Notice: Wiley Online Library will be unavailable on Saturday 27th February from 09:00-14:00 GMT / 04:00-09:00 EST / 17:00-22:00 SGT for essential maintenance. Apologies for the inconvenience.
Hepatic stem cells are activated after liver damage and have a critical role in tissue homeostasis and repair. Characterization of molecular and cellular events accompanying the expansion and differentiation of liver stem cells is essential for understanding the basic biology of stem cells and for facilitating clinical application of the stem cells. We assessed whether in vitro differentiation of putative hepatic progenitor (rat liver epithelial [RLE]) cells toward hepatocytic lineage affects the response to TNFα-mediated cytotoxicity, a common determinant of liver injury. The data show that 50% of differentiated cells underwent apoptosis after 6 hours of TNFα treatment whereas control RLE cells were resistant. Both cell types displayed mitochondrial depolarization and release of cytochrome c but the TNFα treatment resulted in activation of caspases 9 and 3 and the execution of apoptosis only in differentiated RLE cells. Apoptotic death was associated with increased ROS production and depletion of glutathione. Antioxidants completely prevented both glutathione depletion and apoptosis induced by TNFα in differentiated RLE cells. Conversely, glutathione-depleting agents sensitized control RLE cells to TNFα induced apoptosis. In conclusion, efficient antioxidant defense system involving glutathione renders hepatic progenitor cells resistant to TNFα-mediated apoptosis and acquisition of sensitivity to death stimuli is an implicit feature of the differentiation process. Supplementary material for this article can be found on the HEPATOLOGY website (http://interscience.wiley.com/jpages/0270-9139/suppmat/index.html). (HEPATOLOGY 2004;40:590–599.)
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Hepatic stem cells have attracted great attention as potential candidates for liver-directed gene therapy and as a tool for regenerative medicine. It is well established that hepatic stem cells participate in liver regeneration when proliferation of hepatocytes is inhibited and/or when mature parenchyma is damaged by toxic insults.1 Despite extensive efforts to use hepatic progenitor cells as an ideal source for liver repopulation,2, 3 the mechanisms controlling lineage commitment and response of hepatic stem cells to environmental signals remain largely unknown.
Activation of hepatic stem cells has been observed not only in animal models of chemically-induced liver injury but also in a number of human liver diseases, including hepatitis, cholestasis, and alcoholic and nonalcoholic fatty liver disease.4–6 A common feature of these adverse conditions is marked induction of tumor necrosis factor (TNF)α and increased apoptosis resulting in continuous hepatocyte death. TNFα is a pleiotropic cytokine critical in tissue homeostasis. In general, TNFα is not cytotoxic for liver cells, but rather contributes to hepatocyte proliferation and is essential for liver regeneration.7, 8 However, during liver injury, hepatocytes become vulnerable to this cytokine, probably due to impairment of defense mechanisms.9, 10 Similarly, TNFα induces apoptosis of both hepatocytes and hepatoma cells in vitro when RNA or protein synthesis is inhibited.11
During the last few years, understanding of the apoptotic pathway initiated by TNFα in hepatocytes has greatly advanced. Hepatocytes belong to the category of cells with type II phenotype, in which binding of TNFα to the TNFα receptor (TNFR) results in caspase 8 activation triggering mitochondrial apoptotic signaling via release of cytochrome c and subsequent interaction with apaf-1 and culminating in activation of caspases 9 and 3.12 Current evidence points to TNFα-mediated release of mitochondrial reactive oxygen species (ROS) as one of the major signaling events involved in TNFα cytotoxicity.13–16 In agreement with this result, ROS are considered to play an important role in the pathogenesis of many forms of liver disease.17
Recently, it has been reported that TNFα signaling is required for proliferation and differentiation of committed hepatocyte precursors.18–20 This observation prompted us to examine how induction of hepatocytic differentiation would affect the cellular response to TNFα-mediated cytotoxicity. To induce hepatocyte differentiation, we applied a protocol consisting of sequential treatment with a 5-aza-2′-deoxycytidine followed by a combination of oncostatin M (OSM) and dexamethasone (Dex) (Schroeder et al., unpublished observations, 2003) to rat liver epithelial (RLE) cells, an adult liver-derived stem cell line.21 We show here that differentiated but not control RLE cells die by apoptosis in response to TNFα and cycloheximide treatment. We also provide evidence that activation of antioxidant defense systems involving glutathione (GSH) protects against mitochondrial ROS generation and apoptosis induced by TNFα in adult liver-derived stem cells.
RLE cells were isolated and cultured as described earlier.22 For differentiation studies, 1×106 cells were plated per 10-cm dish. Chemically induced differentiation was achieved by culturing RLE in a 10% serum DMEM/Hams' F12 (1:1) medium containing 1% ITS+ culture supplement (BD Bioscience, Bedford, MA), 0.1 nM MEM nonessential amino acids solution (Life Technologies, Grand Island, NY), 5 mM MEM sodium pyruvate (Life Technologies), 10 mM nicotinamide (Sigma, St. Louis, MO) and 0.2 mM L-ascorbic acid-2-phosphate (Sigma). The following inducing factors were added to the medium in this sequence: 1 μM 5-Aza-2′-deoxycytidine (5AC) (Sigma) for 2 days and 1×10−6 M Dex (Sigma) plus 10 ng/mL OSM (R&D Systems, Minneapolis, MN) for 4–8 days. Control cells were maintained in culture for the same period of time in regular 10% serum Hams' F12 medium, without inducing factors and additional supplements. To induce apoptosis, medium was replaced by serum-free Hams' F12 medium supplemented with 20 ng/mL TNFα and 0.5. μg/mL Cycloheximide (CHX). For inhibition of caspase activation, a broad-spectrum caspase inhibitor, z-VAD-fmk (60μM) (Molecular Probes, Inc., Eugene, OR) was added to the cells 1 hour before TNFα treatment. For glutathione depletion experiments, either 7 μM diethyl maleate (DEM) (Sigma) or 2mM buthionine sulfoximine (BSO) (Sigma) were added to the cells 1 hour before TNF+CHX. Pyrrolidine dithiocarbamate (PDTC) at 50 μM concentration or α-lipoic acid (αLA) at 5 μM were used for inhibition of ROS production.
For visualization of mitochondria, cells were observed under a Zeiss LSM 510 Confocal Microscope (Carl Zeiss Inc, Thornwood, NY) after incubation with 30 ng/mL Mito Tracker Red CMXRos (Molecular Probes, Inc.) in the presence of 5 μg/mL Hoechst 33342 for 30 minutes at 37°C.
Assessment of Apoptosis.
Apoptotic cells were examined by Fluorescence Microscopy or by Flow cytometry (FACS) after propidium iodide staining as previously described.23, 24 Apoptotic indices were determined after counting 1,000 cells per treatment or after collection of 10,000 events using a FACS Calibur (Becton Dickinson) and CellQuest software.
Analysis of Mitochondrial Membrane Potential (ΔΨmito).
Changes in mitochondria potential were examined using JC-1 (5,5′,6,6′-tetrachloro-1,1′,3,3′- tetraethylbenzimidazocarbocyaniniodide) (Molecular Probes).25 Cells were loaded with 5μg/mL JC-1 for 15 minutes at 37°C in growing medium, washed and analyzed in a Zeiss LSM 510 Confocal Microscope (Carl Zeiss Inc). For nuclear staining we used Hoechst 33342 as described earlier in this section.
Cells were washed with cold phosphate-buffered saline, scrapped and lysed in 0.2% Triton X-100. After taking an aliquot for determination of the protein concentration, lysates were depleted of protein and used for spectrophotometric determination of both total glutathione (GSH+GSSG) and glutathione disulfide (GSSG) as previously described.24 Data are expressed as nmol/mg protein. Protein quantification was done using Bradford Reagent (Bio Rad, Hercules, CA) and bovine serum albumin as standard.
Measurement of ROS.
Cells grown in 96 well-plates were treated with TNF+CHX for indicated time intervals and loaded with 20 μM H2DCFDA (2′,7′-dichlorodihydrofluorescein diacetate) (Molecular Probes) for 30 minutes at 37°C. Fluorescence emission was monitored with a CYTOFLUOR 4000 fluorometer, (PerSeptive Biosystems, Framingham, MA). Values were normalized by cell number after staining with crystal violet.24
Cells were lysed with M-PER buffer (Pierce Chemical Co., Rockford, IL) supplemented with 1 mM PMSF and protease inhibitor cocktail (Roche, Brandenburg, NJ). Fifty micrograms of protein were separated by SDS-PAGE on 4%–20% gradient gels and blotted on polyvinylidene difluoride membranes (Invitrolon, Invitrogen, Carlsbad, CA). For cytochrome c release assay, cytosolic protein extracts were prepared as previously described26 and separated on 10%–20% gradient gels. Primary antibodies used were: anti-caspase 3 (H-277) (Santa Cruz Biotechnology, Inc., Santa Cruz, CA); rat specific anti-caspase 9 (Cell Signaling, Beverly, MA); anti-cytochrome c (clone 7H8.2C12) (BD Pharmingen, San Diego, CA) and anti-PARP (clone 42) (BD Transduction Laboratories).
Total cellular RNA was prepared using the RNeasy Kit (Qiagen, Valentia, CA) with on-column DNase treatment (RNase-free DNase Set, Qiagen) for 45 minutes to avoid genomic DNA contamination. RNA yields and purity were analyzed using a spectrophotometer (DU 640, Beckman Coulter, Fullerton, CA). 2.5–3.5 μg total RNA were reverse transcribed into complementary DNA, using the SuperScript First-Strand Synthesis System for reverse transcriptase-polymerase chain reaction (RT-PCR) (Invitrogen) with oligo(dT) as a primer. Complementary DNA was used as a template for amplification of selected markers in a PCR reaction. Specific primer sequences and PCR conditions are summarized in Table 1. PCR products were electrophoresed on 2% agarose gels and visualized using ethidium bromide.
Table 1. Primer Sequences and PCR Conditions
Primer Sequences (5′–3′)
Annealing Temperature (°C)
Product Size (bp)
Gene Array Analysis.
Microarray analysis in untreated and treated RLE cells was performed as previously described.27
In Vitro Differentiation of RLE Cells Toward the Hepatocyte Lineage.
A two step-induction protocol consisting of 2-day pretreatment with 1μM 5AC followed by 4–8-day treatment with 1 μM Dex and 10 ng/mL OSM was developed (Schroeder et al., unpublished observations, 2003) and applied to adult rat liver–derived stem cells. This treatment resulted in remarkable enlargement in cell size, increase in organelle number and complexity and acquisition of a more polygonal shape with reduced nucleus to cytoplasm ratio (Fig. 1A). More importantly, RT-PCR analysis revealed an induction of liver specific transcription factors (hepatocyte nuclear factors, HNF 1α, HNF 1β and HNF 4), and hepatocyte specific markers such as tyrosine aminotransferase (TAT), transthyretin (TTR) and connexin 32 (Cx32) (Fig.1B), consistent with activation of hepatocyte differentiation program. In addition, gene-expression profiles of RLE cells before and after treatment demonstrated a significant induction of a number of enzymes characteristic of hepatocyte function (Fig. 1C). Therefore, cells submitted to the differentiation treatment were thereafter referred to as “differentiated RLE” or cells with “differentiated phenotype” as opposed to “control RLE” or “control phenotype.”
Differentiated RLE Cells Acquire Sensitivity to TNFα-Mediated Apoptosis.
To examine the susceptibility to TNFα treatment, control, and differentiated RLE cells were treated with TNFα (20 ng/mL) alone or in combination with CHX (0.5 μg/mL) known to inhibit protein synthesis. FACS analysis revealed that TNFα or CHX alone had no effect on cell viability. Significantly, combination of these two factors induced cell death only in differentiated but not control cells, as demonstrated by the appearance of a subdiploid peak (Fig. 2A). The presence of apoptotic cells was confirmed by fluorescence microscopy of propidium iodide–stained cultures. Again, only differentiated RLE cells treated with TNFα+CHX showed nuclear features consistent with apoptosis (Fig. 2B). The apoptotic process was very rapid. The first apoptotic cells were readily detectable at 1–2 hours after treatment and by 6 hours, about 50% of cells died (Fig. 2C). Significantly, the acquisition of sensitivity to TNFα killing depended on the duration of treatment with 5AC+[Dex+OSM]. The number of cells undergoing apoptosis in response to the cytokine progressively increased along with the duration of the differentiation treatment and reached maximum levels after 8 days (Supplementary Fig. 1A) (Supplementary material for this article can be found on the HEPATOLOGY website (http://interscience.wiley.com/jpages/0270-9139/suppmat/index.html). An acute treatment for1–2 hours with each of the differentiating agents did not affect the resistance of control RLE cells to TNFα (data not shown). Furthermore, the sensitization to TNFα-mediated apoptosis was an irreversible process. When differentiated RLE cells were replated and grown in regular medium for additional 4–6 days, cells maintained both the differentiated phenotype and sensitivity to TNFα+CHX (Supplementary Fig. 1B). Together, these data suggest that the capacity of RLE to respond to TNFα-induced apoptosis is linked to the differentiation state.
TNFα-Mediated Apoptosis in Differentiated RLE Cells Is Caspase Dependent.
The expression levels of TNFR1 were comparable in control and differentiated RLE cells (data not shown) excluding the possibility that resistance of control cells to TNFα was due to a lack of TNFR1 expression. We then conducted a step-by-step analysis of the apoptotic signaling pathway induced by TNFα. In differentiated RLE cells, the levels of procaspase-9 decreased concomitantly with appearance of cleaved fragments after 2–4 hours of TNFα treatment (Fig. 3A). In contrast, cleavage was barely detectable in control cells, and the levels of procaspase-9 remained constant. Similarly, activation of caspase 3 and subsequent cleavage of target substrates such as PARP, occurred only in differentiated RLE cells (Fig. 3A). Finally, pretreatment of differentiated cells with a broad-spectrum caspase inhibitor, z-VAD, resulted in a complete inhibition of apoptosis (Fig. 3B), demonstrating that TNFα-mediated apoptosis is caspase dependent.
Alterations in Mitochondria During TNFα Treatment.
To determine if apoptosis induced by TNFα in RLE involves mitochondrial dysfunction, ΔΨ mito was analyzed by confocal microscopy using the fluorescent dye JC-1. Strikingly, dissipation of the ΔΨ mito was detected in both phenotypes as early as 30 minutes after TNFα treatment (Fig. 4A, b and f), as visualized by the loss of red fluorescence and increase in green fluorescence. This phenomenon persisted and intensified after 2 (Fig. 4A,c and g) and 4 hours of treatment (Fig. 4A,d and h). However, apoptotic cell death was found only in cultures of differentiated cells (Fig. 4A, g and h). Since depolarization of the mitochondria leads to release of apoptogenic proteins, including cytochrome c, into the cytosol,28 we analyzed the release of cytochrome c in RLE cells by Western blotting. An increase in the cytoplasmic levels of cytochrome c was observed in both phenotypes, consistent with the timing of mitochondrial damage (Fig. 4B). Furthermore, z-VAD completely blocked the release of cytochrome c (Fig. 4C), indicating that the TNFα-mediated changes in mitochondria are caspase dependent. Together, these results demonstrate that both control and differentiated RLE cells suffer mitochondrial dysfunction upon TNFα treatment. However, only differentiated RLE cells were able to complete the apoptotic pathway through successful activation of caspases. Recently, the inhibitors of apoptosis (IAPs) have been shown to be major regulators of the caspase cascade.29 Since these proteins act downstream of mitochondria by direct binding and inhibiting caspases, we tested whether they could provide an explanation for the differential response to TNFα-mediated apoptosis. No differences were found either in the protein levels for IAP1, IAP2, XIAP and Survivin, or in the protein expression and subcellular distribution of their negative regulators, Smac/DIABLO and XAF-1 (Supplementary Fig. 2A–D).
TNFα Induces Oxidative Stress in Differentiated RLE Cells.
We examined whether ROS production played a role in TNFα-induced apoptosis in differentiated RLE. Both cell types displayed a transient increase in ROS upon treatment with TNFα, which peaked at 10 and 20 minutes in control and differentiated cells, respectively (Fig. 5A). However, the differentiated cells produced about an order of magnitude more ROS than control RLE suggesting that an overproduction of ROS might account for sensitization of differentiated RLE to TNFα-mediated apoptosis. Accordingly, 30 minutes pretreatment with 50 μM PDTC or 5 μM; α-LA blocked the apoptotic response demonstrating that ROS played a causal role in the induction of apoptosis (Fig. 5B).
Since association between the decrease in hepatocellular glutathione and the sensitization of hepatocytes to TNFα cytotoxicity has been recently established,16, 30, 31 we measured the glutathione content in RLE cells. After TNFα treatment, differentiated RLE cells displayed a persistent depletion of total glutathione that resulted in a significant increase in the GSSG/GSH+GSSG ratio (Fig. 6A). In contrast, control RLE cells maintained the intracellular levels of glutathione. These data suggested that depletion of glutathione may be responsible for the critical accumulation of ROS and for the disruption of the redox balance observed in the differentiated RLE cells. Significantly, glutathione content was maintained in differentiated RLE cells in the presence of antioxidants concomitantly with the resistance to TNFα-induced apoptosis (Fig. 6B). These results indicate that TNFα-mediated apoptosis in RLE occurs through an oxidative stress-dependent mechanism, and suggest that disruption of cellular defense mechanisms against oxidative stress might be responsible for the conversion from a resistant to a sensitive phenotype.
Depletion of Glutathione Sensitizes Control RLE Cells to TNFα-Mediated Apoptosis.
To directly address the significance of glutathione depletion in sensitizing RLE cells to TNFα-induced apoptosis, control cells were incubated either in the presence of DEM, which acutely depletes glutathione, or BSO, an inhibitor of γ-glutamyl cysteine synthetase, the rate limiting enzyme in the glutathione biosynthesis pathway. Noncytotoxic doses of DEM and BSO, 7 μM and 2mM, respectively, provoked a 30%–50% depletion in intracellular glutathione content and increased about 2-fold the percentage of cells undergoing apoptosis following TNFα+CHX treatment (Fig. 6C). These results demonstrate the key role for glutathione-dependent antioxidant defense mechanisms in the protection against oxidative damage and apoptosis induced by TNFα in RLE cells.
We have used an in vitro system of a hepatic stem-like cell line to examine whether sensitivity to TNFα-induced cytotoxicity depends on the differentiation state. The data demonstrate that concomitantly with the differentiation process, RLE cells acquire the properties of a type II cells and undergo a ROS-dependent apoptotic death after TNFα treatment. Furthermore, we show that efficient maintenance of the intracellular pool of glutathione is a key factor in the resistance of immature hepatic cells against TNFα-mediated cell death.
The fact that hepatic stem cells can repopulate the injured liver when the functions of the residual mature hepatocytes are compromised suggests that the progenitor cells possess properties required for survival and growth under adverse conditions. Consistent with this idea, genetic modification of hepatocytes conferring resistance to apoptosis results in selective advantage for liver repopulation during Fas-mediated hepatotoxicity.32 Furthermore, Roskams et al.6 have recently described a positive correlation between the degree of oval cell activation and extent of oxidative damage in mice and humans with fatty livers suggesting that oval cells possess survival advantage against oxidative damage. Our finding that a hepatic stem-like cell line is resistant to oxidative stress and apoptosis induced by TNFα is in agreement with these in vivo observations. Resistance to apoptosis as a mechanism for adaptation to environmental stress seems to be a common trait of various progenitor cell types. Thus, expression of negative regulators of apoptosis is selectively induced during early stages of myogenesis to allow myoblasts to overcome apoptosis and differentiate into mature myocytes.33 Similarly, neural progenitor cells in culture exhibit higher resistance to drug-induced neurotoxicity than primary neuronal cells.34 On the other hand, induction of apoptosis also plays an important role in stem cell biology and tissue homeostasis. Self-renewal of hematopoietic and gastrointestinal stem cells is counterbalanced by apoptosis in order to maintain a stable stem cell population.35, 36 Similarly, during oval cell expansion induced by a single treatment with 2-AAF, equilibrium between mitosis and apoptosis allows the maintenance of a constant number of ductal cells in the periportal areas.37 These observations clearly indicate that the choice between death and survival for progenitor cells is subjected to tight regulation. We hypothesize that the acquisition of sensitivity to apoptosis by differentiated RLE cells is not only an attribute of a mature phenotype but also may contribute to the regulatory mechanisms ensuring liver homeostasis. It is noteworthy that although the mechanisms of antiapoptotic adaptation are beneficial during early stages of activation/expansion of progenitor cells, the failure to apoptose can be detrimental resulting in among other things tumor development.38
The mechanisms evolved by cells to resist apoptosis are innumerable. Our experiments show that control RLE cells possess a functional TNFR. This is demonstrated by control RLE cells that, similarly to differentiated cells, respond to TNFα by induction of downstream events, such as mitochondrial depolarization and cytochrome c release.
However, despite a marked mitochondrial dysfunction, ROS accumulation upon TNFα treatment is significantly lower in control than in differentiated RLE cells. Previous studies have shown that mitochondrial damage or cytochrome c release do not inevitably lead to caspase activation suggesting that at this stage cell death may be reversed by appropriate survival signals.39, 40 Therefore, we hypothesized that efficient defense mechanisms might be operating in control RLE to prevent TNFα-dependent ROS accumulation and subsequent toxicity. Glutathione redox cycle is a well-known mechanism of cellular defense against oxidative stress.41 Glutathione is normally present at high levels in hepatocytes, and numerous studies have shown that this tripeptide modulates hepatocyte sensitivity to TNFα-induced cell death.15, 30, 31, 42, 43 In this study, we found that control RLE cells possess a striking capacity to maintain the intracellular glutathione content rendering them resistant to TNFα. This conclusion is based on the following findings: (1) apoptosis occurs only when the intracellular glutathione pool is depleted, (2) antioxidants preserve glutathione content and protect against TNFα killing, and (3) enforced reduction of glutathione levels sensitizes otherwise resistant cells to TNFα-induced apoptosis. Glutathione, apart from being a general regulator of intracellular redox state, works as a substrate for detoxification.44 Interestingly, studies on transcriptional profiling of stem cells revealed that one of the hallmarks of “stemness” appeared to be a high resistance to stress.45 Among the mechanisms responsible for this property was up-regulation of genes involved in detoxification systems, including MDR1 and Gsta4. MDR1 is a member of the ABC transporter family or MRPs (multidrug resistance proteins), involved in drug resistance,46 and Gsta4, an isoform of glutathione S-transferase, is an enzyme critical for the glutathione conjugation reaction. It has been proposed that the physiological function of MRP1is the extrusion of endogenously formed glutathione-conjugates,47, 48 suggesting that the MRP/glutathione system plays a role in the cellular defense system against oxidative stress which is required to prevent cellular damage induced by inflammatory stimuli. Accordingly, Ros et al.49 have found high levels of several members of the ABC superfamily of membrane transporters, including MDR1, MRP1 and MRP3, in the hepatic progenitor cell compartment in diseased liver. The same authors have also described high expression of MRP1 and MRP3 in cultured RLE cells.50 Based on this information and our data, we hypothesize that glutathione works as part of a detoxification system intrinsic to hepatic progenitor cells. Further studies are necessary to determine the mechanisms involved in regulation of intracellular glutathione levels in these cells.
In conclusion, we have shown that glutathione, a principal intracellular antioxidant agent, is a major determinant for the resistant phenotype exhibited by liver-derived progenitor cells in response to stress and that acquisition of sensitivity to TNFα-mediated apoptosis is an important feature of hepatic progenitor cell differentiation. However, extending these studies to an in vivo system in which the sensitivity to TNFα mediated apoptosis can be assessed as a function of the differentiation stage of the hepatic stem cell progeny is now needed. Nevertheless, these results provide new insights into the potential mechanisms used by hepatic stem cells and early progenitors to overcome the cytotoxic conditions in the liver microenvironment.
The authors thank Susan Garfield for assistance with Confocal Microscopy, Ju-Seog Lee for microarray analysis, and Tanya Hoang for help with tissue culture.