Proc. Nati. Acad. Sci. USAVol. 89, pp. 7541-7545, August 1992Biochemistry

Identification of a peroxisome proliferator-responsive elementupstream of the gene encoding rat peroxisomal enoyl-CoAhydratase/3-hydroxyacyl-CoA dehydrogenase

(fatty acid -oxidation/hypolipidemic drugs/liver cels/transcriptional induction/protein-DNA interactions)

BAOWEI ZHANG*, SANDRA L. MARCUS*, FEREYDOUN G. SAJJADIt, KEITH ALVARESt, JANARDAN K. REDDYt,SURESH SUBRAMANIt, RICHARD A. RACHUBINSKI*§, AND JOHN P. CAPONE*§*Department of Biochemistry, McMaster University, Hamilton, Ontario, Canada; tDepartment of Biology, University of California, San Diego, La Jolla, CA92093; and tDepartment of Pathology, Northwestern University Medical School, Chicago, IL 60611

Communicated by Morris E. Friedkin, May 7, 1992

ABSTRACT Ciprofibrate, a hypolipidemic drug that actsas a peroxisome proliferator, induces the transcription of genesencoding peroxisomal (oxidation enzymes. To identify cis-acting promoter elements involved in this induction, 5.8 kilo-base pairs of promoter sequence from the gene encoding ratperoxisomal enoyl-CoA hydratase/3-hydroxyacyl-CoA dehy-drogenase (EC 4.2.1.17/EC 1.1.1.35) was inserted upstream ofa luciferase reporter gene. Transfection of this expressionvector into rat hepatoma H4IIEC3 cells in the presence ofciprofibrate resulted in a 5- to 10-fold, cell type-specificincrease in luciferase activity as compared to cells transfectedin the absence of drug. A peroxisome proliferator-responsiveelement (PPRE) was localized to a 196-nucleotide region cen-tered at position -2943 from the transcription start site. ThisPPRE conferred ciprofibrate responsiveness on a heterologouspromoter and functioned independently of orientation or po-sition. Gel retardation analysis with nuclear extracts demon-strated that ciprofibrate-treated or untreated H4IIEC3 cells,but not HeLa cells or monkey kidney cells, contained sequence-specific DNA binding factors that interact with the PPRE.These results have implications for understanding the mecha-nisms of coordinated transcriptional induction of genes encod-ing peroxisomal proteins by hypolipidemic agents and otherperoxisome proliferators.

Peroxisomes are organelles that perform biological oxidativefunctions, notably the H202-generating 13-oxidation of fattyacids (1). Administration of a diverse group of xenobioticchemical compounds referred to as peroxisome proliferatorsresults in a dramatic increase in both the number and themetabolic capacity of hepatic peroxisomes (2, 3). Within thisgroup are various hypolipidemic drugs used in the treatmentof ischemic heart disease and in controlling elevated levels ofcirculating triacylglycerols and cholesterol (4-6). Drugs likeclofibrate and ciprofibrate are peroxisome proliferators thatelicit a number of related pleiotropic effects, including he-patomegaly and hepatocarcinogenesis in rodents (2-4, 7).These agents are classified as non-genotoxic carcinogens, asthey are not directly mutagenic and do not damage DNA (8,9).Concomitant with drug-induced peroxisome proliferation

is the rapid and coordinated induction of the enzymes of theperoxisomal fatty acid (-oxidation system: H202-generatingfatty acyl-CoA oxidase (AOX), enoyl-CoA hydratase/3-hydroxyacyl-CoA dehydrogenase (HD) bifunctional enzyme,and 3-ketoacyl-CoA thiolase (10-12). The increased activitiesof these enzymes are related to the rapid transcriptionalactivation of the nuclear genes encoding these enzymes

(13-15). Transcriptional activation is apparently mediatedthrough members of the nuclear hormone receptor super-family that can be activated by peroxisome proliferators(16-18) and have been designated peroxisome proliferator-activated receptors (PPARs).We report the presence of an enhancer-like peroxisome

proliferator-responsive element (PPRE) upstream of the HDgene and demonstrate that this PPRE interacts with liver-specific factors in a sequence-dependent mannery

MATERIALS AND METHODSCells. Rat hepatoma H4IIEC3 cells were cultured as mono-

layers in Dulbecco's modified minimal essential medium plus10% horse serum and 5% fetal bovine serum. BSC40 cellswere maintained in Dulbecco's modified minimal essentialmedium plus 10% calf serum. HeLa cells were maintained insuspension in Joklik's modified medium plus 5% fetal bovineserum.

Plasmids and Construction of Promoter Deletion Mutants.pHDLuc was constructed by inserting a 5.8-kilobase-pair(kbp) promoter fragment from the rat HD gene as a KpnI-Apa I fragment into the luciferase expression vectorpSVOALA5' (19). The HD promoter fragment extends to +22from the transcription start site (20). For the construction ofdeletion mutants, the Kpn I-Apa I fragment was inserted intopGEM-7Zf(+) (Promega). Deletion mutants were con-structed using selected restriction sites and reinserted intopSVOALA5' as Kpn I-Apa I fragments. pCPSLuc is a lu-ciferase expression vector containing the minimal promoterfor the gene encoding rat liver carbamoyl-phosphate synthe-tase (CPS; ref. 21). Subfragments of the HD promoter wereinserted into the BamHI site of pCPSLuc. DNA fragmentswere amplified by the polymerase chain reaction (PCR) usingcombinations of oligonucleotide primers 1-6 (shown in the 5'-- 3' direction).Forward primers

1. ATTAGATCTGTAATGGAATCTGATG (-3159)2. .ATTAGATCTGTGACCACTAG (-3040)3. ATTAGATCTACATTTGAGTGCCCCG (-2928)

Reverse primers4. ATTGGATCCATTCTAAACTCACCACTGTA (-2751)5. ATTGGATCCTGAAATTTATTTACTT (-2845)6. ATTJGATCCCAAAGTCTCGGGGCA (-2900)

Abbreviations: AOX, fatty acyl-CoA oxidase; CPS, carbamoyl-phosphate synthetase; HD, enoyl-CoA hydratase/3-hydroxyacyl-CoA dehydrogenase; PPAR, peroxisome proliferator-activated re-ceptor; PPRE, peroxisome proliferator-responsive element.§To whom reprint requests should be addressed.IThe sequence presented in this paper has been deposited in theGenBank data base (accession no. M97197).

7541

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.

Proc. Natl. Acad. Sci. USA 89 (1992)

The numbers in parentheses refer to the positions of the 5'nucleotides in the amplification target (see Fig. 3). Theunderlined sequences were added to incorporate either a BglII site (forward primers) or a BamHI site (reverse primers).pRSVcat and pCH11O (Pharmacia) are reporter plasmidsexpressing the genes encoding Escherichia coli chloramphen-icol acetyltransferase and f-galactosidase, respectively (22,23).

Transfections. Transfections were done by a modificationof the calcium phosphate procedure (24) with H4IIEC3 cellsat 60-70%o confluence in 10-cm dishes. A calcium phosphate/DNA coprecipitate suspension (2 ml) was made containingplasmid at 30 pg/ml and sonicated salmon sperm DNA at 10pg/ml. The suspension was divided into 0.5-ml aliquots andadded dropwise to cells incubated in the presence of 10 ml offreshly added medium containing either 0.5 mM ciprofibrate(100 mM stock solution in dimethyl sulfoxide) or 0.5%dimethyl sulfoxide alone. After incubation at 370C for 16 hr,medium was replaced with appropriately supplemented freshmedium, and incubation was continued for an additional 24hr. Cells were washed with phosphate-buffered saline, andextracts were prepared by suspension ofthe cell pellets in 100,.l of lysis solution [25 mM Tris phosphate, pH 7.8/2 mMdithiothreitol/2 mM 1,2-diaminocyclohexane-NN,N',N'-tetraacetic acid, 10%o (wt/vol) glycerol, 1% (wt/vol) TritonX-100], followed by vortex mixing. Cell debris was removedby centrifugation, and the supernatant was assayed for lu-ciferase activity with a luminometer (model 1253, Bio-Orbit

Oy, Turku, Finland) and a luciferase activity kit (Promega)(19). Chloramphenicol acetyltransferase and ,3-galactosidasewere assayed as described (25, 26). Protein was determinedby the method of Bradford (27).

Gel Retardation Analysis. Nuclear extracts from HeLa cellswere prepared from cell suspensions (28). Nuclear extractsfrom BSC40 and H4IIEC3 cells were prepared from mono-layer cultures (29, 30). Nuclear extracts were also preparedfrom H4IIEC3 cells cultured for 48 hr in the presence ofeither0.5 mM ciprofibrate or 0.5% dimethyl sulfoxide.Gel retardation assays were performed as described (31,

32). Probe DNA [promoter coordinates -3040 to -2845 (seeFig. 3)] was purified from agarose gel, end-labeled with[a-32P]dATP by the Klenow fragment of DNA polymerase,and extracted with buffer-saturated phenol. Binding reactionmixtures contained 1 ng of labeled probe DNA, 10 mMHEPES (pH 7.9), 120 mM NaCl, 1 mM EDTA, 7% (vol/vol)glycerol, 5 jug of bovine serum albumin, 4 jtg of nonspecificcompetitor DNA [a 1:1 mixture of poly(dI-dC)poly(dI-dC)and sonicated salmon sperm DNA], 150 juM phenylmethyl-sulfonyl fluoride, 1 mM dithiothreitol, and 10 pg of nuclearextract in a final volume of 20 p1. Reaction mixtures wereincubated at 37°C for 20 min. Electrophoresis was performedat 4°C on a pre-run 3.5% polyacrylamide gel (30:1 acrylam-ide/N,N'-methylenebisacrylamide weight ratio) with 22 mMTris base/22 mM boric acid/1 mM EDTA as running buffer.

Miscellaneous. DNA sequencing was performed using Se-quenase (United States Biochemical; ref. 33). Oligonucleo-

AXb H Xh Rog Sm r

* 10

8

oC-I

4

002

0

Plasmid:

Fold Induction:

- ciprofibrate

+ ciprotibrate

WT a b c d * f 9 h I j k

6.22 6.31 3.06 4.06 0.68 7.67 6.62 3.39 1.34 1.62 6.46 1.29

FIG. 1. Ciprofibrate-mediated induction of the HD promoter requires an upstream cis-acting site. (A) Diagram of promoter deletions. Arestriction map of the 5.8-kbp promoter region of the HD gene is shown at top. Mutants with sequential 5' and internal deletions (designatedby A) are shown below the wild-type (WT) promoter and designated according to the restriction enzymes used to generate them: Bg, Bgl II;H, Hpa I; N, Nde I; S, Sal I; Sm, Sma I; Xb, Xba I; Xh, Xho I. (B) The HD promoter requires sequences between the Nde I and Bgl II sitesfor responsiveness to ciprofibrate. Constructs were transfected into H4IIEC3 cells in the absence or presence of ciprofibrate. Results (± SEM)were normalized to activity obtained from transfection oftheWT promoter in the absence ofciprofibrate, which was taken as 1. Absolute activityof the WT promoter ranged from 0.5 to 1.5 integrated light units. Fold induction is the ratio of the value obtained from ciprofibrate-treated cellsto the value obtained from untreated cells. Background luciferase (LUC) activity with the promoterless plasmid pSVOLA5' was negligible.

WTa) AXb/Hb) AXb/Xhc) AXb/Nd) AXb/Bge) AH/Xhf) AXh/Sg) A Xh/Nh) AXh/Bgi) AS/Bgi) AS/Nk) AN/Bg

B 12

F - -I I I I I

7542 Biochemistry: Zhang et al.

I

Proc. Natl. Acad. Sci. USA 89 (1992) 7543

tides were from the Central Facility, Institute of MolecularBiology and Biotechnology, McMaster University.

Ndel Hindill BamHi-3178 -2676 -2281

1 1

RESULTS

A Liver-Specific Element Responsive to Ciprofibrate is Pres-ent Upstream of the Rat HD Gene. Treatment of H4IIEC3cells with ciprofibrate results in marked peroxisome prolif-eration and the concomitant transcriptional induction of thegenes encoding the P-oxidation enzymes (34). Transfection ofH4IIEC3 cells with pHDLuc, which contains 5.8 kbp ofupstream sequence of the HD promoter, resulted in a lowlevel of luciferase activity. This activity was induced 5- to10-fold by ciprofibrate, indicating that sequences mediatingthe response to this peroxisome proliferator were containedwithin the 5.8-kbp fragment (Fig. 1B). Ciprofibrate treatmenthad no effect on expression of pHDLuc in BSC40 cells, a

monkey kidney cell line, indicating that the ciprofibrateresponsiveness of the HD gene is specific to liver cells (datanot shown). Expression from pRSVcat or pCH110 was notincreased by ciprofibrate in either H4IIEC3 or BSC40 cells(data not shown).A restriction map of the 5.8-kbp upstream region of the HD

gene was used to create deletion mutants that permitted apreliminary localization of the PPRE (Fig. 1A). Because ofthe relatively low transfection efficiency ofH4IIEC3 cells, alltransfections were carried out in duplicate and repeated atleast four times, with wild-type pHDLuc as a positive controlin each case and with data normalized for protein concen-tration.As seen in Fig. 1, sequential deletion of the HD promoter

up to the Nde I site at position -3178 had little effect onciprofibrate inducibility, which ranged from 3- to 6-fold forthe deletion constructs versus U6-fold for the full-lengthwild-type HD promoter (plasmids Xb/H, Xb/Xh, Xb/N,WT). The absolute level of ciprofibrate induction varied as afunction of the basal activity of each construct; basal activityof the constructs varied -2-fold. However, the relativeinduction remained fairly constant. Further deletion to theBgl II site at -1623 completely abolished inducibility (plas-mid Xb/Bg), indicating that some critical element lies be-tween the Nde I and Bgl II sites.The necessity of this region for induction was further

demonstrated by a series of internal deletions. The regionbetween the Nde I and Bgl II sites was essential for ciprofi-brate induction, since every deletion spanning this region wasunresponsive to ciprofibrate, whereas deletions retaining thisregion were inducible (compare plasmids Xh/Bg, S/Bg, andN/Bg with plasmids Xh/S, Xh/N, and S/N). Plasmid N/Bg,in which the region spanning the Nde I and Bgl II sites wasdeleted, was unresponsive to ciprofibrate. Therefore, if ad-ditional PPREs exist upstream of the Nde I site, they cannotfunction in the absence of sequences between the Nde I andBgl II sites.The PPRE of the HD Gene Confers Ciprofibrate Respon-

siveness on a Heterologous Promoter. To determine whetherthe region between the Nde I and Bgl II sites was bothnecessary and sufficient for ciprofibrate-mediated induction,and to delimit further the PPRE, this region was sequenced,and subregions were inserted into the BamHI site at the 5'end of the minimal promoter of the gene encoding CPS, apromoter that is unresponsive to peroxisome proliferators.As shown in Fig. 2, the region between -3178 and -2281

conferred inducibility on the minimal CPS promoter. There-fore, this region contains a positively acting PPRE thatfunctions independently of other regions ofthe HD promoter.The PPRE displayed enhancer-like properties in that it func-tioned in a position- and orientation-independent manner(plasmid bin Fig. 2). Two copies ofthis region led to a greaterratio of induction relative to a single copy (compare plasmids

a)b) ,C)

d)e) 00

f)9) (-3169/-2761)

h) (-3169/-2845)i) -- (-3169/-2900)j) (-2928/-2761)

k)l)

m)

-> (-3040/-2761)

- (-3040/-2846)

3.914.255.98

*. 0.273.880.394.703.131.290.427.593.12

C-P-S 0.54

FIG. 2. An enhancer-like PPRE is located between -3040 and-2845 upstream of the HD transcription start site. Subfragmentsspanning the Nde I-Bgl II region of the HD promoter were insertedinto the BamHI site upstream of the minimal promoter for the CPSgene, linked to the luciferase gene. Constructs were transfected asdescribed in Fig. 1. Constructs a-f were made by inserting appro-priate restriction fragments into pCPSLuc. Constructs g-l weregenerated by PCR amplification of appropriate subregions. Numbersin parentheses correspond to the endpoints of the regions amplified.Wild-type pHDLuc was used as a positive control and pCPSLuc(construct m) as a negative control in each case. Values represent theaverage induction ratio of transfections of ciprofibrate-treated cellscompared with untreated cells. Absolute basal activity of pCPSLucranged from 0.2 to 1 integrated light unit.

c and a). Basal expression of the CPS promoter was notsignificantly affected by any subregion of the HD promoter inthe absence of ciprofibrate.

Further subcloning localized the PPRE to between -3178and -2676. The nucleotide sequence (Fig. 3) was used todesign PCR primers to amplify and clone subregions of thissegment. A minimal PPRE was localized to a 196-bp regionbetween -3040 and -2845 (Fig. 2, plasmid 1). A fragmentcontaining the 5'-most 141 bp (Fig. 2, plasmid i) of thisminimal PPRE was induced only 1.3-fold by ciprofibrate,whereas the 196-bp minimal PPRE was induced 3.1-fold(compare plasmids i and 1). A construct containing the3'-most 84 bp of this minimal PPRE (plasmid j) was unre-sponsive to ciprofibrate, and like other unresponsive plas-mids (plasmids d, f, and m), showed reduced activity in thepresence of ciprofibrate. Therefore the 5'-most 141 bp of the196-bp PPRE are essential for induction; however, the entire196-bp fragment is required for maximal response. These datasuggest that either a site within the PPRE spanning the regionoverlapped by plasmids i and j or two distinct sites in the196-bp PPRE are required for maximal induction.

Liver-Specific Cellular Factors Interact with the PPRE.Mobility-shift experiments were performed to determinewhether cellular factors interact with the PPRE. Incubationof a probe (-3040 to -2845, Fig. 2) with nuclear extract fromH4IIEC3 cells resulted in the formation oftwo major protein-DNA complexes, designated C1 and C2 (Fig. 4A, lane d).Two other, minor complexes were occasionally observed.No complexes were observed with nuclear extract fromHeLa or BSC40 cells (lanes b and c, respectively), suggestingthat the DNA-binding factors are liver cell-specific. The C1and C2 complexes were also observed with nuclear extractfrom ciprofibrate-treated H4IIEC3 cells, but interestingly,the C2 complex was more abundant in comparison to extractsfrom untreated cells (compare lanes d and e). The integrity ofeach extract was assessed with a DNA probe containing thebinding site for the ubiquitous transcription factor Oct-i (32).The pattern of protein-DNA complexes formed with theOct-1 probe was the same for ciprofibrate-treated and un-

-gi1I-1623

INDUCTIONRATIO

Biochemistry: Zhang et aL

Proc. Natl. Acad. Sci. USA 89 (1992)

Kdol-3178MATWGTOG CTCACAACCA TCTGTAATCG AATCTGATGC TCTCTTCTGGTGTGTCTGAAGTATACCCC GAGTGTTOGT AGACARTACC TTAGACTACG GGAGAAGACC AC&C&GACTT-3118GACAGCTACA GTGTACTCAT ATACATAA&T AAATCTTAAA AAaAAAMP GGAOGAAGAACTGTCGATGT CACATGAGTA TATGTATTTA TTTAAATTT Tc ccTCTTcT

A PPRE Probe

a b c d ei_ _ f

B )ctaMrnr Pr obe

ai e t c,

AmAU.A.( .I.P.LO .OI.iX2X.2 XX.892XiXf X.(L.1.X.L_ Or i

~~~~~- ----

-2878CCTCACaAC AIAOM A&Am=ATTCr.ACTTZ TC AmTTTTTTT TTATTfTAAMC

qM9AATA&TT CTAGGCTTTT CTTTCCKCATIR ATTAA GATCCGAAA GAAAGGTGTA

-2818CTAGTMCCCT G---CTTAC CTATTAAAAG A&CTTCTTTT TAAAGTACRG TOGTGAaTTTCATCACGGGA COGAAAATG GATAATTTTC TTGAAGAAA& ATTTC&TGTC ACC&CTCAA&

-2758AGATGGAA GTAA=C TAGAOGA& TTT&ACAA AAA AGTCTTACA

TCTTACCTTT CTAC TTTT aTCTCCT AAATGTCTTT TsumT----- TChaTAATOT

-2698 NindIIITC~CX:TCTC TTG&ATGTTT TGAA TCTTATAGC CCA&aGTGGT CTCAA&CTCAGGGAAAC AACGTACAA ACTTOGAAQT AaAC&TATCG QGTCC&CCA CAGTTTGACT

-2638GAGTC&TCCT cCTCTCT CCTCCTG&T ATTGAAATTG CAGCCTTGA& CCCAT

GTCAGTAaGA OGACOGAGAC GGagcTc TAACTTTA&C GCGGRACTT GTGcTGTch

-2578ACC&TAACC CAGTTCAGC AG TG CCAAAMCACC GTOOTTTTGA GAGTGGACTTTGGTATTGTG TACAATCG TCTCNITSACUGTTTTTCCACT CTCACCTGAA

-2518ACTGTAAAAC ACAAGRTGAA AAChaTGATC AAAACGGTc& chTATTA&TA TTQTTTTAGT

TGACATTTTG TGTTCTACTT TTGTC TGCCAGT GTATAATTAT AACAAAATCR

-2458ATATTCTTAC TTTTGTCAQ STA CTTAC&TAAT GCAAAGTTAA AATGAGA&CGTATAACAATG AAAACCGT ACGAACTTT? GAATGTATTA CGTTTCAAT TTACTCTC

-2398GTCATTAGTG ATCCATGAAT ATTTTOCTTC CTGTCGOTGA OACTGAATTT AAACTACOCCCAGTAATCAC TGAGTACTTA TAAAAG GACAGGCACT CTGACTTAA& TTTGATGcG

-2338 BPHIGAAOCCT TGGATTCTTT aOCTGACMGO ARRChOATTG AGCTTTTCAG TOGATAGM

TTCCSTOGA ACCTARGAAA COGACTOTOC TCGAAAAGTC ACCTATCCCT

-227819G="ATGTTICTTCTAC

FIG. 3. Nucleotide sequence of the PPRE. The PPRE is under-lined. The consensus half-site TGACCT of several steroid hormonereceptors is shaded and is present in direct repeats (filled arrows).Other sequences similar to this consensus half-site are indicated byopen arrows.

treated H4IIEC3 cells (Fig. 4B, lanes b and c), suggesting thatthe quantitative and qualitative differences in protein-DNAinteractions observed in these extracts with the PPRE probewere related to treatment with ciprofibrate and not to differ-ences in the integrity of the extracts. HeLa and BSC40extracts also were capable of forming Oct-l-containing com-plexes (Fig. 4B, lanes a and d), supporting the observationthat these cells do not contain specific factors capable ofinteracting with the PPRE.

Competition experiments were carried out to assess thespecificity of these protein-DNA interactions. A 50- or100-fold molar excess of unlabeled -3040 to -2845 probeDNA reduced the amounts of both the C1 and C2 complexes(Fig. 4C, lanes c and d); however, nonspecific yeast DNA(lanes i and j) was unable to inhibit complex formation. Thus,liver nuclear factors interact with the PPRE in a sequence-specific manner.Competitor DNA corresponding to -3159 to -2900 (Fig.

4C, lanes e and f), which overlaps the 5' end of the PPRE by141 nucleotides, blocked formation of both the C1 and C2complexes as effectively as did the homologous -3040 to-2845 DNA. In addition, competitor DNA -2928 to -2751,which overlaps the 3' end ofthe PPRE by 84 nucleotides, wasalso able to inhibit complex formation, although less effi-ciently (lanes g and h). The 3' end of the -3159 to -2900

IF.. C 1

a.,

y - C

:....

c

C c. m pet c.

a bi c d i e f Ig h :i j

FIG. 4. Sequence-specific interaction of liver cell nuclear factorswith the PPRE of the HD promoter. (A) Gel retardation analysis ofa labeled DNA fragment containing the HD PPRE (coordinates-3040 to -2845) incubated with nuclear extracts prepared fromHeLa (lane b), BSC40 (lane c), untreated H4IIEC3 (lane d), andciprofibrate-treated H4IIEC3 (lane e) cells. Lane a, probe incubatedin the absence ofextract. Major protein-DNA complexes are labeledC1 and C2. (B) Nuclear extracts from BSC40 (lane a), untreatedH4IIEC3 (lane b), ciprofibrate-treated H4IIEC3 (lane c), and HeLa(lane d) cells were incubated with a labeled synthetic duplex DNA,5'-GATCCCGTGCATGCTAlGATATTCT-3', containing a bind-ing site for the octamer-binding protein Oct-i (underlined; ref. 32).The panel is a composite autoradiogram of the same gel in which thelane containing the HeLa extract was underexposed relative to theother lanes due to the large amount of Oct-i present in HeLa cells.(C) Specificity of protein-DNA interactions as determined by com-petition analysis. Nuclear extract from ciprofibrate-treated H4IIEC3cells was incubated with labeled probe containing the PPRE as in A,in the absence (lane b) or presence of 50- and 100-fold molar excessof unlabeled probe (-3040 to -2845, lanes c and d) or competitorcorresponding to HD promoter coordinates -3159 to -2900 (lanes eand f) or coordinates -2928 to -2751 (anes g and h). For lanes i andj, reactions were carried out in the presence of 50- or 100-fold molarexcess of a nonspecific 210-bp yeast genomic DNA fragment. Lanea, probe DNA incubated in the absence of extract. For the abovereactions, competitor DNA was incubated with extract for 5 minprior to the addition of labeled probe.

-3058AATGAGTAAAWTACTCAMM

-4

-2998GP glasm 66FJAGM

GACGAGOWT QGCGRTGCM GTOGAGAaM

-2938Maui= nagma UUMN A=mm MaKwUlkakilsoCTTUATAATO GATOTAAACT TQAAACCCTC QACTRTVCMG AAT7MTARA

7544 Biochemistry: Zhang et al.

I -Z.-:)l --el

Q)

.1 .11 Z.COI cI --k

Proc. Natl. Acad. Sci. USA 89 (1992) 7545

fragment overlaps the 5' end of the -2928 to -2751 fragmentby 29 nucleotides (-2928 to -2900), suggesting that thisregion may be critical for protein-DNA interactions. How-ever, as shown in Fig. 2, the -3159 to -2900 fragment wasinsufficient to confer maximal ciprofibrate responsiveness onthe CPS promoter, whereas the -2928 to -2751 fragmentwas unresponsive to ciprofibrate. Thus, in addition to nucle-otides -2928 to -2900, it is likely that both the 5' and 3'sequences flanking this region play some important role in thefunction of the 196-bp PPRE.

DISCUSSIONThe elucidation of the mechanism by which ciprofibrate andother peroxisome proliferators modulate gene expression andthe relationship of this altered expression to hepatic tumor-igenesis requires the identification of promoter elements andtranscription factors responsible for mediating the biologicalresponse to these non-genotoxic agents. Toward this goal wehave identified a PPRE upstream of the rat HD gene. Thiselement functions independently of orientation and position,can confer responsiveness on a heterologous promoter, andinteracts with liver cell-specific factors in a sequence-specificmanner. Recently, the gene encoding rat liver peroxisomalAOX has been shown to contain a PPRE (18, 35) thatfunctions in a manner similar to that described here for theHD gene PPRE.

Recently, several members of the steroid hormone recep-tor superfamily that are activated by peroxisome prolifera-tors have been cloned from mouse (16) and Xenopus (17).These PPARs were shown to transactivate expression ofgenes containing the rat AOX PPRE in a peroxisome-proliferator dependent manner (17, 18) and to recognize thesequence TGACCT, the consensus half-site for several nu-clear hormone receptors (36,37). Interestingly, the HD PPREcontains several sequences similar to this consensus half-site,which are present in the direct repeat array TJGAC(J.A/TTGAaCTA/TtACCTA (nucleotides -2948 to -2926; Fig.3, stippling).While the TGACCT direct repeat array in the HD PPRE

may be required for induction, it is by itself insufficient formaximal induction (Fig. 2, construct i). Therefore, if thesePPAR(s), which have been demonstrated to bind to and toactivate the AOX PPRE, also bind to the HD PPRE, maximaltranscriptional induction for the HD PPRE may depend oncooperativity with other proteins bound to distal sites in thePPRE. The fact that both the -3159 to -2900 and the -2928to -2751 promoter region could compete for factor bindingbut were insufficient for maximal ciprofibrate-mediated in-duction suggests that the properjuxtaposition of several sitesmay be required for full response. In this context, it will beof interest to determine the nature of the protein-DNAinteractions on the HD PPRE and the possible interplay ofligand-activated receptors both in protein-DNA complexformation and in the coordinated transcriptional activation ofthe other members of the j-oxidation pathway.

We thank Dr. G. Shore for pCPSLuc and Sterling Drug forciprofibrate. The contributions of members of the Rachubinski andCapone laboratories are gratefully acknowledged. This work wassupported by the Heart and Stroke Foundation of Ontario (R.A.R.and J.P.C.). S.S. was supported by Grant DK41737 from the Na-tional Institutes of Health. S.L.M. held a Natural Sciences andEngineering Research Council summer studentship. J.P.C. is aResearch Scholar ofthe National Cancer Institute ofCanada. R.A.R.is a Medical Research Council of Canada Scientist.

1. Lazarow, P. B. & Fujiki, Y. (1985) Annu. Rev. Cell Biol. 1,489-530.

2. Reddy, J. K. (1990) Biochem. Soc. Trans. 18, 92-94.3. Rao, M. S. & Reddy, J. K. (1991) Environ. Health Perspect.

93, 205-209.4. Reddy, J. K., Azarnoff, D. L. & Hignite, C. E. (1980) Nature

(London) 283, 397-398.5. Havel, R. J. & Kane, J. P. (1973) Annu. Rev. Pharmacol. 13,

287-308.6. Sirtori, C. R., Catapano, A. & Paoletti, R. (1977) Atheroscle-

rosis Rev. 2, 113-153.7. Reddy, J. K. & Lalwani, N. D. (1983) Crit. Rev. Toxicol. 12,

1-58.8. Warren, J. R., Simmon, V. F. & Reddy, J. K. (1980) Cancer

Res. 40, 36-41.9. von Daniken, A., Lutz, W. K., Jackh, R. & Schlatter, C. (1984)

Toxicol. Appl. Pharmacol. 73, 373-387.10. Hess, R., Stdubli, W. & Riess, W. (1965) Nature (London) 208,

856-858.11. Lazarow, P. B. & de Duve, C. (1976) Proc. Natl. Acad. Sci.

USA 73, 2043-2046.12. Lazarow, P. B. (1977) Science 197, 580-581.13. Furuta, S., Miyazawa, S. & Hashimoto, T. (1982) J. Biochem.

(Tokyo) 92, 319-326.14. Reddy, J. K., Goel, S. K., Nemali, M. R., Carrino, J. J.,

Laffler, T. G., Reddy, M. K., Sperbeck, S. J., Osumi, T.,Hashimoto, T., Lalwani, N. D. & Rao, M. S. (1986) Proc. NatI.Acad. Sci. USA 83, 1747-1751.

15. Sharma, R. K., Lake, B. G., Makowski, R., Bradshaw, T.,Earnshaw, D., Dale, J. W. & Gibson, G. G. (1988) Eur. J.Biochem. 184, 69-78.

16. Issemann, I. & Green, S. (1990) Nature (London) 347, 645-650.17. Dreyer, C., Krey, G., Keller, H., Givel, F., Helftenbein, G. &

Wahli, W. (1992) Cell 68, 879-887.18. Tugwood, J. D., Issemann, I., Anderson, R. G., Bundell,

K. R., McPheat, W. L. & Green, S. (1992) EMBO J. 11,433-439.

19. De Wet, S. R., Wood, K. V., Deluca, M., Helsinki, D. R. &Subramani, S. (1987) Mol. Cell. Biol. 7, 725-737.

20. Ishii, N., Hijikata, M., Osumi, T. & Hashimoto, T. (1987) J.Biol. Chem. 262, 8144-8150.

21. Howell, B. W., Lagacd, M. & Shore, G. C. (1989) Mol. Cell.Biol. 9, 2928-2933.

22. Gorman, C., Merlino, G., Willingham, M., Pastan, I. & How-ard, B. (1982) Proc. Natl. Acad. Sci. USA 79, 6777-6781.

23. Hall, C. V., Jacob, P. E., Ringold, G. M. & Lee, F. (1983) J.Mol. Appl. Genet. 2, 101-109.

24. Graham, F. J. & Van Der Eb, A. J. (1973) J. Virol. 59, 456-467.25. Gorman, C. M., Moffat, L. F. & Howard, B. H. (1982) Mol.

Cell. Biol. 2, 1044-1051.26. Webster, N. J. G., Green, S., Jin, J. R. & Chambon, P. (1988)

Cell 54, 199-207.27. Bradford, M. M. (1976) Anal. Biochem. 72, 278-284.28. Dingam, J. D., Lebovitz, R. M. & Roeder, R. G. (1983) Nu-

cleic Acids Res. 11, 1475-1489.29. Andrews, N. C. & Faller, D. V. (1991) Nucleic Acids Res. 9,

2499.30. Lee, K. A., Bindereif, A. & Green, M. R. (1988) Gene Anal.

Tech. 5, 22-31.31. Singh, H., Sen, R., Baltimore, D. & Sharp, P. A. (1986) Nature

(London) 319, 154-158.32. Werstuck, G. & Capone, J. (1989) J. Virol. 63, 5509-5513.33. Zhang, H., Scholl, R., Browse, J. & Somerville, C. (1988)

Nucleic Acids Res. 16, 1220.34. Osumi, T., Yokota, S. & Hashimoto, T. (1990) J. Biochem.

(Tokyo) 108, 614-621.35. Osumi, T., Wen, J.-K. & Hashimoto, T. (1991) Biochem.

Biophys. Res. Commun. 175, 866-871.36. Evans, R. (1988) Science 240, 889-895.37. Umesono, K., Murakami, K. K., Thompson, C. C. & Evans,

R. M. (1991) Cell 65, 1255-1266.

Biochemistry: Zhang et al.