Frequent Loss of Imprinting of IGF2 and MEST inLung Adenocarcinoma

Masakazu Kohda,1 Hidetoshi Hoshiya,1 Motonobu Katoh,1 Isao Tanaka,2 Ryo Masuda,2

Tamiko Takemura,3 Mutsunori Fujiwara,3 and Mitsuo Oshimura1*

1Core Research for Evolutional Science and Technology (CREST) Program of the Japan Science and Technology (JST)Corporation, Department of Molecular and Cell Genetics, School of Life Sciences, Tottori University, Tottori, Japan2Department of Surgery, Japanese Red Cross Medical Center, Tokyo, Japan3Department of Pathology, Japanese Red Cross Medical Center, Tokyo, Japan

Genomic imprinting is a parental origin±specific chromosomal modification that causes differential expression ofmaternal and paternal alleles of a gene. Accumulating evidence suggests that deregulation of imprinted genes,including loss of imprinting (LOI), plays a role in oncogenesis. In the present study, we investigated allelic expression ofsix imprinted genes in human lung adenocarcinomas as well as in matched normal lung tissue. Informative casesshowing heterozygosity for the gene of interest were selected from 35 patients. LOI of the insulin-like growth factor 2gene (IGF2) and mesoderm-specific transcript (MEST, also known as paternally expressed gene 1) was noted in 47%(seven of 15) and 85% (11 of 13) of informative cases, respectively. Monoallelic expression was maintained in all thematched normal tissues examined. LOI of IGF2 was seen more frequently in moderately to poorly differentiatedadenocarcinomas. In contrast, H19, small nuclear ribonucleoprotein±associated polypeptide N gene (SNRPN ), necdingene (NDN ), and long QT intronic transcript 1 (LIT1) exhibited consistent monoallelic expression in all the informativesamples. These findings indicated that independent deregulation took place in imprinted genes and suggested thataberrant imprinting of IGF2 and MEST was involved in the development of lung adenocarcinoma. ß 2001 Wiley-Liss, Inc.

Key words: genomic imprinting; biallelic expression; monoallelic expression

INTRODUCTION

Genomic imprinting is an inherited epigeneticform of gene regulation that results in preferentialexpression of an allele, depending on the parentalorigin [1]. This epigenetic modi®cation involvesparental origin±speci®c DNA methylation at CpGsites [2], which is established through gametogen-esis and is potentially reversible in somatic cells andin the next generation. Genomic imprinting playsfundamental roles in normal embryonic growth andbehavioral development [3,4]. In addition, recentstudies indicated that abnormalities in imprintingmay be involved in oncogenesis [5±8]. This notionwas ®rst revealed in the preferential loss of hetero-zygosity that has been seen in selective parentalchromosomal loci or in the loss of constitu-tive imprinting at certain genes in several tumors[9±14].

Imprinting abnormalities may contribute totumorigenesis either by activation of a transcrip-tionally repressed allele of a growth-promotingoncogene or by inactivation of a normally expressedallele of an imprinted tumor suppressor gene [15±18]. Alternatively, inactivation of an imprintingcontrol center could cause aberrant expression ofmultiple imprinted proto-oncogenes and/or tumorsuppressor genes, because expression of imprintedgenes is coordinately regulated within a chromo-somal domain [19±22]. A cluster located on chro-

mosome 11p15.5, which contains imprinted geneswith growth-promoting (insulin-like growth factor2 gene, or IGF2) and growth-controlling (cyclin-dependent kinase inhibitor and H19) functions[13,23], was ®rst examined in relation to Beckwith-Wiedemann syndrome, a congenital overgrowthdisorder, as well as in some embryonal tumors,including Wilms' tumor [24]. Loss of imprinting(LOI) of IGF2 has been reported in more than 20different types of adult tumors, indicating itsfundamental signi®cance in oncogenesis [25,26].

Expression of IGF2 and H19 is reciprocallyregulated, and the coupling of biallelic IGF2 geneexpression with H19 inactivation has been observedin Wilms' tumors [27,28] and in H19 knockout

MOLECULAR CARCINOGENESIS 31:184±191 (2001)

ß 2001 WILEY-LISS, INC.

*Correspondence to: Department of Molecular and Cell Genetics,School of Life Sciences, Faculty of Medicine, Tottori University,Nishimachi 86, Yonago, Tottori 683±8503, Japan

Received 21 February 2001; Revised 2 May 2001; Accepted 14May 2001

Abbreviations: IGF2, insulin-like growth factor 2 gene; MEST,mesoderm-speci®c transcript; LOI, loss of imprinting; SNRPN, smallnuclear ribonucleoprotein±associated polypeptide N gene; NDN,necdin gene; LIT1, long QT intronic transcript 1; CDKN1C, cyclin-dependent kinase inhibitor 1C; KvLQT1, potassium voltage-gatedchannel long QT syndrome 1; RFLP, restriction fragment lengthpolymorphism; RT, reverse transcription; PCR, polymerase chainreaction.

mice [29]. Previously we proposed that long QTintronic transcript 1 (LIT1), on proximal 11p15.5,may be a regulator that directs allelic expression ofimprinted genes, including potassium voltage-gatedchannel long QT syndrome (KCNQ1, also known asKvLQT1), cyclin-dependent kinase inhibitor 1C(CDKN1C), and KCNQ1 downstream neighbor(KCNQ1DN) in the proximal imprinting cluster on11p15.5 [19]. LOI of LIT1 has been reported incolorectal cancer [30], and frequent LOI of meso-derm-speci®c transcript (MEST) on chromosome7q32 has been reported in breast cancer and color-ectal cancer [31]. Another prominent imprintedgene cluster on 15q11-q13 may be involved in thepathogenesis of the Prader-Willi and Angelmansyndromes, which are associated with de®cienciesin sexual development and growth and mentalproblems, including retardation [32,33]. Theimprinted small nuclear ribonucleoprotein±asso-ciated polypeptide N gene (SNRPN) and necdin gene(NDN) are present in this cluster, but aberrantimprinting of these genes in tumors has not beenreported.

Abnormalities in epigenetic regulation ofimprinted genes may be an alternate mechanismto irreversible genetic alterations (such as deletionsor mutations) that leads to tumor formation. Toelucidate the involvement of aberrant imprinting intumorigenesis, we assessed the expression status ofsix representative imprinted genes (IGF2, H19,SNRPN, NDN, LIT1, and MEST) in 35 lung adeno-carcinomas. LOI of growth-promoting genes (IGF2and MEST) was noted at a high frequency. Thus,nonrandom abrogation of imprinted gene expres-sion might contribute to tumor formation.

MATERIALS AND METHODS

Tissue Samples

Lung adenocarcinomas, diagnosed by histopatho-logic examination, and matched normal tissues of35 unrelated patients were analyzed. All tissuesamples were collected at the Department ofPathology, Japanese Red Cross Medical Center. Alltissues were quickly frozen in liquid nitrogen andstored at ÿ80�C until analysis. Genomic DNA wasisolated with the PUREGENE kit (Gentra SystemsInc., Mineapolis, MN), and RNA was isolated withthe RNeasy Mini Kit (QIAGEN, Valencia, CA)according to the manufacturers' instructions.

Identi®cation of Polymorphisms

Polymerase chain reaction (PCR) primers foridenti®cation of polymorphism by PCR-restrictionfragment length polymorphism (RFLP) weredesigned in previous studies. The sequences of theprimers are as follows: 5 0-CTTGGACTTTGAGT-CAAATTGG-3 0 and 5 0-CCTCCTTTGGTCTTGAAGGCTGCT-3 0 (IGF2) [34], 5 0-TACAACCACTGCAC-

TACCTG-3 0 and 5 0-TGGAATGCTTGAAGGCTGCT-3 0 (H19) [35], 5 0-CTACTCTTTGAAGCTTCTGCC-3 0

and 5 0-TGAAGATTCGGCCATCTTGC-3 0 (SNRPN)[36], 5 0-GCCCAATACGAGTTCTTTT-3 0 and 5 0-CACACATCATCAGTCCCATA-3 0 (NDN) [37], 5 0-CAGCACAAAGAGGTTTTTGACAG-3 0 and 5 0-GAGTT-TAAAACACGTGTGTGCATT-3 0 (LIT1 with RsaIpolymorphism) [38], 5 0-AAGAAAGTGTTGAGTGG-TAA-3 0 and 5 0-CAAGCAGTACTGTTTCAGAA-3 0

(LIT1 with NsiI polymorphism) [39], and 5 0-CACT-GATGCAGAAAGACGTTC-3 0 and 5 0-CAGCACCATTTGCTCATAGG-3 0 (MEST) [40].

PCR conditions were as follows: for IGF2, 30 cyclesat 95/55/72�C for 30/30/30 s; for H19, 30 cycles at95/60/72�C for 30/30/30 s; for SNRPN, 30 cyclesat 95/62/72�C for 30/30/50 s; and for NDN, 30cycles at 95/62/72�C for 30/30/30 s. For LIT1 withRsaI polymorphism, a step-down protocol was used,that is, nine cycles at 95/60/72�C for 30/30/30 s,followed by reduction in the annealing temperatureby 2�C every three cycles to a ®nal annealingtemperature of 56�C. Twenty-six subsequent cycleswere carried out at 95/56/72�C for 30/30/30 s. PCRconditions for LIT1 with NsiI polymorphism were30 cycles at 95/50/72�C for 20/20/50 s and condi-tions for MEST were 30 cycles at 95/62/72�C for 40/40/40 s. Each PCR product was digested with anappropriate restriction enzyme to detect the poly-morphism.

Restriction enzymes used were as follows: ApaI(NIPPON GENE, Tokyo, Japan) for IGF2, RsaI(NIPPON GENE) for H19, BstUI (New EnglandBiolabs, Inc., Beverly, MA) for SNRPN, NdeII (NIP-PON GENE) for NDN, RsaI or NsiI (NIPPON GENE)for LIT1, and A¯III (New England Biolabs, Inc.) forMEST. In all experiments, genomic DNA of homo-zygotes, indicating the presence of the restrictionsite, was included as control for the speci®city andcompleteness of restriction digestion. LIT1-NsiI wasthe exception, because a sample homozygous forthe NsiI-containing allele was not available.

Allele-Speci®c Expression Analysis of Imprinted Genes

Total RNA was treated with RNase-free DNase(NIPPON GENE). First-strand cDNA was synthesizedwith an oligo(dT)15 primer and Moloney murineleukemia virus reverse transcriptase (Gibco BRL,Rockville, MD). To exclude the possibility of DNAcontamination in RNA samples, mock cDNA synth-esis reactions were performed without reversetranscription (RT), as negative controls. Allele-speci®c expression analysis with RT-PCR and diges-tions by restriction enzymes were performed underthe same conditions as for detecting genomicpolymorphisms. Restriction fragments wereresolved by polyacrylamide gel electrophoresis,stained with SYBR green (Takara, Kyoto, Japan),and quanti®ed with the ¯uorescent image-analyz-ing system FMBIO II Multi-View (Takara). The

IMPRINTING ANALYSIS IN LUNG ADENOCARCINOMA 185

threshold for scoring LOI was a ratio of less than 3:1between the more abundant and less abundantalleles [9].

RESULTS

Frequent Loss of Imprinting of IGF2 andMEST in Lung Adenocarcinoma

First we analyzed the genomic DNA of 35 patients,taken from normal lung tissue adjacent to lungadenocarcinomas, for polymorphisms in siximprinted genes by PCR-RFLP. The number ofinformative cases is summarized in Table 1 ForIGF2, H19, SNRPN, NDN, LIT1 and MEST, weobtained 15, 16, 17, 17, 5, and 13 informative cases,respectively. Allele-speci®c expression of the siximprinted genes was analyzed in the informativeheterozygous cases. Representative results of allelicexpression analysis are shown in Figure 1. The allelicratio was determined by comparing the intensity ofthe restriction fragments resolved in gel electro-phoresis representing the two alleles. LOI was scoredfor a case showing the ratio of the less abundant tothe more abundant allele to be larger than thearbitrarily chosen threshold of 1:3 applied by Cui etal [9]. Seven of 15 (47%) informative cancers showedbiallelic expression of IGF2, while none of 15matched normal lung tissues showed biallelicexpression of IGF2. It was noted that LOI of IGF2was seen more frequently in moderately and poorlydifferentiated adenocarcinomas (LOI of IGF2occurred in six of nine cases) than in well-differ-entiated adenocarcinomas (LOI of IGF2 occurred inone of six cases), suggesting a correlation betweenLOI and histologic classi®cation (Table 2).

H19, SNRPN, and NDN consistently exhibitedmonoallelic expression in all the informative can-cers and matched normal lung tissues examined,indicating the dissociation of these imprinted genesfrom cancer formation. The lung adenocarcinomasand normal lung tissues showed monoallelic expres-sion of LIT1, except for one case that showed nodetectable expression. LOI of MEST was seen in 11 of

13 (85%) lung adenocarcinomas, while all the 13matched normal lung tissue showed monoallelicexpression. In contrast to IGF2, correlation betweenLOI and histologic classi®cation was not evident forMEST. Of the four cases informative for both IGF2and MEST, two showed concomitant LOI at bothloci (cases 18 and 43), and one showed LOI at MESTbut not at IGF2 (case 46). The remaining case (case93) exhibited LOI at MEST and loss of heterozygosityat IGF2.

Loss of Imprinting of MEST Was Not an ArtifactResulting From Lymphocytic Response

MEST has been reported to be expressed bialleli-cally in adult lymphocytes [41]. To ensure that theLOI observed in the adenocarcinomas was not anartifact resulting from lymphocytic response to thetumor, all tissue samples were stained with hema-toxylin and eosin, and the number of lymphocyteswas assessed. Figure 2 shows a representative slide ofthe lymphocytic response in monoallelic andbiallelic expression cases. The degree of lymphocy-tic in®ltration was negligible, irrespective of allelicexpression of the MEST gene. These results indicatedthat a trace of lymphocyte in®ltration did notcontribute to substantial biallelic expression in theLOI cases.

DISCUSSION

To test whether aberrant imprinting is involved inthe pathogenesis of tumors, we investigated theexpression status of six imprinted genes (IGF2, H19,SNRPN, NDN, LIT1, and MEST) in lung adenocarci-nomas and adjacent normal lung tissues. LOI forIGF2 and MEST was found in 47% and 85% oftumors, respectively, while all corresponding nor-mal tissue displayed monoallelic expression of thesegenes. The role of IGF2 as an autocrine growth factorwas ®rst suggested by the observation that lungcancer cells secrete IGF2, which can stimulate theproliferation of cancer cells in vitro [42±44]. Inaddition, IGF2 is expressed abundantly in fetal lungunder tight regulation and is implicated in lung

Table 1. Expression Patterns of Six Imprinted Genes in Informative Cases Among 35 Lung Adenocarcinomas

Tumor tissue Normal tissueInformative

cases Mo* Biy NDz Mo* Biy NDz

IGF2 (n� 15) 8 7 0 15 0 0H19 (n� 16) 16 0 0 16 0 0SNRPN (n� 17) 17 0 0 17 0 0NDN (n� 17) 17 0 0 17 0 0LIT1 (n� 5) 5 0 0 4 0 1MEST (n� 13) 2 11 0 13 0 0

*Monoallelic expression.yBiallelic expression.zNot detected.

186 KOHDA ET AL.

IMPRINTING ANALYSIS IN LUNG ADENOCARCINOMA 187

organogenesis [22]. An overdose of IGF2 by LOItherefore may provide a growth advantage totransformed cells.

The ®nding of frequent LOI in lung cancer in thisstudy and in others [45] supports the substantial roleof IGF2 in the pathogenesis of lung cancer. It wasnoted that there seemed to be a relationshipbetween LOI of IGF2 and the histologic classi®ca-tion of adenocarcinomas. The tendency of LOI ofIGF2 was biased toward poorly and moderatelydifferentiated rather than well-differentiated ade-nocarcinomas. This is consistent with the role ofIGF2 as a growth-promoting factor.

Recently we reported LOI of IGF2 and MEST in asubset of colorectal cancers [46]. LOI was seen inboth cancer tissue and matched normal mucosa,suggesting an imprinting polymorphism with apredisposition to colorectal cancer. In contrast tocolorectal cancer, LOI of these genes was noted inlung cancer but not in matched normal tissue. Thismay imply that LOI of IGF2 plays a role at a differentstage of tumor formation, possibly at the stage ofprogression. In contrast to the ®ndings of a previousreport describing LOI of H19 in lung adenocarcino-mas [47], H19 transcripts maintained monoallelicexpression in our study. This discrepancy is unclear,but the present ®ndings and those of others [48]

suggest that regulation of IGF2 and H19 is notnecessarily coordinated.

Functional inactivation of the paternal allele ofLIT1 by targeted deletion of its 5 0 upstream CpGislands resulted in activation of normally silentpaternal alleles of CDKN1C, KvLQT1, andKCNQ1DN, suggesting that LIT1 acts as a negativeregulator in cis for the coordination of imprinting ofthe proximal imprinted domain on 11p15.5 [19].One of the affected genes, CDKN1C, is a member ofthe p21 cyclin-dependent kinase inhibitor familyand is thought to regulate the cell cycle negatively atthe G1 checkpoint.

Assuming that LOI of LIT1, if it occurs, leads toinactivation of CDKN1C, we investigated allelicexpression of LIT1 in lung cancer. In fact, loss ofmaternal-speci®c methylation at the KvLQT1 differ-entially methylated region 1 (KvDMR1) locus inhepatocarcinoma correlated with abnormal expres-sion of CDKN1C and IGF2 [20]. Contrary to ourassumption, LIT1 exhibited monoallelic expressionin all the informative lung adenocarcinomas.Frequent selective loss of the active maternal alleleof CDKN1C has been reported in lung cancer casescarrying 11p15 deletions [49]. In lung cancer, loss ofthe functional maternal allele of CDKN1C may be adominant genetic aberration rather than an indirect

Figure 1. Representative results for allele-speci®c expressionanalysis of six imprinted genes in paired normal (N) and tumor (T)samples. M indicates the lane for the size marker. LOI was scored fora case in which the ratio of the less abundant to the more abundantallele was larger than the arbitrarily chosen threshold of 1:3 appliedby Cui et al. [9] (A) Expression pattern of IGF2. Case 40 maintainedimprinting in the abnormal tissue. In contrast to monoallelicexpression in the corresponding normal lung samples, relaxation of

IGF2 imprinting was observed in case 101. (B, C, D, and E) Expressionpatterns of H19, SNRPN, NDN, and LIT1, respectively. Imprintedmonoallelic expression was seen in both N and T in all casesexamined. (F) Expression pattern of MEST. Imprinted monoallelicexpression was observed in N from cases 93 and 115, whereas clearbiallelic expression was noted in the matched lung cancer. The ratioof allelic expression is indicated below the panel in the LOI cases (Aand F). These results are summarized in Table 1.

Table 2. Summary of Patients and Allele-Speci®c Expression of IGF2 and MEST in Lung Adenocarcinomas

IGF2 MEST

Histologic HistologicCase no. Age/sex Allelic usage characteristics* Case no. Age/sex Allelic usage characteristics*

43 66/M Biallelic P.D. 43 66/M Biallelic P.D.51 56/M Biallelic P.D. 90 66/M Biallelic P.D.89 59/M Monoallelic P.D. 118 47/M Biallelic P.D.

104 64/M Monoallelic P.D. 88 66/M Monoallelic M.D.96 69/F Monoallelic P.D. 93 60/M Biallelic M.D.23 57/M Biallelic M.D. 18 59/F Biallelic W.D.38 69/M Biallelic M.D. 36 58/F Biallelic W.D.99 69/M Biallelic M.D. 46 73/F Biallelic W.D.

101 58/M Biallelic M.D. 50 76/M Monoallelic W.D.18 59/F Biallelic W.D. 92 64/M Biallelic W.D.40 54/F Monoallelic W.D. 113 58/M Biallelic W.D.41 53/F Monoallelic W.D. 115 81/M Biallelic W.D.46 73/F Monoallelic W.D. 119 45/F Biallelic W.D.85 61/F Monoallelic W.D.97 57/M Monoallelic W.D.

*P.D., poorly differentiated; M.D., moderately differentiated; W.D., well differentiated.

188 KOHDA ET AL.

epigenetic regulation. The integrity of the 11p15.5region itself, including the CDKN1C gene and allelicexpression of CDKN1C, remains to be tested beforediscussing the functional role of LIT1 as an imprint-ing regulator and its involvement in tumor suppres-sion through CDKN1C.

MEST is an imprinted gene identi®ed on chromo-some 7, and its biological function is yet to bedetermined [40,50]. LOI of MEST has been reportedin invasive breast cancer [31]. Because the putativeprotein shares amino acid homology with the a/b

hydrolase fold family, which also includes thelysosomal enzyme cathepsin A, a possible role ofMEST is in the degradation of the extracellularmatrix in the invasive state of tumor development.On the other hand, Mest knockout mice have showngrowth retardation and have smaller placentas inaddition to abnormal maternal behavior [4]. Thisgenetic study implies that Mest may be involved inthe positive regulation of growth of the fetus and/orplacenta. Frequent LOI of MEST in lung cancers,together with the previous report of LOI in breast

Figure 2. Histopathologic analysis of the lymphocytic response totumor by hematoxylin and eosin staining. (A) Case 36: well-differentiated, MEST expressed biallelically (�400). (B) Case 50:well-differentiated, MEST expressed monoallelically (� 400). The

degree of lymphocytic in®ltration present in the tumors appearedunlikely to account for the biallelic expression pro®les of the MESTgene.

IMPRINTING ANALYSIS IN LUNG ADENOCARCINOMA 189

and colorectal cancers, is consistent with a supposedgrowth-promoting function of the MEST gene.

The most intensively studied epigenetic modi®ca-tion is methylation of genomic DNA [51]. It hasbeen proposed that tumor-speci®c increase or loss ofmethylation plays a prominent role in cancer. It isknown that both global hypomethylation andregional hypermethylation are found regularly inthe same tumor [52]. Silencing of tumor suppressorgenes by promoter hypermethylation in cancer hasbeen established from examination of the retino-blastoma 1 gene p16, the breast cancer 1 gene, andthe von Hippel±Lindau gene. On the other hand,increased expression of oncogenes by hypomethy-lation is still the subject of controversy. Althoughthe relationship to methylation is yet to beexplored, our present results and those of othersindicate that LOI is a nonrandom and speci®caberration associated with the development of lungadenocarcinomas and other tumors. Because dozensof imprinted genes have been identi®ed, compre-hensive examination of allelic expression ofimprinted genes in several types of cancers is away toward understanding the role of epigeneticregulation in cancer.

ACKNOWLEDGMENTS

This work was supported by grants from CREST ofthe Japan Science and Technology Corporation andby The Second Comprehensive 10-Year Strategy forCancer Control from the Ministry of Health andWelfare and the Ministry of Education, Science,Sports and Culture, Japan.

REFERENCES

1. Solter D. Differential imprinting and expression of maternaland paternal genomes. Annu Rev Genet 1988;22:127±146.

2. Surani MA, Allen ND, Barton SC, et al. Developmental con-sequence of imprinting of parental chromosomes by DNAmethylation. Philos Trans R Soc LondBiol1990;326:313±327.

3. Surani MA, Barton SC, Norris ML. Development of recon-stituted mouse eggs suggests imprinting of the genomeduring gametogenesis. Nature 1984;308:548±550.

4. Lefebvre L, Viville S, Barton SC, Ishino F, Keverne EB, SuraniMA. Abnormal maternal behaviour and growth retardationassociated with loss of the imprinted gene Mest. Nat Genet1998;20:163±169.

5. Cattanach BM, Beechey CV. Autosomal and X-chromoso-mal imprinting. In: Monk M, Surani A, editors. Genomicimprinting. Development (Suppl.) Cambridge: The Com-pany of Biologists; 1990. p 63±72.

6. Dechiara TM, Roberson EJ, Efstratiadis A. Parental imprint-ing of the mouse insulin-like growth factor II gene. Cell1991;64:849±859.

7. Swain JL, Stewart TA, Leder P. Parental legacy determinesmethylation and expression of an autosomal transgene: Amolecular mechanism for parental imprinting. Cell 1987;50:719±727.

8. Feinberg AP. Imprinting of a genomic domain of 11p15 andloss of imprinting in cancer: An introduction. Cancer Res1999;59:1743±1746.

9. Cui H, Horon IL, Ohisson R, Hamilton SR, Feinberg AP. Lossof imprinting in normal tissue of colorectal cancer patientswith microsatellite instability. Nat Med 1998;4:4469±4474.

10. Fukuzawa R, Umezawa A, Ochi K, Urano F, Ikeda H, Hata J.High frequency of inactivation of the imprinted H19 gene in``sporadic'' hepatoblastoma. Int J Cancer 1999;82:490±497.

11. Hibi K, Nakamura H, Hirai A, et al. Loss of H19 imprinting inesophageal cancer. Cancer Res 1996;56:480±482.

12. Kim HT, Choi BH, Niikawa N, Lee TS, Chang SI. Frequent lossof imprinting of the H19 and IGF-II genes in ovarian tumors.Am J Med Genet 1998;80:391±395.

13. Dao DD, Schroeder WT, Chao LY, et al. Genetic mechanismsof tumor-speci®c loss of 11p DNA sequences in Wilmstumor. Am J Hum Genet 1987;41:202±217.

14. Oda H, Kume H, Shimizu Y, Inoue T, Ishikawa T. Loss ofimprinting of IGF2 in renal-cell carcinomas. Int J Cancer1998;75:343±346.

15. Barlow DP. Competition: A common motif for the imprint-ing mechanism? EMBO J 1997;60:6899±6906.

16. Neumann B, Kubicka P, Barlow DP. Characteristics ofimprinted genes. Nat Genet 1995;9:12±13.

17. Peterson K, Sapienza C. Imprinting the genome: Imprintedgenes, imprinting genes, and a hypothesis for theirinteraction. Annu Rev Genet 1993;27:7±31.

18. Mitsuya K, Sui H, Meguro M, et al. Paternal expression ofWT1 in human ®broblasts and lymphocytes. Hum MolGenet 1997;6:2243±2246.

19. Horike SI, Mitsuya K, Meguro M, et al. Targeted disruptionof the human LIT1 locus de®nes a putative imprintingcontrol element playing an essential role in Beckwith-Wiedemann syndrome. Hum Mol Genet 2000;9:2075±2083.

20. Schwienbacher C, Gramantieri L, Scelfo R, et al. Gain ofimprinting at chromosome 11p15: A pathogenetic mechan-ism identi®ed in human hepatocarcinomas. Proc Natl AcadSci U S A 2000;97:5445±5449.

21. Vu TH, Li T, Nguyen D, Nguyen, BT, et al. Symmetric andasymmetric DNA methylation in the human IGF2-H19imprinted region. Genomics 2000;64:132±143.

22. Leighton PA, Saam JR, Ingram RS, Stewart CL, Tilghman SM.An enhancer deletion affects both H19 and Igf2 expression.Genes Dev 1995;9:2079±2089.

23. Chung WY, Yuan L, Feng L, Hensle T, Tycko B. Chromo-some 11p15.5 regional imprinting: Comparative analysis ofKIP2 and H19 in human tissues and Wilms' tumors. HumMol Genet 1996;5:1101±1108.

24. Lee MP, DeBaun MR, Mitsuya K, et al. Loss of imprintingof a paternally expressed transcript, with antisense orienta-tion to KVLQT1, occurs frequently in Beckwith-Wiedemannsyndrome and is independent of insulin-like growthfactor II imprinting. Proc Natl Acad Sci USA 1999;96:5203±5208.

25. Tycko B. Genomic imprinting and cancer. Results Probl CellDiffer 1999;25:133±169.

26. Nakagawa H, Chadwick RB, Peltomaki P, Plass C, NakamuraY, de La Chapelle A. Loss of imprinting of the insulin-likegrowth factor II gene occurs by biallelic methylation in acore region of H19-associated CTCF-binding sites in color-ectal cancer. Proc Natl Acad Sci USA 2001;98:591±596.

27. Hao Y, Crenshaw T, Moulton T, Newcomb E, Tycko B.Tumour-suppressor activity of H19 RNA. Nature 1993;365:764±767.

28. Webber AL, Ingram RS, Levorse JM, Tilghman SM. Locationof enhancers is essential for the imprinting of H19 and Igf2genes. Nature 1998;391:711±715.

29. Leighton PA, Ingram RS, Eggenschwiler J, Efstratiadis A,Tilghman SM. Disruption of imprinting caused by deletion ofthe H19 gene region in mice. Nature 1995;375:34±39.

30. Tanaka K, Shiota G, Meguro M, et al. Loss of imprinting oflong QT intronic transcript 1 in colorectal cancer. Oncology2001;60:263±273.

31. Pedersen IS, Dervan PA, Broderick D, et al. Frequent loss ofimprinting of PEG1/MEST in invasive breast cancer. CancerRes 1999;59:5449±5451.

190 KOHDA ET AL.

32. Bartolomei MS, Tilghman SM. Genomic imprinting inmammals. Annu Rev Genet 1997;31:493±525.

33. Nicholls RD, Saitoh S, Horsthemke B. Imprinting in Prader-Willi and Angelman syndromes. Trends Genet1998;14:194±200.

34. Uyeno S, Aoki Y, Nata M, et al. IGF2 but not H19 shows lossof imprinting in human glioma. Cancer Res 1996;56:5356±5359.

35. Hashimoto K, Azuma C, Koyama M, et al. Loss of imprintingin carcinoma. Nat Genet 1995;9:109±110.

36. Reed ML, Leff SE. Maternal imprinting of human SNRPN, agene deleted in Prader-Willi syndrome. Nat Genet 1994;6:163±167.

37. MacDonald HR, Wevrick R. The necdin gene is deleted inPrader-Willi syndrome and is imprinted in human andmouse. Hum Mol Genet 1997;6:1873±1878.

38. Mitsuya K, Meguro M, Lee MP, et al. LIT1, an imprintedantisense RNA in the human KvLQT1 locus identi®ed byscreening for differentially expressed transcripts usingmonochromosomal hybrids. Hum Mol Genet 1999;8:1209±1217.

39. Higashimoto K, Soejima H, Yatsuki H, Katsuki T, Mukai T.An Nsi I RFLP in the human long QT intronic transcript 1(LIT1). J Hum Genet 2000;45:96±97.

40. Kobayashi S, Kohda T, Miyoshi N, et al. Human PEG1/MEST,an imprinted gene on chromosome 7. Hum Mol Genet1997;6:781±786.

41. Riesewijk AM, Hu L, Schulz U, et al. Monoallelic expressionof human PEG1/MEST is paralleled by parent-speci®cmethylation in fetuses. Genomics 1997;42:236±244.

42. Nakanishi Y, Mulshine JL, Kasprzyk PG, et al. Insulin-likegrowth factor-I can mediate autocrine proliferation ofhuman small cell lung cancer cell lines in vitro. J Clin Invest1988;82:354±359.

43. Reeve JG, Brinkman A, Hughes S, Mitchell J, Schwander J,Bleehen NM. Expression of insulinlike growth factor (IGF)and IGF-binding protein genes in human lung tumor celllines. J Natl Cancer Inst 1992;84:628±634.

44. Brown AL, Graham DE, Nissley SP, Hill DJ, Strain AJ, RechlerMM. Developmental regulation of insulin-like growth factorII mRNA in different rat tissues. J Biol Chem 1986;261:13144±13150.

45. Suzuki H, Veda R, Takahashi T, Takahashi T. Alteredimprinting in lung cancer. Nat Genet 1994;6:332±333.

46. Nishihara S, Hayashida T, Mitsuya K, et al. Multipointimprinting analysis in sporadic colorectal cancers with andwithout microsatellite instability. Int J Oncol 2000;17:317±322.

47. Kondo M, Suzuki H, Ueda R, et al. Frequent loss ofimprinting of the H19 gene is often associated with itsoverexpression in human lung cancers. Oncogene 1995;10:1193±1198.

48. van Roozendaal CE, Gillis AJ, Klijn JG, et al. Loss ofimprinting of IGF2 and not H19 in breast cancer, adjacentnormal tissue and derived ®broblast cultures. FEBS Lett1998;437:107±111.

49. Kondo M, Matsuoka S, Uchida K, et al. Selective maternal-allele loss in human lung cancers of the maternallyexpressed p57KIP2 gene at 11p15.5. Oncogene 1996;12:1365±1368.

50. Kaneko-Ishino T, Kuroiwa Y, Miyoshi N, et al. Peg1/Mest imprinted gene on chromosome 6 identi®ed bycDNA subtraction hybridization. Nat Genet 1995;11:52±59.

51. Tycko B. Epigenetic gene silencing in cancer. J Clin Invest2000;105:401±407.

52. Issa JP. Aging, DNA methylation and cancer. Crit Rev OncolHematol 1999;32:31±43.

IMPRINTING ANALYSIS IN LUNG ADENOCARCINOMA 191