A mutation in keratin 18 that causes caspase-digestion resistance protects homozygous transgenic mice from hepatic apoptosis and injury

Keratin 8 and keratin 18 (KRT8 and KRT18, respectively; hereafter referred to as K8/K18) in the liver are a pair of intermediate filament proteins that consist of a central α-helical rod domain linked to the non-α-helical N-terminal head and C-terminal tail domains (Herrmann et al., 2009; Omary et al., 2009). They form obligate noncovalent heteropolymers that assemble into filaments and generate the cytoplasmic filament network that maintains cellular integrity (Coulombe et al., 2009). Keratin filaments are highly dynamic structures that can disassemble into nonfilamentous particles in mitotic cells or under various stresses (Eriksson et al., 2009).

Apoptosis occurs in several tissues during development and disease to remove harmful and unneeded cells; its distinct morphological changes include cytoskeletal and nuclear envelope disruption, chromatin compaction, cell shrinkage, cell fragmentation and formation of apoptotic bodies (Earnshaw, 1995; MacFarlane et al., 2000). Caspases, which are cysteine-dependent aspartate (Asp) proteases, cleave their substrates at Asp residues and play essential roles in apoptosis. There are two classes of caspases: initiator caspases that include caspases 2, 8, 9 and 10, and effector caspases that include caspases 3, 6 and 7 (Riedl and Shi, 2004). Initiators cleave and activate the effectors, which in turn cleave other substrates to promote apoptosis.

Various intermediate filament proteins, such as lamins (Lazebnik et al., 1995; Rao et al., 1996; Ruchaud et al., 2002), desmin (Chen et al., 2003), vimentin (Byun et al., 2001) and keratins (Caulín et al., 1997; Ku et al., 1997, 2016; Ku and Omary, 2001) are known caspase substrates. Caspase-dependent digestion of intermediate filaments leads to breakdown of the filament network and may facilitate morphological changes in cells that are associated with apoptosis. Resistance of caspase cleavage is essential in protecting the cells from apoptosis; for example, nuclear lamins − which are main elements of nuclear lamina − have caspase cleavage sites, i.e. VEID²³⁰ in lamin A and VEVD²³¹ in lamin B. Interrupting the caspase-dependent cleavage of lamins delays nuclear events during apoptosis, such as nuclear shrinkage, DNA fragmentation and onset of apoptosis (Lazebnik et al., 1995; Rao et al., 1996; Ruchaud et al., 2002). In the case of the muscle-specific intermediate filament protein desmin, VEMD²⁶³ is a caspase-6 cleavage site (Chen et al., 2003). Cells transfected with caspase-resistant desmin D263E mutant manifest alleviated apoptosis as compared with wild-type (WT) desmin-transfected cells (Chen et al., 2003). Consistent with these findings, desmin that cannot be cleaved by caspase protects cardiomyocytes from apoptosis and improves heart function during TNFα-mediated stress (Panagopoulou et al., 2008).

K18 has two caspase cleavage sites, VEVD²³⁸ and DALD³⁹⁷, and is digested during apoptosis. By contrast, K8 does not have the caspase cleavage consensus sequences and is, therefore, relatively resistant to apoptotic degradation (Caulín et al., 1997; Ku et al., 1997; Leers et al., 1999). VEVD²³⁸ or similar sequences are found in the rod domain of other intermediate filament proteins, whereas DALD³⁹⁷ is found only in the K18 tail domain (Ku and Omary, 2001). Apoptotic signals activate caspase-9, which is responsible for DALD³⁹⁷cleavage as an early event (Ku and Omary, 2001; Riedl and Shi, 2004). The breakdown of the keratin network into large aggregates is due to VEVD²³⁸ cleavage by activated caspase-6 in a later stage of apoptosis, with the consequent movement of the large keratin aggregates into cytoplasmic blebs (Schutte et al., 2004). Mutations of two Asp residues in K18 (D238/D397) to Glu (K18-D238/397E) renders K18 resistant to caspase-mediated digestion under cell culture conditions (Ku and Omary, 2001). Recently, it has been reported that the human K18-D238/397E double mutant, when overexpressed as a transgene in mice that retain endogenous mouse WT K18, results in keratin hypo-phosphorylation, reorganization and increased susceptibility to necrosis (Weerasinghe et al., 2014). However, the function of uncleavable K18 without endogenous mouse WT K18 has, so far, not been investigated in mice.

We, therefore, studied here the significance of K18 caspase cleavage by generating K18-D238/397E mice in a K18-null background (K18-D238/397E;mK18-null) by crossbreeding K18-D238/397E mice with K18-null mice. The K18-D238/397E;mK18-null mice showed remarkable resistance to liver injury when compared with control mice, based on mouse mortality, histopathological and biochemical analysis, terminal deoxynucleotidyl transferase dUTP nick-end labeling (TUNEL) assay and caspase activation. In addition, the non-cleavage form of K18 is more stable than WT K18 when tested in cells treated with cycloheximide or in transgenic mice treated with antibody against the tumor necrosis factor receptor superfamily member 6 (Fas), hereafter referred to as anti-FasAb. Therefore, caspase cleavage-resistant K18 mutation enhances resistance to keratin network disruption under stress, thereby providing protection of liver injury induced by treatment with anti-FasAb.

To investigate the biological significance of K18 cleavage at D238 and D397 during apoptosis (Fig. 1A), we used transgenic mice that overexpress human K18-D238/397E in an FVB/N mouse-strain background, which contains endogenous mouse K18. Expression of K18-D238/397E in mouse liver was confirmed by the lack of the two apoptotic fragments of human K18 (p43 and p29) that are observed in transgenic liver overexpressing human K18-WT as the result of anti-FasAb-mediated apoptosis (Fig. 1B). The mice did not show abnormal phenotypes under basal conditions (not shown). We then compared the mouse lethality of nontransgenic FVB/N, K18-WT and K18-D238/397E mice after anti-FasAb administration to induce apoptosis in the liver. After anti-FasAb injection, the death of mice starts after 8 h but no further death is noted beyond 50 h (Fig. 1C). Mouse mortality was slightly higher in K18-D238/397E mice when compared with nontransgenic FVB/N and K18-WT mice (53% compared with 44% and 42%, respectively; Fig. 1D). The trend of these findings parallels a previous study, which demonstrated that the K18-D238/397E mutation enhanced the susceptibility to anti-FasAb-mediated liver injury (Weerasinghe et al., 2014). However, in our study, the difference of lethality between the K18-D238/397E mice and control mice was not statistically significant (P>0.58) (Fig. 1D).

To generate K18-D238/397E mice that lack endogenous mouse K18 (K18-D238/397E;mK18-null), we crossbred K18-D238/397E mice with K18-null mice. As a control, we also generated K18-WT;mK18-null mice that expresses human K18-WT but not endogenous mouse K18-WT. No mouse K18 was observed in K18-D238/397E;mK18-null or K18-WT;mK18-null mice livers, as shown in Coomassie Blue-stained high-salt-extracted samples − which generate a highly enriched keratin fraction (Ku et al., 2004) − and immunoblots of total liver lysates (Fig. 2A). Given that the genetic background of the generated K18-WT;mK18-null mice is K18-null, we first tested the susceptibility of K18-null mice to anti-FasAb-induced liver injury. The mortality is dramatically increased in K18-null mice when compared with the nontransgenic FVB/N mouse (94% vs 56%, P<0.04; Fig. 2B,E). This enhanced mortality is probably due to the lack of a keratin filament network in liver of K18-null mice as described previously (Magin et al., 1998). The increased mortality of K18-null mice is alleviated by expression of human K18 WT in K18-null mice (K18-WT;mK18-null), which is comparable to the lethality of nontransgenic FVB/N mice (57% vs 56%) (Fig. 2E). These findings indicate that the reintroduction of human K18-WT rescues K18-null mice from liver injury.

We then compared the susceptibility to liver injury mediated by anti-FasAb of K18-D238/397E;mK18-null mice with that of K18-WT;mK18-null mice and found that mortality is markedly reduced in K18-D238/397E;mK18-null mice (23% vs 57%, P<0.001) (Fig. 2C,E). Similar to the pattern of mortality after anti-FasAb administration, we also observed a reduced rate of mortality of the mice after injection of microcystin-LR (MLR) (Fig. 2D,E). The phosphatase inhibitor MLR is a known hepatotoxin (Jochimsen et al., 1998; Ku and Omary, 2006). These results clearly demonstrate that expression of K18-D238/397E in K18-null mice protects the mice from liver injury.

Liver histology was carried out to determine the extent of tissue damage. Liver sections were prepared 6 h post treatment with anti-FasAb or vehicle (PBS). We observed no tissue damage in any mice injected with vehicle only (Fig. 3A). Administration of anti-FasAb caused severe hemorrhage after 6 h in liver of K18-WT;mK18-null mice, whereas the histological change was significantly attenuated in that of K18-D238/397E;mK18-null mice (Fig. 3A). These histological findings were supported by significantly lower levels of alanine transaminase (ALT) in K18-D238/397E;mK18-null serum (Fig. 3B). Given that involvement of apoptotic pathways is the dominant feature in anti-FasAb-mediated liver injury (Ogasawara et al., 1993), we compared the extent of apoptotic cell death by using a TUNEL assay. Neither group of mice injected with vehicle (PBS) alone showed TUNEL-positive apoptotic cells (Fig. 3C, left panels). However, after treatment with anti-FasAb, high numbers of apoptotic cells were observed in liver of K18-WT;mK18-null mice, whereas significantly fewer apoptotic cells were detected in that of K18-D238/397E;mK18-null mice (Fig. 3C, right panels). Thus, the K18-D238/397E mutation in livers of mK18-null mice protects them from liver apoptosis mediated by anti-FasAb.

Given the keratin filament network is crucial in maintaining cell integrity (Coulombe et al., 2009; Omary et al., 2009), we examined the extent of perturbation in the keratin network by keratin immunostaining of the liver sections (Fig. 3D) and the primary cultured hepatocytes (Fig. S1). Anti-FasAb treatment caused a more-prominent collapse of the keratin network in K18-WT;mK18-null hepatocytes than in K18-D238/397E;mK18-null hepatocytes. Collectively, these findings indicate that the caspase cleavage-resistant mutation of K18 helps to maintain hepatocyte integrity and protects liver of transgenic mice from apoptosis.

The protective role of K18-D238/397E mutation was further confirmed by immunoblotting, which shows markedly low levels of cleaved caspase 7 in most of the tested liver from K18-D238/397E;mK18-null mice as compared with that of K18-WT;mK18-null mice after injection with anti-FasAb (Fig. 4A). Notably, one of four livers of K18-D238/397E;mK18-null mice had a high level of cleaved caspase 7 that was probably caused by variation between individual mice as shown by the 23% mortality of K18-D238/397E; mK18-null mice in Fig. 2E. Overall, immunoblotting of cleaved caspase 7 suggested that one of the four K18-D238/397E mice underwent severe apoptosis whereas all four K18 WT mice show signs of apoptosis in their liver.

We examined the phosphorylation and expression of some apoptosis-related proteins−namely Akt, p42MAPK and p44MAPK (also known as MAPK1 and MAPK3, respectively), p90RSK (also known as RPS6KA1), p38MAPK (also known as MAPK14), JNK1 (also known as MAPK8), JNK2 (also known as MAPK9), NF-κB in anti-FasAb-treated livers. The phosphorylation and protein levels of NF-κB and the tested protein kinases were significantly increased and then were maintained in livers of K18-D238/397E mice. However, in livers of K18-WT mice, they had rapidly diminished 8 h after treatment with anti-FasAb (Fig. 4A), which is likely to be related to rapid protein degradation due to massive apoptosis in livers of K18-WT mice. Nonetheless, the reduced phosphorylation of proteins but not the overall reduced levels of proteins, might also have been involved in the substantially increased levels of apoptosis in the liver of K18 WT mice. For example, the protein levels of JNK1/2 8 h after treatment with anti-FasAb are similar in livers of K18-WT and K18-D238/397E mice, whereas levels of phosphorylated JNK1/2 in livers of K18-WT mice is more dramatically reduced when compared with those in livers of K18-D238/397E mice (Fig. 4A). In addition, other apoptosis-associated proteins, such as the Fas-associated protein with death domain (FADD) and Fas itself, are barely detectable in livers of K18 WT mice after stimulation with anti-FasAb, whereas levels of actin and Hsc70 are maintained (Fig. 4A). This might be due to the different accessibility of caspases, depending on their substrates. Specifically, actin forms a filament network and, thus, might be relatively stable under certain apoptotic conditions at which other cytosolic proteins (such as protein kinases described in Fig. 4A) are degraded. HSC70 (known as HSPA8) protein is also stable under apoptotic conditions and is used as a loading control protein.

We investigated the significance of specific protein degradation under apoptosis by using an ex vivo primary culture system. Primary hepatocytes were isolated from mouse liver that had been perfused and were treated with anti-FasAb. We then assessed the level of apoptosis and the amount of signaling proteins in both K18 WT and K18-D238/397E hepatocytes. Consistent with the protective role of K18-D238/397E mutation in vivo (Fig. 4A), we observed lower levels of cleaved caspase-7 and cleaved PARP in primary cultured K18-D238/397E hepatocytes as compared with K18-WT hepatocytes after treatment with anti-FasAb; levels of cleaved caspase-3 and caspase-8 were also reduced, albeit to a lesser extend (Fig. 4B). Notably, levels of the cell survival factor NF-κB are more stably maintained in both K18-D238/397E livers in vivo (Fig. 4A) and K18-D238/397E hepatocytes ex vivo (Fig. 4B). Indeed, NF-κB seemed more stable in K18-D238/397E hepatocytes compared with WT hepatocytes after treatment of cycloheximide, an inhibitor of protein biosynthesis (Fig. 4C). Therefore, under stress conditions, the keratin network might help to maintain the protein level of NF-κB, and NF-κB might play a role in the delayed apoptosis in K18-D238/397E hepatocytes. For example, we have previously shown binding between NF-kB and K8/K18 by coimmunoprecipitation (Lee et al., 2013), although the detailed molecular mechanism of their association remains to be determined.

A previous study demonstrated that the K18-D238/397E mutation in transgenic mice containing endogenous mouse K18-WT predisposed to hepatocyte necrosis during anti-FasAb-induced apoptotic liver injury (Weerasinghe et al., 2014). We compared the effect of K18-D238/397E mutation in transgenic mice without or with endogenous mouse K18-WT during hepatocyte necrosis. Primary hepatocytes from transgenic mice were prepared and treated with anti-FasAb and then the necrosis marker, high-mobility group box 1 protein (HMGB1) (Štros, 2010), was analyzed in hepatocyte culture medium. Under the tested condition, HMGB1 was not detected in the culture medium of K18-D238/397E hepatocytes from mice with an mK18-null background (K18-D238/397E;mK18-null mice), whereas increased levels of HMGB1was observed in the medium of control hepatocytes (Fig. S2). Consistent with the resistance of K18-D238/397E;mK18-null mice to liver injury in vivo (Fig. 2), this result indicated that K18-D238/397E mutation in mK18-null mice protects hepatocytes from necrosis during anti-FasAb-induced apoptotic liver injury. However, we could not observe a dramatic difference of HMGB1 levels in control hepatocytes and K18-D238/397E hepatocytes from transgenic mice containing endogenous mouse K18-WT, although HMGB1 levels in the mutant hepatocyte culture medium were slightly increased (Fig. S2).

Since K18-D238/397E hepatocytes treated with anti-FasAb show a relatively resistant keratin network (Fig. 3D and Fig. S1), we speculated that the increased protein stability of K18-D238/397E is related to enhanced hepatocyte integrity that protect the cells under stress. We, therefore, compared the stability of K18-D238/397E and K18-WT expressed in BHK cells after treating cells with cycloheximide (Fig. 5A). The stability of K18-D238/397E was increased up to 48 h after cycloheximide treatment, whereas K18-WT levels were reduced at 24 h concomitant with increased levels of the apoptotic K18 fragment (29 kD) (Fig. 5A) in the condition when the cell viability was greater than 90% (Fig. 5B). Consistent with this result, we observed the prolonged stability of K18-D238/397E protein in transfected cells after treatment with anisomycin, another protein synthesis inhibitor and apoptosis inducer (Fig. 5C). We also performed protein stability studies by using isolated primary hepatocytes after liver perfusion. Increased levels of K18 fragment (29 kD) coupled with decreased levels of intact K18 were detected in K18-WT hepatocytes but not in the mutant hepatocytes after cycloheximide treatment (Fig. 4C). Similar findings were made in livers of transgenic mice after administration of anti-FasAb (Fig. 5D and Fig. S3). Taken together, our findings show that the K18-D238/397E mutation enhances the protein stability of K18 (Fig. 5) and resistance to keratin network disruption under stress (Fig. 3D and Fig. S1), thereby providing hepatocyte integrity, leading to increased survival (Fig. 2).

Cytoskeletal filaments, such as microtubules, intermediate filaments and microfilaments, provide structural scaffolding to cells in order to maintain their structure and cellular morphology (Goldman, 1971; Jordan and Wilson, 1998; Kawauchi and Hoshino, 2008; Ku et al., 1999; Pollard and Cooper, 2009; Rao and Li, 2004). In apoptotic cells, dramatic morphological changes occur, such as membrane blebbing, cell shrinkage and fragmentation of the nucleus and cell (Earnshaw, 1995; MacFarlane et al., 2000), with caspases playing crucial roles in these morphological changes (Cohen, 1997). Since cytoskeletal proteins, such as actin, actin-binding proteins, tubulin, tubulin-associated proteins and various intermediate-filament proteins, are cleaved by caspases in the presence of apoptotic signals and are also involved in cell morphology, proteolysis of these cytoskeletal proteins is likely to be related to apoptotic morphological changes in response to apoptosis (Taylor et al., 2008).

Intermediate filaments share a highly conserved caspase consensus sequence/caspase cleavage site (Table 1). Prior studies have demonstrated that caspase cleavage-resistant forms of intermediate filaments have a protective role after induction of an apoptosis stimulation. The first identified caspase-cleaved intermediate filament is lamin A, which forms the nuclear lamina and plays a structural role in the nucleus (Davidson and Lammerding, 2014; Worman and Schirmer, 2015; Zhang et al., 2001). Caspase cleavage sites in lamins include VEID²³⁰ in lamin A and VEVD²³¹ in lamin B (Lazebnik et al., 1995; Rao et al., 1996). Inhibition of lamin cleavage by caspases prevents nuclear envelope disruption, nuclear and DNA fragmentation, and chromatin condensation and delays apoptosis (Rao et al., 1996; Ruchaud et al., 2002; Taylor et al., 2008; Zhang et al., 2001). Vimentin in mesenchymal cells has two caspase cleavage sites, DSVD⁸⁵ and IDVD²⁵⁹, that are substrates of caspases-3/7 and caspases-6/8, respectively (Belichenko et al., 2001; Byun et al., 2001). Caspase-dependent vimentin cleavage is involved in vimentin filament disruption and can induce apoptosis (Belichenko et al., 2001; Byun et al., 2001). Desmin, the muscle-specific intermediate filament protein, is cleaved at VEMD²⁶³ by caspase 6 in vitro and its fragmentation promotes apoptosis in myogenic cells treated with TNFα (Chen et al., 2003). However, caspase cleavage-resistant desmin mutation (D236E) inhibits the desmin fragmentation and reduces the TNFα-induced apoptosis as compared with desmin WT in cell culture (Chen et al., 2003). These results were confirmed in corresponding transgenic mouse models. For example, comparison of cardiomyocyte apoptosis in mice that express endogenous WT desmin and mice that overexpress desmin D263E in a desmin-null background has shown that expression of D263E desmin delays TNFα-induced cardiomyocyte apoptosis and protects them from heart failure (Panagopoulou et al., 2008).

K18, with its partner K8, makes up the intermediate filament cytoskeleton in adult hepatocytes, has two caspase cleavage sites, VEVD²³⁸ in the rod-domain and DALD³⁹⁷ in the C-terminal domain. K18 fragmentation occurs during apoptosis (Caulín et al., 1997; Ku et al., 1997; Leers et al., 1999), and fragmented K18 reorganizes into keratin aggregates that is released into the cell culture medium (Schutte et al., 2004). Indeed, caspase-cleaved K18 fragments are released into the blood of patients with liver diseases (Kramer et al., 2004; Yilmaz, 2009). The level of K18 fragments in blood is correlated with hepatic apoptosis, and the apoptotic K18 fragment is a useful biomarker to determine the degree of liver damage (Linder et al., 2010; Tsutsui et al., 2010).

Here, we studied the significance of K18 apoptotic cleavage under several stresses in transgenic mice that express the caspase cleavage-resistant K18 mutant K18-D238/397E in a mouse K18-null background. The phenotype of these mice clearly shows that expression of caspase cleavage-resistant K18 attenuates hepatic apoptosis and enhances the survival of the mice following liver damage induced by anti-FasAb or MLR, and indicates that an intact keratin network is crucial in order to protect cells from apoptotic stress. Our results agree with previous studies of cleavage-resistant lamins, and cleavage-resistant vimentin and desmin in cultured cells, and with a study of cleavage-resistant desmin in transgenic mice (Table 1). However, Weerasinghe et al.−by using K18-D238/397E-overexpressing transgenic mice that also express endogenous WT mouse K18 − have shown enhanced susceptibility to anti-FasAb-induced injury together with predisposition to hepatocyte necrosis (Weerasinghe et al., 2014). In our study, we also observed that the mutant transgenic mice that express endogenous mouse K18 show a slightly enhanced lethality in response to anti-FasAb-mediated liver damage as compared with control mice, but the difference was not statistically significant (Fig. 1). The phenotypic differences between the two mouse models (i.e. the heterozygous versus the homozygous effect) shown in Figs 1 and 2 are probably due to the endogenous mouse K18 in the heterozygous mutant transgenic mice attenuating the effect of the mutation. Our results clearly show enhanced protein stability of the mutant K18 compared with WT − both in vitro (cultured cells) and in vivo (transgenic livers) (Fig. 5) − that allows protection of the keratin network from liver damage (Fig. 3D). Indeed, the significance of an intact keratin network has initially been demonstrated in studies that used a mutated form of skin keratin 14 (K14-R125C) − that caused keratin filament disruption and, consequently, skin disease (Coulombe et al., 1991, 2009). Transgenic mice with the equivalent mutation of K18 (K18-R90C) have disrupted filaments and increased susceptibility to liver injury (Ku et al., 1995; Omary et al., 2009). Both cases highlight the significance of an intact keratin network in protecting cells from damage. Our finding provides evidence that the enhanced protein stability of caspase cleavage-resistant keratin 18 contributes towards maintaining the keratin network under stress.

Antibodies against keratins have been previously described (Ku et al., 2004). We also used antibodies against: Fas (anti-FasAb: BD Biosciences, San Jose, CA), phosphorylated or nonphosphorylated-protein kinases or transcription factors (Cell Signaling Technology, Beverly, MA) such as Akt, p42MAPK and p44MAPK (also known as MAPK1 and MAPK3, respectively), p90RSK (also known as RPS6KA1), p38MAPK (also known as MAPK14), JNK1 (also known as MAPK8), and JNK2 (also known as MAPK9), against cleaved caspase 3, 7 or 8 and PARP (Cell Signaling Technology, Beverly, MA), Hsc70 (Santa Cruz Biotechnology, Santa Cruz, CA), FADD (Upstate Biotechnology/Millipore, Billerica, MA), HMGB1 (Abcam, Cambridge, MA), and actin (Lab Vision/Thermo Scientific, Fremont, CA) (see Table S1 for details on antibodies). Other reagents included: MLR (used at 30 μg/kg body weight; Alexis Corporation, San Diego, CA), anisomycin (10 μg/ml; CalBiochem, La Jolla, CA) and cycloheximide (100 μg/ml; Sigma-Aldrich, St Louis, MO).

Heterozygous transgenic mice expressing human K18-WT or human K18-D238/397E in an FVB/N background (Weerasinghe et al., 2014) were crossbred with K18-null mice. By using PCR, the genome of mouse progenies was screened for the presence of human K18 gene and the absence of mouse K18 as described (Ku et al., 1998; Magin et al., 1998). K18-null mice were kindly provided by Dr Thomas Margin (University of Leipzig, Leipzig, Germany) (Magin et al., 1998) and, as they originally had been of a mixed strain background, were backcrossed ten generations to an FVB/N background. All animal experiments were performed in accordance with the Korean Food and Drug Administration (KFDA) guidelines. Protocols were approved by the Institutional Animal Care and Use Committee (IACUC) of the Yonsei Laboratory Animal Research Center (YLARC).

Mice were starved overnight and intraperitoneally injected with anti-FasAb (0.15 mg/kg body weight), or with MLR (30 μg/kg body weight) as previously described (Ku et al., 2004). The mice were assessed for three days for mortality experiments. Alternatively, mice were killed at 6 or 8 h after injection with anti-FasAb, and their livers were isolated and used for biochemical analysis, histological analysis or immunostaining. For biochemical analysis, livers were homogenized in ice-cold phosphate-buffered saline (PBS) containing 5 mM EDTA, sodium fluoride (NaF), sodium pyrophosphate (NaPP), a protease inhibitor (25 mg/ml aprotinin, 10 mM leupeptin, 10 mM pepstatin and 0.1 mM PMSF) and the phosphatase inhibitor okadaic acid. Total liver lysates and high-salt extracts were prepared by using total liver homogenates as described (Ku et al., 2004). For tissue staining, livers were either fixed with 10% formalin, embedded in paraffin and sectioned for hematoxylin-eosin staining, or embedded in optimum cutting temperature compound and sectioned for immunostaining.

Primary hepatocytes from K18-WT or K18-D238/397E mice were isolated by the liver perfusion method as described previously (Ku and Omary, 2006). Isolated hepatocytes were cultured in HepatoZYME serum-free medium (Gibco BRL Life Technologies, Grand Island, NY) with 10% fetal calf serum for 24 h and then treated with anti-FasAb (0.5 μg/ml or 1.5 μg/ml for 6−36 h) or cycloheximide (100 μg/ml). Cultured hepatocytes were used for immunostaining or preparations of total cell lysates were used for immunoblotting.

Baby hamster kidney (BHK)-21 cells (American Type Culture Collection, Rockville, MD) were cultured in medium as recommended by the supplier. Transient transfection was performed by using Lipofectamine (Invitrogen, Carlsbad, CA) according to the manufacturer's instructions. Transfected cells were treated with anisomycin (10 μg/ml) or cycloheximide (100 μg/ml).

Total lysates and high-salt extract samples were analyzed by SDS-PAGE followed by Coomassie Blue staining or immunoblotting with antibodies as indicated. Proteins were then visualized using enhanced chemiluminescence.

Statistical comparisons were performed using Fisher's exact test in StatView software. P-values of ≤0.05 were considered significant.

We thank Dr Thomas Magin for providing us with the K18-null mice and Dr Bishr Omary (University of Michigan, Ann Arbor, USA) for providing us with the heterozygous K18-WT and K18-D238/397E transgenic mice.