT ± )PCR hetero-zygous samples were digested with RsaI. For both DNA andcDNA only the two upper bands (the undigested upperallele and the largest band of the digested lower allele) wereincluded in calculating ratios. For DNA, a measured ratioof the intensity of the bands with lengths of 949 bp (Adna)and 714 bp respectively (Bdna) of 1 : 0.75 represents a 1 : 1ratio of the alleles. The ratio of DNA alleles=Adna/Bdna61.3=x. For cDNA, in case of biallelic expression, ameasured ratio of the intensity of the bands with lengths of788 bp (Acdna) and 553 bp (Bcdna) respectively of 1 : 0.7represents equal amounts of RNA expression from bothalleles. The ratio of expression of the alleles=Acdna/Bcdna61.43=y. The quotient of y/x indicates the relativeamount of expression per parental allele (1=similarexpression; 41=preferential expression from the upperallele; 51=preferential expression from the lower allele).In DNA samples from normal control tissues this resulted asexpected in a near 1 : 1 ratio of both alleles. Optical densitiesof PCR products on ethidium bromide stained gels weremeasured using the Bio-pro®l 4.6 software from Vilberlour-mat.

AbbreviationsCIS=carcinoma in situ; SE=seminoma; NS=nonsemino-matous-TGCT; EC= embryonal carcinoma; TE-teratoma;YS=yolk sac tumor; CH=choriocarcinoma.

AcknowledgementsWe thank all clinicians and pathologists employed at theDutch hospitals for their co-operation in providing tumorsamples and control tissues. This research was supportedby the Dutch Cancer Society (NKB) to AJMHV (NKB-DDHK project no. 94-867), MCD, RJHMvG, AJMG,JWO and LHJL and by a grant of the US Israel BinationalScience Foundation and a grant (no. 2429) of the Ministryof Health, State of Israel, Chief Scientist's o�ce to IA, TS,NdG and AAH.

References

Ariel I, Lustig O, Oyer CE, Elkin M, Gonik B, RachmilewitzJ, Biran H, Goshen R, de Groot N and Hochberg A.(1994). Gynecol. Oncol., 53, 212 ± 219.

Ariel I, Lustig O, Schneider T, Pizov G, Sappir M, de GrootN and Hochberg A. (1995). Urology, 45, 335 ± 338.

Bartolomei MS, Webber AL, Brunkow ME and TilghmanSM. (1993). Genes Dev., 7, 1663 ± 1673.

Barton SC, Ferguson-Smith AC, Fundele R and Surani MA.(1991). Development, 113, 679 ± 688.

Biran H, Ariel I, de Groot N, Shani A and Hochberg A.(1994). Tumor Biol., 15, 123 ± 134.

Bosman FT, De Bruine A, Van der Wur� A, Ten Kate J andDinjens WW. (1993). Int. J. Dev. Biol., 37, 203 ± 211.

Brannan CI, Dees EC, Ingram RS and Tilghman SM. (1990).Molec. Cell Biol., 10, 28 ± 36.

Brunkow ME and Tilghman SM. (1991). Genes Dev., 5,1092 ± 1101.

Burke AP and Mosto® FK. (1993). J. Urol. Pathol., 1, 21 ±32.

Cooper MJ, Fischer M, Komitowski D, Shevelev A, SchulzeE, Ariel I, Tyckocinski ML, Miron S, Ilan J, de Groot Nand Hochberg A. (1996). J. Urol., in press.


106

Cummings OW, Ulbright TM, Eble JN and Roth LM.(1994). Hum. Pathol., 25, 54 ± 59.

De Groot N and Hochberg A. (1993). Mol. Reprod. Devel.,36, 390 ± 406.

Dekker I, Rozeboom T, Delemarre J, Dam A and OosterhuisJW. (1992). Cancer, 69, 993 ± 996.

Dugimont T, Curgy JJ, Wernert N, Delobelle A, Raes MB,Joubel A, Stehelin D and Coll J. (1995). Molec. Cell, 85,117 ± 124.

Donjacour AA and Cunha GR. (1991). Cancer Treat. Res.,53, 335 ± 364.

Douc-Rasy S, Barrois M, Fogel S, Ahomadegbe JC, StehelinD, Coll J and Riou G. (1996). Oncogene, 12, 423 ± 430.

Eble JN. (1994). Hum. Pathol., 25, 1035 ± 1042.Elkin M, Shevelev A, Schulze E, Tyckocinsky M, Cooper M,Ariel I, Pode D, Kopf E, de Groot N and Hochberg A.(1995). FEBS Lett., 374, 57 ± 61.

Enoch T, Zinn K andManiatis T. (1986).Molec. Cell Biol., 6,801 ± 810.

Feil R, Walter J, Allen ND and Reik W. (1994).Development, 120, 2933 ± 2943.

Feinberg AP. (1993). Nature Genet., 4, 110 ± 113.Ferguson-Smith AC, Cattanach BM, Barton SC, BeecheyCV and Surani MA. (1991). Nature, 351, 667 ± 670.

Fundele RH and Surani MA. (1994). Develop. Genet., 15,515 ± 522.

Gibbs RA, Ngyuen P-N, McBride LJ, Koepf SM and CaskeyCT. (1989). Proc. Natl. Acad. Sci. USA, 86, 1919 ± 1923.

Goodfellow PN and Andrews PW. (1982). Nature, 300, 107 ±108.

Hall JG. (1990). Am. J. Hum. Genet., 46, 857 ± 873.Hao Y, Crenshaw T, Moulton T, Newcomb E and Tycko B.(1993). Nature, 365, 764 ± 767.

Hibi K, Nakamura H, Hirai A, Fujikake Y, Kasai Y,Akiyama S, Ito K and Takagi H. (1996). Cancer Res., 56,480 ± 482.

Howlett AR and Bissell MJ. (1993). Epithel. Cell Biol., 2,79 ± 89.

Jùrgensen N, Giwercman A, MuÈ ller J and Skakkebñk NE.(1993). Histopathol., 22, 373 ± 378.

Jùrgensen N, Rajpert-De Meyts E, Graem N, Mueller J,Giwercman A and Skakkebñk NE. (1995). Lab. Invest.,72, 223 ± 231.

Kondo M, Suzuki H, Ueda R, Osada H, Takagi K andTakahashi T. (1995). Oncogene, 10, 1193 ± 1198.

Latham KE. (1995). Di�erentiation, 59, 269 ± 282.Leighton PA, Ingram RS, Eggenschwiler J, Efstratiadis Aand Tilghman SM. (1995a). Nature, 375, 34 ± 39.

Leighton PA, Saam JR, Ingram RS, Stewart CL andTilghman SM. (1995b). Genes Dev., 9, 2079 ± 2089.

Lothe RA, Hastie N, Heimdal K, FossaÊ SD, Stenwig AE andBùrresen AL. (1993). Genes Chromos. Cancer, 7, 96 ± 101.

Lustig O, Ariel I, Ilan J, Lev-Lehman E, De Groot N andHochberg A. (1994). Molec. Reprod. Develop., 38, 239 ±246.

Maniatis T, Fritsch EF and Sambrook J. (1982). Isolation ofhigh molecular-weight, eukaryotic DNA from cells grownin tissue culture. Molecular Cloning, New York, ColdSpring Harbor Laboratory, p. 280.

Moore T and Haig D. (1991). Trends Genet., 7, 45 ± 49.Mosto® FK, Sesterhenn IA and Davis CJJ. (1987). Sem.

Diagn. Pathol., 4, 320 ± 341.Moulton T, Crenshaw T, Hao Y, Moosikasuwan J, Lin N,Dembitzer F, Hensle T, Weiss L, McMorrow L, Loew T,Kraus W, Gerald W and Tycko B. (1994). Nature Genet.,7, 440 ± 447.

Mutter GL, Stewart CL, Chaponot ML and Pomponio RJ.(1993). Am. J. Hum. Genet., 53, 1096 ± 1102.

Nakashima N, Murakami S, Fukatsu T, Nagasaka T,Fukata S, Ohiwa N, Nara Y, Sobue M and Takeuchi J.(1988). Hum. Pathol., 19, 1144 ± 1154.

Ohlsson R, Hedborg F, Holmgren L, Walsh C and EkstroemTJ. (1994). Development, 120, 361 ± 368.

Rachmilewitz J, Goshen R, Ariel I, Schneider T, de Groot Nand Hochberg A. (1992a). FEBS Lett., 309, 25 ± 28.

Rachmilewitz J, Gileadi O, Eldar-Geva T, Schneider T, deGroot N and Hochberg A. (1992b). Molec. Reprod.Develop., 32, 196 ± 202.

Rachmilewitz J, Elkin M, Rosensaft J, Gelman-Kohan Z,Ariel I, Lustig O, Schneider T, Goshen R, Biran H, deGroot N and Hochberg A. (1995).Oncogene, 11, 863 ± 870.

Rajpert-De Meyts E and Skakkebñk NE. (1994). Int. J.Androl., 17, 85 ± 92.

Reed ML and Le� SE. (1994). Nature Genet., 6, 163 ± 167.Reik W, Brown KW, Slatter RE, Sartori P, Elliott M andMaher ER. (1994). Hum. Molec. Genet., 3, 1297 ± 1301.

Romanenko AM and Persidskii YV. (1983). VoprosiOnkologii, 19, 61 ± 66.

Rosenberg S, Berard C, Brown B, Burke J, Dorfman R,Glatstein E, Hoppe R and Simon R. (1982). Cancer, 49,2112 ± 2135.

Rossant J. (1993). Current Biol., 3, 47 ± 49.Sasaki H, Ferguson-Smith AC, Shum ASW, Barton SC andSurani MA. (1995). Development, 121, 4195 ± 4202.

Skakkebñk NE, Berthelsen JG, Giwercman A and MuÈ ller J.(1987). Int. J. Androl., 10, 19 ± 28.

Solter D. (1988). Ann. Rev. Genet., 22, 127 ± 146.Solter D. (1992). Curr. Opinion Biotechn., 3, 632 ± 636.Steenman MJC, Rainier S, Dobry CJ, Grundy P, Horon ILand Feinberg AP. (1994). Nature Genet., 7, 433 ± 439.

Svensson K, Walsh C, Fundele R and Ohlsson R. (1995).Mech. Develop., 51, 31 ± 37.

Szabo PE and Mann JR. (1994). Development, 120, 1651 ±1660.

Szabo PE and Mann JR. (1995). Genes Dev., 9, 1857 ± 1868.Talerman A, Fu YS and Okagaki T. (1984). Lab. Invest.,51(3), 343 ± 349.

Taniguchi T, Sullivan MJ, Ogawa O and Reeve AE. (1995).Proc. Natl. Acad. Sci. USA, 92, 2159 ± 2163.

Tycko B. (1994). Am. J. Pathol., 144, 431 ± 443.Van Gurp RJLM, Oosterhuis JW, Kalscheuer V, MarimanECM and Looijenga LHJ. (1994). J. Natl. Cancer Inst., 86,1070 ± 1075.

Walsh C, Miller SJ, Flam F, Fisher RA and Ohlsson R.(1995). Cancer Res., 55, 1111 ± 1116.

Zhang Y, Shields T, Crenshaw T, Hao Y, Moulton T andTycko B. (1993). Am. J. Hum. Genet., 53, 113 ± 124.

Zhang Y and Tycko B. (1992). Nature Genet., 1, 40 ± 44.


107

Documents

Unique expression patterns of H19 in human testicular cancers of different etiology