Critical Role for Transcription Coactivator Peroxisome Proliferator-activated Receptor (PPAR)-binding Protein/TRAP220 in Liver Regeneration and PPARα Ligand-induced Liver Tumor Development*

Disruption of the gene encoding for the transcription coactivator peroxisome proliferator-activated receptor (PPAR)-binding protein (PBP/TRAP220/DRIP205/Med1) in the mouse results in embryonic lethality. Here, we have reported that targeted disruption of the Pbp/Pparbp gene in hepatocytes (PbpΔLiv) impairs liver regeneration with low survival after partial hepatectomy. Analysis of cell cycle progression suggests a defective exit from quiescence, reduced BrdUrd incorporation, and diminished entry into G2/M phase in PbpΔLiv hepatocytes after partial hepatectomy. PbpΔLiv hepatocytes failed to respond to hepatocyte growth factor/scatter factor, implying that hepatic PBP deficiency affects c-met signaling. Pbp gene disruption also abolishes primary mitogen-induced liver cell proliferative response. Striking abrogation of CCl4-induced hepatocellular proliferation and hepatotoxicity occurred in PbpΔLiv mice pretreated with phenobarbital due to lack of expression of xenobiotic metabolizing enzymes necessary for CCl4 activation. PbpΔLiv mice, chronically exposed to Wy-14,643, a PPARα ligand, revealed a striking proliferative response and clonal expansion of a few Pbpfl/fl hepatocytes that escaped Cre-mediated gene deletion in PbpΔLiv livers, but no proliferative expansion of PBP null hepatocytes was observed. In these PbpΔLiv mice, none of the Wy-14,643-induced hepatic adenomas and hepatocellular carcinomas was derived from PBPΔLiv hepatocytes; all liver tumors developing in PbpΔLiv mice maintained non-recombinant Pbp alleles and retained PBP expression. These studies provide direct evidence in support of a critical role of PBP/TRAP220 in liver regeneration, induction of hepatotoxicity, and hepatocarcinogenesis.

Transcription cofactors/coregulators consist of corepressors, coactivators, and coactivator-or corepressor-associated proteins, which participate in nuclear receptor-directed tran-scription (1)(2)(3). The functional significance for the existence of Ͼ200 nuclear receptor cofactors is not readily evident, but there is an increasing recognition of the general importance of some of these molecules in gene expression, embryogenesis, cell growth, and oncogenesis, as well as energy and xenobiotic metabolism (1,3). Emerging gene knock-out mouse models show that some of the coactivators that directly bind to transcription factors to enhance gene expression are essential for embryonic growth and survival (see Ref. 3 for review). For example, the disruption of a coactivator gene such as peroxisome proliferated-activated receptor (PPAR) 2 -binding protein (PBP; also known as TRAP (thyroid hormone receptor-associated protein) 220/DRIP (vitamin D 3 receptor-interacting protein) 205)/Med1 (Mediator 1)) is embryonically lethal between gestational days 11.5 and 12.5, implying that this coactivator is widely involved in the transcriptional activity of many transcription factors (4,(5)(6)(7). To define the in vivo role of this coactivator, we used conditional mutagenesis in mice and found that deletion of the Pbp/Pparbp gene in liver parenchymal cells (Pbp ⌬Liv ) results in the abrogation of PPAR␣ ligand-induced pleiotropic effects, indicating that PBP is essential for PPAR␣ signaling (8). Deletion of the Pbp gene in liver also abolished the responses induced by the constitutive androstane receptor (CAR) activators, phenobarbital or 1,4-bis[2-(3,5-dichloropyridyloxy)]benzene (TCPOBOP) and of acetaminophen-induced hepatotoxicity (9,10).
In this study, we have explored the potential role of PBP in liver regeneration and liver tumor development to test the hypothesis that PBP deficiency in hepatocytes affects the function of many genes that control these complex and vital processes (11)(12)(13). We have demonstrated a severe impairment of liver regeneration in Pbp ⌬Liv mice subjected to partial hepatectomy. Analysis of immediate early gene expression and cell cycle progression indicates a deficit in exiting from G 0 after partial hepatectomy in PBP mutant livers. We have also found that conditional PBP mutant mice do not respond to the cellproliferating response of mice to TCPOBOP, a well known primary mitogen (14,15). Impairment of the cell-proliferative response of Pbp ⌬Liv hepatocytes leads to enhanced proliferative expansion of an occasional PBP ϩ/ϩ hepatocyte present in these conditional mutant livers, in response to the PPAR␣ ligand Wy-14,643. Accordingly, none of the liver tumors developing in Pbp ⌬Liv mice chronically treated with Wy-14,643 was derived from cells with Pbp null genotype. All liver tumors in these conditional null mutant livers exhibited PBP positivity, implying that PBP is essential for hepatocarcinogenesis.

EXPERIMENTAL PROCEDURES
Generation of PBP Conditional Null Mutation in Liver (Pbp ⌬Liv ), Partial Hepatectomy, and Treatment with CAR and PPAR␣ Agonists-Homozygous mutant mice lacking PBP in hepatocytes (Pbp ⌬Liv ) were generated as described elsewhere (8). Mice were housed in a pathogen-free animal facility under a 12-h light/12-h dark cycle and maintained on standard rodent chow and water ad libitum. Partial hepatectomy was performed under anesthesia to remove 70% of the hepatic mass (16). TCPOBOP was administered intraperitoneally at a single dose of 3 mg/kg body weight. Wy-14,643 (0.125% weight/weight) was given in powdered diet for 1 week, 4 weeks, and 3 months. For long term studies, Wy-14,643 was administered in the diet at a concentration of 0.05% (weight/weight) for one year. To induce liver necrosis, CCl 4 (400 mg/kg body weight) was injected intraperitoneally either into phenobarbital-pretreated or untreated mice (100 mg/kg intraperitoneally daily for 3 days (Sigma). To assess hepatocyte proliferation, BrdUrd (0.5 mg/ml) was administered in drinking water for 3 days and given a single intraperitoneal dose (100 mg/kg body weight) 2 h before killing. Mice were killed by cervical dislocation, and blood collected from the inferior vena cava was used for assaying serum alanine aminotransferase (ALT) activity using an ALT assay kit (Sigma). The Northwestern University Animal Care and Use Committee approved all animal studies.
Liver slices were fixed in 10% formalin or 4% paraformaldehyde, processed for embedding in paraffin, sectioned, and stained with either hematoxylin and eosin or processed for immunohistochemical localization of PBP, BrdUrd, or peroxisomal enoyl-CoA hydratase/L-3-hydroxyacyl-CoA dehydrogenase (L-bifunctional enzyme; L-PBE), the second enzyme of the inducible peroxisomal fatty acid ␤oxidation system (8,9). The antibodies used were anti-PBP (catalog number sc-5334, Santa Cruz Biotechnology), anti-BrdUrd (BD Biosciences), and anti-L-PBE (from T. Hashimoto). The volume occupied by large hepatocytes was estimated using Scion Image software.
Microarray Approach-Total RNA isolated from livers using TRIzol reagent (Invitrogen) was used for preparing biotin-labeled cRNA and purified, fragmented, and hybridized to 430 2.0 arrays (Affymetrix, Santa Clara, CA) as described (17). After hybridization, bound cRNA was fluorescently labeled using R-phycoerythrin streptavidin (Molecular Probes), and the fluorescence was intensified by the antibody amplification method. Data were collected and analyzed using the Agilent 2100 expert software.
Northern and Immunoblot Analyses-Total RNA isolated from liver was glyoxylated, separated on 0.8% agarose gel, transferred to nylon membrane, and probed with selected cDNAs. Quantitative reverse transcription-PCR was performed using the primers (forward primer) 5Ј-GTGCAAGGAGGCCCTAG-TGAAGTC-3Ј and (reverse primer) 5Ј-CGGTCTTGAATTC-GGATACCTTCG-3Ј specific for the c-met proto-oncogene (18). Whole liver proteins were subjected to 10% SDS-PAGE, transferred to nitrocellulose membrane, and immunoblotted using rabbit polyclonal antibodies against cyclins. The rabbit polyclonal antibody against phospho-retinoblastoma protein (Ser-780) and phospho-c-Jun (Ser-63) were from Cell Signaling Technology.
Hepatocyte Isolation, Culture, and Flow Cytometry-Hepatocytes were isolated from mouse livers using the collagenase perfusion method (19). Hepatocyte motility in response to hepatocyte growth factor/scatter factor (HGF/SF) was assessed as described previously (20). For flow cytometric analysis, liver cells were stained with propidium iodide (Invitrogen), and fluorescence was measured using a Beckman Coulter XL100 flow cytometer equipped with a 488-nm laser. Cell doublets were excluded from the analysis performed using Modfit software (Verity Software House, Topsham, ME).

RESULTS AND DISCUSSION
PBP Is Required for Normal Liver Regeneration-Increased mortality was observed in Pbp ⌬Liv mice usually between 2 and 20 h following partial hepatectomy. The liver-to-body weight ratio, reflective of the regeneration index after partial hepatectomy, increased progressively in wild-type mice, whereas conditional PBP mutant mice failed to show similar increases (Fig.  1A). Liver cell proliferation, as assessed by BrdUrd incorporation, was minimal in Pbp ⌬Liv mice at all times after partial hepatectomy, and this inhibition persisted until 120 h after surgery (Fig. 1B). In contrast, in Pbp fl/fl wild-type controls, BrdUrd incorporation was prominent, with a marked increase in the number of labeled hepatocyte nuclei, with a peak labeling at 36 and 48 h after partial hepatectomy (Fig. 1, B and C). In these livers, cessation of cell proliferation occurred by 120 h after surgery (Fig. 1B). As expected, all hepatocyte nuclei in Pbp fl/fl wild-type livers stained positively for PBP (Fig. 1E). In contrast, PBP expression in Pbp ⌬Liv mouse livers was absent in hepatocytes, and these PBP-negative hepatocytes, in general, appeared smaller in size as compared with hepatocytes in Pbp fl/fl wildtype littermate controls (Fig. 1, D and F). In the Pbp ⌬Liv mouse, an occasional hepatocyte in the centrilobular region of the liver that escaped Pbp gene disruption exhibited PBP staining in the nucleus (Fig. 1F). These cells with abundant cytoplasm are conspicuous in their appearance as compared with the smaller sized PBP null hepatocytes (Fig. 1, D and E). In these Pbp ⌬Liv mouse livers, BrdUrd staining was seen only in these rare PBPexpressing hepatocytes (Fig. 1D), which corresponded to cells expressing PBP (Fig. 1F). Oil red O-stained liver sections obtained 24 -72 h after partial hepatectomy showed a moderate degree of macrovesicular steatosis in Pbp ⌬Liv mice as compared with minimally visible microvesicular steatosis associated with early stages of normal regenerative response in wild-type mice (data not shown). The exaggerated hepatic steatosis in the Pbp ⌬Liv mouse suggests that fatty acids influxed into the residual liver in partially hepatectomized mice are oxidized efficiently by the fatty acid-sensing system in PBP ϩ/ϩ -regenerating livers, but this system appears to be defective in Pbp ⌬Liv mouse livers (21,22). It is well known that PPAR␣ functions as a lipid sensor in liver, and in the absence of this nuclear receptor, there is increased hepatic steatosis in response to starvation (21). It is now known that PBP is essential for PPAR␣ function (8), and the absence of PBP therefore likely abrogates the PPAR␣ lipidsensing mechanism after partial hepatectomy. The increased mortality observed in Pbp ⌬Liv mouse livers after partial hepatectomy is attributed to the failure of hepatocytes to sense and metabolize fatty acids transported into the residual liver.
Gene Expression Changes in Control and PBP-deficient Livers Following Partial Hepatectomy-We examined the changes in the gene expression profile in the liver of the 0-h control and 3 h after partial hepatectomy by using the microarray approach (14,23). The data revealed that ϳ25 genes are up-regulated 6-fold or higher 3 h after partial hepatectomy in Pbp fl/fl mice as compared with Pbp ⌬Liv mice ( Table 1). Many of the genes up-regulated in the Pbp fl/fl mouse liver in response to partial hepatectomy are immediate early genes known to participate in cell cycle, cell growth, apoptosis, and signal transduction (Table 1; Fig. 2A). These include insulin-like growth factor 1 (Igf1), IGFbinding protein 1 (IGFBP1), E2F transcription factor, metallothionein 1 and 2, follistatin, heparin-binding epidermal growth factor, nuclear receptors Nurr1 and NOR1, growth arrest and DNA damage-inducible 45␥ (GADD45␥), tumor necrosis factor receptor, interleukin-1 receptor, and suppressor of cytokine signaling 3 (SOCS3), among others (13,14,(23)(24)(25)(26). Northern blot analysis confirmed the elevated levels of expression of SOCS3, IGFBP1, and follistatin in Pbp fl/fl livers 3 h post-hepatectomy ( Fig. 2A). For example, follistatin, which is known to be a positive regulator in liver regeneration after partial hepatectomy (26), showed higher levels of expression in PBP ϩ/ϩ mice but not in Pbp ⌬Liv mice ( Fig. 2A). Although we did not ascertain the levels of induction of these genes during the later stages of liver regeneration, with IGFBP1, we noted sustained higher levels of expression of this gene up to 48 h in Pbp fl/fl livers (Fig. 2C). IGFBPs are important positive regulators of liver regeneration PBP immunohistochemical staining in Pbp fl/fl (E) and Pbp ⌬Liv (F, arrows) shows that, in Pbp ⌬Liv mouse liver, BrdUrd labeling is confined to an occasional Pbp fl/fl liver cell (D, arrows) in these liver-conditional nulls. G, lack of PBP causes impairment of hepatocyte motility in response to HGF. Linear scrape wounds were made in subconfluent monolayers of primary hepatocytes 4 h after plating. They were allowed to heal over a 48 h period in serum-free medium in the presence of HGF. Healing did not occur in Pbp ⌬Liv mouse hepatocytes suggesting a lack of c-met signaling in these cells. H, quantitative PCR determination of c-met mRNA levels in hepatocytes isolated from Pbp fl/fl and Pbp ⌬Liv livers.  (25), and these immediate early genes are involved in the priming of liver cells after partial hepatectomy for DNA synthesis and replication (11)(12)(13). Lower levels of these cell cycle and cell growth regulatory factors in Pbp ⌬Liv mouse liver reflect the loss of hepatocellular proliferative response following partial hepatectomy. In Pbp ⌬Liv mouse liver, 13 genes were up-regulated 6-fold or higher 3 h following partial hepatectomy as compared with wild-type mice (Table 1). These include transforming growth factor-␤ (TGF-␤) type II receptor (tgfbr2) and Smad7 known to inhibit cell proliferation. Smad7 induces G 0 /G 1 cell cycle arrest by inhibiting the expression of G 1 cyclins (27). TGF-␤, a potent inhibitor of cell proliferation, plays a role in limiting liver-regenerative response at later periods after partial hepatectomy (11,12). We found that its receptor tgfbr2 is expressed at higher levels in Pbp ⌬Liv mouse liver as early as 3 h after partial hepatectomy, implying that the active TGF-␤ sig-naling may interfere with liver regeneration in Pbp ⌬Liv mice (Table 1 and Fig. 2B).

PBP/TRAP220/MED1 in Liver Regeneration and Liver Tumors
To further understand the mechanisms responsible for defective liver regeneration, we analyzed the expression levels of proteins involved in cell cycle progression. Expression of cyclin D1 (which correlates with exit from G 0 ) and of cyclin E (which controls S-phase entry) peaked between 30 and 48 h after partial hepatectomy in Pbp fl/fl mice, but these changes were diminished and delayed in their onset in Pbp ⌬Liv mice (Fig.  2D). Because cyclin D1 is thought to stimulate entry into S-phase by phosphorylating phospho-Rb (23), we examined the changes in the phosphorylation state of phospho-Rb and its family member p107 (Fig. 2D). In the Pbp fl/fl mouse liver, the levels of phospho-Rb phosphorylation and p107 were higher than in the Pbp ⌬Liv mouse. The expression of cyclins D and E is generally under the control of immediate early genes, such as c-jun, and their expression reflects exit from the G 0 phase (28). Cyclins A and B regulate S/G 2 transition and M phase progression, respectively, of the cell cycle (28). In Pbp fl/fl livers, the levels of these two proteins increased, and such increases were not readily apparent in Pbp ⌬Liv mouse liver after partial hepatectomy (Fig. 2D). The phosphorylated c-jun reached peak levels ϳ6 h after partial hepatectomy in Pbp fl/fl mouse liver, and this increase was not prominent in the Pbp ⌬Liv mouse liver. Finally, flow cytometric analysis revealed ϳ86% liver cells in G 0 /G 1 phase and ϳ8% in G 2 /M in Pbp ⌬Liv mouse livers 36 h after partial hepatectomy. This was in contrast to 54% G 0 /G 1 and 40% G 2 /M in the wild-type control (data not shown). These results, together with changes in immediate early gene expression and differences in BrdUrd incorporation, establish that disruption of Pbp gene in liver results in reduced G 0 /G 1 transition and a block in entry to the G 2 /M phase of the cell cycle.
We also noted that primary hepatocytes obtained from Pbp ⌬Liv mice failed to migrate in response to HGF/SF in a standard wound-healing assay (Fig. 1G). HGF/SF exerts its effects through its receptor c-met, a proto-oncogene, and defective c-met signaling has been implicated in impaired liver regeneration and in the etiology and progression of certain human cancers (18,20,29,30). Quantitative PCR data revealed a reduction in c-met mRNA level in Pbp ⌬Liv hepatocytes, suggesting a possible defect in HGF/c-met signaling (Fig. 1H). The HGF/c-met signaling pathway is important for liver regeneration, and it appears that PBP deletion affects this signaling mechanism.
Impaired Hepatocellular Proliferation in Pbp ⌬Liv Mouse in Response to a Primary Mitogen-Hepatomitogen TCPOBOP, a ligand and activator for the nuclear receptor CAR, induces robust hepatocellular proliferation (14,15). We undertook a detailed examination of hepatocellular proliferation in Pbp fl/fl and Pbp ⌬Liv mice 24, 30, 36, 48, and 96 h after a single intraperitoneal injection of TCPOBOP. Increases in liver to body weight ratio as well as BrdUrd labeling indices were evident in Pbp fl/fl mice commencing at ϳ24 h after TCPOBOP injection, but Pbp ⌬Liv mice showed almost no such alterations (Fig. 3, A  and B). We also examined the effect of TCPOBOP on the inducibility of mRNA expression of certain CAR target genes in liver between 0 and 96 h after injection (Fig. 3C). Deletion of Pbp gene significantly attenuated CAR-mediated induction of its

PBP/TRAP220/MED1 in Liver Regeneration and Liver Tumors
target genes after TCPOBOP treatment, whereas TCPOBOP treatment resulted in the induction of hepatic CYP3A11, CYP2B10, CYP1A2, and GST mRNAs that occurred in Pbp fl/fl mice (31)(32)(33). Reduction in Mrp3 (multi drug resistance protein 3) and CAR mRNA levels was noted in TCPOBOP-treated Pbp ⌬Liv mice (Fig. 3C). To further understand the deficit in liver proliferation after TCPOBOP injection, we analyzed the cell cycle-associated proteins. In Pbp fl/fl mice, both cyclin A and cyclin D1 appeared to be up-regulated in liver at 48 and 96 h after TCPOBOP injection but not in Pbp ⌬Liv mice. E2F, which controls cell growth, showed a slightly higher level of expression at 30, 36, and 48 h after TCPOBOP treatment in Pbp fl/fl mice than in Pbp ⌬Liv mice. In the livers of PBP ϩ/ϩ mice, the level of phospho-Rb phosphorylation was higher than Pbp ⌬Liv mice (Fig. 3D). These observations suggest that the absence of PBP in hepatocytes affects the function of CAR and that CAR ligand TCPOBOP does not elicit hepatocellular proliferative response and fails to induce xenobiotic metabolizing enzymes.
PBP LivϪ/Ϫ Mice Exhibit Resistance to CCl 4 -induced Hepatotoxicity-To further address the issue of hepatic regeneration and injury in Pbp ⌬Liv mice, we used a single injection of CCl 4 to induce hepatic necrosis to evaluate the regenerative response. CCl 4 is metabolized by CYP3A, CYP2B, and possibly by CYP1A2 to form a toxic and highly reactive trichloromethyl radical, CCl 3 (31). In Pbp fl/fl mice not pretreated with phenobarbital, CCl 4 induced liver cell proliferation and only a mild degree of hepatic necrosis in the centrilobular regions at 2 days with resultant increases in serum ALT levels (Fig. 4, A-C). Pretreatment with phenobarbital resulted in profound amplification of CCl 4 -induced necrosis in Pbp fl/fl livers with significant elevation of serum ALT levels ( Fig. D-F). Hepatocellular proliferation in these Pbp fl/fl mice was higher starting at 1 day after CCl 4 injection, and this increase persisted for 7 days (Fig. 4D). Complete repair of CCl 4 -induced liver injury was apparent at 7 days in these wild-type mice (Fig. 4, C and F). In contrast, increases in hepatocellular proliferation were not evident in Pbp ⌬Liv mice given CCl 4 without or with phenobarbital pretreatment (Fig. 4, A and D). Also of interest is that CCl 4 failed to induce hepatocellular necrosis in Pbp ⌬Liv mice even with phenobarbital pretreatment (Fig.  4F). These results indicate that, in the absence of PBP, phenobarbital apparently fails to increase CCl 4 metabolism, resulting in the abrogation of hepatotoxicity. We previously reported that PBP is involved in the regulation of hepatic CAR function and the induction of drug-metabolizing enzymes and that PBP deficiency in liver abrogates acetaminophen hepatotoxicity (9,10). Previously, we reported that the absence of PBP in liver cells prevents translocation of the xenobiotic receptor CAR into the hepatocyte nucleus under in vivo and in vitro conditions, even in the presence of excess exogenous CAR (9, 10). CAR target gene transcription requires the presence of CAR in the hepatocyte nucleus, and coactivator PBP, by virtue of its pivotal role as an anchor for TRAP/DRIP/ ARC/Mediator complex, is required  for the retention and concentration of CAR in the nucleus to elicit target gene transcription (10). Regulation of CAR activity by the transcription coactivator PBP may be an important clinical strategy for preventing or minimizing drug-induced liver injury. Activators and inhibitors of nuclear receptor coactivator PBP function may serve as useful tools to modulate drug-induced liver injury. PBP Is Necessary for Wy-14,643-induced Hepatocyte Proliferation and Tumorigenesis-Wy-14,643, a potent peroxisome proliferator and a rodent liver carcinogen, exerts its effects by activating the nuclear receptor PPAR␣ (34 -36). In view of the refractoriness of Pbp ⌬Liv hepatocytes to liver regeneration induced by partial hepatectomy and TCPOBOP treatment, we tested whether PBP is required for Wy-14,643-induced hepatocyte proliferation and more importantly hepatic tumorigenesis. Administration of Wy-14,643 to Pbp ⌬Liv mice resulted in the proliferation of liver cells with large cytoplasms, and these cells were positive for PBP. BrdUrd labeling as well as PBP and L-PBE immunostaining established that these proliferating cell clusters, which appeared to expand rapidly and progressively in response to Wy-14,643, are indeed the residual hepatocytes that escaped PBP gene deletion (Fig.  5A). Of interest is that these PBPpositive cells exhibit a profound increase in their ability to proliferate in a milieu where the majority of hepatocytes do not express PBP and are thus refractory to the mitogenic stimulus (Fig. 5A). These results suggest that few PBP-expressing hepatocytes present in Pbp ⌬Liv mouse exhibit profound growth advantage, whereas the PBP-negative hepatocyte population fail to show cell proliferation and induction of peroxisomal ␤-oxidation enzymes (35). Immunohistochemical staining for L-PBE, the second enzyme of the fatty acid oxidation system (35), demonstrates that L-PBE induction occurs only in PBP-expressing hepatocytes and not in PBP-negative liver cells. These data clearly show that PBP is required for Wy-14,643-induced proliferative expansion of hepatocytes and for the induction of the L-PBE enzyme. Next, we estimated the relative area occupied by large Pbp-positive hepatocytes using hematoxylin-and eosin-stained sections of Pbp ⌬Liv mouse liver at 1 week, 4 week, and 3 months of treatment with Wy-14,643 by using Scion Image software (Fig. 5C). The large hepatocytes in the Pbp ⌬Liv mouse without treatment showed only 0.58 Ϯ 0.06% but at the 3-month treatment period increased to 58 Ϯ 17%. In view of this dramatic differential response in Pbp ⌬Liv mouse livers, we wanted to determine whether PBP is required for mouse liver tumor development in response to Wy-14,643, a non-genotoxic hepatocarcinogen (35). After 52 weeks of Wy-14,643 (0.05% w/w) treatment, 15 Pbp ⌬Liv mice and 15 Pbp fl/fl were killed, and their livers were analyzed for the presence of tumors. The livers of both Pbp ⌬Liv and Pbp fl/fl mice revealed multiple, grossly visible tumors that were randomly distributed among all liver lobes (Fig. 5B). The liver tumor load was similar in both groups (8.63 Ϯ 2.63 in Pbp fl/fl versus 7.8 Ϯ 2.68 in Pbp ⌬Liv mice). We examined ϳ50 tumors from each group (Pbp fl/fl and Pbp ⌬Liv mice) for the expression of PBP to ascertain whether any of the tumors developing in Pbp ⌬Liv mice were derived from PBP-negative hepatocytes (Fig. 5B, a and b). Interestingly, all adenomas and hepatocellular carcinomas that developed in Pbp ⌬Liv mice were PBP-positive. The surrounding non-tumor portions of liver in Pbp ⌬Liv mice did not express PBP (Fig. 5B, c-f). None of the tumors developing in Pbp ⌬Liv mice were PBP-negative, imply- ing that Pbp ⌬Liv hepatocytes are resistant to malignant change. All tumors in the Pbp fl/fl mouse liver stained positive for PBP, and in these livers, hepatocytes in all non-tumor areas also expressed PBP (Fig. 5B, e). From these results, we conclude that long term stimulation with Wy-14,643 induces rapid and sustained proliferation of Pbp fl/fl (Pbp-positive) hepatocytes in Pbp ⌬Liv mice, and because all of these Pbp-positive hepatocytes respond to the inductive effects of this PPAR␣ ligand on fatty acid oxidation systems, there is an additional burden of oxidative stress. These two features impart the potential to acquire neoplastic changes. In contrast, dramatic increases in Pbp-positive hepatocytes were not seen in Pbp ⌬Liv mouse livers due to the relatively short treatment period (5 days after partial hepatectomy; 4 days treatment with TCPOBOP). It should be noted that only a few cells in the Pbp ⌬Liv mouse liver are Pbp-positive. This paucity further underscores the importance of the magnitude of cell proliferation of these Pbp fl/fl cells to overwhelm the liver. Of note is that PBP-positive hepatocytes in Pbp ⌬Liv mouse liver are very rare, accounting for Ͻ1.0% of the hepatocyte population. Nonetheless, these cells seem to be overtly sensitive for liver tumor development in a milieu where the total population is non-responsive to PPAR␣ ligand-induced hepatocellular proliferation and the induction of reactive oxygen species generated by fatty acid-metabolizing enzymes as a result of PBP deficiency (34 -36). It is known that Ppar␣-null mice fed Wy-14,643 in the diet for 11 months fail to develop tumors, whereas 100% of wild-type (PPAR␣-positive) mice fed this compound develop multiple liver tumors, thus underscoring a critical role played by PPAR␣ in peroxisome proliferator-induced hepatocarcinogenesis (34,36). Thus, it is reasonable to conclude that Pbp ⌬Liv mice mimic Ppar␣null mice with regard to the phenotypic changes induced by PPAR␣ ligands (35). Taken together, these data indicate that PBP is essential for Wy-14,643-induced hepatocyte proliferation and tumorigenesis. However, it would be important to ascertain whether the PBP null hepatocytes function as targets for genotoxic chemical carcinogens such as diethylnitrosamine (37).

CONCLUSION
Transcription coactivator PBP/TRAP220/MED1 functions as an anchor for TRAP⅐DRIP⅐Mediator complex, and disruption of this gene in the mouse results in embryonic lethality. Cre/loxP-mediated gene targeting showed that PBP is essential for the function of nuclear receptors PPAR␣ and CAR in liver. A critical role for PBP in the xenobiotically induced transcriptional activation of certain nuclear receptors in liver is exemplified by its requirement in PPAR␣and CAR-regulated gene transcription (8,9). Responses of Pbp ⌬Liv mice to PPAR␣ and CAR ligands are similar to mice lacking the respective receptor (32,36). The results described here provide unequivocal evidence for the essential role of PBP in liver regeneration induced by partial hepatectomy. Mice lacking PBP in liver cells exhibited no DNA synthesis and failed to exit the G 0 /G 1 phase of the cell cycle. Hepatocellular regeneration was also not seen in these PBP null livers when exposed to CAR ligand TCPOBOP. Of considerable interest is that, although Pbp ⌬Liv mice devel-oped liver tumors when chronically treated with PPAR␣ ligand Wy-14,643 (a non-genotoxic hepatocarcinogen (34 -36)), none of the tumors originated from PBP null hepatocytes. All liver tumors expressed PBP, and their rapid growth and sensitivity was attributed, in part, to the predominantly PBP-negative milieu in these Pbp ⌬Liv livers. These observations implicate, for the first time, the involvement of a transcription coactivator in hepatocellular regenerative response and in hepatocarcinogenesis. These Pbp fl/fl mice should provide an opportunity to explore the role of this coactivator in controlling receptor-specific target gene expression in a cell-specific need-based demand.