Identification of the CREB-binding Protein/p300-interacting Protein CITED2 as a Peroxisome Proliferator-activated Receptor (cid:1) Coregulator*

Like other nuclear receptors, the peroxisome prolif-erator-activated receptors (PPARs) use a wide variety of protein-protein interactions to properly regulate transcription of target genes. In an attempt to identify novel PPAR-interacting proteins, a cDNA expression library was screened with bacterially expressed PPAR (cid:1) . One of the genes identified as a PPAR (cid:1) -associated protein by interaction cloning was the CREB-binding protein/p300-interacting transactivator with ED-rich tail 2 (CITED2, also called p35srj/mrg1/msg1). This coactivator inter-acted directly with PPAR (cid:1) in the presence or absence of ligand predominantly via the ligand binding domain of the nuclear receptor. In transient transfection reporter assays, CITED2 acted as a dose-dependent coactivator of PPAR (cid:1) -dependent transcriptional regulation in the presence of several exogenous ligands. CITED2 also increased PPAR (cid:2) -dependent regulation of reporter genes but had no effect on PPAR (cid:3) activity. To determine whether CITED2 affects endogenous gene expression, this protein was stably overexpressed (CITED2 (cid:4) ) or repressed by small inhibitor RNA (CITED2 (cid:5) ) in immortalized mouse hepatocytes. Relative to the control stably transfected or CITED2 (cid:5) cells, CITED2 (cid:4) cells had an increased rate of cell proliferation. Microarray penicillin/streptomycin. Mouse SV40-immortalized hepatocytes (MuSHs) grown in modified Eagle’s medium (cid:1) modification supplemented with 4% fetal bovine serum, 1% penicillin/streptomycin, and 0.1 (cid:5) M dex- amethasone. HepG2 cells were maintained in modified Eagle’s medium supplemented with 10% fetal bovine serum, 1% penicillin/streptomycin, 1% non-essential amino acids, and 1 m M sodium pyruvate. Stable cell lines were generated by plating 50,000 cells into individual wells on a 24-well plate. Cells were allowed to recover overnight and were transfected with 1 (cid:5) g of plasmid using LipofectAMINE according to the man- ufacturer’s protocol (Invitrogen). After transfection, cells were allowed to recover overnight, and individual wells were moved to 10-cm dishes. Selection of stable transfectants was started 24 h after initial plating in the 10-cm dishes. Selection medium was replaced every 3 days until individual colonies were seen. Individual colonies were harvested and expanded in addition to a line of pooled stable transfectants. All stable lines (clonal and pooled) were analyzed for CITED2 expression via Western blot. Stably transfected cell cultures were maintained in normal cell medium with G418 (Invitrogen) at 500 (cid:5) g/ml, substituted for penicillin/ streptomycin, for mouse hepatocytes. Transient Transfection— Cells were plated in 24-well plates and allowed to recover overnight at 37 °C. Mammalian two-hybrid studies were carried out with 400 ng/well rPPAR (cid:1) fused to the GAL4 DNA binding domain (pM-rPPAR (cid:1) ) with 400 ng/well empty pVP16 or pVP16- CITED2, 100 ng/well GAL4-responsive reporter (pFR-luciferase), and 100 ng/well transfection control (pRLTK). Cells were transfected using LipofectAMINE (Invitrogen) according to the manufacturer’s protocol.

The peroxisome proliferator-activated receptors (PPARs) 1 are members of the nuclear receptor (NR) superfamily and exist as three distinct subtypes (␣, ␤/␦, and ␥; NR1C1-3 (1)). The receptors, so named for their ability to respond to a class of compounds deemed peroxisome proliferators, have been implicated in a wide range of biological processes such as cell growth, lipid homeostasis, apoptosis, and response to xenobiotics and endogenous compounds. In particular, PPAR␣ has a role in carcinogenesis with peroxisome proliferators being potent hepatocarcinogens in rodents (for a review, see Ref. 2).
The regulation of gene expression by PPARs follows the classic NR mechanism whereby the receptor binds to ligand, translocates to the nucleus, heterodimerizes (retinoid X receptor ␣), binds to a conserved response element (DR1) in the promoter region of the target gene, and regulates transcription. One aspect of this mechanism that has drawn increasing attention is the potential protein-protein interactions that lead to the proper regulation of the target gene. Several proteins, often called coregulators, have been identified in the nuclear receptor transcription complexes that affect target gene expression. The family of coregulator proteins can be divided into two basic classes, coactivators that serve to increase the level of transcriptional activation of a gene and corepressors that do the opposite. These proteins come in many forms and may or may not contain enzymatic activity such as histone acetyltransferase (e.g. CREB-binding protein (CBP)/p300) or histone deacetylase (e.g. nuclear receptor co-repressor). Those coregulators that do not contain enzymatic activity are thought to serve as a protein bridge that span the distance between the transcription factor and the elements of the complex that act directly upon the target gene. It is these bridging proteins that serve to further regulate transcription by helping to create distinct transcriptional complexes in response to particular chemicals whether endogenous or exogenous. Many coactivators have been initially isolated as PPAR coactivators such as PPAR-binding protein (3), PPAR␥ coactivator-1 (4), and PPAR receptor interaction protein (same as thyroid receptor-binding protein (5)).
The CBP/p300-interacting transactivator with ED-rich tail (CITED) family of proteins contains four members to date. This class of protein has a highly acidic carboxyl-terminal end that interacts with CBP/p300. The CITED proteins affect the transcriptional activity of many transcription factors ranging from AP2 (6), estrogen receptor (7), and hypoxia-inducible factor 1␣ (HIF1␣) (8). One member of this family, CITED2 (also called p35srj, mrg1, and msg1), is involved in the transcriptional repression of HIF1␣ (8). Previous studies have suggested that CITED2 plays a role in the development of the nervous system since the CITED2 Ϫ/Ϫ mouse died in late gestation and showed malformation in the mid and hind brain (9). CITED2 can repress INK4a/ARF via regulation of Bmi1/Mel18 leading to decreased cell proliferation (10). This is significant since the main phenotype of the CITED2 Ϫ/Ϫ mouse is defects in neural tube closure that is thought to be caused by a loss of cell proliferation and increased cell death.
In these studies, CITED2 was identified as a PPAR␣-associated protein in an interaction cloning screen. Due to the role of CITED2 as a coactivator and a CBP/p300 bridging factor, we hypothesized that it would affect the transcriptional activity of PPAR␣. Using a variety of in vivo and in vitro assays, we demonstrated that CITED2 directly interacts with PPAR␣ and acts as an exogenous ligand-dependent and -independent coactivator. Also we demonstrated that CITED2 can affect cell cycle regulation and gene expression in hepatocytes in the absence of exogenous PPAR␣ activators.
Cell Culture-COS-1 cells were maintained in modified Eagle's medium ␣ modification supplemented with 8% fetal bovine serum and 1% penicillin/streptomycin. Mouse SV40-immortalized hepatocytes (MuSHs) (13) were grown in modified Eagle's medium ␣ modification supplemented with 4% fetal bovine serum, 1% penicillin/streptomycin, and 0.1 M dexamethasone. HepG2 cells were maintained in modified Eagle's medium supplemented with 10% fetal bovine serum, 1% penicillin/streptomycin, 1% non-essential amino acids, and 1 mM sodium pyruvate. Stable cell lines were generated by plating 50,000 cells into individual wells on a 24-well plate. Cells were allowed to recover overnight and were transfected with 1 g of plasmid using LipofectAMINE according to the manufacturer's protocol (Invitrogen). After transfection, cells were allowed to recover overnight, and individual wells were moved to 10-cm dishes. Selection of stable transfectants was started 24 h after initial plating in the 10-cm dishes. Selection medium was replaced every 3 days until individual colonies were seen. Individual colonies were harvested and expanded in addition to a line of pooled stable transfectants. All stable lines (clonal and pooled) were analyzed for CITED2 expression via Western blot. Stably transfected cell cultures were maintained in normal cell medium with G418 (Invitrogen) at 500 g/ml, substituted for penicillin/ streptomycin, for mouse hepatocytes.
Transient Transfection-Cells were plated in 24-well plates and allowed to recover overnight at 37°C. Mammalian two-hybrid studies were carried out with 400 ng/well rPPAR␣ fused to the GAL4 DNA binding domain (pM-rPPAR␣) with 400 ng/well empty pVP16 or pVP16-CITED2, 100 ng/well GAL4-responsive reporter (pFR-luciferase), and 100 ng/well transfection control (pRLTK). Cells were transfected using LipofectAMINE (Invitrogen) according to the manufacturer's protocol. Mammalian one-hybrid studies were performed using 400 ng/well pM-rPPAR␣ and CITED2 in a mammalian expression vector (pcDNA 3 ) at a 1:1 or 2:1 ratio (PPAR:CITED2) with pFR-luciferase and pRLTK. Cells were treated with Me 2 SO, 50 M Wy-14,643, or 100 M conjugated linoleic acid (CLA) for 6 h. Luciferase activity was measured using the Dual Luciferase kit (Promega). Luciferase activity was corrected for transfection efficiency (pRLTK) and extraction yield (via total protein assay). HepG2 and MuSH transfections were carried out with identical DNA concentrations and technique unless noted. All treatments were for 6 h with 50 M Wy-14,643, 100 M CLA, 100 M ciprofibrate, or Me 2 SO.
Pull-down of in Vitro Translated CITED2-In vitro translated human CITED2 was radiolabeled with [ 35 S]methionine and incubated with bacterially expressed PPAR␣-MBP for 1 h at 4°C in the presence of amylose resin. Resin was collected and washed three times with cold radioimmune precipitation assay buffer. Bound protein was eluted from the resin using 10 mM maltose in radioimmune precipitation assay buffer for 1 min at 4°C. Eluate was resolved on a 12% Tris-glycine gel, dried, and subjected to autoradiography.
Reverse Transcription-PCR-Total RNA from 10 different mouse tissues was purchased (Ambion), and reverse transcription-PCR for CITED2 (see Table I) and ␤-actin was performed essentially as de-FIG. 1. Verification of the interaction between rat PPAR␣ and CITED2 in vitro. CITED2 was radiolabeled with [ 35 S]methionine and incubated with bacterially expressed PPAR␣-MBP fusion for 1 h at 4°C in the presence of amylose resin. Resin was collected and washed three times with cold radioimmune precipitation assay buffer. Bound MBP was eluted from the resin using 10 mM maltose in radioimmune precipitation assay buffer for 1 min at 4°C. Eluate was resolved on a 12% Tris-glycine gel, dried, and subjected to autoradiography. The image is representative of two independent experiments. scribed previously (14). Products were resolved on a 2% high resolution agarose gel (Cambrex) and quantified by densitometry.
Mouse Oligonucleotide Arrays-The Mouse Genome Oligo Set Version 1 was purchased from Operon (Alameda, CA) and contains 6800 optimized 70-mers plus 24 controls with melting temperature normalized to 78°C. Sequences were optimized by the manufacturer using BLAST against all known mouse genes to minimize cross-hybridization. Oligonucleotides were printed onto glass slides using GeneMachines Omnigrid (San Carlos, CA) with additional controls obtained from Stratagene (SpotReport system, Stratagene, La Jolla, CA) at the Pennsylvania State University microarray core facility.
Microarray Analysis-Total RNA was isolated by TriReagent (Sigma) and further purified with RNAEasy (Qiagen) according the manufacturers' instructions. Details of the microarray analysis including reverse transcription, labeling, and hybridization can be found elsewhere (13). In the present experiments, co-hybridization was performed with cDNA from CITED2ϩ-treated cells labeled with Cy5 and those from control-transfected samples labeled with Cy3. Scanning was performed by a GenePix 4000A scanner (Axon Instruments, Inc., Foster City, CA), and image acquisition was performed with Axon GenePix image software. Analysis of gene expression data was performed using GeneSpring software (Silicon Genetics Inc., Redwood City, CA) and PathwayAssist (Ariadne Genomics, Rockville, MD) (see supplemental data).
Statistical Analysis of Microarray-Normalization and analysis of the gene expression profiles were performed as described previously (13). Intensity-dependent normalization was also applied where the ratio was reduced to the residual value of the Lowess fit of the intensity versus ratio curve. Statistical analysis was performed using a Student's t test with a p value of 0.05 with the additional criteria of being either 1.5-fold increased or decreased in CITED2ϩ-versus control-transfected cells. Genes that met these parameters were classified by molecular function using annotations from Silicon Genetics or Ariadne Genomics.
Real Time PCR-Total RNA was isolated as before. The total RNA was reverse transcribed using the ABI High Capacity cDNA archive kit (Applied Biosystems, Foster City, CA). Standard curves were made using serial dilutions from pooled cDNA samples. Real time PCR was performed using the SYBR Green PCR Master Mix (Applied Biosystems) according to the manufacturer's protocol and amplified on the ABI Prism 7000 sequence detection system. RNA Inhibitor Design-A double-stranded RNA inhibitor for CIT-ED2 was designed using the sequence from GenBank TM . The sequence was synthesized and cloned into the pSUPER.neo plasmid (Oligoengine, Seattle, WA). All transfections of RNAi plasmids were performed using LipofectAMINE (Invitrogen) transfection reagent according to the manufacturer's protocol. The sequence chosen for the CITED2 RNAi is as follows: 5Ј-CCACTACATGCCGGATTTG-3Ј.
Total soluble protein (50 g) was separated on a 12% Tris-glycine gel and electrotransferred to a polyvinylidene difluoride membrane (Immobilon P, Millipore). Membranes were washed three times with TBS, 0.1% Tween 20 (TBSϩ), blocked with 5% nonfat dry milk in TBSϩ for 1 h at room temperature, washed three times with TBSϩ, and incubated while rocking at room temperature for 1 h with primary antibodies. Immunoblotting was performed using a mouse anti-CITED2 antibody (Novus Biologicals, Littleton, CO) in TBSϩ, 0.5% dry milk. The blot was washed three times with TBSϩ and incubated at room temperature for 1 h with anti-mouse (Amersham Biosciences) at 1:10,000 dilution in TBSϩ, 0.5% dry milk. The blot was then washed three times with TBSϩ, and visualization was performed using enhanced chemiluminescent visualization (ECL, Amersham Biosciences).
Cell Growth Analysis-Cells of interest were plated at 200 cells/well in a 96-well plate, allowed to recover overnight in normal defined

Identification of CITED2 as a PPAR Transcriptional
Coregulator-The screening of the 3 ϫ 10 6 plaques from a cDNA expression library with PPAR␣-MBP produced seven reproducible positive interacting clones (data not shown). 2 Each clone was sequenced and compared with GenBank TM . One of the interacting clones contained 437 bp that corresponds to the 3Ј-end of the nuclear protein CITED2 (p35srj (15)). The interaction between this partial sequence and PPAR␣ was verified in a yeast two-hybrid assay (data not shown).
To determine whether a direct interaction occurred between CITED2 and PPAR␣, pull-down studies were performed. Full-length human CITED2 in the pcDNA 3 mammalian expression vector was in vitro translated in the presence of [ 35 S]methionine. Full-length rat PPAR␣-MBP was expressed using a bacterial expression system, the in vitro translated human CIT-ED2 was added, and the mixture was passed over an amylose resin, washed, and eluted with excess maltose. As shown in Fig. 1, in vitro translated human CITED2 could interact with PPAR␣ in the presence or absence of a known ligand (Wy-14,643, 50 M). When expressed alone, the MBP epitope tag did not significantly interact with CITED2.
To examine the association in vivo and to map the interaction between PPAR␣ and CITED2, mammalian two-hybrid experiments were performed. COS-1 cells were transiently transfected with full-length rat PPAR␣ or each domain individually fused to the GAL4 DNA binding domain (pM-PPAR␣) and human CITED2 fused to the VP16 activation domain (VP16-CITED2). As shown in Fig. 2, an interaction between PPAR␣ and CITED2 was observed using full-length PPAR␣ (A-F) or just the D domain alone. If other domains were fused to the D domain, such as C or E/F, there was a noticeable decrease in association between PPAR␣ and CITED2, indicating cross-domain interference. Interestingly the basal activity of the pM-AB domain, which contains the ligand-independent activation domain, was significantly decreased by VP16-CITED2. Unlike what was observed in vitro, the association between PPAR␣ and CITED2 was enhanced by ligand treatment in the mammalian two-hybrid experiments. The association of CIT-ED2 to the ligand binding domain (D-F) of PPAR␣ was significantly enhanced by both Wy-14,643 and CLA.
Similar studies were performed using CITED2 without the heterologous transactivation domain. COS-1 cells were transiently transfected with pM-PPAR␣ and pcDNA 3 /CITED2 or empty pcDNA 3 vector (Fig. 3A). The GAL4 response element reporter (pFR-luciferase) was used to assess the effects of CIT-ED2 on PPAR␣ activity. CITED2 increased the PPAR␣-driven reporter activity when exogenous ligands were added (Wy-14,643 and ciprofibrate) and also increased activity in the absence of exogenous ligand. When a PPRE reporter was used (HD-Luciferase), essentially identical results were obtained (Fig. 3B). HepG2 cells were cotransfected with pBK/PPAR␣, pcDNA 3 /CITED2 (or control), and HD-Luciferase. In this study, CITED2 increased the PPRE-driven reporter activity in the presence of exogenous ligands Wy-14,643 and CLA. In these and in several other subsequent experiments, three different PPAR␣ activators were used to examine the potential for differential gene regulation in response to agonists with varying efficacy. The reason for this inclusion is the fact that one potential mechanism by which selective receptor modulators may have differing effects is through affecting the coactivators they recruit to the transcriptional complex.
The ability of CITED2 to act as a PPAR␣ coregulator was further characterized by examining reporter activity in the presence of increasing amounts of CITED2 plasmid (Fig. 4A). As the amount of CITED2 plasmid increased, there was a significant increase in both the exogenous ligand-dependent and -independent PPAR␣ activity. When examining the effects of CITED2 on the dose-response relationship for PPAR␣ activation by a potent ligand (Wy-14,643, Fig. 4B) there was an increase in activity for all doses. The EC 50 value was not significantly changed by CITED2 overexpression, indicating that ligand affinity was not affected.
The potential coactivating ability of CITED2 was also tested with PPAR␤ and PPAR␥ using the chimeric reporter system (pM-PPAR␤ and pM-PPAR␥). As shown in Fig. 5A, CITED2 acted as a strong coactivator of pM-PPAR␥ basal and ligandinduced activity, while pM-PPAR␤ was unaffected in the absence of exogenous ligand and was slightly repressed by CIT-ED2 in the presence of ligand. The enhancement of pM-PPAR␥ activity by CITED2 alone (Me 2 SO-treated) was not further increased by adding the ligand prostaglandin J 2 . Cotransfecting HepG2 cells with a PPRE-driven reporter and pBK/PPAR␣, -␤, or -␥ resulted in a similar subtype difference in the ability of CITED2 to regulate gene expression (Fig. 5B). Overexpressing CITED2 increased the ligand-inducible PPRE-reporter activity of PPAR␣ and PPAR␥ while having no effect on PPAR␤. The ability of CITED2 to affect PPAR␥ activity is currently being explored and will not be discussed in the present report.
Effects of CITED2 on PPAR␣ Ligand Responses in Hepatocytes-Since CITED2 is able to affect multiple transcriptions factors, including PPAR␣ and PPAR␥ in cell culture, whether these proteins are coexpressed in tissues was of interest. Analysis of total RNA from 10 different mouse tissues revealed that CITED2 mRNA was ubiquitously expressed with detectable levels in all tissues tested (Fig. 6). The MuSH cell line (13) and 3T3-L1 mouse preadipocytes also expressed relatively high levels of CITED2 mRNA (data not shown). Thus, CITED2 is found in tissues that express PPAR␣ and PPAR␥.
Undifferentiated 3T3-L1 preadipocytes were transiently transfected with a construct expressing a double-stranded RNA molecule to target CITED2 for selective inhibition (RNAi). Also included in the transfection were pM-PPAR␣ and GAL4 reporter vectors. Subsequently the cells were treated with PPAR␣ ligands and assayed for luciferase activity. As shown in Fig. 7, cells expressing reduced levels of CITED2 were less responsive to Wy-14,643 activation of PPAR␣. In addition, the basal level of PPAR␣ activity was reduced in the presence of the CITED2 RNAi plasmid.
Due to low transfection efficiency, to examine the effects of manipulated CITED2 protein levels on endogenous gene expression, it was necessary to generate cells that harbor a stable inhibition or overexpression of CITED2. Mouse immortalized hepatocytes were stably transfected with CITED2 RNAi (CIT-ED2Ϫ) or CITED2 cDNA (CITED2ϩ). Control lines were also FIG. 6. CITED2 is ubiquitously expressed in mouse tissues. Total RNA from 10 different mouse tissues and the SV40-transformed mouse hepatocytes was examined for CITED2 mRNA using reverse transcription-PCR. Equivalent amounts of total RNA were tested using primers designed for the 5Ј -end of mouse CITED2 mRNA. The graph is representative of three independent experiments. created using empty plasmids for either CITED2 construct. The relative levels of CITED2 protein were examined via Western blot (Fig. 8A). The CITED2ϩ cells contained ϳ1.5-fold the amount of this protein compared with their control counterparts (pcDNA 3 ), while the CITED2Ϫ cells expressed 3-fold less than their control cells (pSuper). As shown in Fig. 8B, modulating CITED2 expression in these cells altered their growth rate. When MuSH cells stably transfected with empty vector were treated with Wy-14,643, there was an increase in the number of viable cells relative to comparable Me 2 SO-treated cells. This Wy-14,643-mediated augmentation in cell number was slightly increased in the CITED2ϩ cells and decreased in the CITED2Ϫ cells. Interestingly, when compared with pcDNA 3 stably transfected cells, CITED2ϩ hepatocytes grew at a faster rate (5 times the number of cells after a 2-day period) in the absence of PPAR␣ ligand.
The effects of altered CITED2 expression on endogenous gene expression was examined (Fig. 9). The genes examined using real time PCR are shown in Table I and were chosen based on previous knowledge regarding PPAR␣ and CITED2 or were identified in gene expression microarrays (see below). Two statistical comparisons were made in these studies. To determine a ligand-dependent regulation (Fig. 9, A-E), Me 2 SOtreated cells were compared with the same cell line treated with a PPAR␣ ligand. To examine CITED2-dependent effects (Fig. 9, F-J), the control cell line was compared with either CITED2Ϫ or CITED2ϩ within the same treatment group.
Angiopoietin-like protein 4 (Angptl4) mRNA was increased by Wy-14,643 and CLA in both control cell lines but most dramatically by CLA (Fig. 9A). In both the CITED2ϩ and CITED2Ϫ cell lines, induction of Angplt4 mRNA was lost in response to Wy-14,643 and ciprofibrate, while the effects of CLA were not significantly altered. A significant difference in Angptl4 mRNA levels was seen comparing Wy-14,643-treated CITED2ϩ cells to their control cell line (Fig. 9F). HIF1␣ mRNA was decreased by Wy-14,643 in the CITED2ϩ hepatocytes (Fig.  9B), whereas CLA-and ciprofibrate-treated CITED2Ϫ cells contained less HIF1␣ mRNA than comparably treated pSuper cells (Fig. 9G). FOXc2 mRNA was increased by Wy-14,643 in both control cell lines and in the CITED2Ϫ cells but not in the CITED2ϩ cells (Fig. 9C). Relative to pcDNA 3 , CITED2ϩ cells treated with ciprofibrate had increased FOXc2 mRNA concentration (Fig. 9H). As shown in Fig. 9D, MAPK phosphatase 1 (MKP-1) mRNA was increased by CLA in pcDNA 3 , CITED2ϩ, and pSuper cells but not CITED2Ϫ hepatocytes. The basal (Me 2 SO-treated) expression of MKP-1 mRNA was significantly lower in CITED2Ϫ cells (Fig. 9I). The concentration of vascular endothelial growth factor D (VEGF-D) mRNA was decreased by all three PPAR␣ ligands tested in the CITED2Ϫ hepatocytes (Fig. 9E). CLA-and ciprofibrate-treated CITED2Ϫ cells contained less VEGF-D mRNA than comparably treated pSuper cells (Fig. 9J). The difference between the gene regulation seen with the pcDNA 3 compared with the pSuper stable cell lines must take into consideration that the rationale for creating two control cell lines was based upon the assumption that the presence of a vector alone may affect gene expression. Each empty vector control is meant to be coupled to the corresponding CITED2 cell line. Comparisons between pcDNA 3 and CIT-ED2ϩ or pSuper and CITED2Ϫ are based on the fact that the cell lines in each matched pair differ only on the presence of the CITED2-related insert. For all studies performed with stable cell lines, pooled stable transfectants representing multiple integration events and expression levels at low passage number post-transfection were used.
Effects of CITED2 Overexpression on Gene Expression in Hepatocytes-Due to the effects of CITED2 overexpression on basal gene expression and cell growth, microarray analysis was performed using a cross-hybridization between untreated empty vector control stable cells and untreated CITED2-overexpressing stable cells. The results from these arrays led to a small list of genes that were significantly affected by exogenous CITED2 (Table II and see supplemental data). Two of these genes, FOXc2 and MKP-1 were verified as being CITED2regulated using real time PCR (Fig. 9). DISCUSSION Similar to other nuclear receptors, PPAR␣ regulates gene expression by a variety of protein-protein interactions. In addition to its heterodimerization partner retinoid X receptor, PPAR␣ associates with Hsp70 (16) , Hsp90 (11, 17), liver X receptor (18), CCAAT/enhancer-binding protein ␣ (19), and growth hormone factor-1 (20 co-repressors (collectively referred to as coregulators) are an important class of receptor-interacting protein (21). PPAR coregulators have been identified by a variety of techniques including yeast two-hybrid assays, transient transfections, and biochemical means. Known PPAR␣ coregulators include the coactivators PPAR-binding protein/TRAP220/DRIP205 (22), steroid receptor coactivator-1 (3), p300 (24), CBP (25), and PPAR␥ coactivator-1 (26) and the corepressor RIP140 (27). In the present studies we demonstrated that CITED2 acts as a PPAR␣ coactivator capable of increasing transcriptional activity from a PPRE or from heterologous promoters. Similar to other coactivators, the CITED2 association occurred within the ligand binding domain of PPAR␣ (D-F), and the interaction was inducible by addition of a variety of ligands.
There are two characteristics of the PPAR␣/CITED2 association that are dissimilar to the classic NR/coactivator interaction. First, unlike steroid receptor coactivator-1 and PPAR␥ coactivator-1, the NR AF-2 domain did not appear to be utilized by PPAR␣ for CITED2 interaction. In fact, the strongest interaction occurred within the hinge region (D domain) of PPAR␣. Second, CITED2 was capable of acting as a ligand-independent coactivator; this is not often seen with classic coactivators.

TABLE I PCR primers
Full-length coding sequences for genes of interest were obtained from GenBank TM (National Center for Biotechnology Information). Suitable real time PCR primers were designed using PrimerExpress (Applied Biosystems). Standard PCR primers (CITED2 only) were designed using PrimerSelect (DNAStar, Madison, WI). All primers are listed 5Ј-3Ј.

Gene
Forward Reverse GenBank TM accession no.
TABLE II Genes significantly regulated by CITED2 overexpression Gene expression microarrays were performed as detailed under ''Experimental Procedures.'' The normalized ratios represent CITED2ϩ/pcDNA 3 stably transfected hepatocytes. Bold text indicates genes found to be linked using Pathway Assist (Version 2.01, Fig. 10). Increasing amounts of CITED2 increased reporter activity in the absence of exogenous ligand from a PPRE-or a GAL4driven reporter. CITED2 is dynamically regulated, being induced by hypoxia (15), shear stress (28), cytokines, and serum (29) and has a distinct pattern of expression during development (30). Thus, under conditions of high CITED2 expression, there may also be increased PPAR␣ activity irrespective of the addition of exogenous ligand. This is important due to the role PPAR␣ plays in lipid metabolism as well as cell cycle regulation.
The biological role of CITED2 has become an area of interest in regard to embryonic development and cancer. Disruption of the gene encoding CITED2 is embryonic lethal due to defects in the development of heart and neural tube (8) Table II) were examined using Pathway Assist (Version 2.01). The pathway was built by looking for common regulators of the genes shown in Table II, and the predominant cluster is depicted. The lines and arrows depict observations on regulation of gene expression (ϩ, increased expression; Ϫ, decreased expression) from the literature. The effects of CITED2 overexpression on the amount of mRNA in the microarray experiments are shown (black, significantly increased; gray, significantly decreased; white, no significant effect observed). Following creation of this cluster, the proteins in the gradient filled ovals were included (PPAR␣, PPAR␥, Angptl4, and HIF1␣), and the connections were determined by Pathway Assist or by manually adding (PPAR␣ and CITED2 interaction). EGF, epidermal growth factor; TGF, transforming growth factor; DCN, decorin; AQP5, aquaporin 5; TNF, tumor necrosis factor; IFG1R, insulin-like growth factor I receptor; IL2, interleukin 2; IGFBP2, insulin-like growth factor-binding protein 2; IFN, interferon; LTBP1, latent transforming growth factor-␤-binding protein 1; EPS15, epidermal growth factor receptor pathway substrate 15; TGM2, transglutaminase 2; UCHL1, ubiquitin carboxyl-terminal hydrolase L1; FIGF, c-fos-induced growth factor; GHRL, growth hormone receptor, long form; IL13R, interleukin 13 receptor; GH1, growth hormone 1; ADCY8, adenylate cyclase 8; TSA, trichostatin A; MGP, matrix ␥-carboxyglutamate protein.
formation in nude mice (29). In the present studies, overexpression of CITED2 led to an increase in cell growth rates, while the use of RNAi to repress this protein had little effect. The lack of effect of the knock-down studies may represent a difference in cell type (fibroblast versus hepatocytes) or the fact that incomplete repression of CITED2 levels was produced by RNAi techniques. Nonetheless it has now been demonstrated that overexpression of CITED2 can increase cell proliferation of hepatocytes as well as fibroblasts.
The role of CITED2 as a coregulator was originally described for HIF1␣ where CITED2 inhibited HIF1␣ transactivation by blocking its interaction with CBP/p300 (15). In addition to modulating the basic helix-loop-helix, Per-Arnt-Sim family member HIF1␣, CITED2 is a coactivator of basic-helix-spanhelix DNA-binding protein AP2 (33). The present studies showed that CITED2 can affect a third class of transcription factors, nuclear receptors, and that CITED2 acts as a coregulator of both PPAR␣ and PPAR␥. Virtually all of the of the CITED2 in the cell is associated with the CBP/p300 complex (15). Thus, CITED2 may affect the ability of a subset of CBP/ p300-associated transcription factors to regulate gene expression in either an inhibitory (HIF1␣) or stimulatory (PPAR and AP2) manner. Since CBP/p300 is an essential component of many transcriptional complexes, there is potential for CITED2 to affect the activity of a wide range of proteins. Interestingly our work has shown some specificity in that PPAR␤ activity was not affected by CITED2 overexpression even though CBP/ p300 is a coactivator of this receptor (34).
Since CITED2 mRNA is increased markedly by hypoxia or deferoxamine, it has been suggested to operate in a negative feedback loop with HIF1␣ (15). HIF1␣ transcribes CITED2 during hypoxia, and the CITED2 that accumulates during restoration of normal oxygen homeostasis will inhibit HIF1␣ transactivation and, in turn, its own production (35). The increase in HIF1␣ target gene products including VEGF-D (8) in CITED2 Ϫ/Ϫ fibroblasts is further evidence for this model (35). Increasingly there are connections being made between PPARs, HIF1␣, and angiogenesis. Hypoxia via HIF1␣ inhibits the expression of PPAR␣ (36) and decreases PPAR␣/retinoid X receptor binding to its response element (37). During adipocyte dedifferentiation PPAR␥ is decreased, while HIF1␣ mRNA expression increases (38). PPAR␥ ligands in particular have received considerable attention as angiogenesis inhibitors (23). As discussed below, several angiogenesis-related genes are regulated by PPAR␣ ligands. The CITED2⅐p300 complex may increase the ability of PPAR to decrease angiogenesis (via regulation of key target genes) while also inhibiting HIF1␣ and its ability to stimulate this process.
Several cell lines were generated to examine the role of CITED2 in endogenous gene expression and in the ability of PPAR␣ ligands to affect certain transcripts in hepatocytes. Previous work has shown that the model hepatocyte system (MuSH) was appropriate for examination of cell proliferation and PPAR␣-dependent gene expression (13). The genes chosen for examination included those potentially involved in angiogenesis and previously shown to be PPAR-regulated (Angplt4 and MKP-1) or CITED2-regulated from previous studies (VEGF-D and HIF1␣) or from the current microarray experiments (FOXc2). The present data confirms that Angplt4 and MKP-1 are regulated by PPAR ligands; FOXc2 can also be added to this list. In most instances CLA exerted a more dramatic effect on gene expression perhaps due to the fact that it is also a ligand for PPAR␥ (32). Although seen in the microarray experiments, using real time PCR we were unable to confirm VEGF-D as being affected solely by CITED2 expression in the hepatocyte model system.
The determination of an endogenous PPAR␣-regulated gene that uses CITED2 as a coactivator remains elusive. Evidence for such a gene would be an increase in ligand inducibility in the CITED2ϩ cells with an absence (or significant diminution) of this regulation in the CITED2Ϫ hepatocytes. Perhaps the gene that comes closest to this pattern is MKP-1 (Fig. 9I), although FOXc2 regulation by CLA and ciprofibrate may utilize CITED2 for coactivation. Interestingly the desired trend of activity in the four cell lines was observed with cell proliferation (Fig. 8B). This suggests that certain growth-regulatory genes may utilize a PPAR␣/CITED2-containing transcription complex.
An examination of the pathways affected by CITED2 supplementation revealed a small group of genes that are coordinately regulated (Fig. 10). Initially genes found to be regulated in the microarray studies were utilized to create the cluster, and subsequently the key proteins in the present study (PPAR␣ and -␥, HIF1␣, and Angplt4) were included. A central signaling molecule in this pathway is tumor necrosis factor, which has been implicated in the regulation of five of the genes found to be affected by CITED2 overexpression, and it in turn affects the activity of PPAR and HIF1␣. In addition to FOXc2 and MKP-1, tumor necrosis factor has previously been reported to increase the expression of WNT10A and decorin while repressing aquaporin. Epidermal growth factor, interleukin 2, and transforming growth factor-␤ also figure prominently in this pathway. Thus, it is attractive to speculate that CITED2 affects, in addition to PPAR, a common factor in the tumor necrosis factor, epidermal growth factor, and transforming growth factor-␤ signaling cascades. PPAR␣ is a regulator of MKP-1 and Angptl4, and the present studies suggest it also affects FOXc2 mRNA accumulation. HIF1␣ regulates Angtl4 and is negatively regulated by CITED2. It is evident from examining Fig. 10 that a reason for the difficulty in finding an obvious endogenous gene being regulated by a PPAR␣⅐CITED2 complex is the multiple, competing pathways that are affecting this cluster of proteins.
The results described here show that the CBP/p300-interacting protein CITED2 is a coactivator of PPAR␣-mediated transcriptional activity. The coactivating ability of CITED2 is present over a wide range of known peroxisome proliferators, and this protein is also a basal coactivator for PPAR␣. This work has also shown that overexpression of CITED2 affects cell proliferation in hepatocytes, a phenomenon previously observed in fibroblasts. Thus, there may be a significant physiological role for CITED2 protein in the transcriptional regulation of PPAR␣ target genes in the liver in particular for those involved in the cell cycle.