Potential conflict of interest: Dr. Parks received funds from University of Washington.
Oval cells are hepatocytic precursors that proliferate in late-stage cirrhosis and that give rise to a subset of human hepatocellular carcinomas. Although liver regeneration typically occurs through replication of existing hepatocytes, oval cells proliferate only when hepatocyte proliferation is inhibited. Transforming growth factor-β (TGF-β) is a key inhibitory cytokine for hepatocytes, both in vitro and in vivo. Because TGF-β levels are elevated in chronic liver injury when oval cells arise, we hypothesized that oval cells may be less responsive to the growth inhibitory effects of this cytokine. To examine TGF-β signaling in vivo in oval cells, we analyzed livers of rats fed a choline-deficient, ethionine-supplemented (CDE) diet for phospho-Smad2. Phospho-Smad2 was detected in more than 80% of hepatocytes, but staining was substantially reduced in oval cells. Ki67 staining, in contrast, was significantly more common in oval cells than hepatocytes. To understand the inverse relationship between TGF-β signaling and proliferation in oval cells and hepatocytes, we examined TGF-β signaling in vitro. TGF-β caused marked growth inhibition in primary hepatocytes and the AML12 hepatocyte cell line. Two oval cell lines, LE/2 and LE/6, were less responsive. The greater sensitivity of the hepatocytes to TGF-β–induced growth inhibition may result from the absence of Smad6 in these cells. Conclusion: Our results indicate that oval cells, both in vivo and in vitro, are less sensitive to TGF-β–induced growth inhibition than hepatocytes. These findings further suggest an underlying mechanism for the proliferation of oval cells in an environment inhibitory to hepatocytic proliferation. (HEPATOLOGY 2007;45:31–41.)
TGF-β plays a significant role both in normal liver and in many hepatic diseases. In normal liver, TGF-β has 2 primary effects on hepatocytes—growth inhibition1–4 and induction of apoptosis.5, 6 Quiescent liver usually contains only modest amounts of TGF-β but injury to the liver results in the production of TGF-β, most prominently by nonparenchymal cells, including hepatic stellate cells and Kupffer cells.7 TGF-β ligands are involved in the terminal phases of hepatic regeneration,3, 8 and TGF-β is considered one of the most potent profibrogenic cytokines leading to hepatic fibrosis.9
Mammals produce 3 TGF-β ligands: TGF-β1, TGF-β2, and TGF-β3. TGF-β1 is the major form in adults, whereas the expression of TGF-β2 and TGF-β3 primarily occurs in developmental contexts. All 3 TGF-β ligands signal through a heteromeric complex of type I and type II TGF-β receptors. Binding of ligand to the type II receptor results in the recruitment and activation of 1 of 2 type I receptors. The activated type I receptor then phosphorylates the intracellular signaling intermediates of the TGF-β pathway, the Smads.10, 11 The TGF-βs and activins constitute one subgroup of the TGF-β superfamily and result in the phosphorylation of Smad2 and Smad3. The bone morphogenic proteins (BMPs) constitute the other subgroup for this superfamily and use a different set of Smads. Phosphorylation of receptor-activated Smad2 and Smad3 results in the nuclear translocation of the phospho-Smad, with incorporation of the phospho-Smad into transcriptional complexes. Smad4, the common mediator Smad, accompanies the phospho-Smads to the nucleus and is also incorporated into transcriptional complexes.
Hepatic progenitor cells or oval cells are rare quiescent cells thought to reside in the canals of Hering12, 13 that may represent a hepatic stem cell. In animals subjected to carcinogenic regimens and in humans in the late stages of cirrhosis, these cells proliferate and are often found in abundance.14, 15 The proliferation of oval cells shares an unusual reciprocal relationship with the proliferation of hepatocytes.14, 16 Oval cells proliferate only in contexts, such as the administration of specific hepatotoxins, in which hepatocyte proliferation is inhibited. Conversely, hepatocyte proliferation is brisk after partial hepatectomy or acute liver injury, situations in which oval cell proliferation is nearly absent. The mechanisms underlying this relationship between hepatocytes and oval cells have not yet been determined.
The stimuli for oval cell proliferation are now beginning to be identified. The best characterized of these are members of the tumor necrosis factor (TNF) family and IFN-γ. Deletion of TNF receptor 1 in mice significantly diminishes the oval cell response to the choline-deficient, ethionine supplemented (CDE) diet.17 Expression of lymphotoxin-β, a member of the TNF family, increases in response to the CDE diet, and deletion of either the lymphotoxin-β receptor or lymphotoxin-β itself results in impaired oval cell production.18 More recent evidence indicates that the TNF family member TWEAK stimulates oval cell proliferation, and deletion of the TWEAK receptor Fn14 results in decreased oval cell production in response to 3,5-diethoxycarbonyl-1,4-dihydrocollidine.19 Components of the IFN-γ pathway are up-regulated by the administration of 2-acetylaminofluorene followed by partial hepatectomy.20 IFN-γ knockout mice show an attenuated oval cell response to a CDE diet,18 and the combination of IFN-γ with either TNF or lipopolysaccharide stimulates oval cell proliferation while inhibiting hepatocyte proliferation.21
The molecules that stimulate oval cell proliferation, aside from TWEAK and IFN-γ, are also mitogenic for hepatocytes, but oval cells and hepatocytes only rarely proliferate simultaneously. To maintain the controlled, context-specific proliferation of these cell types observed in vivo, other regulatory pathways also must operate. In particular, inhibitory pathways must provide regulatory counterbalance to the mitogenic pathways. We have identified the TGF-β pathway as one inhibitory pathway that functions in both oval cell and hepatocytic lineages. Moreover, we have found that oval cell growth is much less sensitive to TGF-β when compared with hepatocytes. We propose that the differential sensitivity of hepatocytes and oval cells to TGF-β serves as a mechanism permissive for oval cell proliferation.
AFP, α-fetoprotein; BMP2, bone morphogenic protein 2; CDE, choline-deficient, ethionine supplemented; EGF, epidermal growth factor; IFNγ, interferon-γ; IL-6, interleukin-6; LPS, lipopolysaccharide; M2-PK, M2 isozyme of pyruvate kinase; PH, partial hepatectomy; TβRI, type I TGF-β receptor; TβRII, type II TGF-β receptor; TGF-β, transforming growth factor-β; TNF, tumor necrosis factor; TWEAK, TNF-like weak inducer of apoptosis.
Materials and Methods
TGF-β1 and BMP2 were purchased from R&D Systems (Minneapolis, MN). Antibodies were purchased from the following vendors: phospho-Smad2 (Ser465/467), Upstate Biotechnology (Charlottesville, VA); Smad2, Smad3, and Smad6, Zymed Laboratories (San Francisco, CA); Smad4, TβRI, and TβRII, Santa Cruz Biotechnology (Santa Cruz, CA); smooth muscle actin and γ-tubulin, Sigma (St. Louis, MO); cyclin D1, CD34, and CD45, BD Biosciences (San Jose, CA); Ki67, Dako (Carpinteria, CA); AFP, Santa Cruz Biotechnology. Tritiated thymidine was purchased from Amersham Pharmacia Biotech (Piscataway, NJ).
The LE/2 and LE/6 rat oval cell lines were cultured as described.22 AML12 hepatocytes, a nontransformed, well-differentiated murine hepatocyte cell line, were maintained as described.23, 24 In all experiments, cells were incubated overnight in 0.5% serum before their use in experiments. Mv1Lu mink lung epithelial cells were obtained from the American Type Culture Collection (Manassas, VA) and were cultured using ATCC-recommended complete growth medium.
Primary Hepatocyte Isolation.
Primary hepatocytes were isolated from wild-type C57/Black/6 mice by collagenase perfusion of the liver.24 Hepatocytes were purified by centrifugation over a Percoll gradient. Cells were plated onto collagen-coated plates and allowed to adhere for 4 hours in attachment medium containing 5% fetal bovine serum. The medium was then replaced with serum-free feeding medium containing 10 ng/ml murine epidermal growth factor. Primary hepatocytes were used within 48 hours of isolation without further subculture.
Equal numbers of cells were plated in 6-well plates and allowed to adhere overnight. The following day, the cells were cultured for 3 hours in reduced serum conditions (0.2% fetal bovine serum), then treated for the indicated times with ligand. After treatment, the cells were washed with cold phosphate-buffered saline and lysed by the addition of 0.2 ml 6× Laemmli buffer. DNA was sheared by passing the cells four times through a 25-gauge needle, and the cells were scraped into microcentrifuge tubes and boiled for 5 minutes. Lysates were separated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis and transferred onto Immobilon-P membranes (Millipore, Billerica, MA) for blotting. Proteins were detected using horseradish peroxidase–conjugated secondary antibodies and visualized by chemiluminescence (Pierce, Milwaukee, WI).
For the immunoprecipitations, the cells were plated in 100-mm dishes and allowed to adhere overnight. Cells were lysed by the addition of 0.5 ml lysis buffer (25 mM HEPES, pH 7.5, 150 mM NaCl, 1% Triton X-100, 5 mM EDTA, and 10% glycerol) plus phosphatase and protease inhibitors. The cells were allowed to incubate in the lysis buffer for 20 minutes on ice and were then scraped into microcentrifuge tubes. After high-speed centrifugation for 15 minutes, an aliquot of the lysate was removed for western blotting, and the remainder was immunoprecipitated overnight with 1.5 μg anti-TβRII rabbit polyclonal antibody (Santa Cruz Biotechnology, Santa Cruz, CA) and 35 μl protein G-Sepharose (80% suspension) (Amersham Pharmacia Biotech, Piscataway, NJ). Lysates and immunoprecipitates were then separated by sodium dodecyl sulfate polyacrylamide gel electrophoresis and transferred onto Immobilon-P membranes (Millipore, Billerica, MA) for blotting. Proteins were detected using horseradish peroxidase–conjugated secondary antibodies and visualized by chemiluminescence (Pierce, Milwaukee, WI).
Cells were plated on glass cover slips and allowed to attach overnight. The next day, the cells were fixed for 5 minutes in cold 3.5% paraformaldehyde and permeabilized in 0.2% saponin for 2 minutes. Fifty millimolar glycine was added to the cells for 5 minutes to quench autofluorescence. The cells were incubated in primary antibody overnight at 4°C. The cells were then incubated in fluorescein isothiocyanate–conjugated goat anti-mouse and tetramethyl rhodamine isothiocyanate (TRITC)-conjugated goat anti-rabbit secondary antibodies for 1 hour. The cells were mounted in Vectashield mounting medium (Vector Labs, Burlingame, CA) containing 4′-6-diamidino-2-phenylindole. Images were captured using a Leica laser scanning confocal microscope.
Tritiated Thymidine Incorporation Assay.
Cells were plated in triplicate. Cells were pulsed with 1 μCi/ml tritiated thymidine for 3 hours, precipitated with 10% trichloroacetic acid, and solubilized with 0.5N sodium hydroxide before quantitation with a scintillation counter.
Active Caspase-3 Assay.
Cells were plated at a density of 1 × 106 cells per 100-mm tissue culture plate. Sixteen hours later, both adherent and floating cells were collected, and protein lysates were prepared as described.23 One hundred micrograms protein was incubated with the fluorogenic caspase-3 substrate, DEVD-AMC (BioMol, Plymouth Meeting, PA) for 1 hour, and enzymatic activity was determined as previously described.23 As a positive control for apoptosis, AML-12 mouse hepatocyte cells were pre-treated for 30 minutes with actinomycin D (250 ng/ml) followed by treatment for 15 hours with 20 ng/ml TNF.23
Staining was performed on 5-μm sections of paraffin-embedded tissue. Sections were deparaffinized in xylene, then rehydrated before heat-induced antigen retrieval in Citrate buffer (pH 6.0, Biogenex, San Ramon, CA). Slides were blocked using an avidin/biotin blocking kit (Zymed Laboratories, San Francisco, CA) followed by 5% normal goat serum (Vector Laboratories, Burlingame, CA). After several phosphate-buffered saline washes, the slides were incubated in primary antibody overnight. Slides were washed and incubated in goat anti-mouse IgG or anti-rabbit IgG secondary antibodies (1:200, Vector Laboratories, Burlingame, CA), followed by 3% peroxide block. After washing, the slides were developed using the ABC/DAB Detection Kit (Ventana Medical Systems, Tucson, AZ) for single staining or, for double staining, the ABC/DAB Detection Kit plus an ABC/AP kit. Methyl green was used as a counterstain (Dako, Carpinteria, CA).
Quantification of Phospho-Smad2 and Ki67 Immunostaining.
Sections of rat liver were co-stained for either phospho-Smad2 and M2-PK or Ki67 and M2-PK. To examine Smad signaling in oval cells, M2-PK–positive oval cells were identified in 10 high-power fields, and the presence or absence of phospho-Smad2 staining was determined. To examine proliferation in oval cells, M2-PK–positive oval cells were identified in 10 high-power fields, and the presence or absence of Ki67 staining was determined. All hepatocytes were identified in these sections, and the presence or absence of phospho-Smad2 or Ki67 staining was likewise determined for these cells. An average of 66 hepatocytes and 306 oval cells were counted per high-power field.
Smad7 Reverse Transcription PCR.
Equal numbers of cells were plated in 100-mm dishes and allowed to adhere overnight. The following day the cells were cultured for 3 hours in reduced serum conditions (0.2% fetal bovine serum), then treated for the indicated times with ligand. After treatment, RNA was isolated using an RNeasy kit (Qiagen, Valencia, CA). Reverse transcription–PCR was performed using primers exactly matching both mouse and rat sequences.25
Experiments with rats fed a CDE diet have been previously described.26 All animal work was done in accordance with the animal care policies at the University of Washington.
All quantitative data are presented as mean ± SEM obtained from independent experiments. Significance was determined using a two-tailed unpaired Student t test. Statistical calculations were done using Microsoft Excel (Seattle, WA); P < 0.05 was considered significant.
Identification of Oval Cells In Vivo.
The livers from rats maintained on a CDE diet were immunostained and compared with livers from untreated mice fed a normal diet. Control and treated livers were examined at two time points, 6 and 11 weeks after the initiation of the CDE diet. Serial sections were stained from each time point. Ki67 staining was performed to assess the proliferative response of each liver. Although control livers showed no staining for Ki67, many Ki67-positive cells were evident at both 6 and 11 weeks (Fig. 1B, E, H). Two markers were used to detect oval cells, alpha-fetoprotein (AFP) and M2 isozyme of pyruvate kinase (M2-PK). No staining for either oval cell marker was detected in control livers (Fig. 1A, C). Significant numbers of both AFP-positive and M2-PK–positive cells were seen for each time point on the CDE diet, with more positive cells noted at 6 (Fig. 1D, F) than at 11 weeks (Fig. 1G, I). For both 6 and 11 weeks, more cells stained for M2-PK (Fig. 1F, I) than for AFP (Fig. 1D, G). Given these results, cells from the 6-week CDE diet stained for M2-PK were used for further analysis.
TGF-β/Smad Signaling of Hepatocytes and Oval Cells In Vivo.
Nuclear immunopositivity for phospho-Smad2 was employed as a marker for TGF-β/Smad signaling. Sections were costained for both M2-PK and phospho-Smad2 (Fig. 2A), and the presence or absence of phospho-Smad2 staining was determined for both hepatocytes and M2-PK–positive oval cells in 10 high-power fields. Although 83% of hepatocytes were immunopositive for phospho-Smad2, only 50% of M2-PK–positive oval cells demonstrated this staining (P < 0.0001), suggesting that hepatocytes respond more intensely to TGF-β than oval cells (Fig. 2C). These same tissues were also dual-stained for both M2-PK and Ki67 (Fig. 2B). Whereas 23% of the M2-PK–positive oval cells stained for Ki67, only 4% of hepatocytes did (P < 0.0001) (Fig. 2D), arguing that the increased phospho-Smad2 staining noted in the hepatocytes is associated with a lower rate of proliferation.
Smad Signaling in Hepatocyte and Oval Cell Lines.
Our in vivo rat liver studies suggest that hepatocytes signal more readily or more strongly in response to stimulation by a ligand of the TGF-β family than do oval cells. Moreover, these studies suggest that this greater Smad signaling intensity in hepatocytes translates into a lower rate of proliferation. To test these findings more definitively, we employed an in vitro experimental system. For these studies, we used 3 well-characterized cell lines (2 oval cell and 1 hepatocyte) and primary hepatocytes. LE/2 and LE/6 are oval cells lines that were derived from rats after 2 weeks and 6 weeks, respectively, on a CDE diet. Both cell lines are capable of differentiating into hepatocytes in vitro.22 AML12 is a nontransformed, proliferative hepatocyte cell line that retains a differentiated phenotype.23, 24 For these experiments the LE/2 and LE/6 oval cell lines were compared with the AML12 hepatocyte cell line and with primary hepatocytes.
To determine whether oval cells signal in response to TGF-β, we examined their Smad signaling and compared these results with hepatocytes. After TGF-β1 treatment, all cells showed a substantial increase in phospho-Smad2 at 30 minutes (Fig. 3A,B), indicating that both hepatocytic and oval cells contain sufficient quantities of receptors and Smads to signal in response to TGF-β. This slow rate of phosphorylation (15 minutes to 1 hour) is typical for Smad phosphorylation in all tested cell types. Smad2 phosphorylation began to diminish in both LE/2 and LE/6 cells at 3 hours (data not shown), and after treatment for 24 hours was at or below the baseline level of unstimulated cells (Fig. 3B). Primary hepatocytes and AML12 cells showed a more prolonged signaling response, with prominent phosphorylation of Smad2 extending to 24 hours (Fig. 3A,B). No changes in the levels of total Smad2, Smad3, or Smad4 were noted over the 24-hour treatment period (Fig. 3A and data not shown).
Smad Nuclear Translocation in Hepatocyte and Oval Cell Lines.
Activation of Smad2 and Smad3 by phosphorylation rapidly leads to the nuclear accumulation of these proteins. The nuclear translocation of the Smads in response to ligand was examined in oval and hepatocytic cells. Nuclear translocation of Smad2 (Fig. 4, red) Smad3 (data not shown), and Smad4 (Fig. 4, green) were detected through indirect immunofluorescence confocal microscopy after treatment with TGF-β1. For each cell type, nearly the entire population of each Smad moved into the nucleus within 30 minutes. Moreover, the Smads colocalized extensively after treatment, as demonstrated by the yellow color on the images (Fig. 4B, C, E, F, H, I).
Growth Inhibition in Response to TGF-β.
These findings demonstrate that exposure of oval cells to TGF-β initiates an intracellular signaling response that proceeds through a Smad pathway and further suggest that hepatocytic cells may signal longer in response to ligand than oval cells. To measure a functional response to TGF-β stimulation, the inhibitory effects on cell growth for each cell type were determined. Cells for this assay were treated with TGF-β1 (up to 2.5 ng/ml), and dose–response curves were generated. Both the dose–response curves and maximal inhibition values demonstrate that the 2 hepatocytic cell types cluster into 1 group while both oval cell lines cluster into a second group. AML12 cells were highly sensitive to growth inhibition by TGF-β1, showing 82% inhibition at the maximal dose (Fig. 5A,B). Primary hepatocytes were likewise highly sensitive to growth inhibition by TGF-β1, showing 72% maximal inhibition (Fig. 5A,B). By comparison, LE/2 and LE/6 oval cells were much less responsive, with maximal growth inhibitions of 48% and 50%, respectively (P < 0.001 for AML12 versus either LE cell line, P < 0.001 for primary hepatocytes vs. LE/2, and P < 0.01 for primary hepatocytes versus LE/6) (Fig. 5A,B). Mink lung epithelial cells (Mv1Lu cells) were chosen as a positive control, because they are highly sensitive to TGF-β–induced growth inhibition. Mv1Lu cells responded to even very small doses of TGF-β with a sharp drop in tritiated thymidine incorporation (Fig. 5A,B), and the maximal growth inhibition after TGF-β treatment is the lowest of any of the cell types at 92%. Treatment of cell lines with BMP2, a member of the TGF-β superfamily that uses a separate Smad pathway (Smad1/5/8) and does not cause epithelial cells to growth arrest, was used as a negative control. Although all cell lines responded to BMP2 by phosphorylating Smad1 (data not shown), growth inhibition did not occur in any of the cell lines (Fig. 5C,D).
Apoptosis in Response to TGF-β.
TGF-β has previously been shown to stimulate apoptosis in murine primary hepatocytes and hepatocytic cell lines. To assess the level of apoptosis induced by TGF-β in these studies, apoptosis was determined using an active caspase-3 assay. AML12 cells treated with actinomycin D and TNF served as the positive control. For each cell type, little to no apoptosis occurred after treatment of the cells for 24 hours with 2.5 ng/ml TGF-β, the maximal dose in the growth inhibition studies (data not shown).
Low-Dose Smad Signaling in Hepatocyte and Oval Cell Lines.
The previous signaling experiments (Figs. 3, 4) were carried out using a standard signaling dose of TGF-β1 (5 ng/ml) in order to demonstrate that the Smad signaling pathway is intact and functional in these cells. The growth inhibition data show, however, that even at much lower doses of TGF-β1, hepatocytic cells demonstrate significantly greater growth inhibition than oval cells. Smad signaling was also examined at a low dose of TGF-β1 (0.3 ng/ml, equal to 1/16th of the signaling dose and corresponding to the dose indicated by the arrowheads in Fig. 5A), to seek differences among these cell lines that might arise in the context of more moderate TGF-β stimulation. At this lower dose, AML12 cells again showed substantially more robust signaling at 24 hours' treatment than did either oval cell line (Fig. 6).
Effects of TGF-β on Cell Cycle Proteins.
TGF-β inhibits cell growth by arresting cells in the G1 phase of the cell cycle. To confirm the differences in growth inhibition noted by tritiated thymidine incorporation, and to determine the specific pathway involved in this TGF-β–mediated growth inhibition, we examined TGF-β–treated cells for changes in the levels of the cyclins involved in G1 arrest. TGF-β caused the expected decrease in cyclin D1 and cyclin D3 at 24 hours for both AML12 cells and primary hepatocytes (Fig. 7). LE/2 and LE/6 cells, conversely, showed only small changes in cyclin D1 and cyclin D3 levels with treatment (Fig. 7). Cyclin E levels remained constant despite treatment in all 4 cell types (Fig. 7).
TGF-β Receptor Levels in Hepatocytes and Oval Cells.
We next sought to identify the mechanism underlying the relative insensitivity of oval cells to TGF-β signaling and growth inhibition. Because our immunohistochemical data (Fig. 2) showed decreased nuclear immunopositivity for phospho-Smad2 in oval cells compared with hepatocytes, this suggested that the difference between these 2 cell types lies at the receptor level. That is, variations in receptor levels or receptor interacting proteins likely underlie the differences in signaling, rather than variations in downstream signaling intermediates or interacting transcription factors. To determine the levels of the TGF-β receptors, lysates from primary hepatocytes, AML12 cells, and both oval cell lines were immunoblotted for TβRI, the canonical type I TGF-β receptor, and for TβRII, the type II TGF-β receptor. All cell types contained abundant TβRI (Fig. 8A), with primary hepatocytes staining particularly strongly for this receptor. TβRII, the type II TGF-β, receptor is typically heavily glycosylated and therefore appears as a diffuse band extending from 80 to 100 kDa in sodium dodecyl sulfate polyacrylamide gel electrophoresis gels. This pattern is evident for both the LE/2 and LE/6 cell lines (Fig. 8A). By western blotting, the AML12 cells and the primary hepatocytes show only a single band at 65 kDa (Fig. 8A). This pattern is characteristic of mature but unglycosylated type II TGF-β receptors. Such receptors have been shown to bind ligand and generate downstream functional responses nearly equally well in the absence of glycosylation.27, 28 Immunoprecipitation of TβRII from AML12 cells resulted in a faint diffuse band at 80 to 100 kDa, the same size as the band seen in the LE/2 cell lane, indicating that the hepatocytic cells produce at least a small amount of the glycosylated form of the receptor (Fig. 8B). Overall, the oval cell lines show abundant receptor levels for the 2 TGF-β receptors, arguing against diminished receptor density as the mechanism underlying the lower sensitivity of these cells to TGF-β.
Inhibitory Smad Levels in Hepatocytes and Oval Cells.
One other likely mechanism underlying the relative insensitivity of oval cells to TGF-β signaling and growth inhibition is a difference in the levels of one of the inhibitory Smads. Smad6 and Smad7 are the 2 inhibitory Smads, and both of these molecules can function at the level of the receptor to dampen TGF-β signaling. Reverse transcription PCR was used to detect Smad7 levels. No differences in Smad7 RNA expression were seen among the untreated cells (Fig. 9A). Treatment for 24 hours with 5 ng/ml TGF-β1 resulted in increased expression of Smad7 RNA by AML12 cells but not by either oval cell line. These results argue that Smad7 does not contribute to the relative insensitivity of oval cells to TGF-β. To detect Smad6 levels, lysates were immunoblotted for Smad6. Smad6 levels were distinctly different between the cell types, with LE/2 and LE/6 cells containing abundant amounts of this protein, whereas AML12 cells and primary hepatocytes had only low levels (Fig. 9B). To demonstrate that increased Smad6 can inhibit TGF-β function, the low Smad6 levels in AML12 cells were augmented through adenoviral-mediated expression of Smad6, and the growth inhibitory response to TGF-β was assessed. LacZ-expressing control AML12 cells showed 60% growth inhibition in response to ligand, whereas expression of Smad6 resulted in a statistically significant suppression of the percent growth inhibition to 43% (Fig. 9C).
In all 4 cell types, untreated cells and cells treated for 24 hours with TGF-β showed similar levels of Smad6 (Fig. 9B). The previous signaling results demonstrated that treatment of these same cells with TGF-β for 30 minutes resulted in comparable levels of Smad2 phosphorylation (Fig. 3B). Only after 24 hours of treatment did the phospho-Smad2 levels vary among the cell types, with distinctly higher levels of phospho-Smad2 in the hepatocytic cells (Figs. 3B, 6). One final question then is how unchanging Smad6 levels can regulate phosphorylation of Smad2 at 24 hours but not at 30 minutes. In unstimulated cells both inhibitory Smads localize to the nucleus rather than to the cytoplasm. Only hours after treatment do these proteins exit the nucleus and inhibit TGF-β signaling. To test whether the timing of Smad6 translocation from the nucleus to the cytoplasm might explain the differences among the cell types in phosphorylation of Smad2 at 24 hours, oval cells were treated for up to 24 hours with TGF-β, and Smad6 localization was assessed by indirect immunofluorescence microscopy. For each cell type, Smad6 retained its sharply defined nuclear localization until 20 hours (Fig. 10A, E). At the 20-hour time point, however, nuclear Smad6 staining was notably less intense, and the clear presence of Smad6 outside of the nucleus became evident (Fig. 10B, F). Moreover, the lighter backgrounds in the TGF-β-treated images (Fig. 10B, F) are indicative of the longer exposure times required to detect Smad6, which had disbursed throughout the cytoplasm and whose staining intensity had therefore diminished.
Modulating the response of the liver to injury remains a major clinical goal. In the broadest terms, the repair process after liver injury can be characterized as using 1 of 2 mechanisms. Typically, hepatocyte destruction leads to the recruitment of the remaining hepatocytes into the cell cycle, with hepatocyte proliferation replenishing the damaged hepatic parenchyma. The prototypical example for this mechanism is the replication that follows two-thirds partial hepatectomy in the rodent. In circumstances in which hepatocyte proliferation is inhibited, however, hepatic injury leads to proliferation of the hepatic progenitor cell compartment, as is seen with several rodent hepatocarcinogenic models, including the CDE diet and the Solt-Farber models.
We examined one of these models, the CDE diet in the rat, and found a striking difference in the TGF-β signaling response of the oval cells in comparison with the surrounding hepatocytes. TGF-β signaling was detectable in a significantly smaller percentage of oval cells than hepatocytes. This finding correlated with a higher rate of proliferation (as assessed by Ki67 staining) in the oval cells. We then further characterized these differences in vitro, determining that oval cells showed greater resistance to TGF-β–induced growth inhibition than did hepatocytes. These findings suggest that, although both hepatocytes and oval cells signal in response to TGF-β, oval cells are considerably less sensitive to this cytokine. Importantly, these findings suggest a possible mechanism for the reciprocal relationship between oval cell and hepatocyte proliferation. In the presence of TGF-β, inhibition of hepatocellular proliferation could impair the regenerative response to liver injury. Under similar circumstances, oval cells may retain the ability to proliferate and potentially repopulate the damaged liver.
We confirmed the in vitro growth inhibition data by examining the cells for changes in the levels of cell cycle proteins after TGF-β treatment. TGF-β induces growth inhibition by arresting cells in the G1 phase of the cell cycle. This arrest is accomplished by down-regulating cyclins D and E and up-regulating the cyclin-dependent kinase inhibitors p15 and p21. Both hepatocytic cell types showed the expected prominent decrease in cyclin D1 and D3 after TGF-β treatment. AML12 cells had a substantially higher basal level of cyclin D1 than primary hepatocytes, an expected finding when comparing a proliferative cell line with primary hepatocytes. The oval cell lines contained the highest levels of cyclin D3, and surprisingly there was no change in the level of cyclins D1, D3, or E in either oval cell line after treatment. Because oval cells do inhibit growth in response to TGF-β (albeit to a lower degree than hepatocytic cells), the absence of a change in cyclin levels may indicate that the change was too small to be detected by our western blotting or that changes in the cyclin-dependent kinase inhibitors p15 and p21 are more important for oval cell growth inhibition.
Another important function for TGF-β is the induction of apoptosis in certain cells, including hepatocytes. Our apoptosis studies were not performed to compare the degree to which TGF-β induces apoptosis in these different cell types but rather to ensure that the growth inhibition we measured was not augmented by apoptosis. As such, the cells in these experiments were only treated with TGF-β for 24 hours, a length of time that may not have been sufficient to induce detectable apoptosis. We currently have studies underway to assess more definitively the induction of apoptosis in oval cells by TGF-β.
The differential sensitivity of oval cells and hepatocytic cells to TGF-β had a striking time dependence. After treatment with TGF-β for 30 minutes (the typical time for the initial response to ligand), both cell types responded with significant and approximately equal levels of Smad2 phosphorylation. After 24 hours of treatment, however, signaling had diminished substantially in the oval cells but remained strong in the hepatocytic cells. In chronic liver injury, in which the duration of TGF-β stimulation is typically prolonged, this differential sensitivity to TGF-β after 24 hours may play a significant role. Hepatocytes likely remain sensitive to the growth inhibitory properties of TGF-β whereas oval cells rapidly become insensitive to this ligand.
To determine the mechanism underlying this differential sensitivity to TGF-β, we first examined the levels of the type I and type II TGF-β receptors in hepatocytic cells and oval cells. Abundant amounts of the type I receptor in all cell types argued against TβRI as the source of the increased sensitivity of hepatocytes to ligand. Surprisingly, oval cells contained higher levels of the glycosylated form of TβRII than did the hepatocytic cell types (although the unglycosylated form was present at high levels in the hepatocytic cells). These results suggest that differences in receptor levels between the 2 cell types do not account for the differences in signaling. Smad7 RNA expression was also equal among the cell types initially and higher in hepatocytic cells after treatment for 24 hours, suggesting that Smad7 cannot explain the prolonged signaling in hepatocytic cells. In contrast, Smad6 was present in much higher amounts in both LE/2 and LE/6 cells compared with either AML12 cells or primary hepatocytes. Smad6 has been shown to inhibit TGF-β signaling by associating with the type I receptor and thereby interfering with Smad2 phosphorylation by the activated receptor complex.29 The significant levels of Smad6 in oval cells provide a means for inhibiting the signaling response to ligand, preventing the phosphorylation and translocation to the nucleus of Smad2. Moreover, because the Smad6 in the oval cells was initially localized to the nucleus (from which it could exert no inhibitory effect), and only translocated to the cytoplasm after 20 hours of TGF-β treatment, the movement of this inhibitory Smad likely explains the time dependence of the differential sensitivity to TGF-β.
In conclusion, these experiments addressed the limited question of how oval cells are able to proliferate in an environment inhibitory to hepatocytes. The greater sensitivity of hepatocytes to TGF-β ligands likely prevents their proliferation in a setting that yet remains permissive for oval cell proliferation. In more general terms, the ability of oval cells to signal and to slow their growth in response to TGF-β stimulation may provide one of the counterbalancing forces to the many mitogenic factors acting on these cells.