Proc. Nati. Acad. Sci. USAVol. 91, pp. 11733-11737, November 1994Medical Sciences

Cytidine methylation of regulatory sequences near the a-classglutathione S-transferase gene accompanies humanprostatic carcinogenesisWEN-HSIANG LEE*t, RONALD A. MORTONt, JONATHAN I. EPSTEINtt, JAMES D. BROOKSt,PEARL A. CAMPBELL*t, G. STEVEN BOVAt, WEN-SON HSIEH*t, WILLIAM B. ISAACS*t,AND WILLIAM G. NELSON*t§*The Johns Hopkins Oncology Center, tJames Buchanan Brady Institute of Urology, and tDepartment of Pathology, The Johns Hopkins University School ofMedicine, Baltimore, MD 21287

Communicated by Paul Talalay, August 16, 1994 (received for review July 18, 1994)

ABSTRACT Hypermethylation of regulatory sequences atthe locus of the a%-el glutathione S-trnsferase gene GSTPIwas detected in 20 of 20 human prostatic carcinoma tissuespecimens studied but not in normal tissues or prostatic tissuesexhibiting benign hyperplasia. In addition, a striking decreasein GSTPI expression was found to accompany human prostaticcarcinogenesis. Immunohistochemical staining with anti-GSTP1 antibodies failed to detect the enzyme in 88 of 91prostatic carcinomas analyzed. In vitro, GSTPI expression waslimited to human prostatic cancer cell lines conting GSTPIalleles with hypomethylated promoter sequences; a humanprostatic cancer cell line containing only hypermethylatedGSTP1 promoter sequences did not express GSTPI mRNA orpolypeptides. Methylation of cytidine nucleotides in GSTPIregulatory sequences constitutes the most common genomicalteration yet described for human prostate cancer.

Human cancer cells typically contain somatically alteredgenomes, characterized by mutation, amplification, or dele-tion of critical genes (1). In addition, the DNA template fromhuman cancer cells often displays somatic changes in DNAmethylation (1-8). However, the precise role of abnormalDNA methylation in human tumorigenesis has not beenestablished (2-8). For human prostate cancer, the mostcommonly diagnosed cancer in men in the United States (9),we have found uniform somatic changes in DNA methylationat the locus of the glutathione S-transferase gene GSTPL.Glutathione S-transferases (GSTs; EC 2.5.1.18) catalyzeintracellular detoxification reactions, including the inactiva-tion of electrophilic carcinogens, by conjugating chemicallyreactive electrophiles to glutathione (10-12). Human GSTs,encoded by several different genes at different loci, havebeen classified into four families, which have been referred toas a, p., x, and 0 (13). We report here that most humanprostate cancers fail to express the ir-class GST, and thatregulatory sequences near the GSTPI gene, which encodesthe human ir-class GST, appear to be commonly hypermeth-ylated during prostatic carcinogenesis.

MATERIALS AND METHODSImmunohistochemical Staiing for GSTP1. Formalin-fixed

paraffin-embedded prostatic tissue sections were stainedwith anti-GSTP1 antiserum (1:100 dilution; Oncor) using animmunoperoxidase method (Vectastain ABC Kit; VectorLaboratories), with either 3,3'-diaminobenzidine (DAB) or3-amino-9-ethylcarbazole (AEC) as the peroxidase substrate.As a control for the specificity of staining, tissue sections

prepared from normal human liver and normal human kidneywere also stained with the anti-GSTP1 antiserum. Each ofthetissues exhibited the characteristic staining pattern previ-ously described (14).Immunoblot Analysis for GSTP1 Polypeptides. Saline-

washed cultured prostatic carcinoma cells (15-19) were col-lected by centrifugation at 14,000 x g for 10 min at roomtemperature and then lysed in a protein extraction buffer [2%SDS/101% (vol/vol) glycerol/10 mM dithiothreitol in 62 mMTris'HCl, pH 7.8] by heating to 950C for 10 min. The DNAcontent of each cell extract was estimated by using a diphe-nylamine assay (20). Equivalent extracts from each of thecultured cell lines were subjected to immunoblot analysis forGSTP1 and DNA topoisomerase I polypeptides, using spe-cific antisera (rabbit anti-GSTP1 from Oncor; human anti-topoisomerase I provided by W. C. Earnshaw, Johns Hop-kins University) as previously described (21).

Northern Blot Analyss for GSTPI mRNA. Total RNAs wereextracted from cultured cells by the method of Chomczynskiand Sacchi (22) and then quantitated by using an orcinol assay(23). Purified RNAs (20 ug) were electrophoresed on 1.5%agarose gels in the presence of2.2M formaldehyde, transferredto Zeta-Probe (Bio-Rad) filters, and then assessed for GSTPIand TOP] mRNA levels by hybridization with specific 32P-labeled cDNA probes (prepared using the Random PrimersDNA Labeling System, GIBCO/BRL). Following hybridiza-tion at 50CC for 16 hr (in 50%o formamide/7% SDS/0.5% nonfatdry milk/heat-denatured salmon sperm DNA at 200 pg/ml/300mM NaCl/2 mM EDTA/20 mM sodium phosphate, pH 7.4),blots were washed with 0.3x SSC (lx SSC is 150mM NaCl/15mM sodium citrate, pH 7.0) and 0.3% SDS for 60 mi at 650Cbefore being exposed to X-Omat AR (Kodak) film at -700C.PlasmidpGSTii1 containing the entire GSTPI coding sequence(24) was obtained from the American Type Culture Collection;plasmid phTOPl-D1 containing 704 bp of human TOP) cDNA(25) was provided by L. F. Liu (Robert Wood Johnson-University of Medicine and Dentistry ofNew Jersey).Southern Blot Analysis for GSTPI Promoter Methylatlon.

DNA was isolated from human prostatic cancer lines andfrom a variety of normal and neoplastic human tissues aspreviously described (26, 27). Normal tissue specimens,including seminal vesicle, esophagus, kidney, liver, lung, andspleen, were recovered at autopsy. Normal prostate tissuespecimens were obtained from autopsies of men of differentages (18 to 69 years) who did not suffer with prostaticdiseases. Neoplastic prostate tissues, along with matchedadjacent normal prostate or matched normal seminal vesicle

Abbreviations: GST, glutathione S-transferase; BPH, benign pro-static hyperplasia.§To whom reprint requests should be addressed.

11733

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Dow

nloa

ded

by g

uest

on

Oct

ober

27,

202

0

Proc. Natl. Acad. Sci. USA 91 (1994)

tissues, were dissected from resection specimens obtainedfrom men treated for localized prostatic carcinoma by radicalprostatectomy. For analysis of GSTPI promoter methyl-ation, purified DNAs were digested first with HindIll andEcoRI and then extensively with the m5C-sensitive restric-tion endonuclease BssHII. Digested DNAs were then elec-trophoresed on agarose gels, transferred to Zeta-Probe (Bio-Rad) membranes, and then hybridized with 32P-labeledGSTPI cDNA as previously described (26).

RESULTS AND DISCUSSIONMany human cancers exhibit increased GSTP1 expressionrelative to their tissues of origin (28). Surprisingly, we foundthat most prostatic cancers contained decreased levels ofdetectable GSTP1 polypeptides relative to normal prostatictissue. To assess GSTPI expression in normal and neoplasticprostatic cells in vivo, we subjected tissue specimens toimmunohistochemical staining using anti-GSTP1 antibodies(Fig. 1). Normal prostatic tissues exhibited intense stainingfor GSTP1 in almost all basal epithelial cells (Fig. 1A). Somenormal glandular epithelial cells also appeared to containimmunoreactive GSTP1, while most contained no detectableenzyme (Fig. 1B). Lobules with GSTP1-positive or withGSTP1-negative glandular epithelial cells did not differ intheir histologic appearance or anatomic distribution. Other-wise, GSTP1-positive epithelial cells were detected in allnormal epithelial structures within prostatic tissue, includingthe verumontanum, the utricle, prostatic ducts, ejaculatoryducts, and the prostatic urethra (not shown). Positive immu-nohistochemical staining for GSTP1 was also seen in foci oftransitional cell metaplasia, squamous cell metaplasia, andhyperplasia associated with benign prostatic hypertrophy(not shown). Despite the presence of abundant GSTP1 inmost normal prostatic epithelial cells as well as in cellsmaking up benign proliferative prostatic lesions, 88 of 91specimens of prostatic carcinoma contained no detectable

GSTP1 (Fig. 1 D-F). Prostatic intraepithelial neoplasia (PIN)has been proposed to represent a premalignant lesion (29). Ofinterest, in some PIN lesions, many epithelial cells were alsodevoid of GSTP1 (Fig. 1C).To better understand GSTPI regulation in human prostate

cancer, five established human prostatic carcinoma cell lines(15-19) were assessed for GSTP1 expression by immunoblotanalysis for GSTP1 (Fig. 2A) and by Northern blot analysisfor GSTPI mRNA (Fig. 2B). In accordance with the lack ofGSTP1 expression by human prostatic cancers in vivo, im-munoblot analysis failed to detect GSTP1 polypeptides in aprotein extract prepared from LNCaP cells (Fig. 2A, lane 1),and Northern blot analysis failed to detect GSTP1 mRNA inRNA recovered from the LNCaP cell line (Fig. 2B, lane 1).In contrast, Du-145 cells, PC-3 cells, PPC cells, and Tsu-Prlcells contained abundant GSTP1 polypeptides and GSTP1mRNA (Fig. 2 A and B, lanes 2-5).Sequence analysis of the 5' regulatory region upstream of

the GSTPI gene has revealed a TATAA box, two potentialSP1 binding sites, and a consensus AP-1 site (refs. 30 and 31;Fig. 3A). Promoter studies have established that a 72-ntsequence located 5' of the transcriptional start site (fromnucleotide -80 to nucleotide -8) exhibits basal promoteractivity when ligated to a promoterless chloramphenicolacetyltransferase (CAT) reporter gene in transient transfec-tion assays (32, 33). The 400 nt lying immediately 5' of thetranscriptional start site contain nearly 72% C and G nucle-otides, with 41 CpG dinucleotides. Methylation of CpGdinucleotides in regulatory regions of mammalian genes hasbeen frequently associated with diminished transcriptionalactivity (34-38). To determine whether differences in GSTPIpromoter methylation correlate with the differences inGSTPI expression detected in the human prostatic cancercell lines, DNA from each of the cell lines was digested withthe m5C-sensitive restriction enzyme BssHII and then sub-jected to Southern blot analysis using a GSTPI probe (Fig. 3).GSTPI promoter DNA from LNCaP cells failed to cut with

FIG. 1. Ir-class GST expres-sion in normal and neoplastic hu-man prostatic tissues. Tissue sec-tions were assessed for GSTPIexpression by immunohistochem-ical staining with anti-GSTP1 an-tiserum. (A) Normal prostatic ep-ithelium with intensely stainingbasal prostatic epithelial cells.(x70.) (B) Normal prostatic tissueexhibiting lobular heterogeneity inGSTPI expression. (x40.) (C) Di-minished GSTPI expression inprostatic intraepithelial neoplasia(PIN). (x70.) (D-F) Prostatic can-cer tissues with no detectableGSTPI expression in the neoplas-ticcells.(D, x40;E, x70;F, x40.)

11734 Medical Sciences: Lee et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

27,

202

0

Proc. Natl. Acad. Sci. USA 91 (1994) 11735

A Immunoblot analysis

-GSTP

____ _ -~~~topo I

1 2 3 4 5

B Northern blot analysis

_ - ~~GSTPI-

V - topol

2 3 4 5

FIG. 2. GSTPI expression in human prostatic cancer cell lines.DNA topoisomerase I was the control. (A) Immunoblot analysisreveals decreased GSTP1 levels in LNCaP cells and abundantenzyme expression by Du-145 cells, PC-3 cells, PPC cells, andTsu-Prl cells. (B) Northern blot analysis of total RNA from each ofthe prostatic cancercell lines demonstrates decreased GSTPI mRNAlevels in the LNCaP cell line.

BssHII (Fig. 3B, lane 6), consistent with cytidine methylationat the restriction enzyme recognition site. The absence ofBssHII digestion of LNCaP GSTPJ promoter DNA did notappear to be a consequence of inadequate LNCaP DNAdigestion by BssHII in general. When the Southern blotswere hybridized with a triose-phosphate isomerase (TPI)gene probe (39, 40), complete digestion at a BssHII site wasevident (not shown). In addition, the failure of BssHII to cutLNCaP GSTPI promoter DNA likely resulted from cytidinemethylation and not from mutation at the BssHII recognitionsite, because GSTP1 promoter sequences amplified fromLNCaP DNA by the polymerase chain reaction (PCR) werereadily digested by BssHII (not shown). Furthermore, LN-CaP GSTP1 regulatory sequences also failed to cut with them5C-sensitive restriction enzymes Not I and Sac II, whichrecognize sites at -132 and -102 in the GSTPI promoter,respectively (not shown), indicating extensive methylation ofcytidines in the GSTP1 regulatory region.

A EcoRI

--2332

BssHlI

While LNCaP cells, which did not express GSTP1 mRNAor polypeptides (Fig. 2 A and B, lanes 1), contained onlyhypermethylated GSTP1 promoter sequences (Fig. 3B, lane6), Tsu-Prl cells, which expressed abundant GSTP1 mRNAand polypeptides (Fig. 2 A and B, lanes 5), appeared tocontain only hypomethylated GSTPI promoter sequences(Fig. 3B, lane 10). In addition, hypomethylated GSTP1 pro-moter sequences were detected in Du-145 cells, PC-3 cells,PPC cells, and Tsu-Prl cells (Fig. 3B, lanes 7-9). Each ofthese cell lines also expressed abundant GSTP1 mRNA andpolypeptides (Fig. 2 A and B, lanes 2-4), suggesting a strongcorrelation between GSTPI transcriptional activity and thepresence ofhypomethylated GSTP1 promoter sequences. Ashas been proposed for some C+G-rich 5' regulatory se-quences in other genes, the GSTPI promoter sequences mayconstitute a methylation-sensitive "CpG island" (41).

Clearly, the data presented thus far have demonstrated thatthe great majority of human prostatic carcinoma cells fail toexpress GSTP1 polypeptides in vivo. In addition, our studieshave revealed that the lack of GSTPI expression by LNCaPcells in vitro appears to be accompanied by hypermethylationof regulatory sequences in all GSTPI alleles. These datasuggest that a possible mechanism for the failure of humanprostate cancer cells to express GSTP1 polypeptides in vivocould be transcriptional inactivation of the GSTPI gene as aresult of methylation of CpG island sequences in the tran-scriptional regulatory region. Alternatively, the absence ofGSTPI expression in human prostate cancer cells could alsoresult from other genomic alterations at the GSTPI locus,from a lack of transcriptional activating factors, from thepresence of transcriptional repressors, or by a post-transcriptional mechanism. Furthermore, even though LN-CaP cells exhibit evidence of methylation-inactivation ofGSTP1 expression, DNA methylation patterns have beenfound to change as mammalian cells are propagated in vitro(42). Thus, to determine whether lack of GSTPI expressionby human prostate cancer cells in vivo might be accompaniedby hypermethylation of GSTP1 regulatory sequences, weundertook an analysis ofGSTPI promoter methylation, usingDNA isolated from a variety of normal tissues as well as fromprostatic carcinomas.GSTPI promoter hypermethylation has not, to the best of

our knowledge, been previously described (43). To ascertainwhether GSTPI promoter hypermethylation occurs duringnormal physiologic cellular differentiation, DNA isolatedfrom normal human tissues with different GSTPI expressionpatterns was subjected to Southern blot analysis followingdigestion with the m5C-sensitive restriction enzyme BssHII(Fig. 4). No methylated GSTPI promoter sequences weredetected in DNA isolated from normal prostate, seminal

HindlII

-298 L b+2068~~

BssHII-298

AtNot- Sac- l >1

n c)0

- 132 -1 02

B Iqb,>Nqf IO§,,EcoRI + HindIII rV+ + + +?

BssHI-kb6.6- PW4.4- ____

2 3 -

I+ 4 + + + FIG. 3. Methylation of 5' reg-ulatory sequences at the GSTPIlocus in prostatic cancer cell lines.(A) Restriction map of the GSTPIpromoter indicating m5C-sensitiverestriction enzyme recognitionsites. (B) Southern analysis of hu-man prostatic cancer cell lineDNA cut with the m5C-sensitiverestriction enzyme BssHl and hy-bridized with a GSTPI cDNAprobe.

_____

__w

IDm wV

1 2 3 4 5 6 7 8 9 10

Medical Sciences: Lee et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

27,

202

0

Proc. Natl. Acad. Sci. USA 91 (1994)

A Normal tissues Ar r,,.---s- r- r- r->A> < > < ><a ><e ><C > < ><C

U ) U)C) o C)C U) CD (1C) cn C) (PU {:

fbl e|to -I X\~ :g XZ

,o C" ", 49 \-A \s---I++ +_( _+_+_.4V

.* :.,'*A * .*

1 2 3 4 5 6 7 8 9 10 11 12 13 14

B Normal prostates

AGE (yrs) 18 20 24 25 28 29 32 43 46 46 51 56 69

EcoRI+ HindIII + + + + + + + + + + + + +

BssHII + + + + + + + + + + + + +

2 3 4 5 6 8 9 1 0 1 1 1 2 1.3 14

r---At - r- r--i -i

0 () 0 C0 0n 0n 0h0 0 0 0 0 0 o 0

B :< : <s :< as:< <LO mOa_ ) CLa_ Cn-Eau a 0 aCod -Orl{

I* m **01"

3 4 6 8 9 10 1 1 1 2 1.3 1 4 5

c0 = 0-,Co O= oX< a: CL <r, co (- L cn I-

0

O:C.-m C

cr,0 Z

Uo

LO a:A

0raICO(-

1 2 3 4 5 6 7 8 9 10 11 12 13

FIG. 4. GSTPI promoter methylation status in vivo in normaladult tissues. Shown are Southern blot analyses using a GSTPIcDNA probe ofDNA isolated from normal human tissues. (A) DNAfrom normal tissues digested first with EcoRI and HindI1 and thenwith (+) or without (-) BssHII. (B) DNA, prepared from normalprostatic tissues recovered at autopsy from men ofdifferent ages whodid not suffer from prostatic diseases, digested with EcoRl, HindIl,and BssHll. Arrow denotes migration position of GSTPI restrictionfragment uncut by BssHII.

vesicle, esophagus, kidney, liver, lung, or spleen (Fig. 4A).To assess whether hypermethylation of GSTPI promotersequences arises during aging in normal prostatic cells, DNAprepared from normal prostate tissue specimens recoveredfrom men of different ages was also analyzed (Fig. 4B).Again, there was no evidence of GSTP1 promoter hyper-methylation in any of the normal prostatic tissues.To discover whether GSTP1 promoter hypermethylation

occurs during prostatic carcinogenesis in vivo, Southern blotanalyses were performed on BssHII-digested DNA isolatedfrom human prostatic carcinoma specimens (Fig. 5). Each ofthe prostate cancers studied displayed evidence of hyper-methylation ofpromoter sequences in GSTPI alleles relativeto matched control DNA prepared from normal tissues (Fig.5 A and B). Prostatic cancer tissue contains an admixture ofnormal stromal cellular elements and carcinoma cells.Whether differences in the abundance of hypermethylatedGSTPI regulatory sequences among the different cancersreflect different abundances of normal and neoplastic cells inthe cancer specimens, or whether some of the carcinomacells contain both hypermethylated and hypomethylatedGSTPI promoter sequences, has not been determined. Nev-ertheless, each of the prostate cancers contained GSTPIalleles with hypermethylated regulatory sequences. Further-more, for prostatic epithelial cells, hypermethylation ofGSTP1 promoter sequences appeared specific to the processofmalignant transformation. For five cases ofprostate cancerremoved by radical prostatectomy, DNA was selectivelyprepared from different areas of the same resection speci-

1_ 3 45 B.a

FIG. 5. GSTPI promoter hypermethylation in prostatic carcino-mas. DNA was isolated from prostatic carcinomas (CA) recoveredby surgical resection, as well as from grossly normal prostatic tissue(PROS) adjacent to cancer tissue, from benign prostatic hyperplasiatissue (BPH), and from normal seminal vesicles (SV) which accom-panied the resection specimen. The DNA was then digested withEcoRI, HindIII, and BssHII and subjected to Southern blot analysisusing a GSTPI probe. (A) Matched seminal vesicle and prostaticcarcinoma DNA. (B) Matched grossly normal prostate and prostaticcarcinoma DNA. (C) Grossly normal prostate, BPH tissue, andprostatic carcinoma from the same surgical resection specimens.Arrow denotes migration position of GSTPI restriction fiagmentuncut by BssHII.

mens containing normal prostate, BPH tissue, or prostaticcancer (Fig. SC). When Southern blots containing BssHII-digested DNA samples were hybridized with a GSTPI probe,apparent hypermethylation of GSTPI promoter sequenceswas detected in each ofthe cancer specimens, but none oftheBPH specimens (Fig. SC). In one ofthe cases, DNA preparedfrom grossly normal prostatic tissue also appeared to containGSTPI alleles with some methylated promoter sequences(Fig. SC, lane 10). Whether this resulted from infiltration ofgrossly normal prostatic tissue by microscopic carcinomacontaining methylated GSTPJ promoters, from GSTP1 pro-moter hypermethylation in prostatic intraepithelial neoplasia(PIN) lesions in the grossly normal prostatic tissue, or fromGSTPI promoter hypermethylation in histologically normalprostatic epithelial cells has not been established. However,hypermethylation of GSTPI promoter DNA nonethelessappears specific to the process of neoplastic transformationin the prostate: normal prostatic tissues from men withoutprostate cancer did not contain detectable methylation ofGSTPI regulatory sequences (Fig. 4B).

All ofthedata collected in this study suggest that decreasedGSTPI expression and GSTPI promoter hypermethylationoccur commonly during the malignant transformation of

EcoRI + HindIIIBssHII

ANML-

.1.1.1w I ..

iduk-dolkw

lbodbOWA90ho-op

-- -..

qwp.-=,r

11736 Medical Sciences: Lee et A

-A

Dow

nloa

ded

by g

uest

on

Oct

ober

27,

202

0

Proc. Natl. Acad. Sci. USA 91 (1994) 11737

prostatic epithelial cells. Regional methylation changes, of-ten manifest as hypermethylation of CpG island sequences,have been proposed to play an important role in carcinogen-esis (2-8). The prevalent finding of hypermethylated 5'GSTPI regulatory sequences in human prostatic cancers, butnot normal tissues, supports this general proposal. Unfortu-nately, the mechanism by which de novo regional hyper-methylation of specific autosomal alleles occurs in cancercells has not been well established. Abnormal DNA methyl-ation accompanying neoplastic transformation has been sug-gested to result from an enhanced DNA methylation capacityin the neoplastic cells. Both increased DNA methyltransfer-ase activity (44) and elevated DNA-MT mRNA levels (45)have been detected in neoplastic cells. Of interest, DNA-MTmRNA increases were observed early in the process ofcolonic carcinogenesis, even in some histologically normalcolonic mucosa specimens (45). Whether an increased DNAmethylation capacity in human prostatic cells results inGSTP1 promoter hypermethylation remains to be estab-lished.GSTs have been proposed to play a critical role in defend-

ing normal cells against electrophilic carcinogens (10-12).Consistent with this role, inherited homozygosity for nullGSTMJ alleles may confer an increased risk of cigarettesmoke-related lung cancer (46-48). In addition, overexpres-sion of a ii-class GST in cultured cells has been shown toreduce toxicity after benzo[a]pyrene exposure (49). An in-triguing implication of the apparent methylation-inactivationof GSTPI expression associated with prostatic carcinogen-esis is the hypothesis that some normal prostatic epithelialcells ight acquire an increased vulnerability to electrophiliccarcinogens by virtue of containing abnormally methylatedGSTP1 promoters. Abundant epidemiologic evidence sup-ports a significant role for environmental factors both inprostatic carcinogenesis and in prostate cancer progressionto a life-threatening disease (50). Normal prostatic tissueappears to contain epithelial cells with diminished GSTP1expression but hypomethylated GSTPI regulatory se-quences. Perhaps, rare cells with hypermethylated GSTPIpromoter sequences arise among these GSTP1-negative epi-thelial cells and undergo clonal expansion as a consequenceofcarcinogen exposure. Were this hypothesis true, a possibleprostate cancer prevention strategy might be the therapeuticaugmentation ofGST activity by using GST inducers. Recentstudies have identified several GST inducers which may beuseful for such an approach (51-53). We hope that furtherstudies of GSTP1 regulation in normal and neoplastic pros-tatic epithelial cells will help engender several new diagnosticand treatment strategies for prostate cancer.

The authors thank Drs. K. R. Cho, D. S. Coffey, and S. B. Baylinfor helpful discussions in the preparation of this manuscript. Thisresearch was supported by the Mary L. Smith Trust and a NationalInstitutes of Health Specialized Programs of Research ExcellenceGrant for prostate cancer.

1. Fearon, E. R. & Vogelstein, B. (1990) Cell 61, 759-767.2. Jones, P. A. (1986) Cancer Res. 46, 461-466.3. Holliday, R. (1987) Science 238, 163-170.4. De Bustros, A., Nelkin, B. D., Silverman, A., Ehrlich, G., Poiesz,

B. & Baylin, S. B. (1988) Proc. Nat!. Acad. Sci. USA 85,5693-5697.5. Jones, P. A. & Buckley, J. D. (1990) Adv. Cancer Res. 54, 1-23.6. Baylin, S. B., Makos, M., Wu, J. J., Yen, R. W., De Bustros, A.,

Vertino, P. & Nelkin, B. D. (1991) Cancer Cells 3, 383-390.7. Makos, M., Nelkin, B. D., Lerman, M. I., Latif, F., Zbar, B. &

Baylin, S. B. (1992) Proc. Nat!. Acad. Sci. USA 89, 1929-1933.8. Ohtani-Fujita, N., Fujita, T., Aoike, A., Osifchin, N. E., Robbins,

P. D. & Sakai, T. (1993) Oncogene 8, 1063-1067.9. Boring, C. C., Squires, T. S., Tong, T. & Montgomery, S. (1994) Ca

Cancer J. Clin. 44, 7-26.10. Pickett, C. B. & Lu, A. Y. (1989) Annu. Rev. Biochem. 58,743-764.11. Coles, B. & Ketterer, B. (1990) Crit. Rev. Biochem. Mol. Biol. 25,

47-70.

12. Rushmore, T. H. & Pickett, C. B. (1993) J. Biol. Chem. 268,11475-11478.

13. Mannervik, B., Awasthi, Y. C., Board, P. G., Hayes, J. D., Di Ilio,C., Ketterer, B., Listowsky, I., Morgenstern, R., Muramatsu, M.,Pearson, W. R., Pickett, C. B., Sato, K., Widersten, M. & Wolf,C. R. (1992) Biochem. J. 282, 305-306.

14. Terrier, P., Townsend, A. J., Coindre, J. M., Triche, T. J. &Cowan, K. H. (1990) Am. J. Pathol. 137, 845-853.

15. Stone, K. R., Mickey, D. D., Wunderli, H., Mickey, G. H. &Paulson, D. F. (1978) Int. J. Cancer 21, 274-281.

16. Kaighn, M. E., Narayan, K. S., Ohnuki, Y., Lechner, J. F. &Jones, L. W. (1979) Invest. Urol. 17, 16-23.

17. Horoszewicz, J. S., Leong, S. S., Kawinski, E., Karr, J. P.,Rosenthal, H., Chu, T. M., Mirand, E. A. & Murphy, G. P. (1983)Cancer Res. 43, 1809-1818.

18. lizumi, T., Yazaki, T., Kanoh, S., Kondo, I. & Koiso, K. (1987) J.Urol. 137, 1304-1306.

19. Brothman, A. R., Lesho, L. J., Somers, K. D., Wright, G. L., Jr.,& Merchant, D. J. (1989) let. J. Cancer 44, 898-903.

20. Burton, K. (1968) Methods Enzymol. 12, 163-166.21. Nelson, W. G. & Kastan, M. B. (1994) Mol. Cell. Biol. 14, 1815-

1823.22. Chomczynski, P. & Sacchi, N. (1987) Anal. Biochem. 162,156-159.23. Schneider, W. C. (1957) Methods Enzymol. 3, 680-684.24. Moscow, J. A., Townsend, A. J., Goldsmith, M. E., Whang-Peng,

J., Vickers, P. J., Poisson, R., Legault-Poisson, S., Myers, C. E. &Cowan, K. H. (1988) Proc. Nat!. Acad. Sci. USA 85, 6518-6522.

25. Juan, C.-C., Hwang, J., Liu, A. A., Whang-Peng, J., Knutsen, T.,Huebner, K., Croce, C. M., Zhang, H., Wang, J. C. & Liu, L. F.(1988) Proc. Natl. Acad. Sci. USA 85, 8910-8913.

26. Carter, B. S., Ewing, C. M., Ward, W. S., Treiger, B. F., Aalders,T. W., Shalken, J. A., Epstein, J. I. & Isaacs, W. B. (1990) Proc.Nat!. Acad. Sci. USA 87, 8751-8755.

27. Sambrook, J., Fritsch, E. F. & Maniatis, T. (1989) MolecularCloning: A Laboratory Manual (Cold Spring Harbor Lab. Press,Plainview, NY), 2nd Ed., Vol. 1, pp. 9.16-9.19.

28. Tsuchida, S. & Sato, K. (1992) Crit. Rev. Biochem. Mol. Biol. 27,337-384.

29. McNeal, J. E. & Bostwick, D. G. (1986) Hum. Pathol. 17, 64-71.30. Cowell, I. G., Dixon, K. H., Pemble, S. E., Ketterer, B. & Taylor,

J. B. (1988) Biochem. J. 255, 79-83.31. Morrow, C. S., Cowan, K. H. & Goldsmith, M. E. (1989) Gene 75,

3-11.32. Morrow, C. S., Goldsmith, M. E. & Cowan, K. H. (1990) Gene 88,

215-225.33. Xia, C. L., Cowell, I. G., Dixon, K. H., Pemble, S. E., Ketterer,

B. & Taylor, J. B. (1991) Biochem. Biophys. Res. Commun. 176,233-240.

34. Keshet, I., Yisraeli, J. & Cedar, H. (1985) Proc. Nat!. Acad. Sci.USA 82, 2560-2564.

35. Keshet, I., Lieman-Hurwitz, J. & Cedar, H. (1986) Cell 44, 535-543.36. Cedar, H. (1988) Cell 53, 3-4.37. Boyes, J. & Bird, A. (1991) Cell 64, 1123-1134.38. Lewis, J. D., Meehan, R. R., Henzel, W. J., Maurer-Fogy, I.,

Jeppesen, P., Klein, F. & Bird, A. (1992) Cell 69,,905-914.39. Maquat, L. E., Chilcote, R. & Ryan, P. M. (1985) J. Biol. Chem.

260, 3748-3753.40. Brown, J. R., Daar, I. O., Krug, J. R. & Maquat, L. E. (1985) Mol.

Cell. Biol. 5, 1694-1706.41. Bird, A. P. (1986) Nature (London) 321, 209-213.42. Antequera, F., Boyes, J. & Bird, A. (1990) Cell 62, 503-514.43. Daniel, V. (1993) Crit. Rev. Biochem. Mol. Biol. 28, 173-207.44. Kautiainen, T. L. & Jones, P. A. (1986)J. Biol. Chem. 261,1594-1598.45. El-Deiry, W. S., Nelkin, B. D., Celano, P., Yen, R. W., Falco,

J. P., Hamilton, S. R. & Baylin, S. B. (1991) Proc. Natl. Acad. Sci.USA 88, 3470-3474.

46. Seidegard, J., Pero, R. W., Miller, D. G. & Beattie, E. J. (1986)Carcinogenesis 7, 751-753.

47. Seidegard, J., Vorachek, W. R., Pero, R. W. & Pearson, W. R.(1988) Proc. Natl. Acad. Sci. USA 85, 7293-7297.

48. Seidegard, J., Pero, R. W., Markowitz, M. M., Roush, G., Miller,D. G. & Beattie, E. J. (1990) Carcinogenesis 11, 33-36.

49. Moscow, J. A., Townsend, A. J. & Cowan, K. H. (1989) Mol.Pharmacol. 36, 22-28.

50. Carter, B. S., Carter, H. B. & Isaacs, J. T. (1990) Prostate 16,187-197.

51. Prochaska, H. J., Santamaria, A. B. & Talalay, P. (1992) Proc.Natl. Acad. Sci. USA 89, 2394-2398.

52. Zhang, Y., Talalay, P., Cho, C. G. & Posner, G. H. (1992) Proc.Nat!. Acad. Sci. USA 89, 2399-2403.

53. Prestera, T., Holtzclaw, W. D., Zhang, Y. & Talalay, P. (1993)Proc. Nat!. Acad. Sci. USA 90, 2965-2969.

Medical Sciences: Lee et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

27,

202

0