Phenotypic characterization of rat hepatoma cell lines and lineage-specific regulation of gene expression by differentiation agents

Differentiation (1998) 63:215–223

© Springer-Verlag 1998

O R I G I NA L A RT I C L E

&roles:Isabel Zvibel · Anthony S. Fiorino · Shlomo Brill Lola M. Reid

Phenotypic characterization of rat hepatoma cell lines and lineage-specific regulation of gene expression by differentiation agents &misc:Accepted in revised form: 7 May 1998

&p.1:Abstract Hepatoma cell lines can be characterized by their expression of hepatocyte- and biliary-specific genes and by their response to differentiating agents in a lineage-dependent manner. These characteristics can be used to map the maturational lineage position of the cell lines. Tissue-specific gene expression and regulation by heparin, dimethylsulfoxide (DMSO), and sodium butyrate (SB) were examined in three rat hepatoma cell lines and two rat liver epithelial cell lines. Based on antigenic profiles and gene expression in serum-supplemented medium, the hepatoma cell lines could be organized in distinct categories of hepatic differentiation. All three hepatomas expressed the following five genes: γ-glutamyl transpeptidase (GGT), glutathione-S-transferase pi (Yp), glutamine synthetase, and α5 and β1 integrin. Cell line H4AzC2 also expressed α-fetoprotein (AFP), albumin, IGF II receptor, and the biliary/oval cell antigens OC.2 and OC.3, a phenotype characteristic of fetal hepatocytes. FTO-2B cells lacked AFP, OC.2, and OC.3 but expressed albumin and IGF II receptor in addition to the five commonly expressed genes, consistent with a more hepatocyte-like phenotype. Cell line H5D.7 expressed neither albumin nor the IGF II receptor, but did express OC.2, OC.3, and α3 integrin in addition to the five commonly expressed genes, characteristic of biliary epithelial cells. Regulation of gene expression by heparin, DMSO, and SB was examined in cells cultured in hormonally defined medium. The patterns of regulation of AFP, albumin, GGT, and Yp were dependent upon the state of differentiation of the cell. FTO-2B cells regulatI. Zvibel (✉) · S. Brill Department of Gastroenterology, Tel Aviv Sourasky Medical Center, 6 Weizmann Str., Tel Aviv, Israel A.S. Fiorino Department of Dermatology, Two Rhoads Pavilion, University of Pennsylvania Medical Center, Philadelphia, PA 19104, USA L.M. Reid Department of Physiology, UNC School of Medicine, Chapel Hill, NC 27514, USA&/fn-block:

ed genes in a manner similar to that of E16 fetal hepatocytes, H4AzC2 regulated genes characteristic of both hepatocytic and biliary lineages, and H5D.7 regulated only biliary genes. Suppression of GGT by DMSO was uniformly observed. The three cell lines expressed equal amounts of HNF-4, but FTO-2B cells expressed more HNF-3β and less HNF-3α, while the reverse was true of H4AzC2 and H5D.7 cells.&bdy:

Introduction Between E10 and E15, the fetal rat liver is composed of bipotential cells capable of differentiation along either the hepatocytic or biliary lineage ([14, 31]; S. Brill et. al. submitted for publication). These fetal liver parenchymal cells, referred to as hepatoblasts, have a characteristic mixed phenotype, expressing an array of genes characteristic of both hepatocytes and biliary epithelial cells, including α-fetoprotein (AFP), albumin [14, 31], γ-glutamyl transpeptidase (GGT; [38]), glutathione-S-transferase pi (Yp; [34]), glutamine synthetase (GS; [1]), and IGF II receptors [3]. As hepatoblasts become committed to either the hepatic or biliary lineage, their gene expression shifts to the adult patterns specific for these two tissues [1, 3, 26, 39]. In the adult liver, only hepatocytes express albumin and a variety of other serum proteins and metabolic enzymes, while bile duct cells express GGT, Yp and cytokeratins 7 and 19 [39]. Other genes, such as cytokeratins 8 and 18, are expressed by cells of both lineages. Many fetal genes, such as AFP, GGT, and Yp, are commonly re-expressed in preneoplastic foci and hepatomas [4, 23, 30]. The mechanism and significance of this re-expression of “oncofetal” genes in neoplastic cells is controversial and its interpretation reflects two mutually exclusive models of hepatocarcinogenesis. Oncogenic events either target mature hepatocytes, which then “dedifferentiate” to express fetal genes, or target the proposed hepatic stem cells, which are blocked from normal differentiation by transformation and thus continue to ex-

216

press fetal genes. Oval cells, which are small, blast-like epithelial cells that proliferate in the livers of carcinogen-fed rats, are proposed to be activated or transformed liver stem cells, and a good deal of evidence has been marshaled to support this contention (reviewed in [12, 19, 32]). The oval cell phenotype is largely shared by rat liver epithelial (RLE) cells, primitive non-parenchymal epithelial cells that can be subcloned from primary cultures of hepatocytes and seem to have an especially plastic phenotype [32, 35]. By examining the regulation of gene expression in hepatoma cell lines and contrasting this regulation with that of fetal and adult liver cells, as well as that of oval and RLE cells, the relationship between hepatomas, oval cells, and normal cells can be somewhat elucidated. We recently examined the regulation of AFP, albumin, GGT, and Yp in embryonal, neonatal, and adult primary rat liver cultures [3], comparing the effects on gene expression of bovine lung-derived heparin, dimethylsulfoxide (DMSO), and sodium butyrate (SB). Heparin proteoglycans were previously shown to induce tissue-specific gene transcription in primary cultures of rat hepatocytes and transcription of autocrine growth factor genes in hepatomas [2, 33]. Both DMSO and SB were shown to either sustain differentiated functions in hepatocytes or to induce increased expression of tissue-specific genes in both hepatocytes, hepatomas and oval cells [3, 14, 15, 41]. In the present study, the degree of differentiation of three hepatoma cell lines was first evaluated by expression of liver-specific genes and antigens after growth in SSM. We then examined the expression of liver-specific genes and transcription factors in three hepatoma cell lines after growth in serum-free HDM alone, and when supplemented with heparin, DMSO or SB. For comparison, the gene expression and regulation of two RLE cell lines were examined in these same conditions. The expression and regulation of genes in the well-differentiated cell line FTO-2B was found to resemble that previously observed in fetal but not adult hepatocytes, while the expression of genes in the cell line H5D.7 was similar to that in biliary epithelial cells. Cell line H4AzC2 was found to have a mixed hepatocytic/biliary phenotype. Neither of the RLE cell lines demonstrated similarities to the three hepatoma cell lines.

Methods Cell lines The cell lines used in these studies were the following: the hepatoma cell lines H4AzC2 ([10]; provided by Dr. B. Nadel-Ginard, Harvard Medical School, Cambridge, Mass., USA), H5D.7 ([19]; provided by Dr. D. Hixson, Brown University, Providence, R.I., USA), and FTO-2B ([21]; provided by Dr. K. Fournier, Fred Hutchinson Cancer Center, Seattle, Wash., USA), and the RLE cell lines WB-F344 ([36]; provided by Dr. M.-S. Tsao, McGill University, Montreal, Quebec, Canada) and RLC ([16]; provided by Dr. H. E. Conrad, University of Illinois, Urbana, Ill., USA). Both H4AzC2 and FTO-2B were developed from the hepatoma cell line H4-II-EC3, derived from the Reuber H-35 hepatoma [27]. Media and culture conditions All reagents were obtained from Sigma Chemical Co. (St. Louis, Mo., USA) unless otherwise stated. The stocks of the cell lines were grown in 75-cm2 tissue culture flasks in SSM, consisting of RPMI 1640 supplemented with 10% fetal bovine serum (Gibco, Grand Island, N.Y., USA), 100 µg/ml penicillin, and 100 µg/ml streptomycin. The HDM was composed of RPMI 1640 supplemented with the following: 50 ng/ml EGF (Upstate Biotechnology, Lake Placid, N.Y., USA), 10 µg/ml insulin, 10 µg/ml glucagon, 10 µg/ml iron-saturated transferrin, 10 mU/ml prolactin, 10 µU/ml growth hormone, 100 pM zinc sulfate, 100 µM copper sulfate, 300 pM selenious acid, 7.6 µeq/ml of a free fatty acid mixture described by Chessebeuf and Padieu [7], 0.1% fatty acid-free bovine serum albumin, 100 µg/ml penicillin, and 100 µg/ml streptomycin. After trypsinization, the cells were plated in SSM overnight. The plates of cells were then washed with PBS, and the media replaced with either SSM, HDM, or HDM supplemented with 100 µg/ml bovine lung heparin, 1.6% DMSO, or 1 mM SB. After 48 h, plates of cells were trypsinized, cytospun, and fixed in 95% cold ethanol, or fixed in situ in 95% cold ethanol (both for immunofluorescence), or total RNA was extracted for Northern blot analysis. Immunohistochemistry The plates with cells grown under various conditions were washed with PBS and either trypsinized, cytospun, then fixed at 4°C for 10 min in 95% ethanol, or fixed directly in situ under the same conditions. The plates were incubated with the following primary antibodies: rabbit anti-mouse AFP (ICN Biomedicals, Costa Mesa, Calif., USA), rabbit anti-rat IGF II receptor (a kind gift of Dr. M. Czech, University of Massachusetts Medical School, Worcester, Mass., USA), and a panel of anti-rat oval cell antibodies (Table 1) developed by Dr. D. Hixson and Dr. R. Faris (Brown UniversityRhode Island Hospital, Providence, R.I., USA; [19]). The secondary antibodies used were PE-conjugated anti-rabbit IgG, PE-conjugated anti-mouse IgM (µ-chain specific), and PE-conjugated anti-mouse IgG (γ-chain specific; Southern Biotechnology, Birmingham, Ala., USA). GGT was detected using a histochemical assay. The cells were visualized using a Nikon microscope with fluorescence attachments.

Table 1 Panel of monoclonal antibodies (Mab) to rat oval cells&/tbl.c:& Antigen

Mab

Specificity

H.2 OC.2 OC.3 HBD.1

258.26 270.38 374.3 270.26

55-kDa protein in late fetal and adult hepatocytes Found in oval, bile duct, and rat liver epithelial (RLE) cells, as well as some fetal parenchymal cells Distribution similar to that of OC.2 46-kDa protein first seen weakly in E12 parenchymal cells; also expressed in adult hepatocytes and weakly on bile ducts Membrane antigen on bile duct and oval cells

&/tbl.:

–

18.11

217 Molecular hybridization assays The cells were washed with PBS and total RNA isolated from the cells by the RNAzol method [8]. The RNA samples were resolved by electrophoresis through 1% agarose, submerged-slab, denaturing formaldehyde gels in MOPS buffer [29]. RNA was transferred to Gene Screen (New England Nuclear, Boston, Mass., USA), and the RNA-containing filters were prehybridized in a hybridization solution containing 100 µg/ml salmon sperm DNA, then hybridized with the appropriate cDNA probes for 48 h at 42°C. The cDNA clones complementary to specific mRNAs were radioactively labeled with [32P] dCTP by primer extension [13]. The blots were washed four times with 1×SSC and 0.1% SDS at 60°C and exposed to X-ray film. cDNAs were generously provided by the following investigators: Dr. M. Manson, MRC Oncology Unit, Manchester, UK (rat GGT); Dr. M. Zern, Brown University, Providence, R.I., USA (rat albumin); Dr. S. Tilghman, Princeton University, Princeton, N.J., USA (mouse AFP); Dr. J. Darnell, Rockefeller University, New York, N.Y., USA, (mouse GS, HNF-3α, and HNF-3β, and rat HNF-4); Dr. I. Listowsky, Albert Einstein College of Medicine, Bronx, N.Y., USA (rat Yp); Dr. M. Hemmler,

Dana Farber Cancer Institute, Boston, Mass., USA (human α3 and α5 integrin); and Dr. R. Hynes, Massachusetts Institute of Technology, Cambridge, Mass., USA (β1 integrin).

Results Screening of hepatoma cell lines The three hepatoma cell lines, H5D.7, H4AzC2, and FTO-2B, were cultured for 2 days in SSM, then total RNA was extracted for Northern blot analysis (Fig. 1). Expression of GGT, Yp, GS, and α5 and β1 integrin was common to all three cell lines. FTO-2B cells expressed albumin and not AFP, H4AzC2 cells expressed both albumin and a low-molecular weight AFP mRNA that corresponds to either the 1.7-kb fetal or 1.4-kb adult variant AFP transcripts [29], and H5D.7 cells did not express albumin but did express both the 2.1-kb fetal and the variant AFP transcripts. Furthermore, H5D.7 cells expressed α3 integrin, found in biliary epithelial cells but not hepatocytes [42]. Expression of biliary markers GGT and Yp was also the strongest in H5.D7 cells (Fig 1). Duplicate plates of the three cell lines were cultured for 2 days in SSM, then fixed in 95% cold ethanol and examined by immunofluorescence with a panel of monoTable 2 Summary of immunofluorescence and Northern blot analysis of hepatoma cell lines cultured in the presence of 10% fetal calf serum. α-Fetoprotein (AFP), albumin, glutamine synthetase (GS), glutathione-S-transferase pi (Yp) γ-glutamyl transpeptidase (GGT) and α1, α3 and α1 were detected by Northern blot analysis. All the markers in the table, with the exception of GS, Yp, and integrins α3, α5 and β1 were also detected by immunohistochemistry [HB hepatoblast (bipotential fetal liver parenchymal cell from before E14), H adult hepatocyte; B adult biliary epithelial cell, – not detected, + weak expression, ++ strong expression]&/tbl.c:& Marker

Cell type specificity

Cell line H5D.7 H4AzC2 FTO-2B

AFP (2.1 kb) (1.7/1.4 kb) Albumin H.2b GS IGF II Receptora, c GGTe Yp OC.2a OC.3a α3 α5 α1 HBD.1a

Fig. 1 Northern blot analysis of 10 µg total RNA from hepatoma cell lines grown in SSM. The blots were sequentially hybridized with 32P-labeled probes for AFP, albumin-ALB), GGT, Yp, GS, and α3, α5, and β1 integrins. Ethidium bromide-stained rRNA is shown to demonstrate loading&ig.c:/f

HB – HB, H HBb, H HB, some H HB, some H HB, B HB, B HB, B HB, B B HB, H, B HB, H, B HB, H, Bf

+ + – – + – ++ ++ + + + + + –

– + + – ++ + + ++ + + – + + +

– – + – + +d + + – – – + + +

a Determined by immunofluorescence. b Appears at E16 c Staining localized to the Golgi d 50% of the cells are positive e Histochemical detection was in agreement with mRNA levels f HBD.1 is found weakly on fetal parenchymal cells and adult bile

ducts, and strongly on adult hepatocytes&/tbl.:

218 Fig. 2 Northern blot analysis of 10 µg total RNA from hepatoma cell lines grown for 48 h in HDM alone (A), supplemented with 100 µg/ml bovine lung heparin (B), 1.6% DMSO (C), or 1 mM SB (D). The blots were sequentially hybridized with 32P-labeled probes for AFP, albumin-ALB, GGT, and Yp. Ethidium bromide-stained rRNA is shown to demonstrate loading (last two rows of gels); the top corresponds to the blot probed for GGT and the one below was hybridized with the three other probes&ig.c:/f

Table 3 Summary of the regulation of liver-specific genes and transcription factors by heparin (Hep), (DMSO), and sodium butyrate (SB; +, up-regulation, –, down-regulation, 0 no effect, ND, no detectable signal)&/tbl.c:& Marker

AFP (2.1 kb) AFP(1.7/1.4 kb) Albumin GGT Yp HNF-3α HNF-3β HNF-4

H5D.7

H4AzC2

FTO-2B

Hep

DMSO

SB

Hep

DMSO

SB

Hep

DMSO

SB

ND – 0 – 0 0 0 0

ND 0 0 – + + + 0

ND 0 0 0 + 0 0 +

ND 0 0 0 0 0 0 0

ND – 0 – + + + 0

ND 0 + 0 + + + +

ND – + + 0 + + +

+ + + – 0 0 + +

+ + + + 0 + + 0

&/tbl.:

clonal antibodies prepared against oval cells (Table 1). H5D.7 cells expressed only OC.2 and OC.3, found on oval cells, hepatoblasts, and biliary epithelial cells. Both of these antigens were expressed on H4AzC2 cells, in addition to HBD.1, expressed faintly on hepatoblasts and biliary epithelium and more strongly on hepatocytes and the IGF II receptor, expressed by hepatoblasts and some adult hepatocytes. In contrast, FTO-2B expressed only HBD.1 and the IGF II receptor. The expression of these genes and markers is summarized in Table 2.

Gene expression in HDM and regulation by heparin, DMSO and SB The pattern of gene expression of the three hepatoma lines was slightly altered in HDM as compared to SSM (Fig. 2). FTO-2B cells, which do not express AFP in SSM, express low levels of the variant transcript in HDM. The faint albumin expression, seen in cell line H4AzC2 when grown in SSM, is not present in HDM. The 2.1-kb fetal AFP transcript (26) is not present when H5D.7 cells

219 Fig. 3 Northern blots of 10 µg total RNA from hepatoma cell lines grown for 48 h in HDM alone (A), supplemented with 100 µg/ml bovine lung heparin (B), 1.6% DMSO (C), or 1 mM SB (D). The blots were hybridized with 32P-labeled cDNAs encoding the transcription factors HNF-3, HNF-3, and HNF4. Ethidium bromide-stained rRNA is shown to demonstrate loading&ig.c:/f

Table 4 Summary of immunofluorescence and Northern blot&/tbl.c:& analysis of RLE cell lines cultured in HDM Marker

AFP Albumin GGT Yp IGF II Receptorc H.2 OC.2 OC.3 HBD.1 18.11 HNF-3α HNF-3β HNF-4

Cell Line RLC

WB-F344

– – –b – – – + + – – + – –

±a – – + + – – – ND + + + –

a

AFP was detectable in WB-F344 by immunofluorescence, but not by Northern blot. SB appeared to increase AFP staining b GGT was up-regulated in RLC by SB, as detected by Northern blot and histochemical staining c Staining localized to the Golgi ND = not detectable &/tbl.:

are cultured in HDM. GGT expression is increased relative to the expression of other genes in all three cell lines. Cell lines H4AzC2 and H5D.7 show more sharply decreased relative expression of Yp compared to FTO-2B. The pattern of gene expression induced by heparin, DMSO, and SB further distinguishes between the three

cell lines (Fig. 2, Table 3). In H5D.7 cells, heparin reduced AFP and GGT. DMSO also inhibited GGT while inducing Yp, which was also induced by SB. In H4AzC2 cells, heparin showed no activity on the genes examined, except a slight down-regulation of GGT. DMSO reduced AFP and GGT expression, while SB weakly induced AFP and albumin and strongly induced Yp. In FTO-2B cells, heparin inhibited AFP expression, while it up-regulated albumin and GGT. DMSO also induced albumin, but also induced AFP, while inhibiting GGT. In contrast, SB induced all three of these genes. None of the agents significantly regulated the low Yp expression in FTO-2B cells. In all cases, the regulation of GGT mRNA abundance corresponded to GGT activity as measured by histochemistry (not shown).

Expression and regulation of liver-specific transcription factors All three hepatoma cell lines expressed the liver-specific transcription factors examined: HNF-3α, HNF-3β, and HNF-4 (Fig. 3, Table 4). The ratio of HNF-3α expression to HNF-3β expression was low in FTO-2B cells, but high in H4AzC2 and H5D.7, whereas HNF-4 mRNA levels were equal in all three cell lines. In H5D.7 cells, heparin failed to regulate the transcription factors examined, while DMSO induced HNF-3α and HNF-3β, and SB induced HNF-4. H4AzC2 cells regulated the tran-

220

scription factors in response to heparin and DMSO in a manner similar to that observed in H5D.7. However, H4AzC2 cells induced all three transcription factors in response to SB. In FTO-2B cells, heparin up-regulated all three transcription factors, while DMSO up-regulated HNF-3β and HNF-4, and SB up-regulated HNF-3α and HNF-3β. Gene expression and regulation in RLE cell lines Two RLE cell lines, RLC and WB-F344, were screened for gene expression using immunofluorescence and Northern blot analysis (Table 4). At confluence, neither cell line expressed albumin, GGT, the hepatocyte-specific antigen H.2, or the transcription factor HNF-4. RLC cells were negative for most markers examined, with the exception of OC.2, OC.3, and HNF-3α. The only regulation seen in this line was the induction of GGT by SB; heparin and DMSO failed to regulate any of the genes or antigens examined. In contrast, WB-F344 cells expressed Yp, IGF II receptor, 18.11, HNF-3, and HNF-3, but did not express OC.2 or OC.3. By immunofluorescence, WB-F344 cells expressed AFP, which was up-regulated by SB; however, no AFP message was detectable by Northern blot analysis.

Discussion The purpose of our study was to determine tissue-specific gene expression and regulation in hepatoma cell lines and compare it to the gene phenotype and regulation in fetal and adult hepatocytes, performed in a previous study [3]. We have examined gene expression and differential regulation of liver-specific genes and transcription factors by heparin, DMSO and SB in three hepatoma cell lines and two RLE cell lines. We phenotypically characterized the hepatoma cell lines cultured in SSM and found these cell lines to resemble subpopulations of fetal liver parenchymal cells. All three hepatoma cell lines expressed GS, GGT, Yp, and α5 and β1 integrins. FTO-2B expressed genes characteristic of the hepatocytic phenotype, including albumin and IGF II receptor, and failed to express biliary markers OC.2 and OC.3. H5D.7 cells expressed biliary markers GGT, Yp, OC.2, OC.3 and α3 integrin, but no albumin or IGF II receptor, found in fetal and adult hepatocytes. Cell line H4AzC2 expressed markers of both hepatocytic and biliary lineages, including hepatocytic markers AFP, albumin and IGF II receptor and biliary markers OC.2 and OC.3. In this respect, H4AzC2 can be labeled as a less differentiated cell line than either H5D.7 or FTO-2B, since co-expression of hepatocytic and biliary markers is characteristic of early fetal liver cells. Both FTO-2B and H4AzC2 were derived from the same parent line through different regimens; these phenotypic differences may be the result of either a selection process or different degrees of genetic alterations.

In order to further detail the phenotype of the cell lines, we examined the expression of selected liver-specific genes under a variety of differentiation conditions. Gene expression in serum-free HDM and regulation by heparin, DMSO, and SB indicated that the hepatoma cell lines of different phenotypes respond distinctly to differentiation agents. The only effect common to all three cell lines was that DMSO suppressed GGT expression and up-regulated HNF-3β. Further confirming the phenotype of the cell lines, H5D.7 cells up-regulated Yp but not hepatocyte-specific genes in response to the differentiation agents DMSO and SB, whereas in response to SB, H4AzC2 cells up-regulate both the hepatocytic marker albumin and the biliary marker Yp. Of the three hepatoma cell lines, only FTO-2B exhibited similarities to regulation previously observed in fetal liver cells: FTO-2B resembled cultured E16 hepatocytes in their gene expression and response to DMSO and SB [3]. This indicates that this cell line may be a reasonable model for E16 hepatocytes and is probably a poor model for adult hepatocytes in studies of gene expression and regulation. Both H5D.7 and H4AzC2 are even poorer models of either fetal or adult hepatocytes. Hepatocarcinogenesis has been proposed to result from the transformation of progenitor cells in the adult liver, which subsequently give rise to tumors expressing fetal genes [12, 19, 32, 35]. In this model, the activation of progenitor cells during carcinogenesis recapitulates the process of hepatocellular differentiation that occurs during liver development. Thus, multiple cell populations in carcinogen-treated rats could, upon transformation, give rise to tumors predicted to have phenotypes reflecting different cellular populations present during liver development. We have identified these characteristics in the cell lines examined (Fig. 4). These results do not exclude the possibility of a dedifferentiation process in which transformed adult cells re-express fetal genes. However, adult rat hepatocytes immortalized with SV40 and then transfected with the ras oncogene, maintain high levels of liver-specific genes [11, 20] and minimal levels of the fetal genes AFP and Yp, in spite of the ability of such cells to form anaplastic tumors in nude mice. These studies indicate that transformation of adult cells does not necessarily induce fetal gene expression and provide indirect support for the assumption that fetal gene expression in hepatoma cells reflects gene expression in the cells that gave rise to the tumors. The three hepatoma cell lines we examined are similar to the hepatoblast, committed hepatocytic, and committed biliary liver cell populations in the fetal liver. We compared gene expression in hepatoma cell lines to that of two RLE cell lines, RLC and WB-F344, which are not similar in their gene expression (Table 4). Cell line RLC, reported to have low levels of tyrosine aminotransferase and phosphoenolpyruvate carboxykinase [40], expressed very few markers, but was positive for OC.2 and OC.3, indicating it may have a biliary phenotype. Cell line WB-F344 appeared to have a more hepatocytic phenotype, although in contrast to previous reports [36, 37],

221

Fig. 4 Stem cell model of liver development and hepatocarcinogenesis (thin arrows replicating populations of cells, thick arrows a precursor-progeny relationship in the direction of increasing differentiation). Liver development (top): Prior to E14, the liver consists of a large number of bipotential hepatoblasts that have begun to express various liver-specific genes. These cells are proposed to be descendents of hepatic stem cells. Later in fetal life, the hepatocellular population increases in both cell number and specialization. Two new cellular subpopulations appear: a population committed to the hepatocyte lineage and a population committed to the biliary lineage, which are descendents of the bipotential hepatoblasts. These groups of committed cells may be heterogeneous, containing cells resembling adult hepatocytes or biliary epithelial cells, as well as morphologically indistinct cells that have become restricted to differentiation along only one liver lineage. With lineage committment, gene expression begins to shift to the adult patterns. The cell populations and expression of the genes and markers studied in this report are illustrated. Hepatocarcinogenesis (bottom): The loss of mature parenchyma to a carcinogenic insult activates stem cell proliferation and generates populations of cells phenotypically resembling those seen in fetal liver development. Oval cell proliferations may represent expanded stem cells, their descendent hepatoblast-like cells, more differentiated committed hepatocytic and biliary cells, or all of these. Transforming events can target any population of mitotically active cells; thus, there are four possible sites of transformation during carcinogenesis (illustrated). Transformation at site 1, the stem cell, would result in mixed and/or metaplastic tumors. Transformation at site 2, the hepatoblast population, would result in tumors with phenotypes resembling hepatoblasts, represented by cell line H4AzC2 and/or tumors of a mixed phenotype. Transformation at site 3 would yield well-differentiated hepatomas, represented by cell line FTO-2B, and transformation at site 4 would yield tumors with a biliary phenotype, represented by cell line H5D.7&ig.c:/f

we found no albumin or GGT expression. Recently, a study of WB-F344 transplantation to syngeneic hosts revealed that these cells incorporate exclusively into hepatocytic plates but not biliary structures and morphologically differentiate into hepatocytes [8], supporting a hepatocytic lineage for this cell line. In addition to differing from each other, both RLE cell lines differed significantly from the hepatoma cell lines examined. Neither cell line expressed GGT or HNF-4, both of which were expressed by the hepatoma cell lines, although GGT was induced by SB in RLC cells. Moreover, these cell lines were largely unresponsive to the differentiation effects of DMSO, SB and heparin. These results indicate that RLE cells, which are immortal but not tumorigenic, represent a subpopulation of hepatic cells distinct from both normal liver parenchymal cells and hepatoma cells. Using the differentiation agents DMSO, SB and heparin, we were not able to significantly induce either cell line toward either the hepatocytic or biliary lineages. Thus, the complex environmental cues present in vivo and driving the morphological differentiation of WB-F344 into hepatocytes after transplantation cannot be reproduced by the addition of DMSO, SB or heparin to cultured RLE cells. It is more likely that a combination of hormonal stimulation and cell-cell contacts, in the proper extracellular matrix environment, will be necessary to induce RLE differentiation in vitro. The mechanisms by which heparin, DMSO, and SB influence differentiation and gene expression are poorly

222

understood. Heparins and other proteoglycans are known to exert a variety of effects on cells in culture through mechanisms that remain undefined [2, 33]. DMSO may foster maintenance of liver-specific functions through scavenging of hydroxyl radicals [41], and DMSO-responsive elements have been reported in an MHC class I gene promoter [6]. SB induces albumin expression in cultured rat hepatoblasts [3, 14] and oval cells [15]. The effects of SB on gene expression in different systems are thought to be mediated by the inhibition of a nuclear deacetylase, leading to increased histone acetylation and a change in the organization of chromatin [22]. Recently, an SB-responsive element has been identified in the 5’ region of the embryonal chicken β-globin p gene [17]. Many liver-enriched transcription factors have been identified, including the C/EBP, HNF-1, HNF-3, and HNF-4 families. Liver-specific genes often contain multiple binding sites for several of these transcription factors in their promoter and enhancer elements. There is a hierarchical regulation of these transcription factors; for instance, both HNF-1 and HNF-4 activate expression of HNF-1 [24]. Significantly, the degree of differentiation in various hepatoma cell variants was shown to depend on the expression of HNF-4 and HNF-1 [18]. HNF-4 is the earliest liver-enriched transcription factor detected to date. In carcinogen-treated rats, biliary epithelial cells up-regulate multiple transcription factors but fail to express HNF-4, oval cells poorly express HNF-4, and hepatocytes strongly express HNF-4 [5]. Thus, HNF-4 may be a master regulator gene, and it is tempting to connect the paucity of liver-specific gene expression in the two RLE cell lines with the absence of HNF-4 expression. Although the three transcription factors we tested were regulatable by heparin, DMSO, and SB, it is impossible to draw conclusions between this regulation and the observed changes in liver-specific gene expression. Further work can directly address in transcription factors associated with shifts in gene expression observed in hepatoma and RLE cell lines exposed to differentiation-enhancing agents. &p.2:Acknowledgements I. Zvibel and A. Fiorino contributed equally to the discussions and writing process through which this article developed. These studies were supported by an American Cancer Society grant (BE-92 C), an NIH grant (DK44266), and a grant from The Council for Tobacco Research (1897 A and 1897B). Shlomo Brill received support from the Department of Molecular Pharmacology, The Richard Molin Foundation, and by an American Physician’s Fellowship. Salary support for Anthony S. Fiorino derived from an NIH M.S.T.P. grant (5T32-GM07288; Betty Diamond, PI). We wish to thank Dinish Williams for technical support, Patricia Holst for laboratory management, and Rosina Passela for secretarial assistance.

References 1. Bennett AL, Paulson KE, Miller RE, Darnell JE Jr (1987) Acquisition of antigens characteristic of adult pericentral hepatocytes by differentiating fetal hepatoblasts. J Cell Biol. 105:1073–1085

2. Brill S, Holst PA, Zvibel I, Fiorino AS, Sigal SH, Somasundaran U, Reid LM (1994) Extracellular matrix regulation of growth and gene expression in liver cell lineages and hepatomas. In: Arias IM, Boyer JL, Fausto N, Jakoby WB, Schachter DA, Shafritz DA (eds) The liver: biology and pathobiology (3rd edn). Raven Press, New York 3. Brill S, Zvibel I, Reid LM (1995) Maturation-dependent changes in the liver-specific gene expression in embryonal versus adult primary liver cultures. Differentiation 59:95– 102 4. Cameron R, Kellen J, Kolin A, Malkin A, Farber E (1978) γGlutamyltransferase in putative premalignant liver cell populations during hepatocarcinogenesis. Cancer Res 38:823–829 5. Cereghini S, Yaniv M, Cortese R (1990) Hepatocyte dedifferentiation and extinction is accompanied by a block in the synthesis of mRNA coding for the transcription factor HNF1/LFB1. EMBO J 9:2257–2263 6. Cerosaletti KM, Woodward JG, Lord EM, Frelinger JG (1992) Two regions of the H-2 Dd promoter are responsive to dimethylsulfoxide in line 1 cells by a mechanism distinct from IFN. J Immunology 148:1212–1221 7. Chessebeuf M, Padieu P (1984) Rat liver epithelial cultures in a serum-free medium: primary cultures and derived cell lines expressing differentiated functions. In Vitro 20:780–795 8. Chomcznyski P, Sacchi N (1987) Single-step method of RNA isolation by acid guanidinium thiocyanate-phenol-chloroform extraction. Anal Biochem 162:156–159 9. Coleman WB, Wennerberg AE, Smith GJ, Grisham JW (1993) Regulation of the differentiation of diploid and some aneuploid rat liver epithelial (stemlike) cells by the hepatic microenvironment. Am J Pathol 142:1373–1382 10. Dannies PS, Tashjian AH Jr (1974) Unexpected production of a mouse serum protein by rat-mouse somatic cell hybrids. In: Davidson RL, Cruz F de la (eds) Somatic cell hybridization. Raven Press, New York, pp 163–171 11. Fang XJ, Keating A, Flowers M, Liew CC, Gupta H, Mills GB, Sherman M (1993) The v-raf oncogene enhances tumorigenicity and suppresses differentiation in vivo in a rat hepatocyte cell line. Carcinogenesis 14:669–674 12. Fausto N (1990) Hepatocyte differentiation and liver progenitor cells. Curr Opin Cell Biol 2:1036–1042 13. Feinberg AP, Vogelstein B (1984) A technique for radiolabeling DNA restriction endonuclease fragments to high specific activity. Anal Biochem 137:266–267 14. Germain L, Blouin MJ, Marceau N (1988) Biliary epithelial and hepatocytic cell lineage relationships in embryonic rat liver as determined by the differential expression of cytokeratins, alpha-fetoprotein, albumin, and cell surface-exposed components. Cancer Res 48:4909–4918 15. Germain L, Noel M, Gourdeau H, Marceau N (1988) Promotion of growth and differentiation of rat ductular oval cells in primary culture. Cancer Res 48:368–378 16. Gerschenson LE, Anderson M, Molson J, Okigaki T (1970) Tyrosine transaminase induction by dexamethazone in a new rat liver cell line. Science 170:859–861 17. Glauber JG, Wandersee NJ, Little JA, Ginder GD (1991) 5’Flanking sequences mediate butyrate stimulation of embryonic globin gene expression in adult erythroid cells. Mol Cell Biol 11:4690–4697 18. Griffo G, Hamon-Benais C, Angrand PO, Fox M, West L, Lecoq O, Povey S, Cassio D, Weiss M (1993) HNF-4 and HNF-1 as well as a panel of hepatic functions are extinguished and reexpressed in parallel in chromosomally reduced rat hepatomahuman fibroblast hybrids. J Cell Biol 121:887–898 19. Hixson DC, Faris RA, Thompson NL (1990) An antigenic portrait of the liver during carcinogenesis. Pathobiology 58:65– 77 20. Isom HC, Woodworth CD, Meng Y, Kreider J, Miller T, Mengel L (1992) Introduction of the ras oncogene transforms a simian virus 40-immortalized hepatocyte cell line without loss of expression of albumin and other liver-specific genes. Cancer Res 52:940–948

223 21. Killary AM, Lugo TG, Fournier RE (1984) Isolation of thymidine kinase-deficient rat hepatoma cells by selection with bromodeoxyuridine, Hoechst 33258, and visible light. Biochem Genet 22:201–213 22. Kruh J (1982) Effects of sodium butyrate, a new pharmacological agent, on cells in culture. Mol Cell Biochem 42:65–82 23. Kuhlmann WD (1978) Localization of α-fetoprotein and DNA synthesis in liver cell populations during experimental hepatocarcinogenesis in rats. Int J Cancer 21:368–380 24. Kuo CJ, Conley PB, Chen L, Sladek FM, Darnell JE Jr, Crabtree GRA (1993) Transcriptional hierarchy involved in mammalian cell-type specification. Nature 355:457–461 25. Lemire JM, Fausto N (1991) Multiple alpha-fetoprotein RNAs in adult rat liver: Cell type-specific expression and differential regulation. Cancer Res 51:4656–4664 26. Panduro A, Shalaby F, Shafritz DA (1987) Changing patterns of transcriptional and post-transcriptional control of liver-specific gene expression during rat development. Genes Dev 1:1172–1182 27. Pitot HC, Peraine C, Morse PA, Potter VR (1964) Hepatomas in tissue culture compared with adapting liver in vivo. Natl Cancer Inst Monogr 13:229–241 28. Saito H, Morizane T, Watanabe T, Kagawa T, Miyaguchi S, Kumagai N, Tsuchiya M (1991) Differentiating effect of sodium butyrate on human hepatoma cell lines PLC/PRF/5, HCCM and HCC-T. Int J Cancer 48:291–296 29. Sambrook J, Fritsch EF, Maniatis T (1989) Molecular Cloning: A Laboratory Manual, 2nd ed. Cold Spring Harbor Press, Cold Spring Harbor, New York 30. Sato K (1989) Glutathione transferases as markers of preneoplasia and neoplasia. Adv Cancer Res 52:205–255 31. Shiojiri N, Lemire JM, Fausto N (1991) Cell lineages and oval cell progenitors in rat liver development. Cancer Res 51:2611– 2620

32. Sigal SH, Brill S, Fiorino AS, Reid LM (1992) The liver as a stem cell and lineage system. Am J Physiol 263:G139–G148 33. Spray DC, Fujita M, Saez JC, Choi H, Watanabe T, Hertzberg EL, Rosenberg LC, Reid LM (1987) Glycosaminoglycans and proteoglycans induce gap junction synthesis and function in primary liver cultures. J Cell Biol 105:541–551 34. Tee LBG, Gilmore KS, Meyer DJ, Ketterer B, Vandenberghe Y, Yeoh GCT (1992) Expression of glutathione S-transferase during rat liver development. Biochem J 282:209–218 35. Thorgeirsson SS (1993) Hepatic stem cells. Am J Pathol 142:1331–1333 36. Tsao MS, Smith JD, Nelson KG, Grisham JW (1984) A diploid epithelial cell line from normal adult rat liver with phenotypic properties of “oval” cells. Exp Cell Res 154:38–52 37. Tsao MS, Earp HS, Grisham JW (1986) The effects of epidermal growth factor and the state of confluence on enzymatic activities of cultured rat epithelial cells. J Cell Physiol 126:167–173 38. Vanderlaan M, Phares W (1981) γ-Glutamyltranspeptidase: a tumour cell marker with a pharmacological function. Histochem J 13:865–877 39. Van Eyken P, Sciot R, Desmet V (1988) Intrahepatic bile duct development in the rat: a cytokeratin-immunohistochemical study. Lab Invest 59:52–59 40. Van Rijn H, Bevers MM, Wijk R van, Wicks WD (1974) Regulation of phosphoenolpyruvate carboxykinase and tyrosine aminotransferase in hepatoma cell culture: comparative studies in H35, HTC, NH1C1 and RLC cells. J Cell Biol 60:181– 191 41. Villa P, Arioli P, Guaitani A (1991) Mechanism of maintenance of liver-specific functions by DMSO in cultured rat hepatocytes. Exp Cell Res 194:157–160 42. Volpes R, Van Den Oord JJ, Desmet VJ (1991) Distribution of the VLA family of integrins in normal and pathological human liver tissue. Gastroenterology 101:200–206