Genomic Imprinting and Igf2 Influence Liver Tumorigenesis ......imprinted genes on distal chromosome 7, Igf2 and ///'>. are re-expressed in most liver tumors from an SV40 T/t antigen

[CANCER RESEARCH 57. 4615-4623. October 15. 1997]

Genomic Imprinting and Igf2 Influence Liver Tumorigenesis and Loss ofHeterozygosity in SV40 T Antigen Transgenic Mice1

Ramsi Haddad and William A. Held2

Department of Molecular and Cellular Biology. Rum-eli Park Cancer Institute. Buffalo. New York. 14263

ABSTRACT

Maternal-specific loss of heterozygosity (LOH) and allelic imbalances[i.e., partial I.oil (pLOH)] observed in SV40 T/t antigen-induced liver

tumors suggests that an imprinted gene on chromosome 7 is involved inliver tumorigenesis. Maternal-specific I.OH/pLOH may reflect the loss of

a maternally expressed tumor suppressor gene or the acquisition of paternally active alÃ-elesof a growth promoter. In addition, two oppositelyimprinted genes on distal chromosome 7, Igf2 and ///'>. are re-expressed

in most liver tumors from an SV40 T/t antigen transgenic line (M11T-G).IK.I'- ¡sa paternally expressed growth promoter, and HI 9 isa maternally

expressed gene that can suppress growth in some tumor cell lines. Westudied the role of Igf2 during liver tumorigenesis by creating Igf2 ( +/—)

M11T-G mice. These mice are essentially null for Igf2 expression becauseimprinting normally precludes maternal IK!- expression. M11T-G, Igf2(+/—) males exhibit a 15-fold reduction in the frequency of large tumors.Igf2 (+/-) tumors do not express maternal Igf2, indicating rigid im

printing control in the liver. LOH/pLOH analysis was performed on thetumors and indicates that acquisition of paternally active IK!- alÃ-elesis amajor selective event for M11T-G liver tumorigenesis. This also implies

the existence of an imprinted, maternally expressed tumor suppressorgene on chromosome 7 that is unlikely to be IIIV.

INTRODUCTION

Cancer represents the accumulation of genetic and epigeneticchanges that lead to aberrant and uncontrolled cell growth. Thepromoter-directed expression of oncogenes, such as the SV40 TAg ' in

transgenic mice, has permitted the construction of mouse transgeniclines that reproducibly develop tumors in the liver with definedkinetics (1,2). Using the mouse MUP promoter to direct the expression of a MUP-TAg hybrid gene, transgene expression commences in

the liver at 3 weeks of age and remains active for the life of the animal(1).

The liver histopathology of MUP-TAg mice follows a defined

pattern divisible into three stages. TAg expression becomes activatedat puberty and leads to dysplasia and apoptosis of the original hepa-

tocyte population (2). At the same time, multiple foci of small cellsappear throughout the liver parenchyma (2). The original hepatocytescontinue to undergo apoptosis as the hyperplastic foci of small cellsgrow to confluence, completely replacing normal cells within theliver. The last step involves the appearance of multiple neoplasticnodules that subsequently encapsulate, vascularize, and become tumors (2).

Tumorigenesis in MUP-TAg transgenic mice, as well as other

Received 4/24/97; accepted 8/13/97.The costs of publication of this article were defrayed in part by the payment of page

charges. This article must therefore be hereby marked advertisement in accordance with18 U.S.C. Section 1734 solely to indicate this fact.

1This research was supported by National Institute of General Medical Sciences Grant

GM51881-02 (to W. A. H.) and by National Cancer Institute Core Grant 5P30CA16056(to Roswell Park Cancer Institute).

2 To whom requests for reprints should be addressed, at Department of Cell and Molecular

Biology. Roswell Park Cancer Institute. Elm and Carlton Streets. Buffalo. NY 14263. Phone:(716)845-3301; Fax: (716)845-8169; E-mail: [email protected].

' The abbreviations used are: TAg. large T antigen: MUP. major urinary protein: LOI.

loss of imprinting: LOH. loss of heterozygosity; pLOH. partial LOH; pLOHM. pLOHmaternal-specific; pLOHP. pLOH paternal-specific; BWS. Beckwith-Wicdemann syn

drome: RIP. rat insulin promoter; MIT. Massachusetts Institute of Technology: SSLP.simple sequence-length polymorphism.

oncogene-bearing transgenic mice, typically have a latent period

between expression of the transgene and noticeable neoplastic growth,suggesting that additional genetic events are required for tumor progression (reviewed in Ref. 3). Many types of genetic changes havebeen characterized in SV40 TAg-induced tumors, including large

DNA deletions, amplifications, and rearrangements (4).A subset of genetic alterations that occur in both human and mouse

tumors are perturbations in genomic imprinting or LOI control (3, 5,6). Genomic imprinting is an epigenetic mark that allows parentalgenomes to be distinguished in the offspring. Imprinting can result inmonoallelic gene expression that can be developmental and tissuespecific (7).

Two lines of evidence indicate the importance of imprinting inMUP-TAg liver tumors:

(a) The distal imprinted region of mouse chromosome 7 exhibitsmaternal-specific LOH and allelic imbalances (i.e., pLOH) in theseliver tumors (8). Maternal-specific LOH/pLOH could indicate the loss

of a maternally expressed tumor suppressor gene. Alternatively, maternal-specific LOH/pLOH could be accompanied by paternal dis-

omies and be driven by selection for an increase in the copy numberof a paternally expressed growth promoter. In fact, both genetic eventscould be occurring simultaneously.

(b) Igf2 and Hi 9, two imprinted genes on chromosome 7 notnormally expressed in adult liver, are re-expressed in most MUP-TAg

liver tumors (8, 9). Similar observations implicating a role for imprinting in liver tumorigenesis are also observed in transgenic miceexpressing SV40 TAg under control of the liver-specific C-reactive

protein (10).Igf2 and H19 are oppositely imprinted genes that are closely linked

on distal mouse chromosome 7 (11). Both genes are primarily activeduring fetal development, and expression in most tissues is undetect-

able after 2 weeks of age (11). Igf2 is a fetal growth factor expressedfrom the paternal alÃ-elein most mouse tissues (12). H19 is a maternally expressed gene with no conserved open reading frame (13),which exhibits tumor suppressor activity in tissue culture (14). Ectopie expression of HI 9 in G401 cells causes the cells to lose tumor-

igenicity. as determined by a reduction in colony formation in softagar and the loss of growth of tumors in nude mice (14).

Thus, the maternal-specific LOH/pLOH observed on distal chromosome 7 in tumors from the MUP-TAg transgenic line Ml IT-G (8)

could be driven by the acquisition of paternal chromosomes with anactive Igf2 alÃ-eleand/or the loss of maternal chromosomes with anactive HÃŒ9alÃ-ele.It remains possible that an unidentified maternallyexpressed tumor suppressor gene or a paternally expressed growth-

promoting gene reside on distal chromosome 7./g/2 is a fetal growth factor that stimulates growth through the

IGF1 receptor (15). The IGF2 receptor, which also serves as thelysosomal mannose-6-phosphate receptor, appears to negatively reg

ulate growth by targeting Igf2 to digestive lysosomes (16). ¡GF2hasbeen implicated in tumorigenesis in human cancers associated withBWS (5). A maternal loss or paternal disomy of human chromosomeIlpl5.5, which is syntenic with distal mouse chromosome 7, isfrequently associated with BWS (17). BWS patients often re-expressIGF2 biallelically in Wilms' tumors, indicating LOI control (6, 18,

19). Hepatoblastomas also re-express IGF2, although LOI control

4615

on July 7, 2021. © 1997 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

http://cancerres.aacrjournals.org/

EFFECTS OF Igf2 DELETION ON LIVER TUMORIOENESIS

occurs infrequently (5, 20-22). Other tumor types such as testicular 5'-GGGTGAGCCATTCTCCTGGG-3'. Two reverse primers were included:

germ cell tumors (23) and malignant breast tumors frequently re-

express ÃŒGF2transcripts biallelically (24).Igf2 is also implicated in tumors derived from transgenic mice

expressing SV40 TAg under control of the RIP (RIP-TAg mice),which develop insulomas that re-express Igf2 (25). Mating RIP-TAgmice with Igf2 null mutants resulted in RIP-TAg, Igf2 (—/-) tumors

that were much smaller than tumors from RIP-TAg. Igf2 (+/+)animals (25). Further analysis demonstrated that RIP-TAg, Igf2(+/—) tumors all re-activate the maternal Igf2 alÃ-eleand are not

significantly different in size than RIP-TAg, Igf2 (+/+) tumors (26).

To investigate the requirement for Igf2 during liver tumorigenesisin MUP-TAg mice, we crossed MI1T-G female mice with Igf2(-/+) heterozygous males (27). Inheritance of a null paternal alÃ-ele

results in mice effectively null for Igf2 because imprinting normallyprecludes maternal Igf2 expression in the liver. Liver tumors fromIgf2 (+/—)heterozygous null. TAg-positive animals were dissected,

measured, and analyzed for Igf2 and HI9 expression. LOH/pLOHanalysis was also performed using SSLP defined by MIT primer pairs(28).

Our results indicate that M11T-G, Igf2 (+/-) mice exhibit adramatic reduction in the frequency of tumors relative to Ml 1T-G,

¡gf2( +/+ ) litter mates and that Igf2 LOI does not occur in most livertumors. In addition, genetic analysis of distal chromosome 7 suggeststhat acquisition of paternally active Igf2 alÃ-elesis a major selectivepressure for LOH/pLOH. We also present evidence that is consistentwith a maternally expressed tumor suppressor gene on chromosome 7and that HI9 is unlikely to be this tumor suppressor.

MATERIALS AND METHODS

The Ml IT-G transgenic line expressing SV40 TAg under control of the

MUP promoter has been described (8). Dissections were performed on animalswhen liver hyperplasia was evident. Tumors were dissected by excising tissuefrom the surrounding liver and subsequently removing the capsule. In this way,tumor tissue was not contaminated by potentially normal tissue that makes upthe capsule. The tissue was then flash frozen in dry ice and stored at -70°C

until needed.Total RNA and total DNA for Northern analysis and PCR LOH/pLOH

analysis was prepared from tumor tissue according to the manufacturer's

protocol for TRIzol Reagent (Life Technologies. Inc.). Probes for Northern andSouthern Blots were synthesized according to the manufacturer's protocol

using either Prime-It II (Stratagene) or Random Primed DNA Labeling Kit(Boehringer Mannheim). |a-'2P]dATP or dCTP was used to label products

(DuPont NEN). The Igf2 probe used was a 0.7-kb £coRIcDNA fragment fromplasmid pMT5 (9). The HÌ9probe used was a 0.35-kb Pst\ cDNA fragmentfrom plasmid pH19 (8). Cck probe was a 0.8-kb Pnill-Pst\ cDNA fragment

from plasmid pCCK.bS (29).Northern blots were performed as described (8). Quantitative Southern

analysis was performed using total DNA prepared as described (8). Blots werequantified on a Molecular Dynamics Storm Phosphorlmager.

LOH/pLOH analysis by PCR was performed in 96-well microtiter plates(Falcon) using a PTC-100 Thermocycler (M. J. Research, Inc.). Reactions

were assembled by adding approximately KM)ng DNA (in 1 fil) to each welland then adding the following solutions: 5.3 /xl water, 1.0 pA(50% sucrose, 1%cresol red). 1 /nl PCR buffer (Boehringer Mannheim). 0.2 /j.1 10 mM de-

oxynucleotide triphosphate (Pharmacia Biotech. Inc.), 0.24 p,l each of forwardand reverse primers (6 mM), 0.05 n\ [10% (w/v) BSA|. and 0.1 /¿ITaqpolymerase (0.5 unit; Boehringer Mannheim). Reactions were then coveredwith mineral oil (Sigma Chemical Co.). Cycling started with 3 min at 94°Cfollowed by 32 cycles of 1 min at 94°C,2 min at 55°C,and 3 min at 72°C.After cycling, the plate was raised to 94°Cfor 1 min, 55°Cfor 2 min, and afinal extension at 72°Cfor 10 min. Samples were separated on 8 or 10% TBE

acrylamide gels.The set of primers used to detect LOH/pLOH at the Igf2lneo locus con

sisted of a forward primer in the first intron of ¡gf2 with the sequence

one hybridized to Igf2 exon 2: and the other hybridized to neo. The reverseIgf2 primer was S'-GTCAACAAGCTCCCCTCCGC-S', and the reverse neoprimer was 5'-CCCCGACTGCATCTGCGTGT-3'. The Igf2 product was 200

bp, and the neo product was 290 bp in length.The Mann-Whitney U test was used to determine if the frequency distribu

tion of tumors for each of the four size classes differed significantly betweenthe M11T-G, Igf2 (+/-) and M11T-G, Igf2 (+/+) animals in Table 2. The

test was performed for every tumor size class listed in Table 2. Each test wasperformed by sorting animals in ascending order with respect to the number oftumors of the particular size class in question. The test statistic was thencalculated by determining the number of instances that an M11T-G, Igf2(+/-) animal precedes an animal of genotype MI1T-G, Igf2 (+/+). The

resultant value of U was then compared to a table of critical values as described(30).

The statistical significance of the data in Table 4 was performed in thestandard manner using the x2 test of goodness of fit (30). The statisticalsignificance of data in Table 6 was determined by the ^2 test of independence,

which is used for frequency distributions of categorical groupings (i.e., noLOH, pLOHM. LOHM, or pLOHP; Ref. 30). Expected frequencies werecalculated from the marginal totals.

RESULTS

Ml IT-G females (8) were mated with Igf2 (-/+) males (27) to

study the influence of ¡gf2on MUP-TAg tumorigenesis. The cross

and the relevant genotype and phenotype of the progeny is shown(Table 1A). One-half of the F, generation are expected to be Ml IT-Gpositive and develop liver tumors. Also, one-half the F, generation areexpected to be Igf2 (+/—), and the remaining one-half to be Igf2

(+/+). The cross will, therefore, be expected to yield 25% Ml IT-G,Igf2 (+/+) and 25% Ml IT-G. Igf2 (+/-) progeny. Inheritance of anull paternal alÃ-eleof Igf2 results in mice effectively null for ¡gj'2

because imprinting normally precludes maternal Igf2 expression inthe liver (27). As expected, all Igf2 (+/—) mice were approximately

60% the size of their Õgf2(+/+) litter mates (data not shown: Ref.27).

Table IB is a summary of phenotypes in the 43 MI1T-G Fl

animals tested. Male and female animals were analyzed separately dueto known sex-specific differences in tumorigenesis in both human and

other mouse models of liver tumorigenesis (31, 32). An approximatelyequal number of M11T-G, Igf2 (+/+) and M11T-G, Igf2 (+/-)

animals were studied (Table IB). The average age at dissection wassimilar for both Igf2 genotypes for males and females, respectively.

Table I Genetic cross and summary of F¡animals analy-fil

A. Genetic cross and expected progeny genotype and phenotype

Genetic crossMI1T-G. Igf2 (+/+) female

Expected progeny:F¡genotype

X Igf 2 (+/-) male

F, Igf 2 phenotype F} tumor phenotype

M11T-G.M11T-G.,

Igf2(+/+), Igf2(+/-)Normal

TumorsNull TumorsNormal No tumorsNull NotumorsB.

Summary of Ml IT-G Fl animalsanalyzedIgf

2 genotype

Males Females

EFFECTS OF IgßDELETION ON LIVER TUMORIGENESIS

Female animals were dissected at a higher average age because therate of Ml 1T-G tumorigenesis is slower in females.

Igf2 Is Not Required for TAg-induced Liver Hyperplasia. As

mentioned previously, TAg expression causes liver hyperplasia, andtumors subsequently appear on the hyperplastic liver background (2).We wanted to determine whether Igf2 has an effect on the TAg-

induced liver hyperplasia by measuring the livenbody weight ratio ofanimals. The livenbody weight ratio of an animal represents thesummation of liver hyperplasia and tumor development. The averagelivenbody weight ratio of a nontransgenic animal is 0.05. The averagelivenbody weight ratio of Ml 1T-G animals is 0.35 to 0.41 and is not

significantly affected by sex or /#/2 genotype (Table IB).Fig. 1 shows the change in livenbody weight over time in Ml IT-G

mice. Nontransgenic, control animals have a constant livenbodyweight ratio of 0.05. Curved lines on the graph reveal the progressivehyperplasia and subsequent tumorigenesis exhibited by M11T-Gmice. The change in livenbody weight in male Ml IT-G, Igf2 (+/+)mice mimics the time course seen in male M l IT-G, Igf2 (+/-) mice.

In a like manner, the change in livenbody weight in Ml IT-G females

follows a similar curve, irrespective of their Ig/2 genotype. As discussed below, tumor development is greatly reduced in Ml IT-G, Igf2(+/-) animals. We, therefore, conclude that Igf2 is not required forthe TAg-induced liver hyperplasia in Ml IT-G mice.

M11T-G, Igf2 (+/—) Animals Have a Reduced Frequency ofLarge Tumors. Dissections revealed that Ml IT-G. Igf2 (+/—) tu-morigenic livers were visibly different from their M11T-G, ¡gf2(+/+) counterparts (Fig. 2). An M11T-G, Igf2 (+/+) male liver

shows extensive liver hyperplasia and many tumors and nodules ofdifferent sizes (Fig. 2. A and B). Under identical magnification, anM11T-G, Igf2 (+/-) male litter mate shows that extensive liver

hyperplasia remains evident, but the number of tumors and nodules isconsiderably reduced (Fig. 2, C and D).

Tumors were measured and counted at dissection to quantify thedifferences in tumor size and frequency. A summary of tumor sizedistribution and frequency for male and female animals is presented(Table 2). Each tumor size class is sorted by the average number oftumors per animal for each Igf2 genotype. As shown in Table 2A, thelargest class of tumors appears 15-fold less frequently in M11T-G.Igf2 (+/-) males relative to M11T-G, Igf2 (+/+) males. Smaller

tumor size classes also show reductions of 3.4-6.6-fold (Table 2A).

In females, the frequency of tumors greater than 8.0 mm is reduced5-fold in Ml IT-G, Igf2 (+/-) relative to their Ml IT-G. Igf2 (+/+)

counterparts. Similarly, tumors 6.1-8.0 mm are reduced 4.2-fold, andtumors of 4.1-6.0-mm diameter appear 3.5-fold less frequently.Ml IT-G, Igf2 (+/-) females manifest the same trend seen in males;

the largest tumor size class exhibits the greatest reduction infrequency.

Igf2 Imprinting Control Is Rigidly Maintained in Liver andLiver Tumors from Ml IT-G, ¡gf2(+/- ) Animals. If Igf2 expres

sion has an absolute causal role in liver tumorigenesis, LOI controlcould result in the activation of the normally silent maternal alÃ-eleintumors from Igf2 (+/—) mice. Because M11T-G. Igf2 (+/—) ani

mals receive a null paternal Igf2 alÃ-ele,the presence of functionalIgf2 transcripts can only be derived from the maternal alÃ-ele.Consequently, any Igf2 expression observed in Ml IT-G, Igf2 (+/—) liver

or liver tumors can only be the result of LOI.We first examined hyperplastic. Ml IT-G livers at 6 and 8 weeks of

age. Because M11T-G. Igf2 (+/-) animals show the same relative

degree of liver hyperplasia as their Igf2 (+/+) litter mates (Fig. 1 andTable IB), an LOI event resulting in maternal Igf2 expression mightbe responsible for the liver hyperplasia seen in these animals. Igf2transcript levels measured by Northern blot in Ml IT-G, Igf2 (+/—)

hyperplastic liver samples do not have detectable levels of Ii>f2 at 6or 8 weeks of age (data not shown). Igf2 expression in Ml IT-G, Igf2

(+/+) liver is also undetectable at 6 weeks of age but begins at 8weeks and remains activated (data not shown). These results indicatethat the liver hyperplasia observed in Ml IT-G, Igf2 (+/-) liver is not

due to ¡gf2expression from the maternal alÃ-ele.Consequently, imprinting control is not affected in preneoplastic. hyperplastic liver. Inaddition. Ml IT-G, ¡gf2(+/+) livers have extensive hyperplasia priorto re-expression of Igf2, further indicating that liver hyperplasia is not

dependent on Igf2 expression.Igf2 transcript levels in the liver tumors were analyzed to deter

mine whether imprinting control had been disrupted. H19 transcriptlevels were also assayed because in Wilms' tumors with biallelic Igf2

expression, H19 expression is lost (18). In addition, the targeteddeletion of the HÌ9gene by homologous recombination results inbiallelic Igf2 expression (33).

Representative Northern blots of tumor RNA were hybridized withan Igf2 cDNA probe (Fig. 3; Ref. 9). The blots were then stripped andsubsequently hybridized with an H19 cDNA probe. Photographs ofethidium bromide-stained RNA preceding blotting indicates equal

loading of RNA (data not shown).The ¡gf2 gene in mice has three discernible RNA transcripts in

0.5

0.4o4-¡(B

Fig. 1. Effect of Igf2 (+/-) alÃ-eleon Ml IT-G liver: >,body weight ralio. Each point represents one individual "5

mouse. —¿�

0.1

00

D Igf2 (+/-) Male

•¿�Igf2 (+/+) Male

o Igf2 (+/-) Female

•¿�Igf2 (+/+) Female

A Igf2 (+/+)/lgf2 (+/-) Control

0| 10 20TAg Expression A (weeks)

Activated

30

4617

on July 7, 2021. © 1997 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

http://cancerres.aacrjournals.org/

EFFECTS OF Igf2 DELETION ON LIVER TUMORIGENESIS

A) M11T-G, Igf2 (+/+), anterior view B) M11T-G, Igf2 (+/+), posterior view

Fig. 2. Visible effect of Igf2(+l-) on the number and size oftumors in an M11T-G liver. A and B, anterior and posterior viewsof an Ml 1T-G. Igf2 (+/+) male at 16 weeks of age. C and D.similar views of an MI 1T-G. Igf2 (+/-) male littermate at 16weeks of age. E and F, two views of a nontransgenic. Igf2 ( +/—)

female littermate at 16 weeks of age. All photographs were takenunder identical magnification.

C) M11T-G, Igf2 (+/-), anterior view D) M11T-G, Igf2 (+/-), posterior view

E) Igf2 (+/-), anterior view F) Igf2 (+/-), posterior view

Northern blots (Fig. 3). M11T-G, Igf2 (+/+) tumors frequently

express ¡gf2transcripts to various levels (Fig. 3). Strikingly, only twoM11T-G, Igf2 (+/—) tumors of 87 analyzed express the maternal

Igf2 alÃ-ele.Tumors 78/4-5 and 29/4-5 express the maternal Igf2 alÃ-ele

Table 2 Effect of Igf 2 {+/—) on the distribution of iunior size and frequency

A. Males

Average tumorno./animalIgf2

(+/+) Igf2(+/-) Igf2 (+/+):lgf2(+/-)Tumordiameter(mm)>8.06.1

-8.04.1-6.03.0

-4.0tf-10"22.25.619.8N

=150.130.331.73.9ratio15"6.6"3

4*5.1"B.

Females

Average tumor no./animal

Tumor diameter(mm)>8.06.1

-8.04.1-6.03.0

-4.0N

=93.84.29.218.9N =80.712.67.92(+/+):Igf2 (+/-)

ratio5"4.2*3.5*2.4

' N. number of animals.' Statistically significant value at P = 0.001 with Mann-Whitney U test.

to similar levels seen in Ml 1T-G, Igf2 (+/+) tumors (Fig. 3). Thefrequency of Hi9 expression in Ml 1T-G tumors did not depend on

Igf2 genotype (Fig. 3). In particular, of the two tumors that expressmaternal 7g/2, 78/4-5 expresses high levels of H19, and 29/4-5

expresses no detectable HI9. This suggests that, contrary to thesituation observed in Wilms' tumor (18), maternal Igf2 expression

does not preclude HI 9 expression in the liver.Table 3 summarizes the frequency of Igf2 and Hi 9 expression

from 87 Igf2 (+/-) and 108 Igf2 (+/+) tumors. The vast majority

(91%) of M11T-G, Igf2 (+/+) tumors express Igf2, whereas onlytwo Ml 1T-G, Igf2 (+/-) tumors (2%) express maternal Igf2. There

fore, loss of Igf2 imprinting control is a very rare event in Ml 1T-G,Igf2 (+/—) tumors. The frequency of HÌ9expression in M11T-G

tumors is independent of Igf2 genotype (Table 3). There is no clearand discernible correlation between tumor size and Igf2 expressionlevels or between Igf2 and H19 expression levels (data not shown).Because Igf2 LOI is common in many other tumor types (6, 18, 19,23, 24, 26), these results suggest that imprinting control in livertumors is much more rigid than in tumors originating in other tissues.

Maternal-specific LOH and Allelic Imbalance on Distal Chromosome 7 Remains Evident in M11T-G, Igf2 (+/—) Tumors atOne-Half the Frequency. Because M11T-G, Igf2 (+/+) liver tumors exhibit maternal-specific LOH/pLOH on distal chromosome 7

(8), we wanted to determine whether LOH/pLOH was altered in the

4618

on July 7, 2021. © 1997 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

http://cancerres.aacrjournals.org/

EFFECTS OF Igf2 DELETION ON LIVKR TUMORIGENESIS

context of an /#/2 (+/—) heterozygous null genetic background. If

chromosome 7 maternal-specific LOH/pLOH represents selection for

the loss of a maternally expressed growth suppressor gene such asHÌ9(14), then M11T-G, Igf2 (+/-) tumors would be expected to

maintain maternal-specific LOH/pLOH. Alternatively, maternal-spe

cific LOH/pLOH may result in paternal disomies and be driven byselection for the acquisition of paternal alÃ-elesof a paternally expressed growth promoter such as Igf2. In this case, M11T-G, Igf2(+/—) tumors would no longer be expected to exhibit maternal-

specific LOH/pLOH. Maternal-specific LOH/pLOH could also rep

resent a combination of both loss of a maternally expressed tumorsuppressor and acquisition of a paternally expressed growth promoter.

LOH/pLOH analysis was performed using SSLPs defined by MITprimer pairs (28). The females used in this cross were generated bybackcrossing M11T-G males with (C57BL/10 X C3H)F, (BCF1)females. Males in this cross of genotype Igf2 (-/+) were derived

from strain 129/Sv. Consequently, primer pairs defining SSLPs arerequired to differentiate the two possible maternal alÃ-eles(C57BL/10or C3H) from the paternal alÃ-ele(129/Sv).

Fig. 4A is a schematic of the map positions of the MIT primer pairsused for the analysis. Fig. 4B shows typical ethidium bromide-stained

acrylamide gels of PCR reactions at three representative MIT loci.MIT primer pairs were chosen based on coverage of the entirechromosome and the ability to differentiate between maternal andpaternal alÃ-eles.As indicated in Fig. 44, Igf2 and HI9 are approximately 67 cM distal to the centromere (11). The Igf2 knockout usedin these experiments was generated by insertion of a neo cassette intothe coding region of Igf2 (27). A set of primers was, therefore, createdthat could differentiate the maternal Igf2 from the paternal neoinsertion in Ml 1T-G, Igf2 (+/—) tumors.

Control PCR reactions were performed with varying ratios ofpurified maternal and paternal strain-specific DNA (data not shown).

MUP/SV40 T Antigen Liver Tumors

Igl2(W-) («/«)

c < T- w n

£ c a a a

.- CM o

Igl2

f «•«

MUP/SV40 T Antigen Liver Tumors

|gf2 (W.)

Fig. 3. Representative Northern hlots of Inf2 and HIV transcripts. Ten ¿ig"f totaltumor RNA were probed with a 0.7-kb EcnR\ ¡g/2cDNA probe or a 0.35-kb Psll HI9cDNA probe. *. M1IT-G. ¡gf2(+/-) tumors that have lost imprinting control andre-express maternal Igf2 transcripts. See text for further details.

Table 3 SumirÃ-an'of the frequency of Igf2 ami H19 niRNA expression in M11T-G.Igf2 !+/-) and MIIT-G, /gf2 (+/+) liver tumors

Expression of Igf2 and HI9 mRNA was determined by Northern blot analysis.Expression in liver tumors"

GenotypeIgf2

EFFECTS OF Igfl DELETION ON LIVER TUMORIGENESIS

Fig. 4. Genetic map and LOH/pLOH analysis oftumors. A. genetic map of chromosome 7 indicatingmap position of Igf2. HI9, and MIT primer pairsused for analysis. Numbers indicate recombinationdistance from the centromere measured in centi-Morgans. B. ethidium bromide-stained acrylamide

gels of typical PCR reactions to determine LOH/pLOH at representative loci D7MÃŒI152.D7MH96,and D7MÃŒI223.P and M. paternal- and maternal-

specific bands, respectively. Lanes labeled 129/Sv,C57BL/IO. and C3H contain DNA from thesestrains. All Igf2 (+/+) and Igf2 (+/-) samples

contain control spleen DNA or DNA from threetumors from mouse 50/1. two tumors from mouse82/2, and four tumors from mouse 29/2. LOH/pLOH is determined by comparing differences inpaternal and maternal band intensity relative tocontrol spleen DNA and relative to a DNA controlseries with varying ratios of purified maternal- orpaternal strain-specific DNA (data not shown). Maternal- or paternal-specific pLOH is scored when

the allelic imbalance is between 1:3 and 1:9(pLOHM and pLOHP, respectively). CompleteLOH is scored when the maternal- or paternal-specific allelic imbalance is greater than 1:9(LOHM and LOHP. respectively).

Igf2 (+/-) Igf2 (+/+

A) 0

10

5053

6770

CM

—¿�D7MÃŒI152

—¿�D7MÃŒ196—¿�D7Mit40

—¿�Igf2/H19~D7Mit223

—¿�telomere

B)

D7MÃŒM52

D7Mit96

D7MÃŽ1223

analysis, alterations in HÌ9band intensity 25% or greater than theaverage band intensity of control tissue were scored as losses oracquisitions. Tumors can have exclusive maternal-specific loss (e.g.,26/2-3, -4), paternal-specific acquisition (e.g., 19/1-4), or a combination of both genetic events (e.g., 19/1-1, -3; Fig. 5). Of 21 tumorsanalyzed, there were no instances of maternal-specific acquisition

(data not shown).The quantitative Southern blotting was used to analyze 21 tumors

as described above, and the results are summarized (Table 5). Thedistribution of genetic alterations indicates that maternal loss of chromosome 7 is as likely a genetic event as paternal acquisition (Table 5).Taken together with the PCR LOH/pLOH analysis, these results areconsistent with maternal-specific LOH/pLOH on chromosome 7 rep

resenting a combination of paternal alÃ-eleacquisitions of functionalIgf2 alÃ-elesand maternal-specific losses of a maternally expressed

tumor suppressor gene.Maternal-specific LOH and Allelic Imbalance Results in In

creased Igf2 Expression, Whereas Reduced H19 Expression IsNot Selected for in Tumors. One expected consequence of maternal-specific chromosome loss or paternal-specific chromosome acquisi

tion would be alterations in the levels of imprinted gene expression.Because Igf2 is a driving force for selection of paternal chromosome7 alÃ-eleacquisition (scored as maternal-specific LOH/pLOH in thePCR analysis), we would expect that tumors with maternal-specific

LOH/pLOH are more likely to express high levels of Igf2 mRNAtranscripts. Similarly, if H19 was the tumor suppressor gene responsible for the selection of maternal chromosome alÃ-eleloss in Igf2

(+/-) tumors, we would expect that H19 expression levels would bereduced in Igf2 (+/-) tumors with maternal-specific pLOH. PCR-

derived LOH/pLOH data were, therefore, combined with Northernblot expression data (Table 6).

Table 6 includes a summary of Ml 1T-G, Igf2 (+/+) tumors that

do not exhibit LOH (Igf2 (+/+) no LOH] and tumors that exhibitpartial maternal-specific LOH [Igf2 (+/+) pLOHM], complete maternal LOH [Igf2 (+/+) LOHM] and partial paternal-specific LOH

[Igf2 (+/+) pLOHP]. These tumors were categorized for Igf2 expression levels as low, medium, or high as described (Table 6). Igf2(+/+) tumors exhibiting LOHM or pLOHM are considerably morelikely to express high levels of Igf2 relative to tumors without LOH(Table 6). The few tumors with paternal pLOH express Igf2 at lowand medium levels (Table 6). Taken together, these results indicatethat Igf2 is approximately proportional to its allelic representation inall tumor types indicated. These results further corroborate the importance of active paternal alÃ-eleacquisitions of Igf2 as the drivingforce for the selection of maternal-specific LOH/pLOH on distal

chromosome 7 in liver tumors.A summary of H19 expression is also included in Table 6. Igf2

(+/+) tumors without LOH tend to express high HI 9 levels, whereasIgf2 (+/+) pLOHM tumors are considerably more likely to expresslow levels of H19 (Table 6). As expected, fgf2 (+/+) tumors withcomplete LOHM all express low HÃŒ9(Table 6). Therefore, H19expression is proportional to active maternal alÃ-elerepresentation inIgf2 (+/+) tumors. H19 expression could be a factor in drivingchromosome 7 LOH/pLOH in Igf2 (+/+) tumors, which would yield

Table 4 Summary of LOH tinti pLOH analysis

Locus"

GenotypeIgf2

(+/+)LOHM or pLOHMLOHP or pLOHPTotal LOH andpLOHLOHM

or pLOHMLOHP or pLOHPTotal LOH and pLOHcD7MH15252/94

(55%)6 /94 (6%)

58 /94(62%)15/70(21%)

9/70(13%)24 no (34%)D7MH9655

/97 (57%)6 /97 (6%)

61/97(63%)22

/89 (25%)9/89(10%)31 189(35%)D7MMO55

/97 (57%)4 /97 (4%)

59/97(61%)22

/89 (25%)9/89(10%)

31 189(35%)Igf2/neohNA

NA25

/89 (28%)8 /89 (9%)

33 /89 (37%)D7MH22358

/97 (60%)3 /97 (3%)

61 /97(63%)22

/89 (25%)7 /89 (8%)

29 /89 (33%)' Values for all genetic loci tested are statistically significant deviations from equal probability of LOHM or pLOHM and LOHP or pLOHP as determined by %2(P = 0.02) with

the exception of D7Mitl52 in Igf2 (+/—) tumors due to the lower sample size.

NA. primer pair is not applicable.' The reduction in total LOH and pLOH in Igf2 (+/—) tumors relative to Igf2 (+/+) tumors is statistically significant as determined by %2 (P —¿�0.001).

4620

on July 7, 2021. © 1997 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

http://cancerres.aacrjournals.org/

EFFECTS OF Igf2 DELETION ON LIVER TUMORIGENESIS

Fig. 5. Representative quantitative Southern blotof tumor DNA to determine the precise nature ofLOHM/pLOHM on chromosome 7. Ten fig of totaltumor or control DNA from (MI1T-G X Cas) and(Cas X M1IT-G) was digested with Pstl and Pralland simultaneously probed with an 0.8-kb Psll-

Pvull fragment of the Cck gene from chromosome9 and a 1.6-kb P.s/1-PraII fragment of the HI9 genefrom chromosome 7. Chromosome 9 is rarely lostin MUP-TAg liver tumors (7), and the Cck probe.therefore, served as a normalization control. Thetumors in this Southern blot do not exhibit chromosome 9 LOH/pLOH (data not shown). Arrows. HI9bands specific to the Cas or BCF-1 alÃ-ele,as well asthe Cck band common to both Cas and BCF-1

alÃ-eles.Numbers beneath each lane represent quantification of either Cas or BCF-1 H19 allele-specific

bands after normalization to the level of the Cckband in each lane. Control DNA is represented byspleen DNA from animal 11/3 as well as tumors26/2-1 and 19/1-2, which show no LOH/pLOH.Allelic loss was scored as such when the alÃ-elebandintensity was less than 75% of the control average.Allelic acquisition was scored when the band intensity was equal to or greater than 125% of the controlaverage. Thus, the normal range of normalized H19band intensity was 2.6-4.0. P+, paternal alÃ-eleacquisition; Po, the paternal alÃ-elewas unaffected.M—,maternal alÃ-eleloss; Mo, the maternal alÃ-ele

was unaffected.

M11T-GXCasLiver Tumors

CasXM11T-G

Liver Tumors

1

CasH19-

BCF H19-

Cck

CasH19 3.1 3.6 3.3 3.7 3.2 4.7 3.0 3.1BCF1 H19 3.1 3.6 2.1 2.3 2.1 2.1 3.2 2.6

Po Po Po P+M- M- M- M-

1.6 0.5 3.23.0 4.6 3.1

2.35.9

2.55.5

Po P+ P+ P+M- M- M- Mo

this result. Alternatively, another tumor suppressor gene may bedriving LOH/pLOH, and Hi 9 expression is altered as a consequenceof its allelic representation.

The distribution oÃ-H19 expression levels in Igf2 (+/-) tumors is

not altered in tumors with pLOHM relative to tumors without LOH(Table 6). and H19 imprinting is rigidly maintained in M11T-Gtumors.5 These results indicate that the maternal-specific LOH/pLOH

in Igf2 (+/-) tumors is not selecting for changes in H19 expression

levels.In tumors without LOH, H19 expression levels are reduced in ¡gf2

(+/-) tumors compared with Igf2 (+/+) tumors (Table 6). That is,

although the frequency of HI 9 expression is unaffected by Igf2genotype (Table 3), the actual levels of H19 expression are reduced.Thus, in Ml 1T-G, Igf2 (+/+) tumors, both Igf2 and H19 expression

levels are proportional to allelic representation. Furthermore, H19expression levels are not altered in Igf2 (+/—) tumors, irrespective of

maternal-specific LOH/pLOH, suggesting that Hi 9 is not drivingselection for maternal-specific LOH/pLOH observed in Igf2 (+/-)

tumors.

DISCUSSION

We have found that Igf2 has a large effect on tumor size andfrequency. Specifically, Ml 1T-G, Igf2 (+/—) males show a 15-fold

reduction in the frequency of the largest tumor size, and all Ml 1T-G,Igf2 (+/—)animals show reductions in the frequency of other tumor

sizes. In addition, Igf2 is not required for the initial TAg-induced liverhyperplasia in MI 1T-G mice. We have also found that ¡gf2imprinting control is rigidly maintained in M11T-G tumors. Maternal-specific LOH/pLOH on chromosome 7 remains evident in Ml 1T-G, Igf2(+/-) tumors at about one-half the frequency, indicating that Igf2 is

driving approximately one-half of the maternal-specific LOH/pLOH

on chromosome 7. In addition, quantitative Southern blots indicatethat PCR-detected LOH/pLOH represents an equal occurrence of

maternal alÃ-eleloss and paternal alÃ-elegain. Our combined genetic

analysis is consistent with maternal-specific LOH/pLOH on distalchromosome 7 being driven by paternal acquisition of Igf '2 alÃ-elesand

not by H19 alÃ-eleloss.Igf2 has been implicated in several other TAg-induced transgenic

mouse models of tumorigenesis, most notably RIP-TAg mice (25),which develop insulomas, and C-reactive protein-TAg mice (10),

which develop hepatocellular carcinomas. Comparing results presented in this report with studies done using RIP-TAg mice (25, 26)

indicates that the effect of Igf2 on tumorigenesis is tissue specific. Inparticular, RIP-TAg, Igf2 (—/—)mice do not maintain the initial islet

of Langerhans hyperplasia seen in ¡gf2(+/+) mice, and tumors thatsubsequently develop are 7% the volume relative to Igf2 (+/+)tumors (25, 26). This differs from our results, indicating that the initialliver hyperplasia is unaffected by Igf2 genotype. Thus. Igf2 is notrequired for liver hyperplasia in TAg-induced liver tumors, whereasIgf2 is required for islet of Langerhans hyperplasia in TAg-inducedinsulomas. The effect of Igf2 on end-stage tumors taken at dissection

appears similar in both the liver and pancreas because we observed adecrease in tumor size and frequency, whereas the average size ofIgf2 (-/-) insulomas decreases considerably (25). Interestingly, inRIP-TAg mice, all Igf2 (+/-) tumors re-express the maternal Igf2

alÃ-ele(26), indicating that Igf2 imprinting is not rigid in the pancreas,contrary to results demonstrated here for liver.

Although many human tumor types exhibit LOI control and re-express IGF2 biallelically as determined by reverse transcription-PCR

(6, 18, 19, 23, 24), it is noteworthy that hepatoblastomas infrequentlyshow LOI control of ÃŒGF2(5, 6, 20, 21, 22). Although reverse

Table 5 Summary of quantitative Southern blot analvsis of (Cas X Ml ÃŒT-G)and(M1IT-G X CaÃ-)liver tumors

Type of maternalallelic imbalance" No. of tumors

P+/MOP+/M-Po/M-PO/M+

51150

5 M. Gobey and W. A. Held, unpublished results." P-K paternal alÃ-eleacquisition; Po, paternal alÃ-elewas unaffected; M —¿�.matemal

alÃ-eleloss; Mo. maternal alÃ-elewas unaffected.

4621

on July 7, 2021. © 1997 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

http://cancerres.aacrjournals.org/

EFFECTS OF Iff! DELETION ON LIVER TUMORIGENESIS

Table 6 Correlation of Igf 2 and H19 expression levéisin tumors wilh LOH análisis

Expression levels were determined by Northern blot analysis.

Igf2(+/+) noLOHIgf2(+/+)pLOHMIgf2(+/+)LOHMIgf2

(+/ +)pLOHPürIgf2

(+/-) noLOHIgf2(+/-)pLOHMIgf2

(+/-)LOHMIgf2(+/-)pLOHPA?Low6178016.22216220.84H

19expression"Medium612I)120702High1760212202Low61027.49Igf2 expression"Medium10961High13242(I

" Low, expression from undetectahle to less than one-fourth neonatal liver: medium, expression from one-fourth neonatal liver to less than 2x neonatal liver; high, expression at

least 2x more intense than neonatal liver.A pairwise comparison of no LOH and pLOHM for all expression levels described. Pairwise comparison of H19 expression between Igf2 (+/+) no LOH and Igf2 (+/—)no

LOH gives a *2 value of 15.05. x2 values of P = 0.001, 13.82 and /> = 0.025. 7.34.

transcription-PCR is more sensitive than Northern blotting (Fig. 3) for

determining LOI control, it is not clear whether there is biologicalsignificance to LOI without high expression levels. Our results indicate that LOI control in mouse liver tumors is similar to that observedin human hepatoblastomas. This similarity must be qualified becausethe human IGF2 gene shows imprinted expression from the threeIGF2 promoters, which are homologous to the three mouse Igf2promoters. However, the fourth human IGF2 promoter, which doesnot have a mouse homologue, is not imprinted in adult liver andexhibits biallelic and constitutive expression (34, 35).

Maternal-specific LOH/pLOH on distal chromosome 7 implies arole for imprinted genes in M11T-G tumorigenesis (8). LOH/pLOHanalysis of tumors from reciprocal genetic crosses of Ml 1T-G with M.castaneus by Southern blotting with a chromosome 7-specific H19probe has demonstrated that 58% of these tumors have maternal-

specific LOH/pLOH (8). This frequency of allelic imbalance is remarkably similar to the LOH/pLOH frequency determined by PCR forM11T-G, Igf2 (+/+) tumors tabulated in Table 4. Because thisLOH/pLOH is halved in Igf2 (+/—) tumors, Igf2 appears to be, in

part, responsible for driving selection for this LOH/pLOH. Thisconclusion is supported by quantitative Southern blotting of tumorsfrom reciprocal genetic crosses of M11T-G and M. castaneus and

from correlations of increased Igf2 expression associated with maternal-specific LOH/pLOH.

The remaining maternal-specific LOH/pLOH likely represents se

lection for loss of a maternally expressed tumor suppressor gene.RNA expression analysis shows that maternal-specific pLOH does notcorrelate with H19 expression in Igf2 (+/—) tumors, suggesting that

loss of H19 expression is not driving the remaining maternal-specific

LOH/pLOH. However, H¡9cannot be rigorously excluded withoutfurther experimentation, such as an analysis of paternal H19 expression. We are not presently able to explain why H19 expression levelsare reduced in Igf2 (+/—) tumors without LOH relative to similar

tumors bearing the Igf2 (+/+) genotype.What imprinted gene remains to drive maternal-specific LOH/

pLOH on chromosome 7? Such a gene would be required to fulfill thefollowing minimal criteria. The gene must be on chromosome 7, beimprinted, and maternally active. The gene product would be expectedto be expressed in normal adult liver. We would also expect that thisputative gene could be inactivated by genetic or epigenetic eventsother than LOH/pLOH. The imprinted region of distal mouse chromosome 7 contains the following known genes: IGF2, H19, Insulin 2,MASH2, and p51K"'2. Insulin 2 is not expressed in the liver and,

therefore, is unlikely to be a candidate gene. MASH2 is a maternallyexpressed imprinted gene required for trophoblast development (36)and has no known tumor suppressor activity. Currently, p51K"'2 is a

candidate tumor suppressor gene responsible for driving selection formaternal-specific LOH/pLOH on distal chromosome 7. p57A'"'2 is a

cyclin-dependent kinase inhibitor (37), the overexpression of which in

transfected human cells causes complete cell cycle arrest (38). Thegene is imprinted and maternally expressed in mouse tissues, including the adult liver (39). p51K"'2 expression patterns are under inves

tigation to examine its role in M11T-G liver tumorigenesis.

It should also be mentioned that the central region of mousechromosome 7 is imprinted and is syntenic to human chromosome15qll, a region that contains genes associated with Prader-Willi

syndrome (40,41 ), and SNRPN, a paternally expressed imprinted gene(42). The proximal region of mouse chromosome 7 is imprinted aswell (43). We cannot exclude the possibility that the imprinted genedriving maternal-specific LOH/pLOH is contained in one of these

imprinted regions of chromosome 7.Multiple pathways to tumorigenesis exist, and Igf2 is involved in

the major pathway in M11T-G mice because we demonstrated that

loss of Igf2 leads to fewer and smaller tumors. Given that tumors mayresult in an Igf2 (+/—) genetic background, however, Igf2 is notcompletely essential for Ml 1T-G tumorigenesis.

In the course of our LOH/pLOH analysis, we have noticed thattumors from an individual animal can exhibit considerable geneticrelatedness. For instance, all tumors from animal 29/2 showed complete LOHM for all chromosome 7 loci tested (Fig. 4B). Althougheach tumor from animal 29/2 was dissected from separate liver lobes(data not shown), they show this striking relatedness. Thus, eachtumor within an animal may not have an independent origin.

The biochemical mechanism by which Igf2 is having its effect inour tumorigenicity model is presently unknown. It is known thatRIP-TAg tumors have an increase in the number of apoptotic bodiesin an Igf2 (—/—)genetic background relative to Igf2 (+/+) insulo-mas (44). It will be necessary to test whether Igf2 (+/—) tumors

exhibit more apoptosis than Igf2 (+/+) tumors in our transgenicmodel.

Our data are consistent with the importance of Igf2 in M11T-G

liver tumorigenesis, providing strong selection pressure for Igf2 expression in liver tumors. To increase Igf2 expression, tumors mayactivate all three paternal Igf2 promoters (27) or deregulate imprinting control and activate the maternal Igf2 gene. As well, tumors mayacquire more copies of active paternal alÃ-elesof chromosome 7, whichwill be manifest as maternal-specific LOH/pLOH in PCR assays.Quantitative Southern blots indicate maternal-specific LOH/pLOH

represent equal frequencies of maternal alÃ-eleloss and paternal alÃ-elegain. We have shown that Igf2 imprinting control in the liver is tightlymaintained. Consequently, there is a strong selection pressure formaternal-specific LOH/pLOH on distal chromosome 7 in M11T-G

4622

on July 7, 2021. © 1997 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

http://cancerres.aacrjournals.org/

EFFECTS OF Igfl DELETION ON LIVER TUMORIGENESIS

tumors. This LOH/pLOH correlates with increased Igf2 expression,further supporting this model. In an Igf2 (+/—) genetic background,Ml IT-G-induced liver tumors exhibit a residual maternal-specific

LOH/pLOH on chromosome 7, indicating the existence of a maternally expressed tumor suppressor gene. H19 is unlikely to be responsible for driving selection of this residual chromosome 7 maternal-

specific LOH/pLOH because HI 9 expression levels remain unalteredby allelic imbalance.

ACKNOWLEDGMENTS

We are indebted to Argiris Efstratiadis for kindly providing us with Igf2(—/+) heterozygous mice. We gratefully acknowledge Allan Kinniburgh,

Scott Pearsall, and Paul Soloway for critical review of the manuscript. We alsothank Ken Manly for help with statistical analysis of the data. We thankKimberly Kearns. Ya Qin Zhang, and Mary Kay Ellsworth for technicalassistance and Margaret Gobey for providing us with unpublished data. R. H.thanks Catherine Le Feuvre for love and support.

REFERENCES

1. Held. W. A.. Mullins, J. J.. Kühn.N. J., Gallagher. J. F.. Gu. G. D.. and Gross, K. W.T antigen expression and tumorigenesis in transgenic mice containing a mouse majorurinary protein/SV40 T antigen hybrid gene. EMBO J., 8: 183-191, 1989.

2. Schirmacher. P.. Held. W. A.. Yang. D.. Biempica. L.. and Rogler, C. E. Selectiveamplification of periportal transitional cells precedes formation of hepatocellularcarcinoma in SV40 large Tag transgenic mice. Am. J. Pathol.. 139: 231-241. 1991.

3. Christofori, G-, and Hanahan. D. Molecular dissection of multistage tumorigenesis intransgenic mice. Cancer Biol., 5; 3-12, 1994.

4. Levine. D. S.. Sanchez. C. A., Rabinovilch. P. S.. and Reid. B. J. Formation of thetetraploid intermediate is associated with the development of cells with more thanfour cenlrioles in the elastase-simian virus 40 tumor antigen transgenic mouse modelof pancreatic cancer. Proc. Nati. Acad. Sci. USA, 88: 6427-6431, 1991.

5. Rainier, S.. Dobry, C. J.. and Feinberg, A. P. Loss of imprinting in hepatoblastoma.Cancer Res., 55: 1836-1838, 1995.

6. Rainier, S., Johnson. L. A.. Dobry. C. J., Ping. A. J.. Crundy, P. E.. and Feinberg.A. P. Relaxation of imprinted genes in human cancer. Nature (Lond.), 362: 747-749,

1993.7. Pearsall, R. S., Shibata. H.. Brozowska, A.. Yoshino. K.. Okuda, K.. deJong, P. J..

Plass. C., Chapman. V. M.. Hayashizaki. Y.. and Held, W. A. Absence of imprintingin U2AFBPL. a human homologue of the imprinted mouse gene U2afbp-rs. Biochem.Biophys. Res. Commun.. 222: 171-177, 1996.

8. Held. W. A., Pazik. J.. Giancola O'Brien. J.. Kearns. K., Gobey, M.. Meis, R., Kenny,

L.. and Rustum, Y. Genetic analysis of liver tumorigenesis in SV40 T antigentransgenic mice implies a role for imprinted genes. Cancer Res.. 54: 6489-6495.1994.

9. Schirmacher, P., Held. W. A.. Yang, D.. Chisari, F. V.. Rustum, Y.. and Rogler, C. E.Reactivation of insulin-like growth factor II during hepatocarcinogenesis in transgenic mice suggests a role in malignant growth. Cancer Res.. 52: 2549-2556, 1992.

10. Casola, S., Ungaro. P., Pedone. P. V., Lazzaro, D., Fattori, E., Ciliberto. G.. Zarrilli.R.. Bruni. C. B., and Riccio. A. Loss of heterozygosity of imprinted genes in SV40t/T antigen-induced hepatocellular carcinomas. Oncogene, //: 711-721. 1995.

11. Zemel. S., Bartolomei, M., and Tilghman. S. Physical linkage of two mammalianimprinted genes. Nat. Genet.. 2: 61-65, 1992.

12. De Chira. T. M.. Robertson. E. J.. and Efstratiadis, A. Parental imprinting of themouse insulin-like growth factor II gene. Cell, 64: 849-859, 1991.

13. Barlolomei. M. S.. Zemel. S.. and Tilghman, S. M. Parental imprinting of the mouseHI9 gene. Nature (Lond.). 351: 153-155, 1991.

14. Hao, Y., Crenshaw, T., Moulton. T., Newcomb, E., and Tycko. B. Tumour-suppressoractivity of H19 mRNA. Nature (Lond.), 365: 764-767. 1993.

15. Baker, J., Liu, J. P.. Robertson, E. J.. and Efstratiadis. A. Role of insulin-like growthfactors in embryonic and postnatal growth. Cell, 75: 73-82, 1993.

16. Oka. Y., Rozek, L. M., and Czech, M. P. Direct demonstration of rapid insulin-likegrowth factor II receptor internalization and recycling in rat adipocytes. J. Biol.Chem., 260: 9435-9442, 1985.

17. Reik, W.. Brown, K. W.. Slatter. R. E.. Sartori. P., Elliot, M.. and Mäher.E. R. Allelicmethylation of H19 and IGF2 in the Beckwith-Wiedemann syndrome. Hum. Mol.Genet.. J: 1297-1301, 1994.

18. Weksberg. R.. Shen. D. R., Fei, Y. L.. Song. Q. L.. and Squire. J. Disruption ofinsulin-like growth factor 2 imprinting in Beckwith-Weidemann syndrome. Nat.Genet.. 5: 143-150. 1993.

19. Ogawa. O.. Eccles, M. R.. Szeto. J.. McNoe. L. A.. Yun, K.. Maw. M. A.. Smith. P. J.,

and Reeve. A. E. Relaxation of inulin-like growth factor II gene imprinting implicatedin Wilms' tumour. Nature (Lond.), 362: 749-751. 1993.

20. Davies. S. M. Maintenance of genomic imprinting at the ICF2 locus in hepatoblas-loma. Cancer Res.. 53: 4781-4783. 1993.

21. Montagna. M., Menin. C., Chicco-Bianchi. L., and D'Andréa.E. Occasional loss of

constitutive heterozygosity at 1Ipl5.5 and imprinting relaxation of the ICFII maternal alÃ-elein hepatoblastoma. J. Cancer Res. Clin. Oncol., 120: 732-736, 1994.

22. Li, X., Adam, G., Cui, H.. Sandstedt, B., Ohlsson. R., and Ekstrom. T. J. Expression,promoter usage and parental imprinting status of insulin-like growth factor II (IGF2)

in human hepatoblastoma: uncoupling of IGF2 and H19 imprinting. Oncogene, 11:221-229. 1995.

23. van Gurp. R. J.. Oosterhuis. J. W., Kalscheuer, V.. Mariaman. E. C.. and Looijenga.L. H. Biallelic expression of the HI9 and 1GF2 genes in human testicular germ celltumors. J. Nati. Cancer Inst., 86: 1070-1075. 1994.

24. McCann, A. H., Miller. N.. O'Meara. A.. Pedersen. I.. Keogh. K.. Gorey. T.. and

Dervan. P. A. Biallelic expression of the IGF2 gene in human breast disease. HumanMolecular Genetics, 5: 1123-1127. 1996.

25. Christofori. G.. Naik. P., and Hanahan, D. A second signal supplied by insulin-likegrowth factor II in oncogene-induced tumorigenesis. Nature (Lond.). 369: 414-417,

1994.26. Christofori, G.. Naik. P., and Hanahan, D. Deregulation of both imprinted and

expressed alÃ-elesof the insulin-like growth factor 2 gene during beta-cell tumorigenesis. Nat. Genet.. 10: 196-201. 1995.

27. De Chiara. T. M.. Efstratiadis. A., and Robertson. E. J. A growth-deficiency pheno-type in heterozygous mice carrying an insulin-like growth factor II gene disrupted bytargeting. Nature (Lond.), 345: 78-80. 1990.

28. Dietrich. W.. Katz, H.. Lincoln, S. E., Shin, H. S., Friedman. J., Dracopoli. N. C.. andLander. E. S. A genetic map of the mouse suitable for typing intraspecific crosses.Genetics, 131: 423-447. 1992.

29. Freidman. J. M.. Schneider. B. S.. Barton, D. E.. and Francke. U. Level of expressionand chromosome mapping of the mouse cholecystokinin gene: implications formurine models of genetic obesity. Genomics. 5: 463-469. 1989.

30. Daniel. W. Biostatistics. a Foundation for Analysis in the Health Sciences. Ed. 6, pp.507-529: 586-591. New York: John Wiley and Sons. Inc., 1995.

31. Saracci. R., and Repetto. F. Time trends of primary liver cancer: indication ofincreased incidence in selected cancer registry populations. J. Nati. Cancer Inst., 65:241-247, 1980.

32. Kemp, C. J., Leary, C. N.. and Drinkwater. N. R. Promotion of murine hepatocarcinogenesis by testosterone is androgen receptor-dependent but not cell autonomous.Proc. Nati. Acad. Sci. USA. 86: 7505-7509. 1989.

33. Leighton. P. A.. Ingram. R. S., Eggenschwiler. J., Efstratiadis. A., and Tilghman.S. M. Disruption of imprinting caused by deletion of the HI9 gene region in mice.Nature (Lond.). 375: 34-39, 1995.

34. Vu. T.. and Hoffman, A. Promoter-specific imprinting of the human insulin-likegrowth factor-II gene. Nature (Lond.). 371: 714-717, 1994.

35. Ekstrom. T.. Cui, H., Li, X.. and Ohlsson. R. Promoter-specific IGF2 imprinting

status and its plasticity during human liver development. Development (Camb.), 121:309-316, 1995.

36. Guillemot. F.. Caspary. T.. Tilghman. S. M., Copeland, N. G.. Gilbert. D. J., Jenkins.N. A., Anderson, D. J.. Joyner, A. L.. Rossant, J.. and Nagy. A. Genomic imprintingof Mash2. a mouse gene required for trophoblast development. Nat. Genet.. 9:235-241. 1995.

37. Matsuoka, S., Edwards. M. C.. Bai, C.. Parker, S., Zhang. P.. Baldini. A.. Harper.J. W., and Elledge, S. J. p57KIP2, a structurally distinct member of the p21c">l Cdk

inhibitor family, is a candidate tumor suppressor gene. Genes Dev.. 9: 650-662,

1995.38. Malsuoka, S.. Edwards. M. C.. Bai. C.. Parker, S.. Zhang, P.. Baldini. A.. Harper.

J. W.. and Elledge. S. J. P57KIP2, a structurally distinct member of the p21c"'' Cdk

inhibitor family, is a candidate tumor suppressor gene. Genes Dev.. 9: 650-662,

1995.39. Hatada. I., and Mukai. T. Genomic imprinting of p57Klp:. a cyclin-dependent kinase

inhibitor, in mouse. Nat. Genet.. //: 204-206. 1995.40. LaSalle. J. M.. and Lalande. M. Domain organization of allele-specific replication

within the GABRB3 gene cluster requires a biparental 1Iql 1-13 contribution. Nat.Genet.. 9: 386-394, 1995.

41. Tycko. B. Genomic imprinting: mechanism and role in human pathology. Am. J.Pathol.. 144: 431-443. 1994.

42. Wevrick. R., Kems. J. A., and Francke. U. Identification of a novel paternallyexpressed gene in the Prader-Willi syndrome region. Hum. Mol. Genet.. 3: 1877-1882, 1994.

43. Kuroiwa, Y., Kaneko-Ishino, T.. Kagitani. F., Kohda, T.. Li. L., Tada. M.. Suzuki. R..

Yokoyama. M.. Shiroishi, T.. Wakana, S.. Barton. S.. Ishino. F.. and Surani. M. Peg3imprinted gene on proximal chromosome 7 encodes for a zinc finger protein. Nat.Genet., 12: 186-190, 1996.

44. Naik, P., Karrim, J., and Hanahan. D. The rise and fall of apoptosis during multistagetumorigenesis: down-modulation contributes to tumor progression from angiogenicprogenitors. Genes Dev., 10: 2105-2116. 1996.

4623

on July 7, 2021. © 1997 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

http://cancerres.aacrjournals.org/

1997;57:4615-4623. Cancer Res Ramsi Haddad and William A. Held Loss of Heterozygosity in SV40 T Antigen Transgenic Mice

Influence Liver Tumorigenesis andIgf2Genomic Imprinting and

Updated version

http://cancerres.aacrjournals.org/content/57/20/4615

Access the most recent version of this article at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected] at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cancerres.aacrjournals.org/content/57/20/4615To request permission to re-use all or part of this article, use this link

on July 7, 2021. © 1997 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

http://cancerres.aacrjournals.org/content/57/20/4615http://cancerres.aacrjournals.org/cgi/alertsmailto:[email protected]://cancerres.aacrjournals.org/content/57/20/4615http://cancerres.aacrjournals.org/