Adenovirus-mediated transfer of CCAAT/enhancer-binding protein-alpha identifies a dominant antiproliferative role for this isoform in hepatocytes.

CCAAT/enhancer-binding protein (C/EBP) isoforms are thought to be important regulators of the hepatocyte phenotype. However, the specific physiological roles of different isoforms are poorly understood because hepatocytes express multiple C/EBPs, and various isoforms have overlapping functions. To identify the functions of C/EBPalpha in mature hepatocytes, replication-defective adenovirus vectors were used to efficiently and homogeneously overexpress the mouse C/EBPalpha gene in a SV40 virus-conditionally transformed rat hepatocyte line that can be induced to express C/EBPbeta and C/EBPdelta but that has little endogenous C/EBPalpha expression. Hepatocytes were infected with a recombinant adenovirus vector carrying the cDNA for C/EBPalpha driven by Rous sarcoma virus promoter elements (AdCEBPalpha) or a similar vector carrying the Escherichia coli lacZ gene (Adbetagal). Staining for beta-galactosidase demonstrated an infection efficiency of 100% at a multiplicity of infection of 25 plaque-forming units/cell and persistence of foreign gene expression for at least 9 days. Cultures infected with AdCEBPalpha had 50-fold higher levels of C/EBPalpha mRNA and protein than those infected with Ad-beta-gal, but similar expression of C/EBP-beta. Infection with AdCEBPalpha inhibited proliferation in cells expressing little C/EBPbeta, even when proliferation was driven by the SV40 transforming antigen, and also blunted mitogenic induction of the c-myc proto-oncogene in nontransformed cells with high levels of C/EBPbeta. Although overexpression of C/EBPalpha consistently increased C/EBPalpha DNA binding activity, it was not sufficient for albumin expression. Infection with AdCEBPalpha only increased albumin mRNA levels in nontransformed cells that also expressed relatively high levels of C/EBPbeta. Thus, in hepatocytes, C/EBPalpha has a dominant antiproliferative function, but must interact with other factors to regulate hepatocyte-specific gene expression.

CCAAT/enhancer-binding proteins (C/EBPs) 1 are members of the basic leucine zipper superfamily of transcription factors. To date, five distinct genes that encode different C/EBP iso-forms have been identified (1)(2)(3)(4)(5)(6)(7)(8)(9). These genes vary in their pattern of tissue distribution. For example, C/EBP␣ is expressed predominately in tissues that regulate metabolic homeostasis (e.g. liver, fat, and intestine), while C/EBP␤ and C/EBP␦ are more ubiquitous (10 -13). Work in several cell lines indicates that, although they bind to similar regulatory elements, unique C/EBP isoforms differ in their ability to activate various tissue-specific promoters. In cells that express more than one C/EBP isoform, different C/EBPs interact to regulate gene expression by forming homo-and heterodimers (14,15). Since the regulatory regions of many tissue-specific genes contain C/EBP-binding sites, such cells are able to variably titrate the transcriptional rates of a wide array of genes by modifying the relative activities of different C/EBP isoforms (5,12,16,17).
Mature hepatocytes express at least three C/EBP isoforms (␣, ␤, and ␦) that interact to regulate their phenotype (12). However, since these different C/EBP isoforms have overlapping functions (14), the specific physiological role of each isoform in these cells is not well understood. Attempts to modulate the relative levels of these isoforms in vitro and to examine the resulting phenotype have been thwarted, in part, because the transfection efficiency of both primary adult and neoplastic hepatocytes is relatively poor, in the range of 20 -30% (18 -20). Furthermore, basal expression of C/EBP␣ is low in these cell lines (21), unlike in mature hepatocytes, where C/EBP␣ is strongly and constitutively expressed (10,11). To overcome these experimental difficulties and to clarify the role of C/EBP␣ in mature hepatocytes, replication-defective adenovirus vectors were used to efficiently overexpress the mouse C/EBP␣ gene in a conditionally transformed rat hepatocyte line that can be induced to express C/EBP␤ and C/EBP␦ but that has little endogenous C/EBP␣ expression. In vitro adenoviral gene transfer has a significant advantage over traditional transient transfection strategies in that up to 100% of the targeted cells can be infected (22). The subsequent analysis reflects changes in a homogeneous population of cells and indicates that C/EBP␣ has a dominant antiproliferative function in hepatocytes, but must interact with other factors to regulate the expression of hepatocyte-specific genes.

EXPERIMENTAL PROCEDURES
Adenovirus Vector Preparation-The coding sequence for the mouse adipose C/EBP␣ gene was provided by M. D. Lane (Johns Hopkins University, Baltimore) (23). A 2200-base pair BamHI-HindIII fragment containing the entire coding sequence for the C/EBP␣ gene was excised from pBluescript (24) and subcloned into the vector pE1RSV (25), a modified version of p⌬E1sp1B (obtained from F. Graham, McMaster University, Hamilton, Ontario, Canada) that incorporated the Rous sarcoma virus long terminal repeat promoter, a multiple cloning site, and the polyadenylation signal from bovine growth hormone between adenovirus E1 recombination sequences. This plasmid, denoted pCEBP␣, was cotransfected with pJM17 (26), containing the full human adenovirus serotype 5 genome, into HEK293 cells using Lipofectamine (Life Technologies, Inc.). As described previously (25,27), homologous recombination between pCEBP␣ and pJM17 replaced portions of the adenovirus E1 regions with the 2200-base pair gene insertion to make AdCEBP␣. We confirmed successful insertion of the C/EBP␣ gene cassette by restriction enzyme digestion and by Northern and Western blot analyses as described below. The Ad␤gal virus containing the Escherichia coli lacZ gene with the human cytomegalovirus immediate early promoter was provided by G. Wilkinson (Department of Medicine, University of Wales College of Medicine, Cardiff, CF4 4XX, United Kingdom).
For large-scale preparations, 9 -18 T-150 flasks of HEK293 cells were infected with AdCEBP␣ or Ad␤gal at a multiplicity of infection (m.o.i.) of 1-5 plaque-forming units/cell. For detection of ␤-galactosidase activity, cells were fixed with 0.05% glutaraldehyde for 5 min at room temperature and stained with 5-bromo-4-chloro-3-indoyl ␤-D-galactoside solution. The cells were lysed by freeze-thawing, and the supernatant was purified on a CsCl step cushion and dialyzed against phosphate-buffered saline as described previously (25).
Cell Culture Experiments-A hepatocyte line (RALA255-10G) derived from adult rat hepatocytes that were conditionally transformed with mutant SV40 virus containing a temperature-sensitive T antigen was obtained from Dr. Janice Chou (National Institutes of Health, Bethesda, MD). At 33°C (permissive temperature), T antigen is expressed, and these cells actively proliferate, but do not express several hepatocyte-specific genes, including albumin. At 37-39°C (restrictive temperature), T antigen is degraded in RALA255 cells, so they are no longer transformed, and their proliferative activity decreases. This is accompanied by increased transcription of albumin and several other hepatocyte-specific genes (28). For experiments performed at 33°C, cultures were grown on plastic dishes to 30 -40% confluence and then exposed to Dulbecco's modified Eagle's medium, 2% fetal calf serum containing varying concentrations (10 -100 m.o.i.) of AdCEBP␣ or Ad␤gal for 2 h. After replacing this medium with Dulbecco's modified Eagle's medium containing 4% fetal calf serum and 0.2 M dexamethasone, cells were cultured for an additional 1-3 days. To measure treatment-related differences in proliferative activity, 5 Ci of [ 3 H]thymidine was added to cultures 2 h before harvesting, and the tritiumspecific activity of DNA was determined. For experiments performed at the restrictive temperature, cultures were grown on plastic dishes coated with type I collagen at 33°C until 70 -80% confluent and then treated with AdCEBP␣ or Ad␤gal (10 -100 m.o.i.) in Dulbecco's modified Eagle's medium containing 2% fetal calf serum for 2 h at 37°C. This medium was replaced with serum-free Chee medium with 1% dimethyl sulfoxide, ITS (insulin (5 g/ml), transferrin (5 g/ml), selenium (5 ng/ml)), and 1 M dexamethasone, and cells were cultured at 39°C for an additional 3 days. Culture under the latter conditions leads to significantly decreased proliferative activity of uninfected RALA255 cells by day 3. In some experiments, Chee medium was replaced with growth medium (Dulbecco's modified Eagle's medium containing ITS, 20 ng/ml epidermal growth factor, Ham's F-12 nutrient mixture (Sigma), 1 mM pyruvic acid, 260 M L-proline, and 24.5 mM Hepes) at the beginning of day 3.
RNA Isolation and Northern Blot Analysis-Total RNA was isolated from cultures according to the protocol of Chomczynski and Sacchi (29). Samples (20 g of RNA/lane) were loaded onto denaturing agarose gels, separated by electrophoresis, and transferred to nylon membranes as described previously (30). After rinsing membranes with methylene blue dye to ensure equivalency of RNA loading and transfer, membranes were hybridized at 42°C overnight with 32 P-labeled cDNAs for C/EBP␣, C/EBP␤, albumin, c-myc, or 18 S RNA. cDNAs for C/EBP␣ and C/EBP␤ were donated by M. D. Lane (23). The albumin cDNA was provided by D. A. Shafritz (Albert Einstein College of Medicine, New York) (31), and Dr. Chi Dang (Johns Hopkins University) provided the c-myc cDNA (32). The probe for 18 S RNA was obtained from Oncogene Science Inc. (Manhasset, NY). Blots were washed under moderately stringent conditions (30) and then exposed to film. Autoradiographs from triplicate experiments were analyzed by scanning laser densitometry. To quantify treatment-related differences in mRNA expression, results on each radiograph were normalized to 18 S RNA expression at the same time point, and differences in normalized mRNA levels were evaluated by analysis of variance.
Protein Extraction and Western Blot Analysis-Whole cell protein was extracted by boiling cell pellets in 60 mM Tris and 1% SDS for 8 min. Nuclear proteins were extracted from parallel cultures according to the method of Ossipow et al. (33) as described previously (34). The protein concentrations in whole cell and nuclear extracts were determined (34). Samples (200 g of whole cell protein/lane or 40 g of nuclear protein/lane) in Laemmli sample buffer were loaded onto SDS-12.5% polyacrylamide gels, separated by electrophoresis, and transferred to nylon membranes (34). After staining the gel and membrane with Coomassie Blue dye to ensure equivalency of protein loading and transfer, immunoblot analysis was performed (34). Blots were incubated with primary antiserum (1:1000 dilution) for C/EBP␣ or C/EBP␤ (gifts from M. D. Lane) (35), and immunoreactive bands were demonstrated by the chemiluminescence detection system (Amersham Corp.). Results of blots from triplicate experiments were analyzed by densitometry (Molecular Dynamics, Inc., Sunnyvale, CA). Densitometric data were evaluated by analysis of variance.

RALA255
Characterization-As expected, uninfected RALA255 cells under the influence of T antigen do not express detectable levels of albumin mRNA at 33°C, but gradually accumulate albumin message during culture at 39°C (Fig. 1). Although others have reported that these condition-dependent differences in albumin expression are transcriptionally regulated and reflect differences in the protection of several DNasesensitive sites in regulatory elements of the albumin gene (36,37), differences in albumin mRNA expression are not associated with comparable differences in C/EBP mRNA expression ( Fig. 1). Very little C/EBP␣ mRNA is detected during culture at either 33 or 39°C. Similarly, stable low levels of C/EBP␦ mRNA are detected only after prolonged exposure (21 days) of blots to film. C/EBP␤ message is more readily detected at both temperatures, but mRNA levels increase Ͻ2-fold when cultures are shifted from 33 to 39°C.
In contrast to the relatively temperature-insensitive message levels, immunoblot analysis reveals that levels of certain C/EBP isoforms increase at 39°C (Fig. 2). Although C/EBP␣ expression varies Ͻ2-fold, the nuclear concentrations of C/EBP␤ and C/EBP␦ increase 5-10-fold when cultures are shifted to the higher temperature (p Ͻ 0.05 for expression at 33°C versus expression during days 1-9 at 39°C). These increases in C/EBP␤ and C/EBP␦ expression are paralleled by increases in DNA binding activity of these isoforms in gel mobility shift experiments with an oligonucleotide probe that contains the C/EBP-binding site in the albumin gene ( Fig. 3) (36). Complex formation with this oligonucleotide is completely eliminated when a mixture of antibodies to C/EBP␣, C/EBP␤, and C/EBP␦ is added to the reaction mixture, but is unaltered by preimmune sera (see Fig. 7, lanes 9 and 10), confirming previous reports (36) that this probe specifically identifies C/EBP binding activity in liver nuclear extracts.
Total C/EBP binding activity is increased in nuclear extracts prepared from 39°C cultures compared with extracts from 33°C cultures. In experiments with extracts obtained from 39°C cultures, addition of anti-C/EBP␣ antibodies has little effect on complex formation; anti-C/EBP␦ antibodies result in a small supershifted band, and anti-C/EBP␤ antibodies disrupt most of the protein-probe complexes. Thus, almost all of the increased binding activity at 39°C is secondary to C/EBP␤. Because C/EBP␤ accounts for most of the C/EBP binding ac-tivity in these cells at both 33 and 39°C, subsequent experiments focused on identifying effects of C/EBP␣ gene transfer on the expression and function of the C/EBP␤ isoform.
Adenovirus-mediated Gene Transfer-Infection of RALA255 cultures with Ad␤gal (m.o.i. ϭ 25) results in expression of the heterologous gene in virtually 100% of the cells within 1 day (Fig. 4). Following exposure to AdCEBP␣ (m.o.i. ϭ 25), steadystate levels of C/EBP␣ mRNA are Ͼ50-fold higher than in cells treated with Ad␤gal. Increased expression of C/EBP␣ mRNA occurs both during culture at 33°C, when T antigen is actively expressed, and after 2 days of culture at 39°C, when T antigen has been completely degraded. In contrast, C/EBP␤ mRNA levels are similar in the two groups at both temperatures (Fig.  5). Infection with AdCEBP␣ increases albumin mRNA levels at 39°C (p Ͻ 0.01 versus Ad␤gal control on the third day of culture at 39°C), but not in cultures maintained at 33°C. However, as shown in a representative immunoblot (Fig. 6), the concentration of C/EBP␣ protein is increased in AdCEBP␣infected cultures maintained at either 33 or 39°C compared with parallel cultures treated with Ad␤gal. Triplicate experiments demonstrate that both whole cell and nuclear concentrations of C/EBP␣ protein are greater in AdCEBP␣-infected cultures than in Ad␤gal-infected cultures at every time point evaluated (i.e. during the first 3 days of culture at 33°C and on the third day of culture at 39°C). Evidence that the treatmentrelated differences in C/EBP␣ expression (illustrated in Fig. 6) are not merely artifacts produced by lane-lane differences in the amount of protein on the blot is provided in Fig. 6 (lower panel). When this representative blot was reprobed to search for treatment-related differences in C/EBP␤ expression, the results demonstrated that C/EBP␤ protein concentrations are similar in Ad␤galand AdCEBP␣-infected cultures. Thus, the results of Northern and Western blot analyses are consistent with each other and demonstrate that adenovirus-mediated transfer of C/EBP␣ results in overexpression of this C/EBP isoform, but does not alter the expression of C/EBP␤, a closely related transcription factor.
Virus-mediated overexpression of C/EBP␣ also has functional relevance. Gel mobility and supershift experiments demonstrate increased complex formation between C/EBP␣ and the C/EBP-binding site in the albumin enhancer in extracts from Nuclear extracts were prepared from cells grown to 70 -80% confluence at 33°C or from similarly confluent cultures maintained for varying periods of time (1-9 days) at 39°C, separated by SDS-polyacrylamide gel electrophoresis, and transferred to nylon membranes. A, C/EBP expression was analyzed by immunoblotting (40 g of nuclear protein/lane). The last lane is a standard and contains reticulocyte-translated C/EBP protein (40 g/ lane) that was prepared by incubating rabbit reticulocytes with RNA prepared from the respective cDNAs for C/EBP␣, C/EBP␤, and C/EBP␦ (44). B, triplicate immunoblots were performed and analyzed by scanning laser densitometry. Levels of each C/EBP isoform at the various time points during culture at 39°C were normalized for the expression of that isoform in 33°C nuclear extracts on the same blot. Results were analyzed by analysis of variance (*, p Ͻ 0.05 versus C/EBP␤ expression at 33°C; #, p Ͻ 0.05 versus C/EBP␦ expression at 33°C).
AdCEBP␣-treated cultures compared with Ad␤gal-infected cultures at both 33 and 39°C. As shown in Fig. 7 In parallel experiments, 70 -80% confluent cultures of RALA255 cells were infected with the same vectors for 2 h at 37°C and then shifted to 39°C and maintained at this temperature for 3 days. By day 3 at 39°C, T antigen had been degraded, and albumin mRNA levels were increased in uninfected cultures (see Fig. 1). A, total RNA was harvested from cultures at various time points, and treatment-related differences in mRNA levels of C/EBP␣, C/EBP␤, and albumin were evaluated by Northern blot analysis (20 g of RNA/lane). The autoradiographs shown were obtained by repeated hybridization of a single blot. The bottom panel illustrates the same blot after staining with methylene blue. This, coupled with normalization of results to 18 S RNA expression at the same time points (data not shown), confirms that differences in C/EBP␣ and albumin expression are not due to uneven RNA loading or transfer. The results shown are representative of triplicate experiments. B, densitometric analysis of these results and those from two other blots confirms that infection with AdCEBP␣ increases C/EBP␣ mRNA levels 50 -80-fold and albumin mRNA levels 4.5-7-fold (p Ͻ 0.01 for Ad-CEBP␣ at days 1-3 at 33°C and day 3 at 39°C versus Ad␤gal at the same time points). f, Ad␤gal; o, AdCEBP␣.

FIG. 3. Temperature-dependent differences in C/EBP binding activity for the albumin enhancer in RALA255 cells.
Gel mobility shift assays were performed using 23-base pair oligonucleotides that included the C/EBP-binding site in the albumin enhancer plus nuclear extracts (8 g of protein/lane) harvested from cells grown to 70 -80% confluence at 33°C or from similarly confluent cultures that had been maintained at 39°C for 3 days. Antisera specific for different C/EBP isoforms were added to some reaction mixtures (lanes 2-4 and 6 -8) to determine which isoforms were binding to C/EBP sites under the different culture conditions. indicate that the fastest moving residual bands contain C/EBP␤. Presumably, the residual slow moving bands in lane 7 also contain C/EBP␤ since addition of antibodies to all three C/EBP isoforms totally eliminates complex formation with this probe (lane 10). Taken together, these experiments indicate that, although infection of cultures with AdCEBP␣ results in increased C/EBP␣ mRNA, increased C/EBP␣ protein, and increased C/EBP␣ DNA binding activity, this is not sufficient to result in increased expression of albumin, a C/EBP-regulated gene, at 33°C. Similarly, C/EBP␤ binding activity is not associated with albumin gene expression at this temperature.
Since C/EBP␣ has been identified as a growth arrest gene in adipocytes, studies were carried out to determine if overexpression of C/EBP␣ could influence the proliferative activity of hepatocytes. As shown in Fig. 8A show results obtained when identical assays were performed with extracts harvested from Ad␤gal-infected cultures maintained at 39°C for 3 days. These are included as a control to demonstrate that addition of a mixture of C/EBP␣, C/EBP␤, and C/EBP␦ antisera (total volume of 3 l) virtually eliminates complex formation with the same probe (lane 10) but that addition of an equal volume of preimmune serum (IgG) has little effect.

FIG. 8. Effect of adenovirus-mediated gene transfer on T antigen-driven proliferation of RALA255 cells at 33°C. A, RALA255
cells were plated at low density (ϳ10% confluence) and infected with either AdCEBP␣ (CEBP␣) or Ad␤gal (␤gal). Photomicrographs were taken at the time of infection (day 0) and 1 or 2 days postinfection. These results are representative of those obtained in three other experiments and demonstrate the antiproliferative effects of AdCEBP␣. B, 30 -40% confluent cultures were infected with either AdCEBP␣ or Ad␤gal as described above. One, two, or three days later, triplicate cultures were incubated for 2 h in medium containing [ 3 H]thymidine and then immediately harvested. DNA was isolated, and treatmentrelated effects on proliferation were measured by comparing differences in the rate of incorporation of [ 3 H]thymidine. Results of four separate experiments are shown. ous after 1 day of culture, when the concentrations of DNA are not different statistically in the two treatment groups, but become less apparent over time, as DNA concentrations increase in the cultures treated with Ad␤gal (Fig. 8B). Infection with AdCEBP␣ does not overtly influence the expression of T antigen (Fig. 9), which is thought to drive proliferation in cultures maintained at 33°C (28).
When nearly confluent RALA255 cells are infected with Ad-CEBP␣ or Ad␤gal and cultures are shifted to 39°C, T antigen expression disappears after 2 days (data not shown) (28), C/EBP␤ is induced (Fig. 6), and proliferative activity declines in both groups (from 4413 Ϯ 176 to 179 Ϯ 2 and from 5198 Ϯ 160 to 169 Ϯ 11 cpm/g DNA, respectively, by the third day of culture at 39°C). As shown in Fig. 10, changing these cultures (which do not express T antigen) to growth factor-enriched medium increases steady-state levels of c-myc 3-5-fold in Ad␤gal-infected cultures (p Ͻ 0.01 versus Ad␤gal cultures without mitogens). AdCEBP␣-mediated expression of C/EBP␣ blunts this mitogenic induction of c-myc by at least 50%, such that c-myc mRNA levels are increased Ͻ1.5-fold (range of 0.9 -1.5-fold greater than without mitogens; p ϭ not significant).
Results of triplicate experiments indicate that, after treatment with growth factors, levels of c-myc mRNA are significantly lower in AdCEBP␣-infected cultures than in Ad␤gal-infected controls (p Ͻ 0.05). DISCUSSION C/EBP␣ has been incriminated as a component of a "differentiation-proliferation switch" based on work in adipocyte cell lines (15,38). Proliferating undifferentiated adipoblasts do not express C/EBP␣, and increased C/EBP␣ expression accompanies hormonal induction of differentiation in these cells (5,10,23,39,40). Transfection of adipoblasts with inducible expression vectors for C/EBP␣ arrests their proliferation and promotes differentiation in the presence of adipogenic hormones (15). In addition, during hormonally induced differentiation of preadipocytes, plasmid vector-directed expression of C/EBP␣ antisense RNA blocks expression of C/EBP␣ and inhibits the induction of adipocyte-specific genes (41,42). However, whether C/EBP␣ regulates proliferation and differentiated gene expression in other highly differentiated cells is less clear. Indeed, in some cells, including myeloid precursors, C/EBP␣ expression actually declines during terminal differentiation (43).
The adult liver expresses more C/EBP␣ than any other tissue (10), yet the physiological role of C/EBP␣ as a regulator of the hepatocyte phenotype remains undefined. In rats, hepatic expression of C/EBP␣ increases during development and is greater postnatally than earlier during embryogenesis (10,11,44). However, hepatocytes continue to proliferate actively for the first several weeks after birth (44). Furthermore, at birth, liver morphology appears normal in C/EBP␣ knockout mice, and these animals express many, albeit not all, hepatocytespecific genes (45). In the adult, hepatic concentrations of C/EBP␣ mRNA and protein remain relatively stable during FIG. 9. Effect of adenovirus-mediated gene transfer on T antigen expression in RALA255 cultures at 33°C. Immunocytochemistry was used to evaluate the expression of T antigen in AdCEBP␣and Ad␤gal-infected cultures. These experiments were performed in parallel with those described in the legend to Fig. 8. Monoclonal antibodies to SV40 T antigen were generous gifts of Thomas Kelly (Johns Hopkins University). Immunohistochemical evaluation of T antigen expression was performed as described by Hayward et al. (61). T antigen expression in cultures maintained at 33°C for 1 day after infection is shown. Identical results were obtained in cultures maintained for up to 3 days at 33°C. liver regeneration after 70% partial hepatectomy (21,30,46,47), although both the rate of hepatocyte proliferation and the transcription of some hepatocyte-specific genes increase significantly (48). Finally, in hepatoma cell lines and hepatocytederived hybridomas, loss of C/EBP␣ does not extinguish the hepatocyte phenotype (49,50). Thus, although C/EBP␣ has a clear antiproliferative effect in adipocytes, its role in the liver appears more complex.
It is conceivable that, in hepatocytes, identification of potentially important C/EBP␣ actions may be obscured by non-␣ C/EBP isoforms. Several metabolic and inflammatory mediators induce expression of C/EBP␤ and C/EBP␦ in hepatocytes, and it has been postulated that the resulting changes in C/EBP dimerization increase the transcription of some genes while decreasing the trans-activation of others (6,8,12,16,17,44). However, since the differences among C/EBP isoform transcriptional activities appear more qualitative than quantitative (14), it has been difficult to attribute unique functions to C/EBP␣. Work in neoplastic hepatocyte lines and primary hepatocytes has not resolved this dilemma because these systems express relatively high levels of C/EBP␤, but very little C/EBP␣ endogenously (21,46). Furthermore, standard transfection techniques transfer exogenous C/EBP␣ to Ͻ20% of the cultured cells (18 -20), making it difficult to identify treatmentrelated differences because of low signal-to-noise ratios. While recently developed C/EBP␣ knockout mice offer exciting opportunities to study C/EBP interactions, progress in this model has been slow because the ␣-null phenotype is lethal within a few days of birth (45).
We have exploited the high efficiency of adenovirus-mediated gene transfer (22,25) to demonstrate that C/EBP␣ inhibits proliferation in hepatocytes, as it does in adipocytes. In hepatocytes, following infection with AdCEBP␣, the antiproliferative actions of C/EBP␣ are particularly apparent when C/EBP␤ expression is low (at 33°C). Indeed, at this temperature, overexpression of C/EBP␣ is sufficient to inhibit proliferation driven by T antigen. Immunohistochemistry indicates that T antigen expression persists in cells that overexpress C/EBP␣. Although this technique is not sensitive enough to detect small treatment-related differences in T antigen levels, our results suggest that C/EBP␣ may interfere with the function of this transforming protein. Given evidence that the mechanism of action of T antigen involves alteration of events that regulate G 1 to S phase transition (51,52), this implies that C/EBP␣ may retard cell cycle progression. Present evidence that mitogenic induction of c-myc expression is inhibited by overexpression of C/EBP␣ in hepatocytes lacking functional T antigen further supports this theory and suggests that C/EBP␣ may regulate entry into early G 1 phase. Since c-myc is expressed in mid-G 1 (53), this could explain why C/EBP␣ and c-myc are considered opposing elements in a differentiationproliferation switch that regulates the phenotype of transformed adipocytes (15,54). Although additional experiments are necessary to confirm this theory, it is tempting to speculate that, in mature hepatocytes, high constitutive expression of C/EBP␣ (46,47,55) may help to maintain these cells in their normal growth-arrested state.
Our results also provide direct evidence that interactions among C/EBP isoforms modulate differentiated gene expression in hepatocytes by demonstrating that albumin mRNA levels rise when C/EBP␣ concentrations are increased in cells expressing high levels of C/EBP␤ at 39°C. Since overexpression of C/EBP␣ does not increase albumin mRNA levels at 33°C, despite inhibiting proliferative activity at this temperature, and some albumin mRNA is expressed in uninfected RALA255 cells at 39°C, it is unlikely the effects of C/EBP␣ on albumin expression are a nonspecific reflection of its ability to block hepatocyte proliferation. Furthermore, extracts obtained from uninfected or Ad␤gal-infected RALA255 cells grown at 33°C demonstrate some binding activity for C/EBP sites in gel mobility shift assays, and this is mainly due to C/EBP␤. However, although complex formation between C/EBP␣ and C/EBP sites is greater in 33°C cultures infected with AdCEBP␣, albumin expression is not increased.
Liu et al. (37) reported that at least five distinct DNasesensitive sites in the albumin enhancer element are differentially protected in RALA255-like hepatocytes grown at permissive and restrictive temperatures. Our data suggest that some of the factors that bind to these other sites may be differentially expressed in RALA255 cells cultured at the two temperatures. Since multiple trans-acting factors interact to promote transcription of the albumin gene (56,57), this may explain why overexpression of C/EBP␣ increases albumin expression only at the restrictive temperature. Adult hepatocytes in the healthy liver normally express the full complement of trans-acting factors necessary to activate albumin transcription (58). In these cells, C/EBP␤ and C/EBP␣ can each trans-activate the albumin gene (6,59,60). However, our findings indicate that albumin transcription is greater when cells express more C/EBP␣.
In summary, efficient adenovirus-mediated introduction of the mouse C/EBP␣ gene into cultures of a rat hepatocytederived cell line under conditions that variably induce C/EBP␤ expression has clarified the functions of these C/EBP isoforms. Proliferative activity is inhibited by overexpression of C/EBP␣. Hence, C/EBP␣ functions as a growth arrest gene in hepatocytes. In nonproliferating hepatocytes, C/EBP␣ also cooperates with other factors to regulate the expression of certain hepatocyte-specific genes, such as albumin. Of note, however, another C/EBP isoform (C/EBP␤) is capable of supporting hepatocytespecific gene transcription when C/EBP␣ expression declines. Thus, induction of C/EBP␤ may be one mechanism whereby adult hepatocytes maintain their differentiated phenotype while proliferating. Taken together, the results of our gene transfer experiments help to explain the significance of the reciprocal variations in C/EBP␣ and C/EBP␤ DNA binding activity that have been noted during liver regeneration (30,46,47) and suggest that C/EBP isoforms interact to regulate the hepatocyte phenotype during a physiological growth response.