Hepatocyte transplantation is restricted by the impaired ability of hepatocytes to engraft and survive in the damaged liver. Understanding the mechanisms that control this process will permit the development of strategies to improve engraftment. We studied changes in liver matrix during acute injury and delineated the mechanisms that perturb the successful adhesion and engraftment of hepatocytes. Collagen IV expression was increased in sinusoidal endothelium and portal tracts of fulminant hepatic failure explants, whereas there were minimal changes in the expression of fibronectin, tenascin, and laminin. Using an in vitro model of cellular adhesion, hepatocytes were cultured on collagen-coated plates and exposed to serum from patients with liver injury to ascertain their subsequent adhesion and survival. There was a rapid, temporally progressive decrease in the adhesive properties of hepatocytes exposed to such serum that occurred within 4 hours of exposure. Loss of activity of the β1-integrin receptor, which controls adhesion to collagen, was seen to precede this loss of adhesive ability. Addition of the β1-integrin activating antibody (TS2/16) to cells cultured with liver injury serum significantly increased their adhesion to collagen, and prevented significant apoptosis. In conclusion, we have identified an important mechanism that underpins the failure of infused hepatocytes to engraft and survive in liver injury. Pretreating cells with an activating antibody can improve their engraftment and survival, indicating that serum from patients with liver injury exerts a defined nontoxic biological effect. This finding has important implications in the future of cellular transplantation for liver and other organ diseases. (HEPATOLOGY 2004;40:636–645.)
Few conditions in medicine are more devastating or more dangerous than fulminant hepatic failure (FHF).1, 2 The only effective treatment, whole-organ orthotopic liver transplantation is limited by the number of donor organs as well as by the requirement for lifelong immunosuppression,3 necessitating the development of alternative treatment strategies.
Cell transplantation of hepatocytes, has been studied extensively in vivo4 and has been shown to contribute to hepatic function, although benefit has been largely limited to metabolic liver diseases.5 However, when hepatocyte transplantation has been used in models of acute liver injury there is a significant reduction in both engraftment and subsequent efficacy.6 Several hours after portal-vein infusion, hepatocytes can be found in 2 main areas: hepatic sinusoids and periportal areas. While more cells are found in the periportal area, within an hour of their arrival there is activation of macrophages and phagocytes, leading to their near total removal in 24 to 48 hours.7 Notably, the sinusoidal bed is capable of sustaining transplanted cells up to 10% or more of the host hepatocyte mass.8 In the sinusoidal bed can be found the optimum microenvironment for an engrafting hepatocyte, in terms of extracellular matrix (ECM) components, growth factors, nutrients, and interactions with other cells. Several hours after cell transplantation, the sinusoidal endothelium is disrupted, facilitating the entry of hepatocytes into the space of Disse; this is partially induced by vascular endothelial growth factor release from transplanted and host hepatocytes.9
Adhesion to ECM is essential for the growth and survival of hepatocytes, so that their displacement from ECM leads to dedifferentiation and apoptosis, also known as anoikis.10 Integrins appear to be the major receptors by which cells attach to ECM and in mediating cell-cell adhesion events.11 Integrins are heterodimeric transmembrane proteins consisting of noncovalently linked α and β chains, each with a large, extracellular, single transmembrane and short cytoplasmic domain.12 Integrins are activated by contact with ECM proteins,13 leading to inhibition of interleukin-1β–converting enzyme expression, thus preventing apoptosis and playing a major role in determining cell survival.14 Therefore, factors that modulate the interaction and activity of integrins play an important role in determining cell adhesion and survival.15, 16 Our understanding of the factors controlling integrin affinity is incomplete, but work suggests that control is exerted in an energy-dependent fashion on their cytoplasmic domain.16 CD98 has been suggested as a potential controller of β1-integrin affinity, its cross-linking leading to activation of the β1 integrin and hence maintenance of integrin survival signaling.17
A clear understanding does not exist of which factors in liver injury modulate the engraftment and survival of transplanted cells. Serum from patients with FHF has been shown to induce a dose-dependent inhibition of proliferation18–24 and protein synthesis,25 and in some studies induces cell death.23, 26, 27 Is there a change in the hepatic microenvironment, making it less receptive to engraftment, or is there a change in the hepatocyte phenotype so that it is less likely to engraft? The aim of this study was to study the mechanisms that impair the engraftment of hepatocytes in the face of FHF.
Serum was pooled from patients with FHF due to paracetamol overdose. All patients were in the intensive therapy unit at time of sampling and fulfilled the poor prognosis criteria for liver transplantation.28 Samples were taken within 48 to 96 hours of overdose, and none had detectable paracetamol at the time of sampling. This study had local ethics committee approval. Normal serum was obtained from healthy individuals.
Reagents were purchased from Sigma Chemical Co. (Dorset, UK) unless otherwise specified. Fluorescent phalloidin was purchased from Molecular Probes Inc. (Eugene, OR). Fetal calf serum (FCS was purchased from Gibco (Strathclyde, UK). Dispase reagent was purchased from Boehringer Mannheim (Lewes, UK). Vector Target Retrieval solution was purchased from Vector Laboratories (Peterborough, UK). Envision Kit and antibodies directed against collagen IV, laminin, fibronectin, and tenascin were purchased from DAKO Laboratories (Ely, UK). Anti-β1– (12g10 and 7gq), α2-, and α6-integrin antibodies were purchased from Serotec Ltd. (Oxford, UK).
HepG2 is a human hepatoma-derived line, and it was routinely grown in mono/multiplayer cultures in Dulbecco's modified eagle medium containing 10% (vol/vol) FCS in an atmosphere of 5% CO2 in air. Cells were subcultured every 7 days at a split ratio of 1:3 using 0.05% (wt/vol) trypsin and EDTA to detach the cells.
Isolation and Primary Culture of Human Hepatocytes
Normal liver tissue was obtained from patients undergoing partial hepatectomy and sequentially perfused with buffers before conversion to a pulp as described previously.29 The hepatocytes were isolated by differential centrifugation at 50g for 3 minutes (3 times) followed by centrifugation through a Percoll gradient. This procedure gave a cell suspension of pure hepatocytes which were 96% to 100% viable. Hepatocytes were resuspended on collagen-coated plates and maintained in humidified air containing 5% CO2 at a temperature of 37°C. Culture media consisted of William's medium supplemented with 2mmol glutamine, penicillin/streptomycin, and 10% FCS.
Cell Necrosis and Apoptosis Experiments
Cell Necrosis (Propidium Iodide).
Cells were grown on 25 cm2 flasks and exposed to media containing 20% (vol/vol) FHF serum, 20% normal serum (vol/vol), or 10% FCS (vol/vol) for 1, 4, 12, 24, 48, and 72 hours. Adherent and nonadherent cells were removed from the flasks, washed with phosphate-buffered solution and incubated with propidium iodide. Samples were then processed in a FACS scanner and their exclusion of propidium iodide was assessed, as previously described.30
Cells were grown on glass slide-flasks (NUNC, Loughborough, UK) and exposed to media containing 20% (vol/vol) FHF serum, 20% normal serum (vol/vol), or 10% FCS (vol/vol) for 1, 4, 12, 24, 48, and 72 hours. Slides were incubated in Bouin's reagent for 2 hours and 1 N hydrochloric acid for 1 hour, and then washed in water prior to 45 minutes of incubation in Schiff's reagent. They were washed again before counterstaining in light green for 10 seconds. Slides were then washed, dehydrated through alcohols, submersed in xylene, and then coverslipped. The number of apoptotic nuclei were counted per 100 cells. This was performed 6 times per slide and each experiment was repeated 9 times. Values were expressed as percentage of apoptotic cells. Slides were counted by 2 blinded experienced observers.
The Enzchek Caspase-3 Assay Kit (Molecular Probes Inc.) was used. Cells were grown on 75 cm2 flasks and exposed to media containing 20% (vol/vol) FHF serum, 20% normal serum (vol/vol), or 10% FCS (vol/vol) for 24 hours. Cells were harvested, lysed, and assayed as described in the kit protocol. Reactions were carried out at room temperature and fluorescence was measured in a fluorescent microplate reader using excitation at 485 ± 12.5 nm after 60 minutes. Values were expressed as the difference between samples incubated with and without the Ac-DEVD-CHO inhibitor to ensure that the values reflected activity of caspase-3–like proteases.
Annexin V Staining.
Cells were grown on 25 cm2 flasks and exposed to media as described previously. They were detached using dispase (Boehringer Mannheim) and resuspended in binding buffer. Samples were then incubated with FITC-labeled annexin V (Sigma Chemical Co.) for 10 minutes on ice before being analyzed on a FACS machine.30
Scanning Electron Microscopy
HepG2 cells were grown on Perspex coverslips according to the same treatment schedule described in the cell culture section. After 24 hours, culture samples were fixed using 2.5% glutaraldehyde and then dehydrated and coated with gold for scanning electron microscopy.
Archival sections from explanted livers from patients with FHF and from healthy livers were obtained. Sections were also obtained from patients with chronic liver disease. Slides were dewaxed in xylene and progressively dehydrated through graded alcohols and water. Antigen retrieval was performed using trypsin (15 minutes at 37°C) or Vector Target Retrieval solution (microwaved for 15 minutes). Slides were then processed with the Envision Kit (DAKO Laboratories) for collagen IV, laminin, fibronectin, and tenascin (DAKO Laboratories).
Cells were grown on 25 cm2 flasks and then exposed to media containing 20% (vol/vol) FHF serum, 20% normal serum (vol/vol), or 10% FCS (vol/vol) for 1, 4, 12, 24, 48, and 72 hours. Adherent and nonadherent cells were removed from the flasks, washed with PBS, and placed onto collagen-coated wells (96-well plate) at a concentration of 2 × 105 cells/100μL for 1 hour. Wells were then washed, fixed with methanol, stained with methylene blue, and lysed with hydrochloric acid. Plates were read on a spectrophotometer to quantify cellular adhesion. Results were expressed as a percentage of the adhesion results obtained with cells cultured with DMEM and 5% FCS (vol/vol).31 This assay tests the ability of cells to adhere de novo to matrix after prior exposure to differing sera.
Cells were grown on 25 cm2 flasks and exposed to media containing 20% (vol/vol) FHF serum, 20% normal serum (vol/vol), or 10% FCS (vol/vol) for 24 and 48 hours.
Cells were removed from these flasks, washed, and incubated with FITC conjugated monoclonal antibodies directed against β1-, α2-, and α6-integrin molecules (Serotec Ltd.). Expression was assessed flow cytometrically.31
β1-Integrin Activation Analysis
Cells were grown on 25 cm2 flasks and exposed to media containing 20% (vol/vol) FHF serum, 20% normal serum (vol/vol), or 10% FCS (vol/vol) for 1, 4, and 24 hours.
Cells were removed from these flasks, washed, and incubated with HEPES buffer. Cell pellet obtained was split into 3 aliquots, which were incubated with HEPES buffer containing the 12g10 antibody (antibody recognizes the active configuration of the β1 integrin). Aliquot 1 (“inhibited”) contained ethylenediaminetetraacetic acid (to inhibit β1-integrin activity), aliquot 2 (“activated”) contained manganese (to activate β1-integrin activity), and aliquot 3 (“native”) had no additions. Samples were processed flow cytometrically to assess β1-integrin activity using the following equation31:
Confocal Actin Microscopy
Cells were cultured on collagen-coated coverslips, fixed with 4% formaldehyde for 20 minutes, rinsed with PBS, and permeabilized with 0.5% Triton X before addition of fluorescent phalloidin for a further 20 minutes. Coverslips were then washed again with PBS prior to mounting and analysis on a confocal microscope.
This antibody is derived from a hybridoma and was used at a concentration of 25 μg/mL in tissue culture (based on flow-cytometric binding data).32
To compare groups, t test assuming unequal variances was used. Where multiple analyses were required, ANOVA with post-hoc analysis was performed. A P value of less than .05 was used to define significance.
Liver Injury Leads to Up-regulation of Collagen IV in the Hepatic Sinusoids and Portal Tracts
We determined the ECM composition of unselected liver specimens from patients with paracetamol-induced FHF by immunohistochemical staining for collagen IV, fibronectin, laminin, and tenascin. There is marked up-regulation of collagen IV in the hepatic sinusoids of patients with paracetamol-induced FHF (Figs. 1B and C) compared with healthy controls (Fig. 1A). This contrasts with the expression of laminin (Figs. 1J-L), tenascin (Figs. 1G-I), and fibronectin (Figs. 1D-F): this expression only occurs in areas of hepatocyte necrosis and reflects nonspecific uptake.
Adhesion of HepG2 Cells Is Impaired by Exposure to Serum From Patients With FHF
In view of the marked up-regulation of collagen expression in damaged liver, we studied the binding of HepG2 cells to collagen-coated plates after previous culture with fulminant serum. Serum from patients with FHF caused a marked loss of adhesive properties on HepG2 cells after only 4 hours of incubation (Fig. 2A ); this contrasted with serum from normal controls, which did not alter adhesive properties (54.9 ± 6.3% vs. 92.7 ± 10.1%; P < .01). The effect of serum from patients with FHF on HepG2 cells was studied further by scanning electron microscopy images, which revealed striking differences in morphology after 24 hours of culture as demonstrated by rounding up of cells (Figs. 3A-D ), in marked contrast to the well-spread cells cultured with serum from normal controls. Further analysis revealed changes to cytoskeletal structure as demonstrated by disruption of actin cytoskeleton (Figs. 3E-H). This key observation that serum from patients with FHF markedly down-regulates the adhesion of cells to collagen is of paramount importance in cellular transplantation.
Cell Death Occurs After the Loss of Cellular Adhesion and Is Due to Caspase-3–Mediated Apoptosis
The next step was to determine whether loss of adhesion was the result of impaired cell viability or whether it led to cell death. Cell death studies (propidium iodide exclusion) demonstrated that while serum from patients with FHF causes cell death, this occurs over 44 hours after the loss of cellular adhesion becomes apparent (Fig. 1B). Studies of early markers of apoptosis, annexin V staining, demonstrate that apoptosis does not occur in cells incubated in serum from patients with FHF until 24 to 48 hours of exposure (Fig. 4A ). Cell death occurs by caspase-3–mediated apoptosis (Figs. 4B and C) and peaks at 6.36 ± 1.34% by 48 hours (vs. 0.87 ± 1.3% for normal serum; P < .001). Levels of apoptosis greater than 2% are considered biologically significant. We have therefore demonstrated that the down-regulation of adhesion occurs much earlier than the first evidence of apoptosis. This suggests that loss of adhesive status is likely to be the cause of apoptosis rather than the result of it. This important experiment illustrates that although only 5% of the cultured hepatocytes are apoptotic, their ability to adhere de novo to new matrix is reduced to 50% of the level of cells cultured in conventional medium.
Exposure of Primary Human Hepatocytes to Serum From Patients With FHF Also Leads to Loss of Their Cellular Adhesion and Precedes the Onset of Apoptosis
Primary human hepatocytes displayed similar responses to HepG2 cells after exposure to serum from normal controls and patients with FHF, with peak reduction in adhesive capacity occurring between 4 and 24 hours of incubation with serum from patients with FHF (Fig. 2C). After 4 hours, adhesion after exposure to normal versus FHF serum was 61.9 ± 1.94% and 48.0 ± 3.2% (P < .01), respectively; at 24 hours, adhesion after exposure to normal versus FHF serum was 72.8 ± 2.8% and 46.0 ± 2.66% (P < .01), respectively. As with HepG2 cells, serum from patients with FHF could be seen to induce apoptosis in primary hepatocytes (Fig. 4D), but only 20 or more hours after the observed loss of cellular adhesion. There was a small increase in apoptosis after 24 hours; this became more significant after 48 hours (8.4 ± 1.5% for FHF serum vs. 1.8 ± 0.3% for normal serum; P < .001). These key observations indicate that primary human hepatocytes react in a similar fashion to HepG2 cells after exposure to serum from patients with FHF.
Down-regulation of β1-Integrin Receptor Activity Rather Than Loss of Total β1-Integrin Receptor Expression Is Associated With Decreased Cellular Adhesion
To identify the mechanisms that control adhesive capacity of HepG2 cells, we studied the expression of those integrin receptors that regulate adhesion to collagen and that are also found on HepG2 cells.33 β1, α2, and α6 are the major integrins controlling the binding to collagen, and these were subsequently studied. After 24 hours of culture with serum from patients with FHF, when adhesion was significantly impaired, there was increased expression of the β1 and α6 integrins (Fig. 5A ). To assess the activity of the β1 integrin, an antibody that recognizes only its active conformational state, 12g10, was used. However, as shown in Fig. 5B, the proportion of β1-integrin receptors in their active conformational state was much lower after culture for 24 hours with FHF serum (33.2 ± 6.2% vs. 68.6 ± 2.4%; P < .01) than after culture with normal serum. Furthermore, Fig. 5C demonstrates that this down-regulation occurs early, between 1 and 4 hours after exposure to serum from patients with FHF. We demonstrate that while the levels of the integrins that mediate binding to collagen are unchanged/increased after exposure to serum from patients with FHF, the activity of the major integrin, β1, is rapidly and significantly down-regulated. Therefore, we conclude that the mechanism behind the loss of cellular adhesion and hence apoptosis is down-regulation of the β1 integrin on engrafting cells.
Activation of β1-Integrin Receptor Prevents the Loss of Cellular Adhesion and Onset of Cellular Adhesion Seen With Serum From Patients With FHF
To see whether modulation of β1-integrin modulation would influence adhesion and apoptosis, we added TS2/16 antibody (β1-integrin receptor activating antibody) to cells incubated with serum from patients with FHF. Importantly, we showed that it almost completely reversed the down-regulation of adhesive ability and also led to a 44.2% decrease in the amount of cellular apoptosis (2.3 ± 0.25% vs. 5.2 ± 0.72%; P < .01) seen at 48 hours (Figs. 6A and B ). This antiapoptotic effect is also supported by measurement of caspase activity (Fig. 6C). This further strengthens the hypothesis that the mechanism behind failure of cellular engraftment in liver injury is a rapid reversible down-regulation of β1-integrin activity.
This study demonstrates that paracetamol-induced liver injury results in the marked up-regulation of collagen IV on sinusoidal cells, known to be the main conduit for infused hepatocytes entering the hepatic plate. By testing adhesion of hepatocytes to these extracellular matrices, we have shown that serum from patients with FHF reduces their adhesive capability by a rapid down-regulation of their β1-integrin activity. This loss of β1-integrin activity and cellular adhesion subsequently results in the later onset of caspase-3–mediated apoptosis. Importantly, we have demonstrated that the effect of liver injury serum is a specific effect and not just a nonspecific toxic effect. This is evidenced first by the absence of any apoptosis at any of the early time points, when integrin affinity and adhesion are compromised. Second, we demonstrate that loss of cellular adhesion and subsequent apoptosis can be reversed by treatment of the hepatocytes with the stimulatory monoclonal antibody TS2/16, again reinforcing the point this is not a nonspecific toxic phenomenon.
While previous studies have examined the expression of different ECM markers in chronically damaged liver injury,34 such data is lacking in acute liver injury; therefore, our study provides novel information on ECM distribution in this setting. Specifically, we have shown a striking increase in the expression of collagen IV that is localized to the sinusoidal endothelium and to the portal tracts. These areas are known to be important portals of adhesion and subsequent entry of potentially engrafting cells into the liver plate.7 Consequently, we studied hepatocyte binding to collagen and looked at the integrin molecules involved in adhesion to collagen,33 namely β1, α2, and α6. Integrin-mediated adhesion is fundamental to cell transplantation because only after this has been successfully negotiated can cells respond to appropriate migratory signals and migrate across the endothelium and into the liver plate.
The increase in β1-and α6-integrin expression seen after 24 hours of culture with serum from patients with FHF is occurring at a time of impaired cellular adhesion and suggests a possible compensatory response by the cells to increase their attachment. This is confirmed by assessment of the activity of their β1 integrin, which is seen to be lowered after 4 hours incubation. Manipulation of integrin activity has been suggested to occur by modification of its cytoplasmic component (inside-out signaling), and several mechanisms have been identified, including covalent modification, the attachment of activator proteins, or the removal of repressor proteins. Serine phosphorylation has been demonstrated in mitotic cells to lead to the rounding and detachment of cells as a consequence of integrin down-regulation. A further possibility is external modulation of integrin affinity (outside-in signaling), although this tends to result in up- rather than down-regulation and in the context of our study may be the result of a loss of such a stimulating signal in serum from patients with FHF. Other important regulatory factors include temperature, extracellular divalent cations,35 and interaction with intracellular signal transduction molecules such as calreticulin36 and members of the small GTPase family such as R-Ras37 and Rho A.38
Once activated, integrins lead to protein tyrosine phosphorylation, which has been shown to play a protective role in the regulation of apoptosis.39, 40 In the setting of small cell lung cancer cells, ECM-mediated protection from etoposide-induced caspase-3 activation can be blocked by either β1-integrin function-blocking antibody or by a tyrosine kinase inhibitor.31, 41 Other groups have demonstrated that ECM regulates apoptosis (mammary epithelial cells) through an integrin-dependent negative regulation of interleukin-1β–converting enzyme (ICE) expression, whereby expression of ICE was correlated with loss of ECM and inhibitors of ICE activity prevented apoptosis.42 Thus integrin activation prevents the ICE-mediated induction of caspase-3 activity. The stimulatory monoclonal antibody TS2/16 acts by directly inducing a conformational change of β1 integrins. Furthermore, while deletion of the cytoplasmic domain of an α subunit of β1 integrins abolishes phorbol ester stimulation, it has no effect on TS2/16 stimulation,43 indicating the effect is independent of intracellular processes. Finally, Fab fragments of the TS2/16 antibody have similar stimulatory properties, excluding the possibility of a receptor cross-linking mechanism.44
The differential control and order of adhesion and antiapoptotic signals by integrin activation is not fully understood, and in our model it is not clear if apoptosis is a result of decreased cellular adhesion or if it occurs independently. Our data suggests that cytoskeletal disruption (loss of adhesion and disruption of actin cytoskeleton) precedes the onset of apoptosis and may therefore play a role in inducing apoptosis.
β1 Integrin also plays an important role in cellular migration; studies with β1-integrin knockout hematopoietic stem cells demonstrated failure to engraft, without impairment of cellular differentiation. Notably, TS2/16 antibody stimulation of CD34+ cells abrogates almost all marrow-directed migration and leads to engraftment in fetal liver.45 This observation is of particular relevance with respect to our findings, suggesting that β1-integrin activity may be important in liver homing. Furthermore, we have shown an up-regulation of collagen IV expression on paracetamol-damaged liver. Collagen IV is a potent ligand for β1 integrin and is a possible entry point for hepatocytes after cellular transplantation. Thus, the down-regulation of β1 integrin we have observed suggests that this not only decreases hepatocyte homing but also prevents some of the collagen-mediated binding on the damaged liver.
Many groups have attempted to identify the important constituents of FHF serum to discover the agents responsible for its antiproliferative and necrotic properties.21, 24, 46 Although some have demonstrated that some of these properties lie within the sub-10 kd ultrafiltrate, it is likely that these properties are the additive effects of many compounds.24 The removal/supplementation of compounds found in high/low concentrations in such patients is therefore unlikely to improve outcomes of hepatocyte transplantation. A more realistic strategy is to identify the deranged cellular mechanisms, consequent to liver injury, and preemptively treat cells to improve engraftment and hepatocyte survival.
This study utilized HepG2 cells, a human hepatoma cell line, and primary human hepatocytes for certain experiments. With respect to the key observations, primary human hepatocytes reacted in a similar fashion to HepG2 cells after incubation with FHF serum: they display reduced ability to adhere to ECM before undergoing apoptosis. The slightly elevated levels of apoptosis seen with primary hepatocytes most likely reflect their nonimmortalized nature compared with HepG2 cells. The HepG2 line has many characteristics in keeping with primary hepatocytes, including moderately differentiated phenotype and similar integrin expression, allowing it to act as a suitable model for primary hepatocytes.47 Modifications of this line are used in bioartificial liver support systems48 and hepatocyte transplantation.49 It is important to note that this is not anoikis, which is apoptosis consequent to cells being nonadherent. In this study, apoptosis is seen to occur in adherent cells, and in this regard differences in anchorage-independent growth between HepG2 cells and primary hepatocytes are less important in the interpretation and extrapolation of data.
In conclusion, we have made the important observation that serum from patients with FHF rapidly decreases the adhesive capacity of hepatocytes as a prelude to apoptosis. Furthermore, this occurs by a rapid down-regulation of β1-integrin activity and is partially reversed by addition of a β1-integrin stimulating antibody, which prolongs hepatocyte adhesion and survival. Preactivation of β1 integrin could therefore offer a means of enhancing the efficacy of cell transplantation.