Differentiation of Rat Oval Cells after Activation of Peroxisome Proliferator-activated Receptor α43 Free

PPs²are a structurally diverse group of molecules with widespread use as hypolipidemic drugs (WY14,643, clofibrate), plasticizers (di-ethylhexyl phthalate), herbicides, or solvents. In rodents, PPs act as nongenotoxic carcinogens and induce liver tumors (reviewed in Ref.1); however, the precise mechanisms responsible for their carcinogenicity have not yet been elucidated. In rodents, PPs cause an increase in the number of hepatic peroxisomes and induce peroxisomal enzymes involved in fatty acid metabolism such as FACO or CYP4A1(1). These effects are mediated through the activation of the nuclear receptor, PPARα (2, 3, 4). PPARs, including the subtypes α, β, and γ, are members of the steroid receptor superfamily and act as ligand-activated transcription factors(5). PPARα heterodimerizes with the retinoid X receptorα and binds to a peroxisome proliferator-responsive element located in the promoter of responsive genes (4, 6). The requirement of PPARα for the hepatocarcinogenicity of PPs has been demonstrated using PPARα homozygous knockout mice; these mice do not show any of the typical hepatic responses associated with treatment with PPs, and dietary WY14,643 fails to induce liver tumors in this model (7). However, the precise role played by PPARα in the liver during PP-induced hepatocarcinogenesis is unclear, and the cellular origin of PP-induced HCCs is still open to debate. In rodents,PPs have been shown to act as tumor promoters (8). They induce basophilic neoplastic nodules and HCCs that do not express the placental form of GST (πGST) or GGT (9, 10). This is in contrast with the characteristics of the neoplastic lesions induced by a variety of structurally unrelated carcinogens, including other tumor promoters (11).

In the present study, we investigated the response of rat oval cells to in vivo treatment with the PP WY14,643, the prototype PPARα activator (12). Oval cells consist of a heterogeneous population of small cells with an ovoid nucleus that originate in hepatic portal areas. Oval cells proliferate after a parenchymal loss in situations where the ability of the surviving hepatocytes to regenerate is lost or severely impaired because of a chemical or viral insult (reviewed in Ref. 13). Although still controversial, the involvement of oval cells during liver carcinogenesis is supported by a number of observations(14): (a) Tg737, a liver tumor suppressor gene, was shown to control the differentiation of oval cells(15, 16); (b) oval cell proliferation has been observed in many models of rodent liver carcinogenesis as well as in hepatitis B virus-associated hepatocarcinogenesis in people (reviewed in Refs. 17 and 18); and (c)malignantly transformed rodent oval cell lines are able to induce cholangiocarcinomas and HCCs when injected into nude mice or newborn rats (19, 20, 21). Such findings suggest a common cell origin for these two malignancies, a hypothesis supported by the finding that the oval cell compartment shows features of both hepatocytes and bile duct cells. Indeed, a subpopulation of oval cells coexpresses markers of adult and fetal hepatocytes such as AFP, aldolase A, aldolase B, and albumin, as well as markers of bile duct cells such as GGT and CK19(reviewed in Refs. 13 and 22). Moreover,because oval cells can differentiate into both hepatocytes and normal duct cells (23, 24), they act as bipotential progenitors for the two main hepatic lineages and are considered to be closely related to liver stem cells (14).

PPARα is strongly expressed in mature hepatocytes; however, its role in oval cells has not been studied. Because PPARγ is involved in the differentiation of preadipocytes (25), we thought that PPARα might play a similar role in the differentiation of oval cells. In support of this idea, prolonged in vitro treatment of immortalized oval cell lines with the PP clofibrate leads to the induction of peroxisome proliferation (26). In the present study, we investigated the PPARα-mediated response of primary oval cells isolated from rats fed a CDE diet [a regimen commonly used for the induction of oval cell proliferation in rodents (27)]in the presence or absence of treatment with WY14,643. Histopathological and immunohistochemical analyses of this in vivo study will be described elsewhere.³We measured the levels of expression of PPARα and those of PPARα-regulated genes encoding FACO and CYP4A1 in rat oval cells. We report the effects of chronic in vivo treatment with the PPARα activator, WY14,643, on the expression of these genes. In addition, we characterized the phenotypic modulation of the oval cell population in response to this treatment.

Male outbred Sprague Dawley rats (Charles River Laboratories,Raleigh, NC) weighing 113–170 g were individually housed in plastic cages with woodchip bedding in environmentally controlled, clean-air rooms with a 12-h light cycle. Use of animals was approved by the Institutional Animal Care and Use Committee of Merck Research Laboratories. Tap water was available ad libitum.

The diet and treatment regimen for each dose group are detailed in Table 1. Rats were fed ad libitum either a certified plain rodent diet (standard diet; groups 1 and 2) or a Lombardi choline-deficient diet (Ref. 27; Dyets, Inc., Bethlehem, PA; groups 3–5). Ethionine was obtained from Acros (Acros Organics, Springfield,NJ), and WY14,643 was from Chemsyn Science Laboratories (Lenexa, KS). To minimize human exposure, ethionine was administered by gavage, and the dose (50–70 mg/kg/day) was chosen to be equivalent to the intake from dietary ethionine at 0.05–0.1% (w/w) of the daily food consumption. WY14,643 was administered by gavage at 100 mg/kg/day. Vehicle for both agents was 0.5% (w/v) methylcellulose in water. Rats not receiving WY14,643 and/or ethionine (groups 1–3) received the vehicle only.

Oval cells or hepatocytes were isolated at various time points from rats fasted 12 h prior to the procedure. Rats were anesthetized by isoflurane inhalation using a vaporizer (Vetequip,Pleasanton, CA). Each liver was perfused in situ via the portal vein using a two-step collagenase perfusion method without recirculation of the medium as described previously (28). Collagenase I was purchased from Worthington Biochemical Corp.(Lakewood, NJ), HBSS was from Life Technologies, Inc. (Grand Island,NY), and EGTA was from Sigma Chemical Co. (Saint Louis, MO).

Oval cells were then isolated from the perfused livers of rats fed the CDE diet according to a method described previously by Yaswen et al. (29) and modified by Pack et al.(30) with some additional changes. Parenchymal cells were removed by one to three rounds of digestion (20 min each, at 37°C in a shaking water bath) in 25 ml of DMEM (Life Technologies, Inc.)containing 0.1% (w/v) collagenase I, 0.004% w/v DNase I (Sigma), and 0.01% (w/v) Protease E (Sigma). After each digestion, the supernatant was decanted through a 70-μm nylon mesh, followed by a 40- μm nylon mesh, and then centrifuged at 400 × g for 5 min at 4°C. The pellet was resuspended in 50 ml of DMEM supplemented with 0.004% w/v DNase I, decanted through a 40-μm nylon mesh, and centrifuged again. The wash-filtration steps were repeated for a total of four times prior to combining pellets from the three digestions and resuspending in 50 ml of elutriation medium (consisting of DMEM supplemented with 0.004% w/v DNase I, 3% v/v heat-inactivated FCS,and maintained at 10°C). Centrifugal elutriation for the purification of oval cells was performed using a JE-6B elutriator rotor with a standard separation chamber in a J-6 M/E Beckman centrifuge (Beckman Instruments, Palo Alto, CA) at 2500 rpm and at 10°C. Six 150-ml fractions were collected at flow rates of 14, 19, 22, 26, 28, and 40 ml/min. A last fraction was collected by shutting down the rotor and applying a maximal flow rate. The cells collected in each fraction were centrifuged at 400 × g for 5 min and resuspended in DMEM supplemented with 10% v/v heat-inactivated FCS. Cell count and viability were assessed using trypan blue exclusion. Cell size was measured under a microscope (BX60; Olympus, Melville, NY)equipped with a metric stage micrometer. Oval cells collected in fraction 6 (flow rate, 40 ml/min) from rats in group 3 (CDE diet alone)after 4 and 6 weeks were designated CDE 4w and CDE 6w, respectively. Oval cells isolated from rats in group 4 (CDE diet and continuous treatment with WY14,643) after 4 and 6 weeks were designated CDE+WY 4w and CDE+WY 6w, respectively. Finally, oval cells isolated from rats in group 5 (continuous CDE diet for 6 weeks, treatment with WY14,643 during the last 2 weeks) were designated CDE 6w+WY w4–6. In addition,fraction 6 was collected from rats fed the standard diet and treated for 9 weeks with WY14,643 (Fr6 STD+WY 9w) or the vehicle (Fr6 STD 9w;groups 1 and 2).

Hepatocytes were isolated from rats fed the standard diet and treated with WY14,643 for 9 weeks (Hep STD +WY 9w) or receiving the vehicle(Hep STD 9w; groups 1 and 2) as described by Seglen (31). After perfusion, the liver cells were teased away from the Glisson capsule and the connective tissue by combing. The resulting cell suspension was then centrifuged at 50 × g,4°C for 5 min; the pellet was resuspended in buffered HBSS, and the hepatocytes were purified by centrifugation at 20,000 × g, 4°C for 10 min, on a layer of 40% Percoll (Pharmacia Biotech, Piscataway, NJ).

Freshly isolated cells were attached to glass slides by cytocentrifugation (Chandon, Inc., Pittsburgh, PA), fixed for 10 min at−20°C in cold methanol (except for GGT activity cytochemistry),air-dried, and kept frozen at −70°C until processing. Prior to staining, cytospin preparations were fixed for 10 s in cold acetone and soaked for 5 min in 1× PBS (Life Technologies, Inc.).

Staining was performed according to the method of Rutenburg(32) with a 40-min incubation.

Immunocytochemistry on cytospin preparations was performed using the indirect immunoperoxidase method. Specific antibodies and working dilutions were the following: monoclonal mouse anti-CK19 (Amersham International, Little Chalfont, England; 1:10); polyclonal goat antirat albumin (Cappel, Durham, NC; 1:1000); rabbit antirat GST Ya (α class;Biotrin, Dublin, Ireland; 1:1000); rabbit antirat GST Yp (π class;Biotrin; 1:1000); rabbit antimouse AFP (ICN, Aurora, OH; 1:300); goat antirat CYP4A1 (Daiichi Pure Chemicals Co., Tokyo, Japan; 1:350); and monoclonal mouse OV-6 antibody (a gift from Dr. S. Sell, Albany, NY;Ref. 33; 1:40). Negative controls were incubated with nonimmune serum from the same host as the primary antibody. All incubations were performed in a humid chamber under a plastic coverslip. Cytospin preparations were first incubated for 30 min at room temperature with PBS containing 10% v/v nonimmune serum from the host of the secondary antibody. Then, they were incubated with the primary antibody diluted in PBS containing 5% v/v nonimmune serum. All incubations were for 30 min at room temperature, except for OV-6(overnight at 4°C). After three washes in PBS, cytospin preparations were incubated for 30 min at room temperature with the appropriate peroxidase-labeled antibody diluted in 1.5% v/v nonimmune serum. The secondary antibodies and working dilution were the following: horse antigoat (Vector Laboratories, Burlingame, CA; 1:200); goat antirabbit(Dako, Carpinteria, CA; 1:200); and sheep antimouse (Amersham; 1:100). The washes were repeated, and the reaction was visualized with diaminobenzidine and hydrogen peroxide (Stable DAB Research Genetics,Huntsville, AL). The slides were washed in 0.3% v/v Tween 20,counterstained for 30 s with Gill n.1 Hematoxylin (Sigma), washed in water, air-dried, and mounted. Slides were observed under a light microscope (AX70; Olympus). A total of 1000 cells were scored in random fields on each slide, and scoring was repeated at least twice.

Freshly isolated cells were lysed with guanidinium isothiocyanate, and RNA isolation was performed using RNeasy Midi Kit according to the manufacturer’s instructions (Qiagen, Valencia, CA). RNA quantification was performed by spectrophotometry. TaqMan Reverse Transcription Reagents (Perkin-Elmer Corporation, Foster City, CA)were used to prepare cDNA from 125 ng of RNA of each sample. The reaction was performed in a final volume of 100 μl of a buffer containing 50 units of Multiscribe Reverse Transcriptase, 40 units of RNase inhibitor, 0.5 mm deoxynucleotide triphosphates, 2.5μ m random hexamers, and 5.5 mmMgCl₂. The reactions were incubated at 42°C for 45 min, 95°C for 5 min, and 4°C for 5 min using a GenAmp PCR System 9700 (Perkin-Elmer Applied Biosystems).

TaqMan PCR has been used to quantitatively monitor mRNA expression and has been described in detail previously(34).TaqMan PCR for rat PPARα, FACO, and CYP4A1 were performed using an ABI Prism 7700 Sequence Detector (Perkin-Elmer). For each gene, an optimal primer pair and an oligonucleotide probe were selected using the ABI Prism Primer Express Software (Perkin-Elmer Corp.) and were synthetized and labeled by Perkin-Elmer. The oligonucleotide probe was labeled with a reporter fluorescent dye at the 5′-end (FAM) and a quencher dye at the 3′-end (TAMRA). The primers and probes used were as follows: rat PPARα gene, forward primer 5′-TTGCTGAAGTACGGTGTGTATGAA-3′, reverse primer 5′-CCTGCAACTTCTCAATGTATCCTATGT-3′, and probe 5′-CCATTGCCGTACGCGATCAGCAT-3′); rat FACO, forward primer 5′-GGCCAACTATGGTGGCACTCA-3′, reverse primer 5′-TACCAATCTGGCTGCACGAA-3′,and probe 5′-CTTGTAGGCTTCTGTCAGGCCCTCCA-3′; and rat CYP4A1,forward primer 5′-CCCGACACAGCCACTCATTC-3′, reverse primer 5′-CTTCATCTCACTCATAGCAAATTGTTT-3′, and probe 5′-GGTTACGTCAAGGAGCGAGGAGGACT-3′. The PCR was performed on each cDNA sample corresponding to 6.25 ng of input RNA in 1× TaqMan Universal PCR Master Mix (Perkin-Elmer) and in a final volume of 25 μl. For each sample, 18S rRNA levels were determined and used as endogenous controls for PCR quantification; PCR was performed on each cDNA sample corresponding to 0.25 ng of input RNA using TaqMan rRNA Control Reagents (Perkin-Elmer) under the conditions recommended by the manufacturer. For PCR, all samples were run in quadruplicate on a 96-well reaction plate (MicroAmp Optical 96-well reaction plate;Perkin-Elmer) using optical caps as lids (Perkin-Elmer). After an initial step at 50°C for 2 min, followed by the activation of AmpliTaq Gold Polymerase (10 min at 95°C), 40 cycles of PCR were performed (95°C for 15 s and 60°C for 1 min).

The relative amount of RNA in each well was calculated by comparing the 18S PCR results to that of a reference well (normal primary hepatocytes, i.e., Hep STD 9w RNA) arbitrarily ascribed to represent exactly 0.25 ng. The amount of PPARα, FACO, and CYP4A1 mRNA was expressed relative to the target quantity measured in primary hepatocytes (Hep STD 9w), which was used as a calibrator. PPARα, FACO, and CYP4A1 expression levels were normalized to the relative amount of RNA determined in the 18S assay.

Oval cells were isolated from rats fed a CDE diet for 4 or 6 weeks. In addition, some rats received a daily dose of 100 mg/kg WY14,643, as detailed in Table 1. Oval cells were purified from the nonparenchymal cell suspensions by centrifugal elutriation. In agreement with previous reports (29, 30), oval cells isolated from rats fed a CDE diet were identified according to the following criteria: (a) cell diameter ranging from 10 to 15μm; (b) immunoreactivity to OV-6; (c)expression of CK19 and πGST; (d) presence of GGT activity;(e) partial expression of albumin and AFP; and(f) absence of peroxidase activity. This expression pattern is unique to oval cells and distinguishes them from other cell types. Analysis of the different elutriated fractions revealed that the greatest number of cells fulfilling these criteria was found in fraction 6, collected at a flow rate of 40 ml/min. Therefore, all of our investigations were performed on fraction 6, which is referred to as the oval cell fraction.

Table 2 and Fig. 1 summarize the characteristics of the oval cell fraction isolated in each dose group and at various time points during the regimen. The oval cell fraction, isolated from rats fed the CDE diet for 4 weeks (CDE 4w cells), had a mean diameter of 12 μm. Eighty-seven % (87.3 ± 0.6%) of these cells showed immunoreactivity to OV-6,82.0 ± 2.8% exhibited GGT activity, 66.2 ± 3.5% expressed albumin, 73.1 ± 0.6% stained positive for AFP, and expression of αGST was detected in <0.2%cells. Light microscopic analysis revealed that <0.03% of these cells were contaminating hepatocytes (data not shown). Finally, this fraction contained <0.1% peroxidase-positive cells (a characteristic of Kupffer cells).

As illustrated in Fig. 2,A, primary oval cells showed relatively low expression levels of PPARα, compared with normal hepatocytes. Interestingly, in vivo treatment with WY14,643 increased the expression of PPARαin the isolated oval cell fraction (compare CDE+WY with CDE). Furthermore, as shown in Fig. 2,B, in vivo treatment with WY14,643 increased the levels of FACO and CYP4A1mRNA in the isolated oval cells. The expression of these PPARα-regulated genes in primary hepatocytes (from untreated rats and rats treated with WY14,643) is shown for comparison. In addition, as indicated in Fig. 3, immunocytochemistry on the isolated oval cell fraction showed that in vivo treatment with WY14,643 increased the percentage of CYP4A1-positive cells as well as the levels of expression of CYP4A1.

The expression of markers of bile duct cells (OV-6, CK19, GGT, andπGST), fetal hepatocytes (AFP), and mature hepatocytes (albumin,αGST) in oval cells (fraction 6) isolated from each dose group was examined by cytochemistry or immunocytochemistry on cytospin preparations. For each marker, the percentage of cells positively stained is shown in Fig. 1.

As illustrated on Fig. 4,A, ∼90% CDE 4w and CDE 6w cells showed reactivity to OV-6,a monoclonal antibody identifying an epitope shared by CK19 and CK14 and expressed by oval cells and bile duct cells (33, 35). Similarly, most cells showed GGT activity and expressed CK19 as well asπGST. Continuous in vivo treatment with WY14,643 decreased the percentage of OV-6-and CK19-positive cells to 40–60% (Fig. 1),whereas GGT activity and πGST expression remained in 65–78% of the cells. A more dramatic effect of WY14,643 was observed in CDE 6w +WY w4–6 cells with <40% remaining positive for OV-6, CK19, and πGST and 55% cells showing GGT activity (see Figs. 1 and 4 B). Thus, in vivo treatment with WY14,643 generally decreased the expression of bile duct markers in the oval cells.

Conversely, in vivo treatment with WY14,643 increased the expression of αGST, a marker of mature hepatocytes, in a subpopulation of oval cells. As shown on Fig. 1, αGST was detected in 4% or less of the oval cells isolated from untreated rats, whereas up to 14.7% of cells were αGST positive in the oval cell population isolated from WY14,643-treated rats (see Figs. 1 and 4, C and D).

As shown in Table 2, oval cells isolated from rats treated with WY14,643 had a larger diameter than their counterparts isolated from control rats. We quantified the cells having a diameter >15 μm,which is approximately the size of the largest cells present in the oval cell fraction CDE 4w. As indicated in Table 2, <4% oval cells isolated from rats fed the CDE diet were <15 μm. This population was increased (up to 16.7%) upon in vivo treatment with WY14,643. As shown in Fig. 5, most cells in this larger subpopulation expressed albumin. A majority stained positive for markers of mature hepatocytes (αGST and CYP4A1)as well as for AFP, a marker of fetal hepatocytes. Only 10–35% of these larger cells expressed bile duct markers. Thus, these larger cells appear as “transitional” oval cells, in the process of differentiating toward the hepatocyte lineage.

Fraction 6 was isolated from rats fed the standard diet, with or without treatment with WY14,643. The cells in fraction 6 isolated from rats treated with WY14,643 were larger than their counterparts isolated from control rats; ∼52% of cells are >15 μm (Table 2). As shown in Fig. 2, the levels of PPARα, FACO, and CYP4A1 mRNA were very low in cells collected in fraction 6 from control rats. The expression of these genes appeared up-regulated upon chronic treatment with WY14,643, although levels remained relatively low compared with those measured in normal hepatocytes. Immunocytochemistry showed that the cells in fraction 6 isolated from control rats expressed markers of bile duct cells exclusively (Fig. 6). As shown in Fig. 6 and illustrated in Fig. 4, E–H,chronic treatment with WY14,643 decreased the proportion of cells expressing bile duct markers, whereas it strongly induced the expression of albumin and AFP in a majority of cells. The expression ofαGST was induced in a very limited number of cells (Fig. 6), in contrast with the large proportion of “transitional” cells(isolated from rats fed the CDE diet) that stained positive for αGST(Fig. 5). No expression of CYP4A1 could be detected by immunocytochemistry on cytospin preparations from Fr6 STD+WY 9w cells(data not shown).

The nongenotoxic class of carcinogens known as PPs induce liver tumors in rodents through a mechanism that requires the expression of the nuclear receptor PPARα (1, 7). However, the role played by PPARα in hepatocarcinogenesis, and more generally in the growth and differentiation of liver cells, is not known. Another PPAR subtype, PPARγ, plays a key role in the differentiation of adipocytes and of other cell types (25); thus, one role of PPARα in the liver might be its involvement in the differentiation of hepatocytes. Indeed, activators of PPARα were shown to inhibit keratinocyte proliferation and to induce their differentiation,suggesting a regulatory role of PPARα in this process(36). In the present study, we investigated the response of oval cells isolated from rats chronically treated with the prototype PPARα activator, WY14,643. Numerous studies have described the involvement of oval cells during experimental protocols of rodent liver carcinogenesis (reviewed in Ref. 17), but to our knowledge, this is the first report characterizing the response of rat oval cells to a PP and investigating the role of PPARα in these cells.

The levels of expression of PPARα in freshly isolated oval cells were measured by real time quantitative PCR. Oval cells isolated from rats fed the CDE diet expressed levels of PPARα that were about three to eight times lower than those measured in primary hepatocytes. Interestingly, oval cells isolated from rats fed the CDE diet for 6 weeks expressed higher levels of PPARα than their counterparts isolated after 4 weeks of the same regimen. They also expressed higher levels of αGST, a finding in agreement with other studies (37). Previously,the changes in the pattern of expression of different pyruvate kinase isotypes had also indicated a progression in the developmental maturity of oval cells during the CDE diet (38). Together, these results suggest that oval cells progress along a maturation/differentiation process during the CDE diet and argue in favor of the ability of these cells to differentiate toward the hepatocyte lineage (24). We observed that in vivo treatment with WY14,643 increased the expression of PPARα and that of two PP responsive element-containing genes, FACO and CYP4A1, in the oval cell population. Therefore, we can conclude that PPARα is expressed as a functional transcription factor in primary rat oval cells; moreover,when rats are administered a PPARα activator, oval cells show a PPARα-mediated response (induction of FACO and CYP4A1) qualitatively similar to the response observed in primary hepatocytes, though of lesser magnitude.

The PPARα-mediated response to in vivo treatment with WY14,643 was accompanied by various phenotypic changes in the isolated oval cell fraction. The PPARα activator down-regulated the expression of typical oval cell markers, including CK19, πGST, GGT,and the antigen identified by OV-6. Conversely, WY14,643 increased the expression of αGST, a marker of mature hepatocytes. These results show that WY14,643 induced the maturation of a subpopulation of oval cells.

In addition to inducing these phenotypic changes, WY14,643 induced a subpopulation of oval cells of larger size compared with the cells isolated from untreated rats (CDE diet alone). Similar larger cells have been described before in a number of studies and have been referred to as “transitional” cells (29, 39, 40, 41). The round- or irregularly-nucleated transitional cells observed in the present study all stained positive for albumin, and >50% expressed AFP. However, unlike oval cells isolated from untreated rats (CDE alone), a majority of transitional cells expressed αGST and CYP4A1,whereas only a small fraction (10–30%) stained positive for bile duct markers. This phenotype resembles that of a maturing population,differentiating toward the hepatocyte lineage. The effect of WY14,643 on the phenotype of the transitional population appeared less marked when the treatment was started once the oval cell proliferation had reached a peak (CDE 6w+WY w4–6 cells; Ref. 27); this might reflect a delay in the maturation-differentiation process of the transitional cells. Another possibility is that oval cells that are further advanced in their maturation process might be less susceptible to the effect of WY14,643 than less mature cells. In conclusion, our data show that chronic in vivo treatment with WY14,643, in combination with the CDE diet, induces a subpopulation of larger oval cells to differentiate toward the hepatocyte lineage.

In an attempt to exclude any effect of the CDE diet, we isolated nonparenchymal liver cells from rats fed the standard diet and collected fraction 6 exactly as done for the isolation of oval cells. The 9-week treatment with WY14,643 decreased the expression of bile duct cell markers among this population. In addition, WY14,643 induced the expression of albumin and AFP, markers that were not detectable in fraction 6 cells isolated from the control rats. A subpopulation of larger sized cells was observed in WY14,643-treated rats only. Most of this subpopulation (80%) expressed albumin and AFP, with a limited number of cells (10%) staining positive for GGT and πGST. This phenotype resembles that of the subpopulation of transitional cells we observed in the CDE diet regimen, with: (a) AFP and albumin being detected in a majority of cells; and (b) bile duct markers being expressed in a limited number of cells, if any. A tempting hypothesis is that chronic in vivo treatment with WY14,643 induced the proliferation of a few oval cells present in normal rat liver (13). These cells may have started differentiating toward the hepatocyte lineage but experienced a block during this process. This maturation arrest would account for the mixed phenotype observed here and, most importantly, for the very limited number of cells expressing αGST, in striking contrast with the large proportion of transitional cells that stained positive for this marker of mature hepatocytes (Figs. 5,6).

The arrested maturation of determined stem cells is a concept opposing that of the “dedifferentiation” of mature hepatocytes for interpreting the cellular origin of HCCs (14). Indeed, the alteration in the control of stem cell differentiation is believed to be a critical change that occurs during neoplastic progression(14). In this respect, the product of the Tg737gene was identified as an oval cell differentiation factor; loss of Tg737 expression results in the proliferation of murine oval cells without concomitant differentiation (15, 16). Moreover, Tg737 acts as a liver tumor suppressor gene and was found rearranged in 40% of chemically induced liver tumors in rats(15, 16). Interestingly, one-third of liver tumors induced in rats by WY14,643 showed a rearranged Tg737 gene(16), arguing for the involvement of oval cells in the hepatocarcinogenicity of WY14,643. These data suggest that the altered control of oval cell differentiation/proliferation and subsequent block in the differentiation process might be an important event during PP-induced tumorigenesis. Thus, PPs might act as tumor promoters by giving indirect selective growth advantage to initiated cells that have become nonresponsive to the differentiation signals delivered through the activation of PPARα; as a result, such nonresponsive cells will not differentiate but rather will proliferate.

Consistent with the involvement of oval cells during PP-induced tumorigenesis is the similarity between the phenotype of the cells promoted here by chronic treatment with WY14,643 (in the absence of the CDE diet regimen) and that of the cells present in the preneoplastic foci typically induced by PPs (8, 9, 10). In particular, the lack of expression of πGST and GGT is unique to neoplastic nodules and HCCs induced by PPs, in contrast to the lesions induced by any other type of carcinogen or tumor promoter (11). Furthermore, a notable feature of the preneoplastic foci promoted by PPs is that they are basophilic; interestingly, oval cells appear to go through a basophilic intermediate stage as they differentiate into hepatocytes (42, 43).

In summary, rat oval cells express PPARα; they show a typical PPARα-mediated response upon in vivo treatment with a PPARα activator and a phenotypic modulation toward the hepatocyte lineage. These findings argue for a role of PPARα in the differentiation of oval cells. In addition, a prototype PPARαactivator and PP promotes a subpopulation of cells with a phenotype reminiscent of that of the cells usually found in neoplastic foci induced by PPs. Together with previous findings (16),these results suggest the involvement of oval cells in the hepatocarcinogenicity of PPs.

We thank Dr. S. Sell for the generous gift of monoclonal antibody OV-6.