Cloning and Identification of Rat Deoxyuridine Triphosphatase as an Inhibitor of Peroxisome Proliferator-activated Receptor α

Peroxisome proliferator-activated receptors (PPARs) are members of the nuclear receptor superfamily that transcriptionally regulate responsive genes by binding to the peroxisome proliferator response elements. Protein(s) interacting with PPAR isoforms (α, δ, and γ) may modulate the PPAR-mediated transcriptional activation. Using a yeast two-hybrid system to screen a rat liver cDNA library, we have identified rat deoxyuridine-triphosphatase (dUTPase, EC 3.6.1.23) as a PPARα-interacting protein. This cDNA encodes a polypeptide of 203 amino acids; the C-terminal 141-amino acid segment of this protein corresponds to the full-length human enzyme, which exhibits 92% identity with human dUTPase; the N-terminal extra 62-amino acid residue region is arginine-rich. In vitro binding assays indicate that rat dUTPase interacts with all three isoforms of mouse PPAR, but not with retinoid X receptor and thyroid hormone receptor. Interaction of PPARα with dUTPase is with the N-terminal 62-amino acid segment of rat dUTPase. Full-length rat dUTPase prevents PPAR-retinoid X receptor heterodimerization resulting in an inhibition of PPAR activity in a ligand-independent manner. Immunostaining of human kidney tsA201 cells, transiently expressing dUTPase showed that this protein is present predominantly in the cytoplasm but translocates into the nucleus with PPARα when PPARα is coexpressed with dUTPase. Northern blot hybridization shows that rat dUTPase is encoded by an abundant 1kilobase mRNA species present in all rat tissues. The identification of dUTPase as a PPAR-interacting protein suggests a possible link between tumorigenic peroxisome proliferators and the enzyme system involved in the maintenance of DNA fidelity.

Peroxisomes are cytoplasmic organelles widely distributed in most animal and plant cells. Although their number and volume density remain fairly constant under various physiological and pathological conditions, they increase dramatically in the liver cells of rats and mice following exposure to certain nonmutagenic chemicals designated as peroxisome proliferators (1). These compounds consist of a broad spectrum of chemicals with little obvious structural similarity except for the presence of an aromatic ring, carboxylic acid, and an aliphatic chain.
Examples include certain widely used hypolipidemic drugs (such as clofibrate, ciprofibrate, and gemfibrozil), phthalate ester plasticizers, herbicides, and some chlorinated hydrocarbons (2). Increases in the peroxisome volume density in liver cells is associated with transcriptional activation of genes encoding for the peroxisomal ␤-oxidation enzymes (3), the cytochrome P-450 CYP4A isoforms (4), and the liver fatty acidbinding protein (5), among others (6). Chronic administration of peroxisome proliferators leads to the development of liver tumors in rats and mice (7). To date, all the compounds with peroxisome-proliferative effect that have been tested in long term studies have been found to be tumorigenic despite the fact that these chemicals neither bind covalently to DNA nor produce somatic mutations directly or after metabolic activation (7). It has been postulated that H 2 O 2 overproduced by the sustained increases in H 2 O 2 -generating peroxisomal fatty acid ␤-oxidation system, results in oxidative stress that contributes to hepatocarcinogenesis in rodents (8).
The induction of peroxisome proliferation is mediated by members of the nuclear receptor superfamily, termed peroxisome proliferator-activated receptors (PPARs) 1 that are closely related to the thyroid hormone receptors and retinoid receptors (9,10). To date, three isoforms of PPARs have been identified in amphibians, rodents, and humans: PPAR␣, PPAR␦ (also called ␤ or NUC-1), and PPAR␥ (11,12). PPAR isoforms display distinct patterns of tissue distribution and appear to have different functions (12,13). PPAR␣ is highly expressed in hepatocytes, cardiomyocytes, enterocytes, and the proximal tubular epithelium of kidney (14) and plays a crucial role in the peroxisome proliferator-induced pleiotropic responses (6,10). Corroborative evidence for the functional role of PPAR␣ in peroxisome proliferator-induced signal transduction comes from the observation that disruption of the mPPAR␣ gene results in the abolishment of the pleiotropic effects of peroxisome proliferators in mice (15). PPAR␦ is expressed ubiquitously and often at higher levels than PPAR␣ and PPAR␥ (12,14). Instead of activation, the human PPAR␦ has been found to repress the activation of PPAR␣, as well as thyroid hormone receptor (16). PPAR␥ is expressed predominantly in adipose tissue and the immune system (14,17) and is activated by prostaglandin 15d⌬ 12,14 PGJ2 (18,19) and thiazolidinediones (20). Expression and activation of PPAR␥ in firoblasts is sufficient to trigger the adipocyte differentiation cascade, implying that PPAR␥ plays a key role in adipogenesis (21,22).
Tissue and species responses to peroxisome proliferators may depend on pharmacokinetics, the relative abundance of the PPAR isoforms and their auxiliary proteins, the nature of peroxisome proliferator response element (PPRE), and to some extent, hormone levels (23). PPARs regulate target genes by forming a heterodimer with RXRs, another subfamily of nuclear receptors (24). RXRs also form heterodimers with other members of the nuclear receptor superfamily, and these interactions appear to influence the PPAR-regulated gene activation because of the competition among various RXR heterodimerization partners for RXR (23). Nuclear receptor actions are also modulated by factors referred to as co-activators or co-repressors (25). Co-activators for glucocorticoid receptors (26), progesterone receptor (27), and TR-RXR (28) identified so far show no common structural features, implying that different nuclear receptors contact with the transcription machinery through receptor specific co-activators. Recently, two structurally related proteins, designated as N-CoR (nuclear receptor co-repressor) and SMRT (silencing mediator for retinoid and thyroid hormone receptors), have been shown to interact with TR and RAR and repress basal transcription in the absence of ligand (29,30). Also, the finding that human RNA polymerase II complex contains transcriptional coactivators as well as DNA repair proteins suggests that the transcriptional activation and DNA repair functions are coordinated (31). In an attempt to identify co-activators or co-repressors for PPAR␣, we utilized a yeast two-hybrid system to screen rat liver cDNA library and identified deoxyuridine-triphosphatase (dUTPase) as a PPAR␣-interacting protein. We show that the rat dUTPase interacts with all three PPAR isoforms and blocks the formation of PPAR-RXR heterodimers, causing repression of PPARmediated transcriptional activation. The identification of rat dUTPase as a PPAR-interacting protein not only provides an additional factor that participates in the PPAR-mediated gene regulation but also suggests a novel link between the tumorigenic peroxisome proliferators and the DNA fidelity enzyme system.
Cloning of dUTPase-encoding cDNA into Prokaryotic and Eukaryotic Expression Vectors-After sequencing, one of the activation domain plasmids encoding a polypeptide showing homologies with the human dUTPase was digested with HindIII and blunt-ended by filling with Klenow and dNTPs. The plasmid was further cut with BamHI, and the insert was subcloned into the BamHI and SmaI sites of the prokaryotic expression vector pGEX-2T (Pharmacia Biotech Inc.). The same fragment was also cloned into the BamHI and SmaI sites of a mammalian expression vector pSG5 (Stratagene). The full-length cDNA was further truncated into 5Ј and 3Ј fragment by polymerase chain reaction amplification with two sets of primers: 5Ј-AATACCACTACAATGGATGA-TGTATAT-3Ј/5Ј -GAATTCGGATCCTCACGTCATCTTCCGCTCGAGC-CC-3Ј and 5Ј-GAATTCGGATCCATGCTTCTCTGCGTTTCGTGCGGC-3Ј/5Ј-GAGATGGTGCACGATGCACAGTTGAAG-3Ј. The polymerase chain reaction products were digested with BamHI and BamHI/BglII, respectively, and subcloned into the BamHI site of pGEX-2T, pGAD424, and pSG5 plasmids. The authenticity and orientation of the amplified fragment cloned in these vectors were confirmed by restriction digestion and sequencing.
Expression and Purification of Rat dUTPase in E. coli-The expression and purification of GST fusion proteins were essentially as described by Smith and Johnson (33). After transformation into E. coli JM109 and induction by isopropyl-1-thio-␤-D-galactoside, bacteria were collected by centrifugation, resuspended 1:10 in NETN (0.5% Nonidet P-40/1 mM EDTA/20 mM Tris, pH 8.0/100 mM NaCl) containing 2 mM phenylmethylsulfonyl fluoride, sonicated, and centrifuged. Fusion protein was then purified on glutathione-agarose beads by incubating 10 ml of supernatant with 1 ml of beads (1:1 in NETN plus 0.5% powdered milk) for 30 min. After extensive washes with NETN, the beads were used for GST pull-down assay as described (27) or further eluted by 5 mM reduced glutathione for gel retardation assay (23) and for antibody generation in goat using standard immunization procedures (34).
In Vitro Protein-Protein Interaction Assays- 35 S-Labeling of mP-PAR␣, mPPAR␦, mPPAR␥, rRXR␣, and hTR␤ were achieved by in vitro translation using [S 35 ]methionine (Amersham Corp.) and a TNT Coupled Wheat Germ Extract System (Promega) according to manufacturer's instructions, and 50 l of labeled receptors were precleaned twice for 1 h at 4°C on 25 l of beads preloaded with GST. The receptors were then incubated with 25 l of beads preloaded with GST-dUTPase for 1 h at 4°C. After three washes with NETN, an equal volume of 1 ϫ SDS-polyacrylamide gel electrophoresis loading buffer was added into the beads and boiled for 5 min. Samples were resolved by SDS-polyacrylamide gel electrophoresis (35). Binding of rPPAR␣ to truncated forms of GST-dUTPase was carried out using the same procedure except an in vivo [S 35 ]methionine-labeled rPPAR␣ was used (32).
Cell Culture and Transfection-The human embryonic kidney cells (tsA 201, a gift from Dr. Richard Horn, Thomas Jefferson University) were maintained in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) containing 10% hormone-depleted fetal bovine serum (36) and 1% penicillin/streptomycin at 37°C in 5% CO 2 . Transfections were done using the cationic liposome-mediated (N[1-(2,3-dioleoyloxy)propyl]-N, N, N-trimethylammonium methylsulfate) method according to the manufacturer's instructions (Boehringer Mannheim) in 24-well plates. Unless otherwise indicated, each transfection contained a total of 200 g of reporter and receptor/dUTPase expression plasmid DNA balanced with pSG5 and an additional 50 g of ␤-galactosidase expression vector pCMV␤ (Clontech). The cells were incubated with DNA for 6 h and washed with phosphate-buffered saline before adding fresh medium containing the indicated ligand. After 40 h of incubation, the cells were processed to assess luciferase activity, and the activity obtained for individual transfections was expressed relative to the ␤-galactosidase activity obtained from the same preparation of lysate.
Immunofluorescence Staining-Cells grown on coverslips coated with collagen in 6-well plates were transfected with 1 g of pSG5-dUTPase and 1 g of either pSG5 or pSG5-PPAR␣, respectively, as described above. Transfected cells were fixed with cold methanol for 10 min and permeabilized with 0.5% Triton X-100, 40 h after transfection. Cells were then incubated with bovine serum albumin for 1 h, followed by incubation with the primary antibody for 1 h and then incubation with an tetramethylrhodamine isothiocyanate-or fluorescein isothiocyanate-coupled secondary antibody for 1 h. Coverslips were mounted onto slides for fluorescence microscopic examination.
Other Methods-Plasmid isolation, endonuclease digestion, ligation, and DNA sequencing were performed according to standard protocols (37). dUTPase enzyme assay was performed as described (38) with minor modifications. Protein concentration was determined by the method of Bradford (39).

Identification of Rat Liver cDNA Clone(s) That
Interact with rPPAR␣ in the Yeast Two-hybrid Screen-The rPPAR␣, expressed in S. cerevisiae strain HF7c as a fusion protein with the DNA binding domain of GAL4, was used as the bait in a two-hybrid screen. Two-hybrid interactions in this strain activate transcription of the HIS3 and lacZ genes. A rat liver cDNA library in pGAD10 was screened for proteins that interact with the pGBT9-rPPAR␣ bait. Approximately 1.5 million transformants were plated on His Ϫ plates, and surviving colonies were further screened by ␤-galactosidase assay for lacZ expression. A total of 41 clones, which activate both the HIS3 and ␤-galactosidase reporter genes, were isolated and reintroduced into yeast strain SFY526 with pGBT9, pGBT9-rPPAR␣, pGBT9-rRXR␣, and pGBT9-hTR␤, respectively. Among these 41 clones, six exhibited strong and specific interaction with rP-PAR␣ (data not shown). These positive clones were digested with restriction enzymes and partially sequenced. Clone number 4, designated PIP4 (PPAR-interacting protein 4), exhibited strongest interaction with rPPAR␣ and was chosen for further analysis.
PIP4 Encodes Deoxyuridine-triphosphatase-Sequence analysis of the ϳ1-kb cDNA insert of PIP4 revealed the presence of an open reading frame encoding a peptide of 203 amino acids with the predicted molecular mass of the protein of 23.9 kDa. The start of the coding sequence was defined by the first ATG, with the surrounding sequences (AGCGCCATGCCC) conform-ing to a consensus sequence for the translation initiation site (40). The 3Ј-untranslated sequence includes a putative polyadenylylation signal AATAA 17 base pairs upstream of a short poly(A) stretch of 18 nucleotides (Fig. 1A). A GenBank data base search revealed that the deduced 203-amino acid sequence has significant homology with the human dUTPase (41). The rat dUTPase is 62 amino acids longer than the human dUTPase (141 residues) with an identity of 92%. The rat dUTPase also contains five distinct amino acid sequence motifs (motifs 1-5, Fig. 1B) that are common to all the dUTPases identified so far (42). The N-terminal sequence (residues 1-62) is arginine-rich (10 of 62 residues) and is composed of two putative protein kinase C phosphorylation sites (RRKQRS and RTRS) (43). The portion of the protein between amino acid residues 24 and 49 has significant homology with mouse RNA-DNA binding protein (57%, X70067), human RNA-binding protein (57%, L37368), and DNA-binding nucleocapsid protein (56%, X77048). These proteins are known to bind with single strand DNA and RNA (48), implying that this domain of dUTPase may exhibit similar functions. Thus, on the basis of the degree of identity shared with these proteins, we conclude that the PPAR␣-interacting PIP4 is a dUTPase. At the N terminus, rat dUTPase is 62 amino acid residues longer than its human counterpart (38,39). To determine whether the isolated plasmid encodes a functional dUTPase, we performed dUTPase enzyme assays on permeabilized cells of yeast strain SFY526 transformed with PIP4. The transformants exhibited 10-fold  (45), yeast (46), and E. coli (47). dUTPase. Rat dUTPase is 62 amino acids longer than the human enzyme. The position of motifs 1-5 are shown in boldface type. These five motifs found in all dUTPases identified so far are functionally important (42). more dUTPase activity than the untransformed control (Table I), indicating that PIP4 did, in fact, encode dUTPase.
Specific Interaction of Rat dUTPase with all Three Murine PPAR Isoforms-The interaction of rat dUTPase with PPARs was confirmed by an in vitro GST pull-down assay. For this purpose, rat dUTPase was first expressed in E. coli as a GST fusion protein. After induction with isopropyl-thio-␤-D-galactoside, the fusion protein, with an apparent molecular mass of 47 kDa, accumulated in the induced E. coli cells. When glutathione-coupled agarose beads were incubated with the induced E. coli lysate and washed extensively, a 47-kDa protein was the main component retained with the beads. Two other minor bands, with an apparent molecular mass of ϳ30 kDa, were observed, and these may represent some other bacterial proteins associated with the GST-dUTPase fusion protein (see "Discussion") ( Fig. 2A). Incubation of preloaded glutathione beads (i.e. beads with attached GST-dUTPase fusion protein) with[ 35 S]methionine-labeled mPPAR␣, mPPAR␦, mPPAR␥, rRXR␣, or hTR␤ revealed specific interaction of all three PPAR isoforms with the GST-dUTPase fusion protein. Both rRXR and hTR␤ failed to bind with the GST-dUTPase fusion protein (Fig.  2B). No binding of PPAR isoforms to GST was observed (data not shown), confirming that PPAR binding to GST-dUTPase preloaded beads is due to the interaction of PPAR with dUTPase.
Arginine-rich N-terminal dUTPase Domain Interacts with PPAR␣-Rat dUTPase was truncated at the junction of arginine-rich N-terminal domain (1-62 residues) and the C-terminal domain (63-203 residues) by amplification of the full-length cDNA using two sets of primers as described under "Experimental Procedures." These amplified cDNA fragments were first cloned into the BamHI site of pGAD424 plasmid and tested for interaction with rPPAR␣, rRXR␣, and hTR␤ in vivo using yeast strain SFY526. In this in vivo assay, the N-terminal domain (1-62 residues) exhibited strong interaction with rPPAR␣ similar to that observed with the full-length dUTPase, whereas the C-terminal domain (63-203 residues) did not interact with rPPAR␣. Neither the full-length nor the N-terminal 62-amino acid segment of dUTPase was able to interact with rRXR␣ and hTR␤ (Fig. 3A). The interaction was further confirmed in vitro in experiments using [ 35 S]methionine-labeled recombinant rPPAR␣ produced in insect Sf9 cells and bacterially expressed truncated forms of dUTPase. Among the labeled insect cell proteins, only rPPAR␣ bound to the full-length dUTPase and to the N-terminal 62-amino acid segment (Fig.  3B). The rPPAR␣ failed to bind to the C-terminal (63-203 residues) dUTPase segment. These results indicate that the interaction is specific and the N-terminal 62-amino acid domain is responsible for interacting with PPAR␣.
dUTPase Prevents Formation of PPAR-RXR Heterodimers-PPARs heterodimerize with RXRs and bind to PPRE to regulate target gene transcription (24). To test whether dUTPase interacts with the PPAR⅐RXR⅐PPRE complex, bacterially expressed full-length and truncated GST-dUTPase fusion proteins were used for gel mobility shift assays (23). Recombinant rPPAR␣ and rRXR␣ form a heterodimer on rat peroxisomal fatty acyl-CoA oxidase PPRE (Fig. 4, lane 1), whereas fulllength dUTPase alone did not bind to this DNA element (Fig. 4,  lane 11). Combination experiments were performed with increasing amounts of GST-dUTPase (Fig. 4, lanes 2-4), GST-⌬ 63-203 dUTPase (Fig. 4, lanes 5 and 6), or GST-⌬ 1-62 dUTPase (Fig. 4, lanes 7-9) added to an equal amount of rPPAR␣⅐rRXR␣⅐PPRE binding reaction mixture. The results revealed that the full-length dUTPase blocked the formation of PPAR⅐RXR⅐PPRE complex, but the two truncated dUTPase fusion proteins showed no effect. We also noted that dUTPase incubated with either rPPAR␣ or rRXR␣ failed to bind to PPRE (data not shown). These results show that dUTPase, instead of interacting with the PPAR⅐RXR⅐PPRE complex to induce a supershift, actually blocks the formation of PPAR⅐RXR⅐PPRE complex. Accordingly, this dUTPase-PPAR␣ interaction may lead to inhibition of transcription.
Rat dUTPase Is a Ligand-independent PPAR Inhibitor-To investigate the functional significance of the interaction of dUTPase with PPARs, transient transfection assays were carried out in mammalian cells using an established chimera system, in which the ligand-binding domains of murine PPAR␣, PPAR␦, and PPAR␥ were fused to the DNA binding domain of yeast transcription factor GAL4 (20). Expression plasmid for GAL4-PPAR␣ chimera was first used to transfect human kidney tsA201 cells together with a luciferase reporter construct containing five copies of the GAL4-binding site upstream of the minimal thymidine kinase promoter, and various amount of dUTPase-expression plasmid (Fig. 5A). As expected, the luciferase reporter gene was efficiently transactivated by GAL4-PPAR␣ chimera in the presence of 10 Ϫ5 M Wy 14, 643, a potent peroxisome proliferator. Co-transfection with dUTPase caused a reduction in the reporter gene activity, and this reduction correlated with increases in the amounts of dUTPaseexpressing plasmid added. In contrast, both truncated forms (⌬1-62, and ⌬63-203) did not exhibit inhibitor activity. Inhibition of PPAR transactivation by dUTPase was further investigated using different concentrations of ligand (Fig. 5B). The PPAR␣ was optimally activated by Wy 14, 643 at a concentration of 1 ϫ 10 Ϫ5 M, whereas PPAR␥ had highest transcriptional activity in the presence of 5 ϫ 10 Ϫ5 of 15-d⌬ 12,14 PGJ2, the PPAR␥-specific ligand. These activities were significantly reduced in the presence of dUTPase at different concentrations of  Wy14, 643 and 15-d⌬ 12,14 PGJ2. Notably, the classic peroxisome proliferator, Wy 14,643 showed very high potency in activating PPAR␣, whereas the newly identified PPAR␥ ligand, 15-d⌬ 12,14 PGJ2, is more effective in activating PPAR␥. Rat dUTPase also inhibited mPPAR␦, although this isoform was only weakly activated by both Wy 14, 643 and 15-d⌬ 12,14 PGJ2 (data not shown). Transfections using native rPPAR␣ and rRXR␣-expressing plasmid and PPRE-HD-Luc reporter system (23) exhibited similar results, indicating that the full-length dUTPase represses PPAR-mediated transcriptional activation in a ligand-independent manner (data not shown).
Colocalization of Rat dUTPase with rPPAR␣ in the Nucleus in Transiently Transfected tsA201 Cells-To confirm that rat dUTPase and rPPAR␣ interact in vivo, we transiently transfected human embryonic kidney cells (tsA201 cells) with rat dUTPase-expressing plasmid, either alone or with PPAR␣expressing plasmid. The cells were then fixed, permeabilized, and immunostained with anti-dUTPase or anti-rPPAR␣ antiserum. In the absence of rPPAR␣, dUTPase was observed predominantly in the cytosol with a weak signal in the nucleus (Fig. 6). In the presence of rPPAR␣, the cellular distribution of dUTPase was significantly different. In these cells, the dUTPase was localized mostly in the nucleus with a trace amount remaining in the cytoplasm. The rPPAR␣ was found almost exclusively in the nucleus of the cotransfected cells. The merger of dUTPase and rPPAR␣ fluorescence patterns in the nucleus strongly suggests colocalization due to interaction between these two proteins in transfected cells.
Tissue Distribution of Rat dUTPase-Northern analysis using dUTPase cDNA as a probe revealed the presence of an abundant 1-kb mRNA species and two minor species of mRNA (1.5 and 3 kb) (Fig. 7). The minor mRNA species may be due to differential splicing of a primary transcript. Each of these mRNA species is widely distributed in rat tissues, but the abundance of these transcripts relative to each other is different in different tissues. The 1-kb transcript, which is the expected size for the dUTPase, is abundant in all organs but appears more in the heart and kidney.

DISCUSSION
Using a GAL4 DNA-binding domain-PPAR␣ fusion in a yeast two-hybrid screen, we isolated several cDNA clones, and one of these clones, PIP4 (peroxisome proliferator-activated receptorinteracting protein-4), has been found to interact strongly with unliganded full-length PPAR␣ but not with other receptors such as RXRa and TR␤. This PPAR␣-interacting protein prevented the formation of PPAR-RXR heterodimers and inhibited the transcriptional activity of all three isoforms of PPAR (␣, ␦, and ␥). Furthermore, when PPAR␣ and PIP4 are transiently coexpressed in human embryonic kidney cells, these proteins appeared to colocalize within the nucleus. Thus it appears that PIP4 interacts with PPAR and that this interaction results in the functional repression of PPAR activity.
Nucleotide sequence of PIP4 showed that it encodes a polypeptide of 203 amino acids with an estimated relative molecular mass of 23.9 kDa. This protein has an arginine-rich N-terminal 62-amino acid segment that has been shown to interact strongly with the PPAR␣. Homology search revealed that the C-terminal 141 amino acid sequence of this protein corresponds to the full-length human dUTPase (141 residues) and exhibits significant identity with this human enzyme (41). The other dUTPases cloned so far, such as the yeast (147 residues) and E. coli (150 residues), are also shorter in length when compared with the rat dUTPase. Evidence also includes the pseudoprotease domains of some retrovirus-encoded dUTPases (42). The presence of dUTPase in widely divergent organisms (i.e. from humans to the retroviruses) suggests that this enzyme may generally perform a vital role in DNA replication (49).  (32). Standard protocols were used for insect cell culture, infection, and in vivo labeling. c, GST pull-down assay. Glutathione-agarose beads preloaded with GST fusion proteins from a were incubated with [ 35 S]methionine-labeled proteins as shown in b using the same procedure as described in Fig. 2.
In vivo, dUTPase hydrolyzes dUTP to dUMP and pyrophosphate ion and prevents accumulation of cellular dUTP. It is widely speculated that dUTPase is essential for cell viability because this enzyme is believed to maintain extremely low levels of intracellular dUTP and thus prevent the synthesis of highly uracil-substituted DNA during replication (50). Although deletion of the dUTPase from herpes simplex virus type 1 did not affect the viral DNA replication, the loss of dUTPase activity resulted in viruses with 5-fold increased relative mu-tant frequency, indicating that the herpes simplex virus type 1 dUTPase has an antimutator function (51). Mutants of E. coli deficient in dUTPase demonstrate an apparent increase in the amount of short, Okazaki-like DNA fragments, which are intermediates of DNA synthesis and are explained by the misincorporation of uracil in DNA as a result of an increase in available dUTP, followed by its rapid excision and repair (52). The presence of high levels of dUTPase mRNA in various normal and developing tissues reflects that dUTPase functions in these tissues to keep low levels of dUTP in order to prevent misincorporation of uracil into DNA and maintain DNA fidelity.
The observation that rat dUTPase interacts with PPAR␣ (and other isoforms of this receptor) and functions as a repressor raises some interesting questions. Rat PPAR␣, when interacting with dUTPase, cannot form heterodimers with rRXR␣ to bind to PPRE (Fig. 4). Activation of target gene expression requires PPARs to heterodimerize with RXR, and that dUTPase acts as a negative regulator is confirmed by transfection assays. PPAR␣ interacts with rat dUTPase with the 62amino acid N-terminal region, but it appears that the C-terminal dUTPase domain is also necessary for the inactivation as seen in the gel mobility shift assay and transfection assays. The 62-amino acid sequence of rat dUTPase is arginine-rich, consists of two putative protein kinase C phosphorylation sites (43) and has significant homology with single strand DNA-RNA binding proteins and viral DNA-binding nuclear capsid pro-  (20) and full-length or truncated dUTPase expression plasmid. Cells were treated with 1 ϫ 10 Ϫ5 M Wy 14, 643 and subsequently assayed for luciferase activity. B, ligand-independent repression of PPAR activation by dUTPase. The same reporter were cotransfected with GAL4-mPPAR␣ or GAL 4 -mPPAR␥ chimeras in the presence or the absence of full-length dUTPase expression vector, respectively, into tsA201 cells. Transfected cells were treated with Wy 14, 643 or 15d⌬ 12,14 PGJ2 at concentrations as indicated and assayed for luciferase activity. pSG5 plasmid DNA and ␤-galactosidase expression vector pCMV␤ were included to normalize transfections. The data represent the means of two independent experiments performed in triplicate. Variations were less than 10%.

FIG. 6. Colocalization of dUTPase and PPAR␣ in tsA201 cells.
Transiently transfected cells grown on glass coverslips, were fixed with methanol and double stained by goat anti-dUTPase and rabbit anti-PPAR␣ (32) antibodies. The antigen-antibody complexes were visualized by staining with a monoclonal fluorescein isothiocyanate-conjugated anti-goat IgG antibody and affinity purified polyclonal tetramethylrhodamine isothiocyanate-labeled anti-rabbit IgG antibodies. Photographs were taken using a Zeiss fluorescence microscope, and the merger of staining patterns was done by a computer Photoshop program. A, tsA201 cells transfected dUTPase-expressing plasmid and stained for dUTPase. B and C, tsA201 cells cotransfected with dUTPase and rPPAR␣-expressing plasmids and stained for dUTPase and rP-PAR␣. D, the fluorescence merger of B and C. Note the patterns of dUTPase staining in A and B are different.
FIG. 7. Tissue distribution of rat dUTPase. A membrane filter containing mRNA from rat tissues (purchased from Clontech) was hybridized with the P 32 -labeled rat dUTPase cDNA probe using the ExpressHyb hybridization system according to the manufacturer's instructions (Clontech). Arrows indicate the size of positive bands in kb.
teins (48). The mouse mammary tumor virus contains a dUTPase domain derived from the pro-open reading frame and a nucleocapsid protein domain at its N-terminal region (53). dUTPase (pseudoproteases) encoded by poxviruses contains 370 residues (42). The functional significance of the extra domain present in these dUTPases remains to be fully elucidated, but our results demonstrate that the extra domain of rat dUTPase is responsible for interaction with PPARs. The human dUTPase lacks this extra 62-amino acid portion (41), implying that it may not interfere with PPAR action. Because the rat dUTPase is only the second mammalian dUTPase cDNA cloned to date, cloning of dUTPases from other species that are responsive to peroxisome proliferators, such as the mouse, will enhance the understanding of the significance of this additional arginine-rich N-terminal sequence present in the rat but not in the human dUTPase.
Although PPARs serve as transcription factors, the interaction of PPARs with dUTPase suggests the possibility that peroxisome proliferators may affect the DNA fidelity by modulating dUTPase activity. In T4 phage-infected E. coli, dUTPase is found in a multienzyme complex in a total mass of about 1500 kDa, and it functionally interacts with other dNTP-synthesizing enzymes (54). Whether PPAR-dUTPase interaction modulates the dNTP-synthesizing enzymes remains speculative at this time. The intricate interrelationships between transcriptional machinery and DNA-repair/synthesis processes are becoming increasingly evident from recent observation (31). dUTPase is a crucial enzyme involved in the maintenance of DNA fidelity (50 -52). During the early Drosophila development, dUTPase activity is modulated by interaction with a regulatory 61-kDa protein (55). The interaction of PPARs with dUTPase suggests that PPARs may also modulate dUTPase activity. Because PPARs are activated by peroxisome proliferators, it would be of interest to investigate whether peroxisome proliferators affect dUTPase activity and alter cellular dUTP levels during sustained induction of the pleiotropic responses.