Critical role of acidic sphingomyelinase in murine hepatic ischemia-reperfusion injury


  • Laura Llacuna,

    1. Liver Unit, Hospital Clínic, Instituto Investigaciones Biomédicas August Pi i Sunyer (IDIBAPS); and the Department of Experimental Pathology, Instituto Investigaciones Biomédicas Barcelona, Consejo Superior de Investigaciones Científicas, Barcelona, Spain
    Search for more papers by this author
    • Laura Llacuna and Montserrat Mari have contributed equally to the work.

  • Montserrat Marí,

    1. Liver Unit, Hospital Clínic, Instituto Investigaciones Biomédicas August Pi i Sunyer (IDIBAPS); and the Department of Experimental Pathology, Instituto Investigaciones Biomédicas Barcelona, Consejo Superior de Investigaciones Científicas, Barcelona, Spain
    Search for more papers by this author
    • Laura Llacuna and Montserrat Mari have contributed equally to the work.

  • Carmen Garcia-Ruiz,

    1. Liver Unit, Hospital Clínic, Instituto Investigaciones Biomédicas August Pi i Sunyer (IDIBAPS); and the Department of Experimental Pathology, Instituto Investigaciones Biomédicas Barcelona, Consejo Superior de Investigaciones Científicas, Barcelona, Spain
    Search for more papers by this author
  • José C. Fernandez-Checa,

    Corresponding author
    1. Liver Unit, Hospital Clínic, Instituto Investigaciones Biomédicas August Pi i Sunyer (IDIBAPS); and the Department of Experimental Pathology, Instituto Investigaciones Biomédicas Barcelona, Consejo Superior de Investigaciones Científicas, Barcelona, Spain
    • Liver Unit, Hospital Clinic i Provincial c/Villarroel, 170, 08036 Barcelona, Spain
    Search for more papers by this author
    • fax: (34) 93-451-5272

    • J. C. Fernandez-Checa and A. Morales share senior authorship.

  • Albert Morales

    Corresponding author
    1. Liver Unit, Hospital Clínic, Instituto Investigaciones Biomédicas August Pi i Sunyer (IDIBAPS); and the Department of Experimental Pathology, Instituto Investigaciones Biomédicas Barcelona, Consejo Superior de Investigaciones Científicas, Barcelona, Spain
    • Liver Unit, Hospital Clinic i Provincial c/Villarroel, 170, 08036 Barcelona, Spain
    Search for more papers by this author
    • J. C. Fernandez-Checa and A. Morales share senior authorship.

  • Potential conflict of interest: Nothing to report.


The molecular mechanisms of hepatic ischemia/reperfusion (I/R) damage are incompletely understood. We investigated the role of ceramide in a murine model of warm hepatic I/R injury. This sphingolipid induces cell death and participates in tumor necrosis factor (TNF) signaling. Hepatic ceramide levels transiently increased after the reperfusion phase of the ischemic liver in mice, because of an early activation of acidic sphingomyelinase (ASMase) followed by acid ceramidase stimulation. In vivo administration of an ASMase inhibitor, imipramine, or ASMase knockdown by siRNA decreased ceramide generation during I/R, and attenuated serum ALT levels, hepatocellular necrosis, cytochrome c release, and caspase-3 activation. ASMase-induced ceramide generation activated JNK resulting in BimL phosphorylation and translocation to mitochondria, as the inhibition of ASMase by imipramine prevented these events. In contrast, blockade of ceramide catabolism by N-oleyolethanolamine (NOE), a ceramidase inhibitor, enhanced ceramide levels and potentiated I/R injury compared with vehicle-treated mice. Pentoxifylline treatment prevented TNF upregulation and ASMase activation. Furthermore, 9 of 11 mice treated with imipramine survived 7 days after total liver ischemia, compared with 4 of 12 vehicle-treated mice, whereas 8 of 8 NOE-treated mice died within 2 days of total liver ischemia. In conclusion, ceramide generated from ASMase plays a key role in I/R-induced liver damage, and its modulation may be of therapeutic relevance. (HEPATOLOGY 2006.)

Hepatic ischemia/reperfusion (I/R) damage, which can occur in diverse settings including liver transplantation, trauma, hemorrhagic shock, or liver surgery, is a serious clinical complication that may compromise liver function because of extensive hepatocellular loss. The mechanisms responsible for hepatic I/R injury are not well understood despite the identification of several intercellular and molecular mechanisms involved in the necrotic and apoptotic death of hepatocytes. For instance, recruitment and activation of inflammatory cells (CD4+ T lymphocytes), as well as Kupffer cells and platelet adhesion in the sinusoidal lining, have been involved in sinusoidal endothelial cell death and hepatic I/R injury.1–4 Molecular events include nuclear factor kappaB (NF-κB) activation, tumor necrosis factor (TNF) generation, JNK activation, mitochondrial permeability transition (MPT), and reactive oxygen species overproduction.5–8

Sphingolipids, ceramide in particular, have emerged as signaling lipid intermediates that play a role in the stress response, and as mediators of apoptosis and autophagic cell death (type II programmed cell death).9–12 Ceramide levels in cells can increase by de novo synthesis in the endoplasmic reticulum from the N-acylation and subsequent desaturation of sphinganine.13 However, early ceramide increases are often associated with sphingomyelinase (SMase) activation, which hydrolyzes sphingomyelin to yield ceramide.11 This pathway contributes to the rapid upregulation of ceramide and may promote specific microdomain formation in the plasma membrane, allowing oligomerization of certain cell surface proteins such as ligand receptors (TNF family).14 Several SMases have been characterized, of which the neutral SMase (NSMase), with an optimum pH at 7.5, Mg2+ dependent and membrane bound, and the acidic SMase (ASMase), with an optimum pH at 4.8, are of relevance in the signaling of diverse cell death stimuli.11, 15–18

Ceramide levels are tightly regulated in cells, and hence, its potential participation in death signaling can be modulated by metabolic pathways that transform ceramide in other derivatives. For instance, ceramide can be metabolized into complex glycosphingolipids by its glucosylation in the cytosolic surface of the Golgi catalyzed by the rate-limiting enzyme glucosylceramide synthase (GCS).19 In addition, ceramide can be transformed into sphingomyelin20 by sphingomyelin synthase (SMS) or deacylated by ceramidases (CDases) into sphingosine.17, 21 The phosphorylation of sphingosine by sphingosine kinase generates sphingosine-1-phosphate, a bioactive lipid with anti-apoptotic properties.17, 18, 22

Previous studies using mice deficient in TNF gene or TNF receptor 1 (TNFR1) have identified TNF as a critical mediator in warm hepatic I/R injury.5, 6, 23 Moreover, increased hepatic ceramide levels have been observed in cold ischemia and warm reperfusion,24 although its consequences and regulation were not further examined. Furthermore, the regulation of SMases has been reported in a rat model of I/R, with NSMase being stimulated during the reperfusion of the ischemic liver, as opposed to ASMase, whose activity decreased.25 However, their impact on ceramide regulation and contribution to hepatic I/R injury were not addressed. Because ceramide overgeneration by ASMase has been shown to contribute to TNF-induced hepatocellular cell death and liver injury,16, 26 the aims of this study were first to examine whether ASMase contributes to warm hepatic I/R injury, because this specific aim has not been previously reported, and second, to analyze the metabolic regulation of ceramide during I/R. We describe here a critical role for ceramide generation through ASMase activation in hepatic I/R injury, which may be of therapeutic relevance because its pharmacological inhibition or suppression by siRNA protects the liver from I/R-induced damage.


I/R, ischemia-reperfusion; NF-κB, nuclear factor-κB; TNF, tumor necrosis factor-α; MPT, mitochondrial permeability transition; NSMase, neutral sphingomyelinase; ASMase, acidic sphingomyelinase; GCS, glucosylceramide synthase; SMS, sphingomyelin synthase; ACDase, acid ceramidase; PTX, pentoxifylline; NOE, N-oleomylethanolamine; NCDase, neutral ceramidase; ALT, alanine aminotransferase; TUNEL, terminal deoxynucleotidyl transferase-mediated nick-end labeling; PCR, polymerase chain reaction; IL-12, interleukin-12; d-threo-PDMP, d-threo-1-phenyl-2-decanoylamino-3-morpholino-propanol HCl.

Materials and Methods

Reagents, Antibodies, and Cells.

See Supplementary Materials and Methods (Available at:

Partial and Total Hepatic Ischemia and Treatments.

Animal studies were approved by the IDIBAPS Animal Care and Use Committee. ASMase knockout mice (C57BL/6 strain) were propagated using heterozygous breeding pairs and genotyped as described previously.16, 26 Hepatic partial warm ischemia was performed during 90 minutes in wild-type or ASMase−/− (8–12 weeks) mice as described previously.6 Animal survival was determined using a model of total hepatic ischemia in male wild-type mice (25–30 g), in which only the ischemic tissue is left in place.7 Control animals were sham operated. See Supplementary Materials and Methods for detailed protocols and treatments.

Enzyme Activities, Ceramide, Ganglioside, and TNF Determinations.

Mg2+-dependent NSMase and ASMase activities were determined from [N-methyl-14C]sphingomyelin (56.6 mCi/mmol, Amersham Biosciences, Piscataway, NJ) as described.27 CDase activities were measured using 5 μmol/L [14C]ceramide (55 mCi/mmol, ARC, St. Louis, MO) or C12-NBD ceramide (Molecular Probes, Eugene, OR) in 10 mmol/L Tris-HCl, pH 8.0 for the neutral/alkaline CDase or 0.5 mol/L acetate buffer, pH 4.5, for the acid CDase.21, 28 The GCS and SMS were determined as described previously.29 Cellular lipids were extracted, dried under N2, and subjected to mild alkaline hydrolysis (1N methanolic KOH, 1 hour, 37°C) to remove glycerol phospholipids. Ceramide levels were quantified from liver samples by the DG kinase assay.10 Sphingolipids were separated in LK6D gel 60A TLC plates using chloroform/acetic acid (90:10). Labeled spots were scraped and quantified using a scintillation counter or a Fuji BAS-2500 PhosphoImager system. Hepatic ganglioside levels during I/R were determined by high-performance thin-layer chromatography as described previously.16 Total TNF (bound and free) levels in serum and hepatic tissue were measured using a mouse-specific competitive enzyme immunoassay kit (Chemicon, Temecula, CA).

Liver Histology and Terminal Deoxynucleotidyl Transferase-Mediated Nick-End Labeling.

See Supplementary Materials and Methods.

Western Blotting and Caspase Activation.

The release of mitochondrial proteins was monitored using cytochrome c (1:2000) and Smac/DIABLO (1:2000) antibodies30 and quantitated by scanning densitometry. JNK and AKT phosphorylation were analyzed in cell extracts using phospho-JNK (1:2500) and phospho-AKT (1:2000) antibodies. Mitochondrial fractions, obtained as previously indicated,10 were used to analyze Bim and phospho-Bim (1:1000) levels. Caspase-3 activation was monitored by immunohistochemistry with an active caspase-3 antibody (clone C92-605), whereas its activity was quantified as described.30

Real-time Polymerase Chain Reaction.

See Supplementary Materials and Methods.

In Vitro siRNA Transfection and In Vivo Treatment.

siRNA transfection into the murine hepatoma cell line H1c1c7 was accomplished with siPORTamine™ cationic solution (Ambion, Austin, TX). siRNA-mediated silencing of ASMase was verified at the enzyme level by measuring ASMase activity and at mRNA level by real-time polymerase chain reaction (PCR) as described in the supplementary Methods. For in vivo experiments, chemically synthesized siRNAs (Curevac, Tübingen, Germany) were delivered into mice using the hydrodynamic transfection method31, 32 by intravenously injecting 1 nmol scrambled or ASMase siRNAs in 1 mL phosphate-buffered saline into the tail vein 24 and 48 hours before surgery. For additional information, see Supplementary Materials and Methods.

Statistical Analyses.

Results were expressed as mean ± standard deviation with the number of individual experiments detailed in figure legends. Statistical significance of the mean values was established by the Student′s t test. Animal survival was evaluated using the Kaplan-Meier method and compared using the log-rank test.


Transient Accumulation of Ceramide by Sequential Activation of ASMase and Acid Ceramidase During I/R.

The levels of ceramide increased transiently during reperfusion compared with sham controls, with two peaks at 30 minutes and 6 hours post-reperfusion (Fig. 1A), returning to basal levels after 6 hours (not shown). Because ceramide levels reflect the balance between its generation and metabolism, we next analyzed the activity of the enzymes involved in ceramide regulation. As seen, only the activities of ASMase and acid ceramidase (ACDase) were stimulated during the reperfusion phase, without changes in NSMase, GCS or NCDase (Fig. 2A–E). The SMS activity, however, decreased between 1 and 3 hours after reperfusion (Fig. 2F). ASMase exhibited two peaks of activity, an early one by 10 to 30 minutes and a later peak by 6 hours after reperfusion (Fig. 2A). ACDase, however, displayed an activation window between 1 and 6 hours post-reperfusion (Fig. 2D). Because TNF activates ASMase,11, 16 we next determined the kinetics of serum TNF release during I/R (Fig. 1B). In agreement with previous findings,5 we observed an early increase by 10 to 15 minutes post-reperfusion that lasted for up to 8 hours (Fig. 1B). Similarly, TNF levels were enhanced in liver samples post-reperfusion (not shown). Moreover, pentoxifylline (PTX), a well-established inhibitor of TNF production in vivo, which prevented I/R-induced hepatic damage,6 suppressed the ASMase activation during I/R (Fig. 1C) and the subsequent increase in ceramide levels (not shown), indicating that TNF release during I/R mediates the activation of ASMase.

Figure 1.

Ceramide regulation during I/R. (A) Mice were subjected to I/R and the hepatic content of ceramide was determined at different times post-reperfusion as indicated. The levels of ceramide in control (sham) mice were 171 ± 18 pmol/mg protein (n = 3). *P ≤ .043 versus control. (B) TNF levels were determined in serum at different times post-reperfusion (n = 3). *P ≤ 0.012 versus sham control. (C) ASMase activity was determined in liver extracts 15 minutes post-reperfusion from vehicle- or pentoxifylline (PTX)-pretreated mice. Control mice were sham operated. (n = 3). *P = .017 versus control. #P = .033 versus untreated I/R. (D) Ceramide levels were quantified in liver 30 minutes post-reperfusion. In some cases, mice were pretreated with ASMase inhibitor (imipramine, 25 mg/kg), ACDase inhibitor (NOE, 100 mg/kg) or vehicle (saline) 30 minutes before surgery. Control mice were sham operated. (n = 3). *P ≤ .011 versus control. #P ≤ .045 versus untreated I/R.

Figure 2.

Activation of ASMase and ACDase during I/R. Liver samples were taken at different times after I/R and homogenized to measure the enzymatic activity of ASMase (A), NSMase (B), GCS (C), ACDase (D), NCDase (E), and SMS (F) (n = 3–4). *P ≤ .009 versus control. #P ≤ .006 versus control. &P ≤ .010 versus control.

Pharmacological Regulation of Ceramide Modulates Hepatic I/R Injury.

To establish the relationship between ASMase/ACDase activation and hepatic ceramide accumulation, we tested the effect of inhibition of both enzymes in ceramide homeostasis during I/R. Imipramine, which prevents ASMase activation,16, 33 and N-oleoylethanolamine (NOE), a potent ACDase inhibitor,21, 34 were administered 30 minutes before surgery. Imipramine, which prevented the I/R-induced ASMase activation, decreased the peak of ceramide levels generated 30 minutes after reperfusion (Fig. 1D). As expected, NOE enhanced the accumulation of ceramide with respect to the group subjected to I/R (Fig. 1D). Neither imipramine nor NOE modulated hepatic ceramide levels of sham-operated mice. Furthermore, we examined the impact of ceramide regulation on hepatic I/R injury. As indicated, I/R caused hepatic injury reflected by serum alanine aminotransferase (ALT) levels and liver histology (Fig. 3A–B). In addition, we observed significant terminal deoxynucleotidyl transferase-mediated nick-end labeling (TUNEL)-positive hepatocytes after I/R (Fig. 3C), consistent with the coexistence of oncotic necrosis and apoptotic cell death.7, 8 Importantly, imipramine pretreatment protected the liver against I/R injury, as seen by ALT levels, histology, and TUNEL staining, whereas NOE enhanced the extent of hepatic injury (Fig. 3A–C). Imipramine pretreatment, however, did not affect the phosphorylation of Akt or the level of nuclear p65 measured in hepatic extracts after I/R (Fig. 3D), indicating that the protective effect of imipramine is unrelated to NF-κB signaling.

Figure 3.

Modulation of hepatic I/R damage by ceramide generation. (A) Serum ALT levels were measured 6 hours post-reperfusion from sham-operated or I/R–exposed mice pretreated with imipramine (IMIP), NOE, or vehicle (n = 4). *P = .022 and #P = .031 versus untreated mice subjected to I/R. (B-C) Representative HE and terminal deoxynucleotidyl transferase-mediated nick-end labeling (TUNEL) images of liver sections 6 hours post-reperfusion (magnification 10×) from mice treated as above. (D) A representative Western blot from four independent experiments showing AKT phosphorylation in cell fractions and p65 levels in nuclear extracts, after 30 and 60 minutes post-reperfusion.

ASMase Down-Regulation by siRNA Protects the Liver Against I/R-Induced Damage.

To further verify the role of ASMase-induced ceramide generation in hepatic I/R damage and discard potential side effects of imipramine, we used two approaches. First we performed hepatic I/R in ASMase−/− mice. However, most (80%–90%) ASMase−/− mice died shortly after anesthesia during ischemia (not shown), because of severe pulmonary dysfunction as characterized previously.35 Therefore, we used an alternate loss-of-function approach using RNA interference, which has been used successfully in silencing gene expression in the liver.31, 32 siRNA efficacy to deplete ASMase mRNA levels was verified in Hepa1c1c7 cells by real time PCR (Fig. 4A), and was accompanied by the reduction of the endogenous ASMase activity (Fig. 4B), whereas the scrambled siRNA did not display any effect. In addition, ASMase-siRNA did not affect the endogenous activities of NSMase or GCS (Fig. 4B), or those of ACDase, NCDase, or SMS (not shown), leading us to examine its effects in vivo.

Figure 4.

Suppression of ASMase activity by siRNA reduces I/R-induced liver injury. Murine hepatoma H1c1c7 cells were transfected with ASMase-siRNAs (closed bars) or a scrambled control siRNA (Sc siRNA, open bars). (A) ASMase mRNA levels were measured after 36 hours (n = 4). *P = .001 versus Sc siRNA-transfected cells. (B) ASMase, NSMase, and GCS activities were assayed in cell extracts (n = 4). *P = .001 versus Sc siRNA-transfected cells. (C) IL-12 serum levels of mice injected with saline, Poly I:C, scrambled siRNA or ASMase-siRNA 24 hours before surgery (n = 3). *P = .005 versus saline-injected animals. (D) Specific suppression of ASMase mRNA in vivo measured by real-time PCR after two intravenous injections with ASMase-siRNA or scrambled siRNA 24 hours before surgery. Results are expressed as percentage of change of the ASMase/GAPDH mRNA ratio in sham-operated animals. (n = 4). *P ≤ .012 versus Sc siRNA-injected animals. Inset shows activity in the same group of mice. (E) Ceramide levels 30 minutes post-reperfusion from liver samples of mice treated with ASMase-siRNA or scrambled siRNA 24 hours before surgery (n = 3). *P = .045 versus Sc siRNA-injected animals. (F) Serum ALT levels 6 hours post-reperfusion from mice treated with ASMase siRNA or scrambled siRNA 24 hours before surgery (n = 4). *P = .042 versus Sc siRNA-injected I/R-exposed animals.

Double-stranded RNA duplexes may elicit immune responses via interactions with Toll-like receptors 3 and 9,36 and hence we investigated this possibility by examining the involvement of immnune responses in the potential effects of ASMase-siRNA on hepatic I/R in vivo. Interleukin-12 (IL-12) levels were measured in mice 2 hours after injection of ASMase-siRNA, scrambled siRNA, or saline with respect to mice treated with Poly I:C, an analog of dsRNA that is recognized by TLR3 and elicits immune responses in mice.36 As seen, in contrast to the effect observed after Poly I:C injection, no significant increase in IL-12 serum levels were detected in mice treated with ASMase-siRNA (Fig. 4C), consistent with recent reports using naked siRNAs.37 Next, mice were injected into the tail 2 consecutive days with either ASMase-siRNA or scrambled siRNA and then subjected to I/R the day after the last dose. ASMase-siRNA decreased hepatic ASMase mRNA levels (Fig. 4D) and inhibited ASMase activity translating in a reduction in the levels of ceramide after reperfusion (Fig. 4E). Importantly, ASMase-siRNA reduced the extent of I/R-induced hepatic damage monitored by serum ALT levels (Fig. 4F), histology, or TUNEL (not shown). These data establish a key role of ASMase in hepatic I/R injury.

Ceramide Regulation During I/R Modulates the Mitochondrial Cell Death Pathway.

Because ceramide is known to play a role in cell death pathways through mitochondria,10, 11 we next examined the mitochondrial release of cytochrome c and Smac/DIABLO during I/R. As seen, I/R caused the release of cytochrome c and Smac/DIABLO after I/R (Fig. 5A). This outcome was accompanied by active caspase 3 detected by fluorescence assay or immunohistochemistry (Fig. 5B–C). Of note, the number of positive cells observed were less than those stained with TUNEL (Fig. 3C). Next, we examined whether the regulatory role of ceramide on hepatic I/R damage was accompanied by appropriate changes in the apoptosome assembly. Although imipramine treatment reduced cytochrome c and Smac/DIABLO release from mitochondria by I/R, NOE enhanced the release of these apoptogenic proteins from mitochondria (Fig. 5A). In addition, caspase 3 activation induced by I/R was reduced or enhanced by imipramine or NOE pretreatment, respectively (Fig. 5B–C).

Figure 5.

Ceramide modulates mitochondrial cytochrome c release and caspase 3 activation during I/R. (A) Representative Western blot showing the presence of cytochrome c and Smac/DIABLO in cytosol 4 hours post-reperfusion with imipramine (IMIP), NOE, or vehicle treatment before surgery. Results are from three independent experiments expressed as percentages with respect to sham-operated animals. *P ≤ .031 versus sham-operated control animals. #P ≤ .048 versus untreated mice subjected to I/R. (B) Caspase 3 activity in liver extracts 6 hours post-reperfusion from mice pretreated with imipramine (IMIP), NOE, or vehicle was measured fluorometrically (ex 380 nm, em 460 nm) (n = 4). *P ≤ .035 versus sham-operated animals. #P ≤ .048 versus vehicle-treated mice subjected to I/R. (C) Representative immunohistochemical staining of active caspase 3 fragment in liver sections 6 hours after reperfusion (magnification 40×) from mice treated with imipramine (IMIP), NOE, or vehicle before surgery.

JNK Activation by I/R Is Suppressed by ASMase Inhibition.

To identify the potential downstream intermediates of ASMase-mediated hepatic I/R injury, we focus on complex glycosphingolipids and gangliosides because in cultured hepatocytes ganglioside GD3 mediates TNF-induced death.16 Gangliosides are synthesized from ceramide on its glucosylation by GCS, which is inhibited by d-threo-1-phenyl-2-decanoylamino-3-morpholino-propanol HCl (PDMP).16, 19 Unexpectedly, pretreatment with PDMP 30 minutes before surgery did not protect the liver against I/R injury (Fig. 6A), consistent with the lack of GCS activation during I/R (Fig. 2C). Moreover, no changes in the pattern of gangliosides were observed during I/R (not shown), thus further discarding a role for gangliosides in the hepatic I/R injury.

Figure 6.

Inhibition of JNK activation by ASMase-induced ceramide but not GCS protects against hepatic I/R injury. (A) Serum ALT levels 6 hours post-reperfusion from mice pretreated with PDMP (75 mg/kg) or vehicle (5% Tween 20 in saline) before surgery (n = 4). (B) Representative Western blot of JNK phosphorylation in liver extracts 30 minutes post-reperfusion. In some cases, mice were pretreated with SP600125 (SP) or imipramine (IMIP) 30 minutes before ischemia. (C) Serum ALT levels 6 hours post-reperfusion from mice pretreated with JNK inhibitor (SP600125, 20 mg/kg) or vehicle (1% dimethylsulfoxide in saline) before surgery (n = 5). *P = .024 versus untreated I/R-injured animals. (D) 1 and 6 hours after I/R liver samples from mice pretreated with vehicle, IMIP or SP600125 were processed for mitochondria isolation to examine the levels of Bim, phospho-Bim, and cytochrome c oxidase by Western blot. Representative blot of three to four independent experiments is shown.

JNK inhibition has been reported to protect the liver against I/R-induced apoptotic/necrotic cell death,7 and the apoptotic role of ceramide is exerted through JNK activation.38–40 Therefore, we evaluated whether the role of ceramide in hepatic I/R damage is mediated by JNK activation. I/R resulted in the phosphorylation of JNK (p46 and p54 isoforms, Fig. 6B), and this was accompanied by phosphorylation of c-Jun (not shown). JNK activation contributed to the hepatic damage induced by I/R since SP600125, a potent and specific JNK inhibitor,41 protected against I/R-induced hepatic damage as reflected by serum ALT levels (Fig. 6C) or histology (not shown). Importantly, imipramine pretreatment, which decreased ceramide generation and protected the liver against I/R (Figs. 1D and 3A), prevented JNK phosphorylation caused by I/R (Fig. 6B). SP600125, however, did not prevent the early increase in ceramide levels during the reperfusion phase, or ASMase activation (not shown), indicating that JNK activation is downstream of ASMase stimulation during I/R. In addition, because ceramide-induced phosphorylation and translocation of Bim to mitochondria are mediated by JNK,40 we analyzed its phosphorylation during I/R. As seen, BimL and its phosphorylated form both translocated to the mitochondria and, more importantly, imipramine and SP600125 suppressed these events (Fig. 6D).

Ceramide Regulation Modulates Survival After Total Hepatic I/R.

We next examined the survival rate in a model of total hepatic I/R in which survival depends on recovery of the ischemic liver after reperfusion. Mice (n = 8–12 per group) were subjected to 60 minutes of ischemia instead of the 90-minute period used in the partial ischemic experiments because this protocol killed all mice within 24 hours of total hepatic I/R. Under these conditions, 8 of 12 vehicle-treated mice subjected to total ischemia died within the first 72 hours post-reperfusion (Fig. 7A), with histological evaluation on autopsies showing massive necrosis in the liver (not shown), consistent with previous reports.7, 8 However, imipramine pretreatment markedly improved the survival rate with 9 of 11 treated mice surviving after 7 days post-reperfusion (Fig. 7A). In contrast, all NOE-treated mice (8 of 8) died within 48 hours postischemia, with 75% of animals dying within 24 hours of reperfusion (Fig. 7A). Thus, these data demonstrate the relevance of the pharmacological regulation of ceramide in hepatic I/R injury.

Figure 7.

Ceramide regulation modulates survival after total hepatic ischemia. (A) Mice pretreated with vehicle (saline), imipramine (IMIP) (25 mg/kg), or NOE (100 mg/kg) before surgery were subjected to total liver ischemia for 60 minutes. Animals were monitored twice per day during 7 days. *P = .035 and #P = .013 versus saline-treated I/R mice by log rank test, n = 8–12 in each group. (B) Schematic diagram summarizing the findings. The generation of ceramide during I/R targets mitochondria through JNK activation and subsequent translocation of BimL to mitochondria, culminating in hepatocellular death. PTX or IMIP, which prevented ASMase activation, abolished JNK activation, and attenuated mitochondrial-dependent hepatocellular death. The dashed line denotes that although ceramide generated from ASMase activation may directly target mitochondria, this possibility remains to be tested.


This study examines the role of ASMase in the homeostasis of ceramide and its contribution to hepatic I/R injury. Here we describe the transient generation of ceramide during I/R due to the sequential activation of ASMase followed by ACDase stimulation. Because ASMase generates ceramide from sphingomyelin hydrolysis whereas ACDase catabolyzes it, the time-dependent pattern in the activation of these ceramide-regulating enzymes accounts for the temporal window in which ceramide peaked during I/R. These observations raise several questions, namely, how these enzymes become stimulated in the liver during I/R and what would be the consequences of their activation for hepatic function in I/R?

TNF Release During I/R Activates ASMase.

TNF upregulation by activated Kupffer cells plays a key role in hepatic I/R injury,5, 6, 23 and TNF-induced hepatocellular death is mediated in part through ASMase activation.11, 16 During I/R we detected an early release of TNF and postulated that this cytokine may contribute to I/R-induced ASMase activation. PTX, a methylxantine inhibitor of TNF production in vivo,6, 42 prevented the activation of ASMase and the consequent increase in ceramide levels observed during I/R. The mechanism for ACDase stimulation by I/R is currently unknown. However, it has been reported that the overexpression of ACDase in human skin fibroblasts led to ASMase stimulation and that ASMase could be co-precipitated with ACDase using anti-ceramidase antibodies,43 suggesting the physical interaction between these acid pH optimum enzymes. Whether this outcome occurs in I/R remains to be established. Moreover, we did not address the role of the ACDase activation during I/R in modulating hepatic injury, although it can be expected that through ceramide deacylation, ACDase would generate sphingosine, which may be further metabolized to sphingosine 1-phosphate, a bioactive lipid with anti-apoptotic properties.17, 18, 22 In line with this, it has been shown that ACDase overexpression prevented TNF-induced death in L929 cells,21 whereas NCDase transfection protected rat hepatocytes from TNF-induced cell death through ASMase activation.44

Ceramide Generation Through ASMase Contributes to Hepatocellular Death.

The next question we addressed was whether ceramide signals hepatocellular death and whether hepatic I/R injury would be modulated by the generation of ceramide. Tissue damage during I/R can proceed through adenosine triphosphate–dependent necrosis or caspase-dependent apoptosis, two forms of cell death that are regulated by MPT.8, 45 The release of cytochrome c from mitochondria is a key event in caspase 3 activation through the apoptosome assembly. Although the release of cytochrome c in I/R we observed may seem modest, it closely correlated with the degree of caspase 3 activation. Even in the case of NOE pre-treatment in which the degree of liver injury estimated by ALT release and histology is rather substantial compared with vehicle-treated mice, the level of cytochrome c released into the cytosol is enhanced from twofold to approximately 3.5-fold. Although this increment is reflected in the activation of caspase 3, it suggests that the release of cytochrome c from mitochondria is highly regulated. Cytochrome c is anchored to the mitochondrial membrane through both electrostatic and hydrophobic interactions with acidic phospholipids, most notably cardiolipin, and recent observations indicated that the releasable pool of cytochrome c is determined by its interaction with cardiolipin, which undergoes a cytochrome c–mediated peroxidation.46 Whether cytochrome c release correlates with the loss of cardiolipin in hepatic I/R injury remains to be established. Accordingly, the contribution of apoptosis to hepatic I/R injury appears to be relatively minor as it normally occurs in isolated hepatocytes. In contrast, necrosis affects extensive areas of parenchymal cells consistent with the massive release of ALT in serum that reflects plasma membrane leakage, indicating that necrosis is the predominant form of liver injury during I/R.8 In line with this, recent studies in cyclophilin D null mice revealed a critical role of MPT in necrotic cell death in heart, liver, and brain.45

To examine the impact of ceramide in hepatic I/R injury, we inhibited ASMase and ACDase before surgery. Imipramine, a tricyclic antidepressant that inhibits the activity of ASMase due to the hydrolysis of the active ASMase form,16, 33 prevented the generation of ceramide post-reperfusion and decreased hepatocellular death. Although, unfortunately, the use of ASMase−/− mice was precluded because of pulmonary dysfunction and sensitivity to ischemia due to accumulation of sphingomyelin in their lungs,35 the suppression of ASMase in vivo by siRNA reduced the activation of ASMase, diminished ceramide generation, and protected the liver from I/R-induced damage, thus verifying the results observed with imipramine. Moreover, PTX, which has been reported to protect the liver against I/R injury,6 prevented the activation of ASMase and the consequent ceramide generation by I/R (Fig. 1C). In contrast, the inhibition of ACDase resulted in enhanced ceramide content and potentiation of the I/R-induced hepatic damage. Of potential clinical significance is the fact that in vivo pretreatment with imipramine extended the survival of mice subjected to total liver ischemia.

Mitochondrial Targeting by ASMase-Induced Ceramide.

Earlier studies in isolated mitochondria indicated that ceramide induced MPT and reactive oxygen species,10, 16 thus implying that the ceramide generation during I/R elicits MPT, with the phenotype of hepatocyte demise modulated by the extent of adenosine triphosphate depletion. The question that arises then is how ceramide targets mitochondria and regulates hepatocyte cell death during I/R. We discard the involvement of glucosylceramide and glycosphingolipids (such as GD3) generation in hepatic I/R injury. Consistent with these observations, I/R did not stimulate GCS, nor increased glucosylceramide or GD3 levels, and PDMP, a GCS inhibitor, failed to protect against I/R. These unexpected findings indicate that although the TNF-ASMase arm occurs in both isolated hepatocytes and hepatic I/R, the ASMase–GCS connection does not seem to occur in the latter. Whether this relates to differences in TNF dosage, signaling, or other factors requires further investigation.

We considered the possibility that ceramide targets mitochondria through activation of downstream effectors such as JNK.38, 39 Our data indicate first that I/R induced JNK phosphorylation, confirming previous findings,7 and second that imipramine prevented JNK activation. Thus, the contribution of JNK to hepatic I/R damage may be mediated through ASMase-induced ceramide formation (Fig. 7B). In agreement with these findings, the potent activation of JNK (eightfold) by exogenous ASMase47 has been reported in primary hepatocytes. Furthermore, the interaction of JNK with mitochondria has been reported during cell death,48, 49 opening the possibility that the role of JNK in hepatic I/R damage may be due to its interaction with mitochondria. However, the cytotoxic role of JNK may be attributable not only to its putative interaction with mitochondria but also to the activation of downstream targets, including BH3-only members of the Bcl-2 family such as Bim, which can be regulated by transcriptional and post-translational mechanisms.40, 50 Indeed, we have observed both the upregulation and phosphorylation of BimL, a Bim splice variant normally associated with microtubules through its interaction with the dynein light chain 1 that prevents its interaction with other Bcl-2 family members.50 The phosphorylation of BimL at threonine56 by JNK unleashes BimL from microtubules, allowing its redistribution and binding to Bcl-2.50 Consistent with this mechanism, we have observed the phosphorylation of BimL and its translocation to mitochondria during I/R, in agreement with previous findings in lung cancer cells,40 and this step was prevented by JNK inhibition (SP600125) or ASMase inhibition (imipramine) (Fig. 6D). Moreover, Bak has been involved in hepatic I/R injury in a JNK-dependent fashion,7 and hence we do not discard the participation of members of the Bcl-2 family of proteins. Thus, overall our findings indicate that the modulation of ceramide generation by ASMase during I/R regulates the mitochondrial pathway of hepatocellular death through JNK activation and subsequent mitochondrial translocation of BimL, suggesting that the modulation of ceramide may be a novel therapeutic approach to prevent post-ischemic liver injury.


The authors thank Susana Núñez for expert technical assistance. M. Marí, and A. Morales are Ramón y Cajal investigators.