Expression of alpha, mu and pi class glutathione S-transferases inoval and ductal cells in liver of rats placed on a choline-deficient,ethionine-supplemented diet
Lisa B.G.Tee1, Patrick CJ.Smith and George C.T.Yeoh
Department of Physiology, University of Western Australia, Nedlands,Western Australia 6009
'To whom correspondence should be addressed
Expression of the alpha, mu and pi class glutathioneS-transferases (GSTs) in hepatocytes, oval cells and ductalcells derived from the livers of rats placed on a choline-deficient, ethionine-supplemented (CDE) diet for 5 weeks wasinvestigated. An overall decrease in the expression of alphaand mu class GSTs and an over-expression of pi class GSTwas observed in the liver after CDE treatment as indicatedby Northern blotting analysis. Massive disruption of the liverwith oval cell infiltration in the sinusoids throughout the lobuleoccurred after 5 weeks CDE treatment. 'Duct-like' structuresconsisting of oval-like cells (ductal cells) with rounder nucleiand more cytoplasm than oval cells within the sinusoids werealso apparent. Immunocytochemteal analysis revealed that thealtered expression of GST in the whole liver is attributed toa differential expression of alpha, mu and pi class GSTs inthe different cell types in the liver, including hepatocytes, ovalcells around the portal region and among the sinusoids, andoval-like cells (ductal cells) in the 'duct-like' structures.In vitro studies using purified oval-ductal cells andhepatocyte populations confirmed the differential expressionof GSTs in the varying cell populations in situ. The expressionof the alpha and mu class GSTs in hepatocytes does notappear to be altered by the CDE diet. Heterogeneity indistribution of pi class GST was observed in the hepatocytepopulation, some hepatocytes were stained strongly while nostaining was observed in others. Oval and ductal cellsrepresent two distinct populations displaying differentexpression of GSTs. Pi class GST was detected in the majorityof oval and ductal cells. Alpha class GST was detected in< 5% of the oval cell population and was found in > 50%of the ductal cell population. In contrast, mu class GST wasabsent in ductal cells and was present in 24% of oval cellsaround the portal region. This supports the view that ductalcells are not of bile ductal origin since mu GST is presentin normal bile duct epithelial cells. Furthermore the changein expression of GSTs in the liver after CDE treatment isattributed to the large increase in oval and ductal cellpopulations.
Introduction
Exposure of rats to hepatocarcinogens results in different cellularresponses in the liver and in the production of cell populationsdisplaying different patterns of expression of liver enzymes. Theemergence of a 'new' cell population that has been morpho-logically defined as oval cells is one of the first cellular changesinduced by hepatocarcinogens (1—3). Extensive studies using
various regimes including the Solt—Farber and choline-deficient,ethionine-supplemented (CDE*) models, revealed that the ovalcell population is heterogeneous with diverse developmentalpotential (4-7). Oval cells have been shown to display fetal livermarkers such as fetal aldolases A and C, fetal* form of pyruvatekinase, and alpha-fetoprotein (4). We have recently demonstratedthat two populations of oval-like cells derived from the CDE ratsare similar to fetal hepatocytes at ~ 15 and 19 days gestationrespectively, in terms of pyruvate kinase and tyrosine amino-transferase expression (unpublished). To date there has beenconflicting evidence regarding the origin of oval cells since thereis no satisfactory cell-lineage marker. The association of ovalcell proliferation arising from the periportal region, with theappearance of 'duct-like' structures has been used as support foroval cells originating from bile duct cells (4—6,8). Studies usingliver-specific enzymes indicate that 'duct-like' structures whichcontain albumin and L-pyruvate kinase differ from normalbile duct cells which are devoid of both enzymes (3,4,9,10).Numerous studies involving azo-dye carcinogenesis provideevidence for the transition of oval cells into hepatocytes (4,10,11).We have recendy established the pattern of alpha, mu and pi classglutathione S-transferase (GST) expression in the developing fetalliver as well as normal adult hepatocytes and hepatoma cell lines(12,13). Pi class GST which is expressed in adult liver onlyduring hepatocarcinogenesis (14—16) is strongly expressed infetal development (12) and therefore displays the classicalcharacteristic of an oncofetal marker such as alpha-fetoprotein(17). In the present study we use a combination of fetal (pi) andadult (mu and alpha) class GSTs to characterize oval cells and'duct-like' structures derived from rats placed on a CDE diet for5 weeks. We also investigate the potential of pi class GST indetecting oval cells and other cell population(s) that are producedas a result of the CDE diet.
Materials and methodsCDE diet
Male albino Wistar rats weighing 120 g were fed a standard laboratory chowwith normal drinking water (control rats) or a choline-deficient diet (product codeTD 79246, Teklad, Madison, Wl) with water supplemented with 0.15% tthionine(CDE rats). Ethionine was given in drinking water rather than as a mixture inthe choline-deficient diet to minimize handling and risk of exposure to thecarcinogen. Measurement of food and water intake of the rats showed thatthis was equivalent to giving the rats a choline-deficient diet supplementedwith 0.07% ethionine.
ElutriationLiver cells from control and CDE rats were isolated by collagenase perfusionusing the method of Seglen (18) and subsequently separated according to sizeand density by centrifugal elutriation as described by Yaswen et ai. (5). Fivefractions were collected, with an additional fraction collected at a flow rate of50 ml/min. Fraction 3 was shown to be free of adult hepatocytes by the followingcriteria. First, cytocentrifuged samples stained with haernotoxylin and chromotrope2R do not reveal the presence of hepatocytes which are two to three timeslarger than oval cells and are distinguished by a large round nucleus. Second,immunocytochemistry performed on this fraction does not show large cells whichare stained strongly positive for albumin.
Histochemistry and immunocytochemistry
For in vivo studies the liver was first perfused in situ with phosphate-bufferedsaline (PBS) via the hepatic portal vein to remove the Wood and thereafter liver
samples of 1 cm3 were fixed in Camoy's solution for 6 h and embedded inparaffin wax. Liver samples were also taken for preparation of RNA (see below).For light microscopy, 4 /un sections of the fixed liver were attached to glass sbdesand all subsequent staining procedures were performed on the glass slide. Liversections were stained with haemotoxylin and chromotrope 2R to verify themorphology of the liver.
For in vitro studies cells from fractions 3 and 5 were attached to glass slidesby cytocentnfugation followed by fixation in a series of 70, 100 and 70% methanolfor 1 min each, at room temperature before analysis by immunocytochemistry.
The indirect immunoperoxidase detection of alpha, mu and pi GSTs wasperformed according to the method of Clement et al. (19). Antibodies to GSTsubunit 1 (cross-reacting with other members of the alpha family including subunits2, 8 and 10), subunit 3 (cross-reacting with other members of the mu familyincluding subunits 4, 6, 9 and II) and subunit 7, were supplied by Medlabs,Dublin, Ireland. Endogenous peroxidase in the samples was Nocked by treatmentwith 2.5% aqueous periodic acid for 5 min and 0.02% sodium borohydride for2 min (20). This was followed by a 1 h incubation with 10% fetal calf serumin PBS. The liver sections were then reacted for 1 h, with a dilution of 1:100of the alpha, mu and pi class GST, and of the non-immune rabbit serumand albumin antibodies.
After three washes with PBS, the samples were exposed to a 1:200 dilutionof the second antibody (peroxidase-coupled goat IgG directed against rabbit IgG)for 1 h. The washing procedure was repeated, followed by a final wash in 0.O5 MTris, pH 7.5. Localization was revealed by reaction with 0.05% diaminobenzidineand 0.01% fyOj in 0.05 M Tris pH 7.5 for 20 min. In all experiments liversections treated with the non-immune serum were incorporated as negativecontrols and those treated with albumin antibodies were used as positivecontrols for hepatocytes.
Isolation of total RNA and hybridization with cDNA representing alpha, mu andpi class GSTs
Total RNA was isolated from rat liver samples, collected as described above,using the method of LeMeur et al. (21). Samples were homogenized in 3 M LiG,6 M urea, 10 mM Na acetate (pH 5.0), 200 jig/ml heparin, 0.1 % SDS and 1 %/3-mercaptoethanol, and stored overnight at 4°C. The precipitated RNA wassedimented by centrifugation at 15 000 g for 20 min, washed three times with4 M LiCl, 8 M urea and 1 % 0-mercaptoethanol in water, dissolved in I % SDSin 50 mMNa acetate (pH 5.0) and extracted with phenol-chloroform (1:1, v/v),and diethyl ether. The RNA in the aqueous phase was then precipitated with ethanoland was finally dissolved in water. The concentration of RNA was estimatedfrom A^o- For Northern analysis, 10 fig of RNA was electrophoresed accordingto the method of Lehrach et al. (22) arid then electrophoretically transferred toGeneScreen using a Bio-Rad Transblot apparatus.
The alpha, mu and pi class GST mRNAs were detected by hybridizationto "P-labelled probes pGSTr 155 (23), JTL9 (S.PemMe and J.B.Taylor,unpublished data) and pGSTr 7 (24) respectively. Radioactivity was visualizedby autoradiography. The cDNA probes were labelled by nick translation.Hybridization in 50% deionized formamide, 5 x SSC, 0.1% SDS, 5 xDenhardt's solution and 250 /tg/ml sonicated salmon sperm DNA was performedat 45°C for 24 h after 18 h of prehybridization. The intensity of the image producedon X-ray film was quantitated using a Bio-Rad model 620 video densitometerand is expressed as a ratio of the values obtained for the 18S and 28S ribosomalRNA, stained with ethidium bromide.
Results
Morphological changes and cellular distribution of GST in theliver after CDE treatmentAlpha and pi class GSTs showed specific localization in thenormal adult rat liver. Alpha class GST is homogeneouslydistributed throughout the parenchyma while pi class GST islocalized only in the bile duct epithelial cells. Mu class GST ispresent in both hepatocytes and bile duct epithelial cells.
A marked change in the morphology of liver of rats placedon a 5 week CDE diet was observed in this study. Extensiveaccumulation of small oval cells around the portal region andinfiltration of these cells across the lobule was apparent (Figurela). Oval cells are easily distinguished from hepatocytes by theirdense oval shaped nuclei and a low cytoplasm to nucleus ratio(Figure lb). Hepatocytes contain pale round nuclei and are abouttwo to three times greater in diameter compared with oval cells(8). Accumulation of 'oval-like' cells forming 'duct-like'structures was also apparent after CDE treatment (Figures laand 5a). These 'oval-like' cells are referred to as ductal cellshereafter. Ductal cells are larger and differ slightly in morphologyto oval cells situated around the portal region and in thesinusoids. They contain rounder nuclei and a relatively higherproportion of cytoplasm.
A marked variation in the pattern of staining in the differentcell populations (hepatocytes, oval cells and ductal cells) wasobserved with alpha (Figure 2), mu (Figure 3) and pi (Figure4) class GSTs. Immunostaining revealed the presence of alphaclass GST in the hepatocytes and its absence in the majority ofoval cells, as marked by the distinct negative staining around theportal vein (Figure 2, 'P'). Less than 5% of the oval cellpopulation was positively stained for alpha GST (Figure 2,arrows). All hepatocytes were strongly stained for the mu classGST while only 24% of the oval cells were positively stained(Figure 3, arrows). In contrast to alpha class GST, the majorityof oval cells stained for pi class GST, as indicated by intensestaining around the portal vein (Figure 4a, 'P'). Heterogeneityin the distribution of the pi class GST was apparent in thehepatocyte population. Some hepatocytes were strongly stainedwhile virtually no staining was observed in others (Figure 4b).A group of —13 'duct-like' structures is depicted in Figure 5(a).The majority of these ductal cells contain pi class GST (Figure5d) and -50% of cells in 'duct-like' structures contain alphaclass GST (Figures 2a and 5b). Thorough screening of liversections revealed that all ductal cells were devoid of mu classGST. Figure 5(c) shows the absence of mu class GST stainingin cells which make up the ductal processes and weak staining
Fig. 1. Haematoxylin and chromotrope 2R staining of liver section of rat on CDE diet for 5 weeks, (a) Accumulation of oval cells was apparent around theportal region. P, portal vein. Magnification bar represents 400 pm. (b) Oval cells (A) adjacent to portal vein (P) contain dense, oval-shaped nuclei whilehepatocytes (H) contain pale round nuclei. Magnification bar represents 100 /un.
in a few oval cells (arrow) in the spaces between the ductalprocesses. Weak staining of oval cells and more intense stainingof hepatocytes was observed with albumin (data not shown). Ovalcells (arrows) were stained for fetal form of pyruvate kinase(M2-PK) and were found to be virtually devoid of adult formof pyruvate kinase (L-PK). All ductal cells contain M2-PK and- 8 0 % contain L-PK (Figure 6).
Expression of GST in isolated liver cellsThe mixture of liver cells isolated from CDE-treated rats,designated as pre-elutriation fraction, consists of oval cells, ductalcells, small hepatocytes, large hepatocytes, endothelial cells andmyeloid cells. Elutriation of mis mixture yielded fivefractions, designated 1-5. The composition of pre- and post-
elutriation fractions was analysed by haematoxylin and chromo-trope 2R staining (Table I) and by immunostaining with albumin.Hepatocytes, oval cells and ductal cells are positive, whileendothelial and myeloid cells are negative for albumin. Fraction3 contains predominantly oval and ductal cells and is free ofhepatocytes (Figure 7a), while fraction 5 is mainly hepatocyteswith clusters of oval cells which collectively possess the samesize and density of individual hepatocytes (Figure 8a). Thesefractions were stained for alpha (Figures 7b and 8b), mu andpi (Figures 7c and 8c) GST arid relative numbers of positive cellsfor each family were determined. These results summarized inTable n reveal that only 7.8% of oval—ductal cells in fraction3, and all hepatocytes in fraction 5 were positively stained foralpha GST, while >90% of oval and ductal cells and <8% ofhepatocytes were stained for pi GST.
Fig. 2. Distribution of alpha class GST in liver of rats after 5 weeks treatment with CDE diet. The majority of oval cells accumulated around the portalregion were not stained, only a few were stained (^ )• Hepatocytes were stained strongly, (a) Magnification bar represents 400 /un. P, portal vein,(b) Magnification bar represents 100 /un.
Fig. 3. Distribution of mu class GST in liver of rat after 5 weeks treatment with CDE diet. Pale staining around portal region. Only a few oval cells ( | )were stained. Hepatocytes were stained strongly, (a) Magnification bar represents 400 /an. P, portal vein, (b) Magnification bar represents 100 /un.
Fig. 4. Distribution of pi class GST in liver of rat after 5 weeks treatment with CDE diet. Uniform staining of oval cells around the portal region andheterogeneous staining of hepatocytes arc seen, (a) Magnification bar represents 400 /un. P, portal vein, (b) Magnification bar represents 100 /un.
Fig. 5. Formation of ductal processes in rat liver after CDE treatment, (a) Haematoxylin and chromotrope 2R staining, and immunostaining of liver sectionwith (b) alpha, (c) mu and (d) pi class GST antibodies. Ductal processes are made up of oval-like cells which are larger and which contain rounder nucleithan oval cells (A)- The liver section stained for alpha class GST is in series with that stained for the pi. Magnification bar represents 200 /un.
Fig. 6. Oval cells (A) are also located amongst 'duct-like' structures. Immunostaining of liver sections in series for (a) M2-PK and (b) L-PK. Oval cells werestained for M2-PK and are virtually devoid of L-PK. All ductal cells contain M2-PK and - 80% contain L-PK. Magnification bar represents 200 nm.
Table I. Composition of cell types isolated from the liver of CDE-treated rats
Pre-elutriarion (P) and five post-elutriation fractions (1 —5), were collected and cell types identified as described in Materials and methods. Results arepresented as percentage of total cells present in the cytocentrifuged samples and are mean ± SEM of three different cell preparations.
Fig. 7. Fraction 3 contains oval cells, (a) Haematoxylin and chromotrope 2R staining, and immunostaining of fraction 3 with (b) alpha and (c) pi class GST.Magnification bar represents 200 (xm.
oFig. 8. Fraction 5 contains mainly hepatocytes and clumps of oval cells, (a) Haematoxylin and chromotrope 2R staining, and immunostaining of fraction 5with (b) alpha and (c) pi class GST. Magnification bar represents 200 jun.
Steady-state mRNA levels of GSTs after CDE treatmentNorthern analysis of RNA extracted from liver of control andCDE-treated rats showed a decrease in steady-state mRNA levelsof alpha and mu class GSTs and an increase in pi class GSTmRNA level after CDE treatment (Figure 9). Densitometricanalysis of the Northern blot autoradiograph indicated that thesteady-state mRNA levels of alpha and mu class GST in rat liversafter CDE treatment were 65.1 and 72.8% respectively, ofcontrol. A 6-fold increase in the steady-state mRNA pi class GSTwas observed after CDE treatment. This differential expressionof the three classes of GST mRNAs in the liver after CDEtreatment was attributed to the differences in expression of theGSTs in the different cell populations as shown in Northernanalysis of RNA extracted from cells in elutriated fractions(Figure 10). A high level of alpha class GST mRNA wasobserved in fraction 5 which contains mainly hepatocytes andrelatively low levels were observed in fractions 3 and 4 whichwere enriched in oval and ductal cells (Figures 7 and 10b). Incontrast high levels of pi class mRNA were observed in fractions3 and 4 and a lower level was observed in fraction 5 (Figure 10c).
DiscussionIn a recent study we have established that pi class GST displaysthe classical pattern of expression of a liver oncofetal protein andthat alpha class GST is predominantly an adult liver protein(12,13). Furthermore, alpha, mu and pi class GSTs exhibit celllocalization specificity in the adult liver. Alpha GST is presentonly in hepatocytes, mu GST in both hepatocytes and bile ductepithelial cells, and pi GST only in bile duct epithelial cells.In the present study, a combination of these three classes of
Table II. Distribution of GSTs in isolated cells in fractions 3 and 5
Antigen
Alpha GSTMu GSTPi GST
Fraction(oval
7.828.491.4
3and ductal cells)
±±+
1.311.43.0
Fraction 5(hepatocytes)
100100
7.8 ± 0.7
Results show the percentages of oval and ductal cells in fraction 3 andhepatocytes in fraction 5 which are positively stained for alpha, mu and piclass GST. Results are mean ± SEM of three different cell preparations.
GST each with a specific pattern of expression during liverdevelopment, and different cell specificity in adult liver, provesto be useful in the study of cell lineages associated with thepathogenesis of liver cancer. In this study, we show thatCDE treatment gives rise to two distinct populations of non-parenchymal cells, oval and ductal cells, which exhibit differentpatterns of expression of the three classes of GST. Oval cellssurrounding the portal region are smaller than ductal cells whichform 'duct-like' structures amongst the parenchymal regions. Ourresults support the view that oval and ductal cells are ofhepatocytic lineage, and indicate that they are most likely atdifferent stages of maturation.
It is now well established that oval cell proliferation representsone of the early changes associated with experimental hepato-carcinogenesis (1-3). However, there is still conflicting evidenceregarding the origin of oval cells. The association of ovalcell proliferation arising from the periportal region, with theappearance of duct-like structures has been used to support theview that oval cells originate from bile duct cells (4-6,8). Studies
Fig. 9. Northern analysis of total RNA prepared from liver of control rats(CON) and rats which were placed on the CDE diet for 5 weeks (CDE).(a) An even loading of the gel is shown by ethidium bromide staining ofribosomal RNA bands. The membrane was sequentially probed for (b)alpha, (c) mu and (d) pi class GST.
Fig. 10. Northern analysis of total RNA prepared from liver cells isolatedby centrifugal elutriation from rats which were placed on the CDE diet for5 weeks. The fractions 1 - 5 and P (pre-elutriation) are indicated asdescribed in Materials and methods, (a) An even loading of the gel isindicated by ethidium bromide staining for the ribosomal RNA bands. Themembrane was sequentially probed for (b) alpha, (c) mu and (d) pi classGST.
using liver-specific enzymes indicated that duct-like structureswhich contain albumin and L-PK differ from normal bile ductcells which are devoid of both enzymes (3,4,9,10). Theseduct-like structures are formed in response to CDE treatment andare different from bile ducts which are associated with the portal
triad in normal liver (25). Furthermore, normal bile duct existsas a single process around the portal region while duct-likestructures appear in groups (Figure 5a) and are dispersed amongstthe parenchyma. In our study mu GST which is normally presentin the bile duct epithelium is absent in the duct-like structuressuggesting that these two elements are derived from different celllineages. We also show by immunocytochemical staining of serialliver sections that these duct-like structures express hepatocyte-specific enzymes including albumin and L-PK (Figure 6). Pi classGST, which is absent in adult hepatocytes and is expressed atsignificantly high levels in fetal hepatocytes (12,13), is presentin virtually all the oval and ductal cells after CDE treatment.Evarts et al. (26) using an in situ hybridization technique,reported the presence of pi GST transcript in both oval and ductalcells in liver of rats exposed to the Solt—Farber regimen. Ourstudies showed that alpha GST which is only detected in < 4 %of hepatocytes in 17 day gestation rat fetuses (12), is virtuallyabsent in oval cells. Collectively, these results support the viewthat oval cells display characteristics which resemble fetalhepatocytes and that ductal cells are of hepatocytic lineage.Indeed our studies suggest that oval cells are more immaturethan ductal cells.
In a recent study, we showed that the level of hepatic alphaGST increased with development and that it is predominantlyan adult enzyme. Immunocytochemical analysis of whole liversections in the present study, show that only some ductal cellscontain alpha GST whereas almost all oval cells are devoid ofthe enzyme thus suggesting that the alpha GST positive ductalcells are more mature than oval cells. This is further confirmedin our in vitro studies using isolated oval and ductal cell fractions,where only 7.8% of cells in fraction 3 were stained positivelyfor alpha class GST (Figure 7b). The differentia] expression ofGSTs in oval and ductal cells suggests that they are two distinctcell types. The majority of oval cells contain fetal (pi) GST whileductal cells contain both fetal (pi) and adult (alpha) GST. It islikely that oval cells are precursors of ductal cells since theoccurrence of oval cells precedes the formation of duct-likestructures. It must be emphasized that although some ductal cellsexpress alpha class GST, some are devoid of the enzyme. Thesealpha GST-negative ductal cells display characteristics closer tooval cells and may represent a transitional population betweenoval and 'mature' alpha GST-positive ductal cells. Further studieswill be required to establish this relationship.
Recent studies by Thorgeirsson et al. (27—29) show thattransformed rat liver epithelial (RLE) cell lines share manycommon cellular markers with the oval cells derived from liverof rats exposed to the Solt —Farber regimen. Markers usedinclude the cytokeratins, lactate dehydrogenase, aldolase andGST. Indeed, these RLE cell lines expressed a reduced level ofalpha GST and increased pi GST expression, similar to our ovaland ductal cells derived from CDE rats. These changes inexpression of GST were detected at both cellular and RNA levelsin our studies using irnmunocytochernistry and Northern analysisrespectively.
In addition to oval cell proliferation and duct-like structureformation, CDE treatment also gives rise to hepatocytes whichare strongly stained for pi GST. We have compared pi GST withother oncofetal markers including alpha-fetoprotein and M2-PKin their ability to detect changes in hepatocytes after CDE diet.Pi GST appears to be the only common marker for both ovalcells and altered hepatocytes. These altered hepatocytes mightbe similar to the basophilic hepatocytes which are described byEvarts et al. (30) as the progeny of oval cells.
In conclusion, results from the present study support the viewthat oval cells resemble immature hepatocytes and that duct-likestructures are not of bile ductular lineage. Oval and ductal cellsmay represent a heterogeneous pool of 'oval-shaped' cells atdifferent stages of development, the oval cells being moreimmature than the ductal cells. These results also indicate thata combination of fetal (pi) and adult (alpha and mu) GSTs asmarkers is useful in the study of changes in cell population(s)during hepatocarcinogenesis. Based on the differential expressionof the three classes of GST in preneoplastic liver after CDEtreatment, we propose that a precursor-product relationshipexists between oval cells and ductal cells.
AcknowledgementsProfessor Brian Kettercr, Cancer Research Campaign Molecular ToxicologyGroup, University and Middlesex School of Medicine, London, provided expertadvice and cDNA probes for alpha, mu and pi GST. Technical assistance fromMr Alan Light, Miss Deborah Faulkner and Miss Alice O'Connor is gratefullyappreciated. This study is supported by the National Health and Medical ResearchCouncil of Australia. Lisa B.G.Tee was a recipient of the Saw Medical ResearchFellowship, University of Western Australia, while this study was in progress.
References1. Farber,E. (1956) Similarities in the sequence of early histological changes
induced in the liver of the rat by ethionine, 2-acetylaminofluorene and3'-methyl-4-dimethylaminoazobenzene. Cancer Res., 16, 142—148.
2.Shinozuka,H., Lombardi.B., Sell.S. and Iammarino.R.M. (1978) Earlyhistological are! functional alterations of ethionine liver carcinogenesis in ratsfed a choline-deficient diet. Cancer Res., 38, 1092-1098.
3. Sell.S. and Leffert.H.L. (1982) An evaluation of cellular lineages in thepathogenesis of experimental hepatocellular carcinoma. Hepaxology, 2, 77—86.
4.Hayner,N.T., Braun.L., Yaswen.P., Brooks,M. and Fausto.N. (1984)Isozyme profiles of oval cells, parenchymal cells, and biliary cells isolatedby centrifugal elutriation from normal and preneoplastic livers. Cancer Res.,44, 332-338.
5. Yaswen.P., Hayner.N.T. and Fausto,N. (1984) Isolation of oval cells bycentrifugal elutriation and comparison with other cell types purified fromnormal and preneoplastic livers. Cancer Res., 44, 324-331.
6. GnshamJM. and Porta.E.A. (1964) Origin and fate of proliferated hepaticductal cells in the rat: electron microscopic and autoradiographic studies. Exp.Mol. Pathol., 3, 242-261.
7. Evarts.R.P., Nagy.P., Nakatsukasa.H., Marsden.E. and Thorgeirsson,S.S.(1989) In vivo differentiation of rat liver oval cells into hepatocytes. CancerRes., 49, 1541-1547.
8. Farber.E. and Cameron,R. (1980) The sequential analysis of cancerdevelopment. Adv. Cancer Res., 31, 125-226.
9. Fausto.N., Thompson,N.L. and Braun.L. (1987) Purification and culture ofoval cells from rat liver. Cell separation. Methods Select. Appl., 4, 45—TJ.
10. Germain,L., Noel,M., Gourdeau.H. and Marceau.N. (1988) Promotion ofgrowth and differentiation of rat ductalar oval cells in primary culture. CancerRes., 48, 368-378.
11. Sell,S. (1978) Distribution of a-fetoprotein and albumin-containing cells inthe liver of Fischer rats fed four cycles of N-2-fluorenylacetamide. CancerRes., 38, 3107-3113.
12. Tee.L.B.G., Gi)more,K.S., Meyer.D.J., Ketterer.B., Vandenberghe,Y. andYeoh.G.C.T. (1992) Expression of glutathione S-transferase during liverdevelopment. Biochem. J., 282, 209-218.
13. Tee,L.B.G. and Yeoh.G. (1991) Hepatocytes differentiation and liver cancer:expression of glutathione S-transferases. Proc. Ausl. Physiol. Pharmacol. Soc.,22, 204-208.
14. Kitahara.A., Satoh.K., Nishimura.T., Ishikawa.T., Ruike.K^ SatoJC,Tsuda.H. and Ito.N. (1984) Changes in molecular forms of rat hepaticglutathione S-transferase during chemical hepatocarcinogenesis. Cancer Res.,44, 2698-2703.
15. Meyer,D.J., Beale.D., Tan.K.H. and Ketterer,B. (1985) Glutathionetransferases in primary rat hepatomas: the isolation of the form with GSHperoxidase activity. FEBS Lett., 184, 139-143.
16. Sato.K., Satoh.K., Hatayama.I., Tsuchida.S., Soma,Y., Shiratori.Y.,Tateoka,N., Inaba.Y. and Kitahara,A. (1987) Placenta glutathione S-transferaseas a marker for (pre)neoplastk tissues. In Mantle,T.J., Pickett.C.B. andHayesJ.D. (eds), Gttaathione S-Transferase and Carcinogenesis. Taylor &Francis Press, London, pp. 127-137.
17. Sell.S., Becker.F.F., Leffert.H.L. and Watabe.H. (1976) Expression of an
oncodevelopmental gene product (a-fetoprotein) during foetal developmentand adult oncogenesis. Cancer Res., 36, 4239-4249.
18. Seglen.P.O. (1981) Preparation of isolated liver cells. Cell, 23, 561-571.19. Clement,B., Rissel.M., Peyrol.S., Mazurier.Y., Grimaud.J.A. and
Guillouzo.A.A. (1985) Procedure for light and electron microscopicintracellular immunolocalization of collagen annd fibronectin in rat liver. J.Histochem. Cytochem., 33, 407-414.
20. Heyderman.E. (1979) Immunoperoxidase technique in histopathology:applications, methods, and controls. J. Clin. Pathol., 32, 971-978.
21. LcMeur.M., Glanville.N., Mandel.J.L., Gerlinger.P., Palmiter.R. andChambon.P. (1981) The ovalbumin gene family: hormonal control of X andY gene transcription and mRNA accumulation. Cell, 23, 561—571.
22. Lehrach.H., Diamond,D., WozneyJ.M. and Boedtker.H. (1977) RNAmolecular weight determinations by gel electrophoresis under denaturingconditions, a critical reexamination. Biochemistry, 16, 4743-4751.
23. TaylorJ.B., Craig.R.K., Beale.D. and Ketterer.B. (1984) Construction andcharacterization of a plasmid containing complementary DNA to mRNAencoding the N-terminal aminoacid sequence of the rat glutathione Yasubunit.Biochem. J., 219, 223-231.
24. Pemble.S., TaylorJ.B. and Ketterer.B. (1986) Tissue distribution of ratglutathione transferase subunit 7, a hepatoma marker. Biochem. J., 240,885-889.
25. Sell,S. (1983) Comparison of oval cells induced in rat liver by feedingN-2-fluorenylacetamide in a choline-devoid diet and bile duct cells inducedby feeding 4,4'-diaminodiphenylmethane. Cancer Res., 43, 1761-1767.
26. Evarts.R.P., Nakatsukasa.H., Marsden.E.R., Hsia,C.-C., Dunsford.H.A.and Thorgeirsson.S.S. (1990) Cellular and molecular changes in the earlystages of chemical hepatocarcinogenesis in the rat. Cancer Res., 50,3439-3444.
27. Garfield,S., Huber.B.E., Nagy.P., Cordingley.M.G. and Thorgeirsson.S.S.(1988) NeoplasOc transformation and lineage switching of rat liver epithelialcells by retrovirus-associated oncogenes. Mol. Carcinogenesis, 1, 189—195.
28. Nagy.P., Evarts.R.P., McMahonJ.B. and Thorgeirsson.S.S. (1989) Roleof TGF-beta in normal differentiation and oncogenes in rat liver. Mol.Carcinogenesis, 2, 345 — 354.
29. Bisgaard.H.C. and Thorgeirsson.S.S. (1991) Evidence for a common cellof origin for primitive epithelial cells isolated from rat liver and pancreas.J. Cell. Physiol., 14, 333-343.
30. Evarts.R.P., Nagy.P., Marsden.E. and Thorgeirsson.S.S. (1987) Aprecursor-product relationship exists between oval cells and hepatocytes inrat liver. Carcinogenesis, 8, 1737-1740.
Received on October 21, 1991; revised on July 20, 1992; accepted on July 21, 1992
