changes in methyl-sensitive restriction sites of liver dna from

Carcinogenesis vol.17 no.12 pp.2711-2717, 1996

Changes in methyl-sensitive restriction sites of liver DNA fromhamsters chronically exposed to hydrazine sulfate

Hua Zheng1 and Ronald C.Shank2

Environmental Toxicology Program, Department of Community andEnvironmental Medicine, University of California, Irvine,CA 92697-1825, USA

'Present address: Hematology/Oncology, UCSD Cancer Center, La Jolla,CA 92093-0182, USA2To whom correspondence should be addressed

Hydrazine sulfate is a genotoxic hepatocarcinogen for thehamster. A study was conducted to follow changes in DNAmaintenance methylation in selected genes in liver DNAduring the 21-month induction of liver adenomas andhepatocellular carcinomas by demonstrating changes inrestriction fragment length polymorphism. Male Syriangolden hamsters were exposed to hydrazine sulfate in thedrinking water at three concentrations (170, 340 and 510mg/1) shown previously to result in a dose-dependentinduction of liver tumors. Liver DNA from animals exposedto the high concentration for 6, 12, 16, 20 and 21 monthsand animals exposed to the low or mid concentration for21 months was digested with EcoRl, Mspl, Hindlll orBamHl, or a combination of one of these endonucleasesand a methyl-sensitive restriction enzyme, HpaH or Hhal.The DNA digests were subjected to Southern analysis usinga c-DNA probe for one of the following genes: DNAmethyltransferase (DMT), c-Ha-ras, c-jun, c-fos, and c-myc proto-oncogenes, p53 tumor suppressor gene or y-glutamyltranspeptidase. Alteration in DNA restriction bymethyl-sensitive endonucleases was detected in four (DMT,c-Ha-ras, p53 and c-jun) of the seven genes examined andas early as 6 months in animals exposed to the highestconcentration of hydrazine sulfate; alteration of recognitionsites in c-Ha-ras was also detected in DNA from animalsexposed for 21 months to the intermediate concentrationof hydrazine sulfate. Early changes in recognition sites,presumed to indicate altered methylation status of DNAcytosine and/or guanine mutations, were seen using c-DNAprobes for DMT, c-Ha-ras and c-jun; in the p53 tumorsuppressor gene alteration of such sites was a late eventrelevant to appearance of liver adenomas and hepatocellu-lar carcinomas. Evidence for hypomethylation in the p53and c-jun genes and hypermethylation of the c-Ha-ras andDMT genes is provided. This study supports the induction ofsite-specific hypomethylation and hypermethylation duringthe course of hydrazine carcinogenesis.

Introduction

Hydrazine is genotoxic and induces liver cancer in the hamster(1). Hydrazine is biotransformed to the methylating inter-mediate, formaldehyde hydrazone (CH2=NNH2) (2), and thecarcinogenicity of hydrazine probably depends on mutations

•Abbreviations: 5mC, 5-methylcytosine; DMT, DNA methyltransferase;GGT, y-glutamyltranspeptidase.

© Oxford University Press

arising from methylation of DNA guanine and altered mainten-ance methylation of cytosine (3). In an earlier carcinogenicitystudy (1) chronic administration of hydrazine sulfate resultedin transient levels of A^7-methylguanine and C^-methylguaninein liver DNA that diminished to undetectable levels after ayear of exposure, then returned to dose-dependent levels justbefore hepatocellular carcinomas became evident. Those resultswere repeated in a recent study (3) that also provided evidencefor a dose-dependent decrease in the level of 5-methylcytosine(5mC*) in liver DNA that correlated well with the developmentof liver tumors, especially adenomas. This paper providesadditional evidence for hypomethylation of two genes in liverDNA in the same tissue as studied by FitzGerald and Shank(3), using restriction of liver DNA by methyl-sensitive endonu-cleases and Southern analysis of the resulting DNA fragments.Evidence is also provided that suggests two other genes maybe hypermethylated.

There is ample evidence for hypomethylation of DNA inneoplastic tissue as compared with normal tissue. Reducedlevels of 5mC have been demonstrated in both benign andmalignant colon tumors (4) and in hamster kidney tumors (5).A number of proto-oncogenes including Ha-ras, Ki-ra^ andc-myc have been reported to be undermethylated in neoplastictissue (6-13). Other studies have provided evidence of hyper-methylation of DNA in tumor tissue (14-17).

Several studies have demonstrated that chemical carcinogenscan alter the DNA methylation system (18-20). These carcino-gens mostly induce a decrease in genomic 5mC level byaltering the recognition sequence for DNA methyltransferase(DMT) by directly inhibiting the activity of the methyl trans-ferase or by depletion of the methyl donor, S-adenosylmethion-ine (21-24). For example, 5-azacytidine, an antimetabolite ofcytidine, is incorporated into DNA and leads to extensivehypomethylation at maintenance CpG sites; this can result inreactivation of methylation-silenced genes and new cellularphenotypes (25,26). Ethionine, an antimetabolite of methionineand a non-genotoxic liver carcinogen, inhibits the synthesis ofS-adenosylmethionine by competing with methionine to form5-adenosylethionine; this ethyl analog inhibits DNA methyl-ation in regenerating rat liver (27), and induces phenotypicchanges in cultured cells by compromising DNA methylation(28). Another mechanism to explain hypomethylation is theslow and inefficient maintenance methylation that occurs inrepair patches (29,30), as discussed in the accompanying paper(3). This is particularly relevant here as hydrazine has beenshown to be genotoxic in hamster liver (1-3). It may be thatthe hypomethylation seen here resulted from the incompletemethylation of deoxycytidine newly incorporated intosequences of DNA after excision repair of methylated guaninenucleotides. Mechanisms for de novo methylation (hyperme-thylation) of DNA are less clear and are discussed later.

Activation of many unexpressed genes, which has beenshown to occur frequently during chemical carcinogenesis, isaccompanied by hypomethylation of cytosine in DNA (31).

2711

Downloaded from https://academic.oup.com/carcin/article-abstract/17/12/2711/345308by gueston 11 April 2018

H.Zheng and R.C.Shank

Since over-expression of certain proto-oncogenes is believedto be a tumorigenic process, studies to evaluate the relationbetween DNA methylation and expression of oncogenes duringcarcinogenesis are important. Wainfan et al. (32) reported thatchanges in methylation levels in DNA correlated well withchanges in expression of c-myc and c-fos and c-Ha-ras mRNAin livers of rats fed methyl-deficient diet. These proto-onco-genes (e.g. c-myc, c-fos and c-Ha-ras) were found to containhigh transcription frequency-like sequences, which are CpGrich islands (33). Methyl-deficient diet induced hypomethyl-ation in high transcription frequency sequences of c-Ha-rasand c-Ki-ras oncogenes in DNA from both neoplastic andpreneoplastic livers of rats (34). This hypomethylation may bedirectly responsible for the over-expression of these onco-genes (35).

In a benzidine-induced liver tumor model of B6C3F1 mice,the methylation status of proto-oncogenes c-Ha-ras and c-Ki-ras tend to be hypomethylated in tumor tissues compared withadjacent normal liver tissue. This was evidenced by the alteredrestriction patterns that were generated by methyl-sensitiverestriction endonucleases in DNA fragments that hybridizedto Ha-ras and Ki-ras probes (12). Methylation levels of rasproto-oncogenes were different in livers from different strainsof mice; hypomethylation corresponded to high susceptibilityof the strains for hepatoma development (such as B6C3F1mice) while hypermethylation of ras genes correlated with alow tendency to develop liver tumors (36). Those studiessuggested that alterations of methylation status of individualoncogenes (i.e. qualitative hypomethylation) may have a directinfluence on oncogene expression and tumor development. Asimilar study, characterizing DNA methylation during tumorinduction up to 21 months, is reported here.

Materials and methodsAnimalsWeanling male Syrian golden hamsters (Charles River Breeding Laboratories,Wilmington, MA) were exposed to hydrazine sulfate in the drinking waterfor up to 21 months in the study described by RtzGerald and Shank (3).Exposure levels were 0, 170, 340 and 510 mg hydrazine sulfate per literdrinking water, solutions were prepared fresh daily. Water consumption wasmeasured at five randomly selected 10-day periods, and animals were weighedevery 5 weeks. Three animals from each exposure group were sedated withCO2 and decapitated after 6, 12, 14, 16, 18, 20 and 21 months exposure. Notall liver DNA samples were examined for this report. Only those tissues thatwere most likely to show changes were examined, that is, DNA from animalsexposed to the highest concentration for 6, 12, 16 and 20 months (todemonstrate a time-related response), and from animals exposed for 21 monthsto all concentrations (to demonstrate a dose-related response).

Chemicals and materials

Restriction endonucleases Hhal, HpaW, Msp\, and BamWl, agarose (ultra-pure), proteinase K, phenol (molecular biology grade) and X HindlU fragmentmarker were purchased from Gibco BRL (Gaithersburg, MD); £coRI andWndlll were obtained from Boehringer Mannheim Corp. (Indianapolis, IN).RNase A and ethidium bromide were purchased from Sigma Chemical Co.(St Louis, MO). Zeta-Probe GT membranes were obtained from Bio-RadLaboratories (Richmond, CA) and Hyperfilm™-MP high performance autorad-iography film was purchased from Amersham Corp. (Arlington Heights, IL).

DNA probes

The mouse genomic clone of the pRSV-S107 myc proto-oncogene contained5.4 kb (exons 2 and 3), 2.4 kb and 1.8 kb (exon 1) gene fragments releasedfrom the vector pRSVcat by BamYW restriction; the proto-oncogene c-fos fromthe BALB/c mouse [pc-/oi(mouse)-3] was a 7.1 kb fragment of the FBJosteosarcoma oncogene inserted in the vector pBR322 from which £coRIreleased a 6.6 kb fragment that, when further cut with Sad, released a 2.0 kbfragment (5' part of the gene containing exon 1). These two clones wereobtained from the American Type Culture Collection (Rockville, MD). Mousec-jun cDNA clone (pBS JUNO), with a 1.6 kb fragment of the mouse c-jun

cDNA inserted in the Pstl sites of a Bluescript I vector (pGem) (37), wasprovided by Dr H.Su and Dr Peter Vogt (USC-Medical Center, Los Angeles,CA). The BS-9 clone, provided by Dr R.Ellis of the National Cancer Institute(Bethesda, MD), was composed of a 460 bp fragment subcloned from aHarvey murine sarcoma virus full-length genome clone (0-5540 bp), the cloneH-l (38); this probe, which included the 5' half of the v-ras gene (3910-4370bp), was inserted at the £coRI sites of plasmid pBR322 and shown tocrosshybridize with identical DNA fragments probed by the rat cDNA probec-Ha-rasl (39). The mouse p53 cDNA clone (p27. la), with a 745 bp fragmentinserted in the Pst\—BgR\ sites of the plasmid, representing the C-terminalsequence of 280 amino acids (two-thirds the length cDNA at the 3' portionof the gene) (40), was provided by Dr A.Bernstein (Samuel LunenfeldResearch Institute, Toronto, ON). A 4.8 kb fragment representing the nearlyfull-length cDNA of mouse DMT gene, cloned in the SmaX site of the vectorpBluescript SK Ml3+ (41), was kindly provided by Dr T.Bestor (HarvardMedical School, Boston, MA).

DNA isolation and restriction

High molecular weight DNA for Southern analysis was isolated from frozenliver samples following the protocol described in Current protocols inmolecular biology (42). Protein in fresh liver homogenate was digested withproteinase K, and DNA was isolated by repeated extraction with phenol/chloroform/isoamyl alcohol and precipitation with ethanol; RNA was removedwith RNase A. DNA restriction was accomplished using the recommendedincubation conditions and buffers supplied by the vendor of each endonuclease;the restriction buffer used in double digestions was the same as used for themethyl-sensitive endonucleases (Hpall or Hhal) Restrictions were taken tocompletion under the same conditions (amount of DNA sample, restrictionvolume, enzyme concentration) and confirmed by agarose gel electrophoresis.

Southern analysis of DNA restriction fragments

Southern analysis of DNA was carried out as described by Ausubel et al.(42). Ten microgram aliquots (in 40-50 \i\) of restricted DNA were loadedinto wells in a 1% agarose gel together with the XHindlU fragment markerand the unrestricted high molecular weight DNA control. Fragments werefractionated by electrophoresis using 1 X TAE [0.04 M Tris—acetate, 0.001M disodium ethylene-diaminetetra-acetate (EDTA), pH 8.0] or 1 X TBE (0.09Tns-borate, 0.002 M EDTA, pH 8.0) running buffer and 45 V for 16-24 h;gels were stained with ethidium bromide and photographed under UV light(302 nm). Transfer of DNA fragments from the denatured gel onto Zeta-Probe GT membranes was performed by paper-mediated capillary transferwith 6 X SSC (0.9 M sodium chloride, 0.1 M sodium citrate, pH 7.0) as themobile phase. Sodium phosphate buffered hybridization solution [0.5 MNaPO4 (2.16:1 Na2HPO4:NaH2PO4)/l mM EDTA/1% bovine serum albumin/7% sodium dodecylsufate/15% formamide, pH 7.2] was used for bothprehybridization (1-2 h) and hybridization (up to 24 h at 63—65°C dependingon the stringency for each individual probe and blot) as described by Churchand Gilbert (43). Washed membranes were exposed to Hyperfilm™-MP withintensifying screens for up to 10 days at —70°C.

Results

Liver DNA was isolated from hamsters that received either 0,170, 340 or 510 mg hydrazine sulfate per liter drinking waterfor 21 months or 510 mg hydrazine sulfate per liter drinkingwater for 6, 12, 16 and 20 months. Changes in restrictionpatterns in the DNA produced by EcoRl, HindUl, BamHl,Mspl and Hpall or combinations of EcoRl + Hpall, BamHl+ Hpall, BamHl + Hpall and HindlU + Hhal were determinedby Southern hybridization to DNA probes of c-jun (1.6 kb),c-fos (2.0 kb), c-myc (1.8 kb), c-Ha-ras (the BS-9 clone), p53(1.1 kb), y-glutamyltranspeptidase (GGT) (1.0 kb) and DMT(1.1 kb) genes. For hybridizations using the GGT, c-fos andc-myc probes, no changes in the methyl-sensitive restrictionpatterns in any of the DNA samples analyzed were detected(autoradiographic data not shown); however, several changesin restriction band patterns were observed when probes forDMT, c-Ha-ras, p53 and c-jun were used.

Digestion of DNA with both EcoRl and Hpall producedrestriction patterns that revealed changes related to the highconcentration (510 mg/1) of hydrazine sulfate at exposures of6-21 months and to the intermediate (340 mg/1) and highconcentrations after 21 months exposure. Restriction patterns

2712


Changes in methyl-sensitive restriction

AaBaCaDaAbBbCbDbAcBcCcDc AdBdCdDd |A» Be

Fig. 1. Autoradiograph for Southern hybridization of hamster DNA withp53 probe. DNA samples in lanes A - D were from livers of hamsters givenhydrazine sulfate in water at doses of 0, 170, 340 and 510 mg/1,respectively, for 21 months. Ten |ig DNA per lane were restricted with eachof the endonuclease sets—a £coRI + HpaU; b: EcoRl only; c: Mspl;d: HpaU only. EcoRl produced one major hybridization band (14.0 kb);double digestion by EcoRl + HpaU restricted this 14.0 kb band into fivesmaller fragments of 11.0, 8.0, 6.2, 5.0 and 2.5 kb. DNA samples restrictedby HpaU alone were not completely digested; however, one 8.0 kb bandwas distinguishable in lanes Ad, Bd and Cd. Both 5.0 kb band (lane Da)and 8.0 kb band (lane Dd) were lost by HpaU digestion of DNA fromanimals that were exposed to the highest concentration of hydrazine sulfate,suggesting that a demethylation of 5mC occurred at the S'-C^GC-S'HpaU recognition sites. No change in hybridization band patterns wasobserved when DNA from high dose animals exposed for less than 21months was restricted by EcoRl, EcoRl + HpaU, HpaU or Mspl (data notshown), indicating the putative demethylation in the p53 gene was a lateevent.

Fig. 2. Autoradiograph for Southern hybridization of hamster DNA withp53 probe. DNA samples in lanes A - D were from livers of hamsters givenhydrazine sulfate in water at doses of 0, 170, 340 and 510 mg/1,respectively, for 21 months. Ten ng DNA per lane were restricted with eachof the endonuclease sets—e: HindlU + Hhal; f: HindlU only; g: BamHl +HpaU; or h: BamHl only. HindlU digestion produced two majorhybridization bands at 8.0 and 3.8 kb in DNA from hamsters exposed tohydrazine sulfate for 21 months (lanes Af-Df). The two HindlU bandswere further restricted by HpaU (lanes Ae, Be, Ce and De), generating fourhybridization bands at 8.0, 5.2, 1.8 and 1.2 kb. The 8.0, 5.2 and 1.8 kbbands were lost, suggesting greater digestion by methyl-sensitive Hpal ofDNA from the high dose animals (lane De). BamHl restriction generatedthree major bands (8.5, 6.0 and 1.8 kb) in all four DNA samples (lanes Ah,Bh, Ch and Dh). Double restriction by BamHl + HpaU (lanes Ag, Bg, Cgand Dg) led to the disappearance of the 8.5 and 6.0 kb BamHl bands andappearance of a new 3.2 kb band; the 1.8 kb band was unchanged. Nochange in the pattern of restriction by BamHl + HpaU or SamHI alone wasdetected among the four DNA samples.

that show a loss of high molecular weight hybridization bandswith an increase in low molecular weight bands indicate morefrequent cutting of the DNA; this would be expected formethyl-sensitive endonucleases such as HpaU and Hhal whenthe inhibitory 5mC moiety is replaced by cytosine in therestriction site (hypomethylation). In this study when restrictedDNA was hybridized with p53 or c-jun probes (Figures 1-4),hypomethylation of DNA was apparent. New HpaU recognitionsites (i.e. more frequent cutting of the DNA with a consequentloss of high molecular weight fragments) were detected in/?53-hybridizing bands in DNA from animals exposed Whydrazine sulfate, but only at the highest concentration andlongest exposure. The putative hypomethylation was detectedusing two restriction systems, EcoRl + HpaU and HindBl +Hhal. In the c-jun system the hypomethylation could bedetected in the DNA from mid-dose animals as well, at thelongest exposure.

Different results were obtained when restricted DNA washybridized with DNA 5'-deoxycytidine methyltransferase(DMT) and c-Ha-ras probes and restricted with £coRI andHpaU (Figures 5-8). Instead of an increase in restrictionfrequency, loss of restriction sites was detected as formationor persistence of high molecular weight bands. HpaU restrictionsites were lost (i.e. a 3.9 kb EcoRl band was not cut by HpaU,Figure 6) in DMT-hybridizing DNA fragments from hamstersexposed to only the highest concentration of hydrazine sulfate,but that site alteration was detectable at the earliest measure-ment time, 6 months exposure (Figure 5). The same result wasseen when the DNA was probed with c-Ha-ras, except thatthe restriction site alteration was detected in DNA from mid-dose animals as well but only at 21 months exposure. Theseresults suggest that the restriction sites in the DMT and c-

Aa Ba Ca Da Ab Bb Cb Db Ac Be Cc Dc Ad Bd Cd Dd i |u»

Fig. 3. Autoradiograph for Southern hybridization of hamster DNA withc-jun probe. DNA samples in lanes A - D were from livers of hamstersgiven hydrazine sulfate in water at doses of 0, 170, 340 and 510 mg/1,respectively, for 21 months. Ten ng DNA per lane were restricted with eachof the endonuclease sets— a: EcoRl + HpaU; b: EcoRl only; c: Mspl; ord: HpaU only. Restriction by EcoRl produced two major c-yun-hybridizingfragments (8.5 and 4.4 kb) in DNA from animals exposed to hydrazinesulfate for 21 months (lanes Ab, Bb, Cb and Db); double restriction withEcoRl + HpaU (lanes Aa-Da) produced as many as four restriction bands(8.5, 6.9, 2.3 and 1.7 kb). The 6.9 kb band was partially lost in DNA fromhamsters exposed to 340 mg hydrazine sulfate per liter for 21 months (laneCa), and both the 8.5 and 6.9 kb bands were lost in the DNA (lane Da)from the high exposure animals.

Ha-ras hybridizing fragments were either hypermethylated(increase in frequency of 5mC in the -CCGG-sites) and/ormutated; either case could account for loss of recognition by

2713


H.Zheng and R.CShank

Fr - Gb Hb Ec Fc Gc He Ed FiJ Gd Hd uncutX Hin dill Ea Fa Ga Ha Eb Fb Gb Hb

- * 23.1

- i 9.4- 1 6.6- i 4.4

- i 2.3

c-jun

Fig. 4. Autoradiograph for Southern hybridization of hamster DNA withc-jun probe. DNA samples in lanes E—H were from livers of hamstersgiven 510 mg/1 hydrazine sulfate in water for 6, 12, 16 and 20 months,respectively. Ten u.g DNA per lane were restricted with each of theendonuclease sets—a: EcoRl + Hpall; b: EcoRl only; c: Mspl; or d: Hpallonly. EcoRl restriction yielded two fragments that hybridized with the c-junprobe. In EcoR\ + Hpall restricted samples (Ea-Ha) these fragmentsdisappeared and a new smaller 1.7 kb fragment became evident; thisindicates complete restriction by Hpall of the EcoRl fragments, the sameresult as seen with the DMT probe (Figure 5, lane Aa). No significantchanges were seen in restriction patterns for any of the six endonucleasesused on any of the DNA samples.

Aa Ba Ca Da Ab Bb Cb Db Ac Be Cc De

DMT

Fig. 5. Autoradiograph for Southern hybridization of hamster DNA withDMT probe. DNA samples in lanes A —D were from livers of hamstersgiven hydrazine sulfate in water at doses of 0, 170, 340 and 510 mg/1,respectively, for 21 months. Ten |ig DNA per lane were restricted withendonuclease sets—a: EcoRl + HpaU; b: EcoRl only; c: Mspl; andd: Hpall only. In lanes Aa, Ba and Ca, methyl-sensitive endonuclease Hpallrestricted the 3.9 kb EcoRl fragment into nonvisible smaller fragments,which led to the total loss of the 3.9 kb hybridization signal. The 3.9 kbband was still present in lane Da (DNA from the highest exposure group),although the signal was less intense than that in the corresponding lane Dbrestricted by EcoRl only. This change indicated a substantial loss of theHpall restriction sites within this 3.9 kb DNA fragment for animals exposedto the highest concentration of hydrazine sulfate. Mspl restriction generatedtwo major bands with hybridization signals at 3.8 and 2.0 kb; there was nodifference between lanes. Hpall alone was not able to restrict DNA well, asthe hybridization signals remained in high molecular weight (=23 kb) DNA.

Hpall and persistence of the high molecular weight bands inthe restriction patterns.

Details of the results are given in the legends to the figures.

Discussion

Southern analysis of DNA isolated from hamsters exposedto hydrazine sulfate in a long-term carcinogenicity study

2714

3.9 kb

DMTFig. 6. Autoradiograph for Southern hybridization of hamster DNA withDMT probe. DNA samples in lanes E—H were isolated from livers ofhamsters given 510 mg/1 hydrazine sulfate in water for 6, 12, 16 and 20months, respectively. Ten ng DNA per lane were restricted with each of theendonuclease sets—a: EcoRl + Hpall and b: EcoRl only. Pattern is similarto that seen in lane Da in Figure 5, indicating a lack of Hpall restriction onthe 3.9 kb EcoRl fragment. Thus, the blocking of Hpall restriction in the3.9 kb £coRI fragment of the DMT gene seen in the highest dose animalsafter 21 months exposure (Figure 5) appears to have begun as early as thesixth month of hydrazine exposure.

demonstrated several changes in methyl-sensitive restrictivepatterns for four of the seven genes examined. The fact thatprobing digested genomic DNA with c-DNA fragments forthe DMT, Ha-ras, p53 and jun genes detected restriction sitechanges related to hydrazine exposure, but probing with c-DNA fragments for/as, myc and y-glutamyltranspeptidase didnot detect such changes, indicates some specificity to thehydrazine-induced DNA damage. This specificity was apparentin tissue from animals exposed for as long as 21 months, and inextensively damaged tissue with adenomas and hepatocellularcarcinomas.

The putative demethylation in the p53 gene in hydrazine-induced carcinogenesis apparently is a late event, as thesechanges were not detectable in the first 20 months of exposurefor high-dose animals. This could argue that the hypomethyl-ation of this gene is probably not part of the transformationprocess; however, it could be that the hypomethylated cell isthe one selected for promotion and progression to the franktumor. This experiment cannot answer that question.

Alterations in restriction included both increased anddecreased frequencies of site recognition. Increased recognitionwas seen in the loss of high molecular weight fragmentsrecognized by two methyl-sensitive restriction endonucleases,Hpall and Hhal. These enzymes cleave DNA at the arrow insequences 5'-CiCGG-3' (Hpall) and 5'-GCGlC-3' (Hlial);however, if any of the cytosine nucleotides is methylated atthe 5-position, the enzymes do not cleave the DNA. Anincrease in the number of restriction sites for these enzymessuggests a reduced occurrence of 5mC in the sites, hencehypomethylation of the DNA. This putative hypomethylationwas seen when the DNA was probed with c-DNA fragmentsfor p53 and c-jun. The genomic sequence of the Syrian hamster



Aa Ba Ca Da Ab Bb Cb ObAc Be Cc De Ad Bd Cd Dd I <»)

c-H-nsFig. 7. Autoradiograph for Southern hybridization of hamster DNA withc-Ha-ras probe. DNA samples in lanes A - D were from livers of hamstersgiven hydrazine sulfate in water at doses of 0, 170, 340 and 510 mg/1,respectively, for 21 months. Ten |_g DNA per lane were restricted with eachof the endonuclease sets—a: EcoRI + HpaU; b: EcoRI only; c: Mspl; or d:HpaU only. Digestion by EcoRI + HpaU (lanes Aa, Ba, Ca and Da)restricted one major EcoRI fragment of 20 kb into three smaller c-Ha-rasbinding bands of 9.5, 7.0 and 6.6 kb. The other major fragment, 3.4 kb,generated by EcoRI restriction (lanes Ab—Db) was further digested byHpaU, which led to the disappearance of this band in DNA from controlanimals and animals exposed to the low concentration of hydrazine sulfate(lanes Aa and Ba). This 3.4 kb band persisted in DNA from animalsexposed to the middle and high concentrations of hydrazine sulfate for 21months (lanes Ca and Da), a result similar to that for the DMT gene. Thehybridization signal was stronger in lane Da than in Ca, which is consistentwith a dose—response effect. No change in restriction patterns was seenamong the DNA samples digested with Mspl or HpaU alone.

E _ F _ G _ H _ E b F b < _ - H _ Ec Fc Ge He Ed Fd Od Hd S

3.4kb » -

mi— u

2.0

c-H-ras

Fig. 8. Autoradiograph for Southern hybridization of hamster DNA withc-Ha-ras. DNA samples in lanes E—H were isolated from livers of hamstersgiven 510 mg/1 hydrazine sulfate in water for 6, 12, 16 and 20 months,respectively. Ten ng DNA per lane were restricted with each of theendonuclease sets—a: EcoRI + HpaU; b: EcoRI only; c: Mspl; or d: HpaUonly. Results similar to those in Figure 7 were observed for HpaUrestriction of the 20 kb EcoRI fragment of DNA from animals exposed tothe high concentration of hydrazine sulfate for 6-20 months (lanes Eb-Hb)into three smaller bands (9.5, 7.0 and 6.6 kb; lanes Ea—Ha; cf. lanesAa-Da in Figure 7). The 7.0 and 6.6 kb bands in lane Ha (longestexposure) were less distinctive compared with those in lanes Ea, Fa and Ga(shorter exposure). The 3.4 kb EcoRI bands (in lanes Eb-Hb) were reducedin intensity, but still detectable, after restriction by HpaU (lanes Ea-Ha).These results are similar to those obtained with the DMT probe, indicatingthat the HpaU restriction site was blocked within the first 6 months ofexposure to the high concentration of hydrazine sulfate.

p53 gene surrounding exons 5-8 has been described by Changet al. (44); there are two HpaU. sites and two Hhal sites, oneeach located in an exon. Tornaletti and Pfeifer (45) have shownthat in the normal human p53 gene (exons 5—8) all CpG sitesare methylated. p53 is a tumor suppressor gene; the effect ofhypomethylation in exons 5-8 in this gene has not yet beencharacterized, but might alter binding of nuclear protein to thegene (44,47). While methylation in promoter regions has beenassociated with decreased gene expression, methylation inexons does not correlate with gene expressibility (45), butincreased expression of a mutated p53 due to hypomethylationwould be expected to decrease tumor suppression activity andincrease oncogenic activity (48,49). '

Probing the DNA digest with DNA fragments that hybridizeto Ha-ras and DMT, however, gave the opposite response; forthese fragments there was a decrease in the frequency ofrestriction manifested by the appearance of high molecularweight DNA bands. The genomic sequence of the Syrianhamster Ha-ras gene was elucidated by Ebert et al. (50); itcontains 12 HpaU sites (10 in exons) and six Hhal sites (fourin exons). A decrease in the recognition of a restriction sitecould be due to either: (i) hypermethylation of cytidinenucleotides in the site which inhibits the endonuclease activityof HpaU (5'-nK:mCGG-3') and Hhal ^ ' - G ^ G 1 ^ ^ ' ) ; or (ii)mutation in the site. The most likely mutations would be fromdeamination of 5mC to thy mine, which would not requirereplication, or a single base mutation of a guanine nucleotide,which would require replication. Hydrazine exposure is knownto result in methylation of DNA guanine in the hamster (1,3).Hypermethylation of the ras gene, unlike hypomethylation ofthe p53 gene, was detected at the first observation point, 6months exposure. Hypermethylated ras was also reported byVorce and Goodman (36), but hypomethylation in GCGCsequences of the c-Ha-ras gene was reported by Rao et al.(51) in DNA from hepatic nodules but not from non-nodularsurrounding liver of rats treated with 1,2-dimethylhydrazineand orotic acid.

HpaU and Mspl have the same restriction recognitionsequence, 5'-C4-CGG-3', but the two endonucleases differ intheir sensitivity to the presence of 5mC in the restriction site.Methylation of either cytosine inhibits the activity of HpaU;for Mspl, however, only methylation of the 5' cytosine inhibitsactivity, and methylation of the 3' cytosine has no effect. Innone of the Southern analyses was a hydrazine-related changein Mspl restriction observed. This strongly suggests that, ifhypermethylation accounts for the inhibition of HpaU activityin the restriction of DMT and Ha-ras sensitive sites, thenthe site of methylation must be the internal cytosine (5'-

Although most of the methylation of cytosine in DNAoccurs in hemi-methylated substrate at the end of replication,de novo methylation of unmethylated DNA has been describedby several investigators (52). The A -̂terminal domain ofmammalian DMT normally inhibits the de novo methylationof unmethylated DNA and loss of this functional terminus hasbeen associated with hypermethylation (53). The fact that thepossible hypermethylation seen in this study was observed inonly two of the seven hybridization systems argues against aneffect of hydrazine on the transferase, but rather on specificsites in DNA. The ability of DMT to methylate unmethylatedDNA is known to be affected by secondary structures in DNA(54), and it may be that the methylation of the 3' guanine inthe -CG- dinucleotide can have such an effect. This would

2715


H.Zheng and R.CShank

suggest, then, that such guanine nucleotides in the DMT andHa-ras genes are methylated to a greater extent during hydra-zine exposure, than are similar guanine nucleotides occurringin the other sites examined in this study.

Restriction polymorphism is neither sufficiently sensitivenor quantitative to answer questions on the mechanisms ofDNA damage by hydrazine. In the accompanying paper (3),analysis of differential rates of incorporation of thymidine andthe methyl moiety of methionine into DNA indicated areduction in maintenance methylation from what would beexpected from the uptake of thymidine. That study lookedonly at incorporation into total liver DNA; the current studywas able to increase sensitivity by narrowing the focus toknown sequences in seven genes and thus detected bothhypomethylation and hypermethylation of target DNA. Studiesare currently underway to determine the methylation status ofeach cytosine in exons 5-8 in the p53 gene of the same DNAas used in this study using ligation mediated PCR genomicsequencing; also PCR amplification of mutational hot spots inthe p53 gene will determine the DNA sequence changes inspecific cell types isolated from these same tissues.

Acknowledgements

This work was supported in part by PHS grants RO1-ES03726 and T32ES07157 from the US National Institutes of Health.

References

l.Bosan.W.S., Shank,R.C, MacEwenJ.D., Gaworski.C.L. and Newbeme,P.M. (1987) Methylation of DNA guanine during the course of inductionof liver cancer in hamsters by hydrazine or dimethylnitrosamine.Carcinogenesis, 8, 439—444.

2. Lambert.C.E. and Shank,R.C. (1988) Role of formaldehyde hydra-zone and catalase in hydrazine-induced methylation of DNA guanine.Carcinogenesis, 9, 65-70.

3. FitzGerald.B.E. and Shank,R.C. (19%) Methylation status of DNA cytosineduring the course of induction of liver cancer in hamsters by hydrazinesulfate. Carcinogenesis, 17, 2703-2709.

4.Feinberg,A.P., Gehrke.C.W., Kuo.K.C. and Ehrlichjvl. (1988) Reducedgenomic 5-methylcytosine content in human colonic neoplasia. CancerRes., 48, 1159-1161.

5,Lu,L.-J.W., LiehrJ.G., Sirbasku.D.A., Randerath.E. and Randerath.K.(1988) Hypomethylation of DNA in estrogen-induced and -dependenthamster kidney tumors. Carcinogenesis, 9, 925-929.

6. Feinberg,A.P. and Vogelstein,B. (1983) Hypomethylation of ras oncogenesin primary human cancers. Biochem. Biophys. Res. Commun., Il l , 47—54.

7.Cheah,M.S.C., WaUace.C.D. and Hoffman,R.M. (1984) Hypomethylationof DNA in human cancer cells: a site-specific change in the c-myconcogene. J. Natl Cancer Inst., 73, 1057-1065.

8.Ramsden,M., Cole.G., Smith J. and Balmain.A. (1985) Differentialmethylation of the c-Ha-ras gene in normal mouse cells and during skintumor progression. EMBO J., 4, 1449-1454.

9.Kaneko,Y., Shibuya,M., NakayamaJ., HayashidaX., Toda,G., Endo.Y.,Oka,H. and Oda,T. (1985) Hypomethylation of c-myc and epidermalgrowth factor receptor genes in human hepatocellular carcinoma and fetalliver. Jpn. J. Cancer Res., 76, 1136-1140

10.Barbieri,R., Piva,R., Buzzoni.D., Volinia,S. and Gambari.R. (1987)Clustering of undermethylated CCGG and GCGC sequences in the 5'region of the Wz-ras-1 oncogene of human leukemic K562 cells. Biochem.Biophys. Res. Commun., 145, 96-104.

ll.Nambu.S., Inoue.K. and Sasaki.H. (1987) Site-specific hypomethylationof the c-myc oncogene in human hepatocellular carcinoma. Jpn. J. CancerRes., 78, 695-704.

12.Vorce,R.L. and GoodmanJ.I. (1989) Hypomethylation of ras oncogenesin chemically induced and spontaneous B6C3F1 mouse liver tumors. MoLToxicol., 2, 99-116.

13.Vorce,RX. and Goodman J.I. (1989) Altered methylation of rat oncogenesin benzidine-induced B6C3F1 mouse liver tumors. Toxicol. Appl.Pharmacol., 100, 398-410.

14.Baylin,S.B., HoppenerJ.W., deBustrosA, Steenbergh,P.H., Lips,CJ. andNetkin.B.D. (1986) DNA methylation patterns of the calcitonin gene inhuman lung cancers and lymphomas. Cancer Res., 46, 2917-2922.

15.SilvermanAL., ParkJ.-G., Hamilton.S.R., GazdarAF., Luk.G.D. andBaylin.S.B. (1989) Abnormal methylation of the calcitonin gene in humancolonic neoplasms. Cancer Res., 49, 3468-3473.

16.Nelkin,B.D., Przepiorka,D., BurkeJU., Thomas,E.D. and Baylin,S.B.(1991) Abnormal methylation of the calcitonin gene marks progression ofchronic myelogenous leukemia. Blood, 77, 2431-2434.

17.Malinen,T., PalotieA-, Pakkala,S., Peltonenl., Ruutu,T. and Jansson.S.-E. (1991) Acceleration of chronic myeloid leukemia correlates withcalcitonin gene hypermethylation. Blood, 77, 2435-2440.

18.RiggsAD. and Jones,P.A. (1983) 5-methylcytosine, gene regulation andcancer. Adv. Cancer Res., 40, 1-30.

19. Krawisz,B.R. and LiebermanJvl.W. (1984) Methylation of deoxycytidine inreplicating cells treated with ultraviolet radiation and chemical carcinogens.Carcinogenesis, 2, 873—878.

20.Jones,P.A. and BuckleyJ.D. (1990) The role of DNA methylation in cancer.Adv. Cancer Res., 54, 1-23.

21.Hoehm,T.LJ. and Drahovsky.D. (1983) Alteration of enzymaticmethylation of DNA cytosines by chemical carcinogens: A mechanisminvolved in the initiation of carcinogenesis. /. Natl Cancer Inst., 71,429-432.

22. Wilson,V.L. and Jones,P.A. (1983) Inhibition of DNA methylation bychemical carcinogens in vitro. Cell, 32, 239-246.

23.Zucker,K.E., Riggs,A.D. and Smith.S.S. (1985) Purification of humanDNA (cytosine-5)-methyltransferase. J. Cell Biochem., 29, 337-349.

24.Wilson,V.L., Smith,R.A., LongoriaJ., LionxM.A., Harper,C.M. andHarris.C C. (1987) Chemical carcinogen-induced decrease in genomic 5-methyldeoxycytidine content of normal human bronchial epithelial cells.Proc. NatlAcad Sci. USA, 84, 3298-3301.

25.BouckJM., KokkinakisJD. and OstrowskyJ. (1984) Induction of a step incarcinogenesis that is normally associated with mutagenesis bynonmutagenic concentrations of 5-azacytidine. Mol. Cell. Biol., 4, 1231—1237.

26.Knust£., Bruggemann.U. and Doerfler.W. (1989) Reactivation of amethylation-silenced gene in adenovirus-transformed cells by 5-azacytidineor by E1A trans-activation. / Virol., 63, 3519-3524.

27.Cox,R. and Irving.C.C. (1977) Inhibition of DNA methylation by S-adenosyl-ethionine with the production of methyl-deficient DNA inregenerating rat liver. Cancer Res., 37, 222-225.

28.Boehm,T.L.I. and Drahovsky.D. (1979) Effect of carcinogen ethionineon enzymatic methylation of DNA sequences with various degrees ofrepetitiveness. Eur. J. Cancer, 15, 1167-1173.

29.Kastan,M.B., GowanJ3J. and Lieberman.M.W. (1982) Methylation ofdeoxycytidine incorporated by excision-repair synthesis of DNA. Cell, 30,509-516.

30.Eremenko,T., Palitti.F., Morelli.F., Whithead,E.P. and Volpe.P. (1985)Hypomethylation of repair patches in HeLa cells. Mol. Biol. Rep., 10,177-182.

31. Doerfler.W. (1983) DNA methylation and gene activity. Annu. Rev.Biochem., 53, 93-124.

32.Wainfan,D.M., Sheikhnejad.G. and CnristmanJ.K. (1990) Rapid andreversible changes in DNA methylation and gene expression in livers ofrats fed cancer-promoting methyl-deficient diets. Pwc. Am. Assoc. CancerRes., 31, 141.

33.Bird,A.P. (1986) CpG-rich islands and the function of DNA methylation.Nature, 321, 209-213.

34.BhaveJvI.R., Wilson.MJ. and Poirier,L.A. (1988) c-H-ras and c-K-rasgene hypomethylation in the livers and hepatomas of rats fed methyl-deficient, amino acid-defined diets. Carcinogenesis, 9, 343-348.

35.Hsieh,L.L., Wainfan.E., Hoshina,S., Dizik.M. and Weinstein,I.B. (1989)Altered expression of retrovirus-like sequences and cellular oncogenes inmice fed methyl-deficient diets. Cancer Res., 49, 3795-3799.

36. Vorce,R.L. and Goodman J.I. (1990) Alterations in the methylation statusof ras oncogenes in B6C3F1 mouse liver tumors. Prog. Clin. BioL Res.,331, 335-343.

37.Lampn,W.W., Wamsley.P., Sassone-Corsi.P. and Verma,I.M. (1988)Induction of proto-oncogene JUNIAp-1 by serum and TPA. Nature, 334,629-631.

38.Ellis,R.W., DeFeo.D., MaryakJ.M., Young.H.A., Shih.T.Y, Chang,E.H.,LowyJ3.R. and Scolnick.E.M. (1980) Dural evolutionary origin for the ratgenetic sequences of Harvey murine sarcoma virus. / Virol., 36, 408—420.

39.ParadaX.F., Tabin.CJ., Shin,C and Weinberg,R.A. (1982) Human EJbladder carcinoma oncogene is homologue of Harvey sarcoma virus rasgene. Nature, in, 474-478.

2716



40. Jenkins J.R., Rudge.K. and Currie,G A. (1984) Cellular mortalization bya cDNA encoding the transformation associated phosphoprotein p53.Nature, 312, 651-654.

41.Bestor,T.H., Laudano.A., Mattaliano.R. and Ingram.V.M. (1988) Cloningand sequencing of a cDNA encoding DNA methyltransferase of mousecells. J. Mol. Biol., 203, 971-983.

42.Ausubel,F.M., Brent,R., Kingston.R.E., Moore,D.D., SeidmanJ.G.,Smith J . A. and Struhl,K. (1989) Current Protocols in Molecular Biology.Wiley Interscience Press, New York (sections 2.1.1-2.2.2; 3.1.1; 3.1.12).

43 Church.G.M. and Gilbert,W. (1984) Genomic sequencing. Proc. NatlAcad.Sci. USA, 87, 1991-1995.

44,Chang,K.-W., Mangold,K.A., Hubchak,S., Laconi.S. and Scarpelli.D.G.(1994) Genomic p53 mutations in a chemically induced hamster pancreaticductal adenocarcinoma. Cancer Res., 54, 3878-3883.

45.Tornaletn.S. and Pfeifer.G.P. (1995) Complete and tissue-independentmethylation of CpG sites in the p53 gene: implications for mutations inhuman cancers. Oncogene, 10, 1493-1499.

46 Meehan.R.R., LewisJ.D., McKay.S., Kleiner.E.L. and Bird,A.P. (1989)Identification of a mammalian protein that binds specifically to DNAcontaining methylated CpGs. Cell, 58, 499-507.

47.Nan,X., Meehan.R.R. and Bird.A. (1993) Dissection of the methyl-CpGbinding domain from the chromosomal protein MeCP2. Nucleic AcidsRes., 21, 4886-4892.

48. Vogelstein.B. and Kinzler.K.W. (1992) p53 function and dysfunction. Cell,70, 523-526.

49.Dittmer.D., Pati.S., Zambetti.G., Chu.S., Teresky,A.K , Moore,M., Finlay.C.and Levme.A J. (1993) Gain of function mutations in p53. Nature Genet.,4, 4 2 ^ 5 .

50.Ebert,R., Wiseman,R.W., BarretU-C, Reiss.E., Rollich,G. andSchiffmann.D. (1992) Charactenzation of the Syrian hamster c-Ha-rasgene and intron-D-exon transcript. Mol. Carcinogen., 5, 254—258.

51.Rao,P.M., Antony.A., Rajalakshmi,S. and Sarma.D.S.R. (1989) Studies onhypomethylation of liver DNA during early stages of chemicalcarcinogenesis in rat liver. Carcinogenesis, 10, 933—937.

52. Bestor.T.H. and Ingram.V.M. (1983) Two DNA methyltransferases frommurine erythroleukemia cells: purification, sequence specificity, and modeof interaction with DNA. Proc. Nail Acad. Sci. USA, 80, 5559-5563.

53. Bestor.T.H. (1992) Activation of mammalian DNA methyltransferase bycleavage of a Zn binding regulatory domain. EMBO J., 11, 2611-2617.

54. Bestor,T. (1987) Supercoiling-dependent sequence of mammalian DNAmethyltransferase. Nucleic Acids Res., 15, 3835-3843.

Received on October 20, 1995; revised on August 19, 1996; accepted onAugust 26, 1996

27'17


changes in methyl-sensitive restriction sites of liver dna from

Documents