Sphingolipid signalling and liver diseases

Authors

  • Montserrat Marí,

    1. Liver Unit and Centro de Investigaciones Biomédicas Esther Koplowitz, IMDiM, Hospital Clinic i Provincial, CIBER-HEPAD, Instituto Salud Carlos III, IDIBAPS, Barcelona, Spain
    2. Department of Cell Death and Proliferation, Instituto Investigaciones Biomédicas de Barcelona, Consejo Superior de Investigaciones Científicas, Barcelona, Spain
    Search for more papers by this author
  • José C. Fernández-Checa

    1. Liver Unit and Centro de Investigaciones Biomédicas Esther Koplowitz, IMDiM, Hospital Clinic i Provincial, CIBER-HEPAD, Instituto Salud Carlos III, IDIBAPS, Barcelona, Spain
    2. Department of Cell Death and Proliferation, Instituto Investigaciones Biomédicas de Barcelona, Consejo Superior de Investigaciones Científicas, Barcelona, Spain
    Search for more papers by this author

Correspondence
Dr. José C. Fernández-Checa, Liver Unit, Hospital Clinic, C/Villarroel, 170, 08036 Barcelona, Spain
e-mail: checa229@yahoo.com

Abstract

Sphingolipids (SLs) comprise a class of lipids with important structural functions and increasing relevance in cellular signalling. In particular, ceramide has attracted considerable attention owing to its role as a second messenger modulating several cell functions such as proliferation, gene expression, differentiation, cell cycle arrest and cell death. Increasing evidence documents the role of SLs in stress and death ligand-induced hepatocellular death, which contributes to the progression of several liver diseases including steatohepatitis, ischaemia-reperfusion liver injury or hepatocarcinogenesis. Furthermore, recent data indicate that the accumulation of SLs in specific cell subcompartments, characteristic of many sphingolipidoses, contributes to the hepatic dysfunctions that accompany these inherited diseases. Hence, the regulation of the cell biology and metabolism of SLs may open up a novel therapeutic avenue in the treatment of liver diseases.

The liver is a multifunctional organ that plays essential roles in metabolism, biosynthesis, secretion and detoxification of xenobiotics, due in part to its multicellular nature. Because parenchymal cells constitute the bulk of the liver, hepatocellular loss has a profound toll in liver dysfunction and disease, and hence the identification of hepatotoxins that contribute to or mediate hepatocellular death may be of clinical relevance in liver pathophysiology. Although sphingolipids (SLs) have been considered mere structural components of biological membranes, recent evidence has indicated a dynamic role of this family of lipids in the regulation of many different processes including gene expression, proliferation, cell growth and cell death (1–4). Among SLs, ceramide has attracted considerable attention owing to its recognized role as key intermediate of many inducing stimuli, particularly stress and death ligands-induced cell death. Death receptors are transmembrane cytokine receptors that belong to the tumour necrosis factor (TNF)/nerve growth factor superfamily and include TNF, FAS ligand and TNF-related apoptosis-inducing ligand (TRAIL). Among death receptors, TNF is of great relevance to liver pathology, as its expression and levels increase in many forms of liver diseases such as alcoholic liver disease, fulminant hepatitis, viral hepatitis and steatohepatitis, and hence regulation of the hepatocellular death caused by TNF may be of clinical relevance (5, 6).

TNF is a multifunctional pro-inflammatory cytokine produced mainly by activated macrophages and in smaller amounts by other cell types, including hepatocytes. TNF in the liver exerts a variety of (patho)physiological effects including cell proliferation, growth, inflammation, liver regeneration and cell death, and this diversity of functions is accounted for by a sophisticated and versatile signal transduction pathway. This signalling network begins with the binding of TNF to two different membrane receptors, TNF-receptor 1 and 2 (TNF-R1 and TNF-R2). Although TNF may potently activate both pro-inflammatory and pro-apoptotic pathways, these signalling pathways interact in a complex network at several levels, and activation of one pathway often depends on the inactivation of the other. Whereas TNF-R1 is efficiently activated by soluble TNF, TNF-R2 activation requires the binding of membrane-bound TNF (7), which adds another level of regulation in the TNF signalling. After TNF binding, TNF receptors undergo a conformational change allowing them to recruit adapter molecules, which then initiate the activation of intracellular signalling pathways (5, 6). Most of the pro-apoptotic function of TNF is exerted through TNF-R1. Several domains within the TNF-R1 cytoplasmic tail are responsible for the recruitment of distinct adaptor proteins and activation of specific enzymes, including sphingomyelinases (SMases), which hydrolyse sphingomyelin (SM) from membranes to generate ceramide (3, 4). Recent evidence has positioned this SL as a key mediator of TNF-induced hepatocellular death (8, 9). The hydrolysis of SM upon activation of SMases constitutes an effective means for the rapid generation of ceramide, which has been recognized to contribute to cell death induction in response to different stimuli other than TNF signalling (1–4, 10).

In addition to the prominent recognized role of ceramide in apoptosis signalling, other members of this family of lipids, particularly sphingosine 1-phosphate (S1P), signal survival pathways and hence can antagonize the cytotoxic effects of ceramide. Therefore, the balance between ceramide/S1P modulates apoptosis pathways, and thus strategies targeting the metabolic regulation of these lipids may offer potential relevance in the treatment of liver diseases. In this review, we will summarize the role of ceramide in TNF-mediated hepatocellular death, which may be of paramount relevance in liver pathophysiology owing to the importance of TNF in liver diseases. In addition, we will bring up to date recent data on the current knowledge of sphingolipidoses, inherited diseases characterized by accumulation of SLs, which also results in liver disease.

SL metabolism

Ceramide is the prototypic SL that has been most intensively studied in relation to cell death induction and in stress response. Cellular ceramide levels can increase by several means. Ceramide synthesis by the de novo pathway (Fig. 1) is widespread among cell types and tissues. Ceramide de novo synthesis occurs in the endoplasmic reticulum (ER) with the condensation of l-serine and palmitoyl-CoA to form 3-ketosphinganine, catalysed by the pyridoxal phosphate-dependent enzyme serine palmitoyl transferase (SPT), the rate-limiting enzyme in ceramide biosynthesis (11). 3-ketosphinganine is then reduced to the sphingoid base sphinganine and acylated to generate dihydroceramide by the (dihydro)ceramide synthase. Dihydroceramide is then oxidized to ceramide catalysed by dihydroceramide desaturase, which introduces a trans-4,5 double bond. The de novo ceramide synthesis is necessary for many cellular functions as inferred from the detrimental consequences of its inhibition. Fumonisins, mycotoxin contaminants of maize, inhibit (dihydro)ceramide synthase and are known to cause cancer, leucoencephalomalacia, pulmonary edema, and liver and kidney toxicity (11, 12). By inhibiting (dihydro)ceramide synthase, fumonisins cause the accumulation of sphinganine in tissues, serum and urine, which is widely used as a biomarker of fumonisin exposure. The accumulation of sphinganine appears to be responsible for many of the deleterious effects of these mycotoxins, although depletion of complex SLs needed for optimal membrane functions, such as gangliosides, may contribute to the manifestations of toxicity.

Figure 1.

 Sphingolipid metabolism.

In addition to the de novo synthesis through activation of SPT or ceramide synthetase, ceramide can arise from hydrolysis of SM by SMases (3, 4, 10). This pathway may be of significance in promoting specific macrodomain formation in the plasma membrane, allowing oligomerisation of certain cell surface proteins such as ligated receptors (Fas) (13, 14). Several SMases have been characterized of which two types are of relevance in signalling. The membrane-bound neutral SMases (NSMase) exhibit an optimum pH of 7.4 and have been characterized at the molecular level, with three genes cloned so far (15–18). On the other hand, acidic SMases (ASMases) exhibit a pH optimum of around 4.8, of which several isoforms have been described (19, 20). Compared with the de novo synthesis, the generation of ceramide through SMase activation is quick and transient, and has been involved in the apoptosis induced by apoptotic stimuli, such as death ligands (e.g. Fas and TNF), chemotherapeutic agents or ionizing radiation (4). Although the precise cellular location at which the activated NSMase and ASMase act hydrolysing SM is not completely understood, these SMases account for the ability of the inducing stimuli to generate ceramide with different kinetics, and possibly at distinct intracellular sites. In line with this, the domains within the intracytoplasmic region of the death ligand receptor responsible for the activation of NSMase and ASMase are distinct (1, 3, 10).

Ceramide also provides the carbon source for glycosphingolipids (GSLs) synthesis and gangliosides in the Golgi network (Fig. 1), in a process coupled to the exocytotic vesicle flow to the plasma membrane, one of the predominant destinations of GSLs in cells. Gangliosides are a subfamily of GSLs that are distinguished from SLs by the presence of carbohydrates and sialic acid residues in the carbon backbone of ceramide. This group of SLs arises by the action of specialized glycosyltransferases that transfer a glucose or galactose residue in an α-glycosidic linkage to the C1-hydroxyl of ceramide to produce glucosylceramide (GluCer) or galactosylceramide (21). GSLs and gangliosides have been implicated in fundamental cell processes such as growth, differentiation, adhesion and cell signalling (22).

In addition to the generation of ceramide by de novo synthesis or SM hydrolysis, a novel pathway of ceramide generation entails the processing of GLSs at the plasma membrane (23). The detachment of sugar units from GM3 at the plasma membrane of fibroblasts results in the formation of ceramide that correlated with the expression of the plasma membrane ganglioside sialidase Neu3. Furthermore, recent evidence has disclosed the ability of mitochondria to generate ceramide through the activity of ceramide synthase and a reverse ceramidase (24, 25), although the contribution of these mitochondrial pathways of ceramide formation in apoptotic stimuli remains to be further characterized. These studies demonstrating the involvement of plasma membrane/mitochondria in the metabolism of ceramide along with its almost negligible solubility support the hypothesis that the topology of ceramide formation could determine its functions. Nevertheless, despite the hydrophophic nature of ceramide that limits its diffusion, ceramide can traffic among cell subcompartments by vesicle-dependent and independent mechanisms that contribute to the final destination of this lipid to modulate cell-death pathways (4).

The conversion of ceramide into several other metabolites modulates its potential apoptotic function (Fig. 1). The deacylation of ceramide by ceramidases (CDases) yields sphingosine, which may be phosphorylated by sphingosine kinase (SK) to S1P, another SL with an important role in cell survival (1, 2, 4). Moreover, ceramide may be phosphorylated by ceramide kinase (CK) generating ceramide 1-phosphate (C1P), which in turn may be dephosphorylated back to ceramide by the C1P phosphatase. Similar to S1P, C1P has mitogenic properties and promotes cell survival (26), and hence the balance between the generation of ceramide and its conversion into S1P/C1P determines the fate of hepatocytes in response to TNF and stress.

SM-ceramide pathway and hepatocellular death

In contrast to glycerophospholipids, which have been long known to play an important role in lipid-mediated signal transduction, membrane SLs such as SM, ceramide and gangliosides have been classically regarded as stable structural components of the membrane and hence of little relevance in cell signalling. However, accumulating evidence from the last 15 years has indicated that SLs are more than just structural components of biological membranes. Indeed, many different stimuli such as death ligands (e.g. TNF, Fas or TRAIL), bacterial and viral infections, bile acids, chemotherapy or ionizing radiation generate or stimulate the upregulation of ceramide, either through de novo synthesis in the ER or by the activation of SMases leading to SM breakdown (4). Unlike the former process, which contributes to the slow but sustained generation of ceramide, the latter pathway provides a fast strategy for the generation of ceramide within minutes to hours, mediating diverse cell responses including cell-cycle arrest, differentiation, gene expression or cell death. This pathway is known as the ‘SM-ceramide pathway’, and in addition to the quick generation of ceramide it also results in the formation of other derivatives that modulate the biological activities of ceramide and hence regulate cell fate (1, 2, 27).

Hepatocytes represent the major cell population in the liver, and their susceptibility to apoptotic stimuli, particularly to TNF, contributes to the progression of several liver diseases, and therefore the understanding of the mechanisms underlying the hepatocellular susceptibility to TNF would allow the design of novel therapeutic approaches (5, 6). Because enhanced hepatocellular death by stress or TNF determines the progression of steatohepatitis (alcoholic and non-alcoholic) or ischaemia-reperfusion injury among other liver diseases, the goal of intervention will be to reduce or prevent the hepatocellular susceptibility to stress and TNF, whereas in the case of hepatocellular carcinoma, the final outcome will be to design strategies that kill hepatoma cells or that enhance their susceptibility to current therapy. Recent evidence indicates that the modulation of ceramide fits into these two extremes of cell-death regulation.

SMases and hepatocellular susceptibility to TNF

The TNF signalling is a complex process involving protein–protein interactions and second messengers that ultimately results in the simultaneous stimulation of survival and death signals, the balance of which determines the fate of hepatocytes to TNF (5, 6). A central pathway that controls the susceptibility of hepatocytes to TNF is the activation of NF-κB, a transcription factor that induces a cadre of genes that orchestrate the cellular survival by antagonizing the death signals induced by TNF at various levels including the neutralization of caspases, sustained JNK activation, or downregulation of reactive oxygen species (ROS) from mitochondria (5, 6). In examining the hepatocellular susceptibility to TNF, we observed that hepatocytes lacking ASMase were resistant to TNF-mediated apoptosis/necrosis and to fulminant liver failure in vivo caused by d-galactosamine/lipopolysaccharide (LPS) challenge (8, 9). Detailed analyses of transducing intermediates indicated that the signalling events upstream of mitochondria including NF-κB activation and mitochondrial Bax translocation were preserved in ASMase-deficient hepatocytes, yet the generation of mitochondrial ROS was defective in ASMase−/− hepatocytes sensitized to TNF/Fas through mitochondrial GSH depletion (28), suggesting that the onset of local oxidative changes within mitochondria are necessary for TNF-induced hepatocellular death. Although these observations do not support a role for NSMase in TNF-mediated hepatocellular death, previous studies have shown that FAN, an adapter factor associated with NSMase activation, is involved in TNF-induced apoptosis in fibroblasts and in LPS-induced lethality in d-galactosamine sensitized mice (29, 30). However, whether the lack of FAN, in addition to the modulation of NSMase, regulates the production of other cytokines such as IL-8 or granulocyte/macrophage colony-stimulating factor secreted in response to TNF that may affect the course of liver damage remains to be established. Therefore, ASMase contributes to TNF-mediated hepatocellular apoptosis through a dual mechanism involving the targeting of ganglioside GD3 to mitochondria and the downregulation of methionine adenosyltransferase 1A (MAT 1A) with subsequent S-adenosyl-l-methionine depletion (Fig. 2).

Figure 2.

 Role of tumour necrosis factor (TNF)-derived sphingolipids in hepatocellular cell death. Cer, ceramide; GD3, ganglioside GD3; CDase, ceramidase; SIP, sphingosine 1-phosphate; SAM, S-adenosylmethionine; GSH, glutathione; ROS, reactive oxygen species.

Furthermore, in addition to a direct role for ceramide in the regulation of TNF-induced cell death, ceramide provides the carbon backbone for the synthesis of complex GSLs, such as gangliosides. Ganglioside GD3 has emerged as a cell-death effector activating the mitochondrial-dependent apoptosome through sequential mitochondrial reactive oxygen species (ROS) stimulation, cytochrome c release and caspase activation (31, 32), and this cell death function seems to be modulated by the acetylation state of GD3 (33). As previously observed with ceramide, the cellular GD3 levels increase in response to specific apoptosis stimuli such as Fas, TNF or expression of α2,8-sialyltransferase (31–33), while the downregulation of GD3 synthase, the enzyme responsible for GD3 synthesis from its precursor ganglioside GM3, prevents TNF-cell death (34). In addition, the active role of GD3 in promoting apoptosis is potentiated by interfering with the nuclear translocation of active members of NF-κB, thus suppressing the activation of NF-κB-dependent gene induction (35). Of note, it was shown using GSL derivatives that while the N-fatty acyl sphingosine moiety, common to both ceramide and GD3, is necessary for its ROS-stimulating effect, the presence of sugar residues in the backbone of ceramide is required to block NF-κB nuclear translocation (35). Hence, these findings provide evidence that the efficiency of GD3 in promoting hepatocyte cell death involves the lethal combination of mitochondrial ROS generation followed by the subsequent release of apoptotic factors and the suppression of NF-κB-dependent survival pathway.

SLs in steatohepatitis, ischaemia-reperfusion injury and viral hepatitis

As alluded above, hepatocytes are normally resistant to TNF owing to the early activation of survival pathways dependent on NF-κB, although TNF overexpression has been associated with the progression of several liver diseases including steatohepatitis, both alcoholic and non-alcoholic, and ischaemia-reperfusion injury. Interestingly, hepatocytes isolated from alcohol-fed rats develop an unusual susceptibility to TNF exposure as first described by Colell et al. (36) and then confirmed subsequently (37, 38), in the absence of any other sensitizing factor such as NF-κB, which remained activated in alcohol-fed hepatocytes. In examining the mechanism of susceptibility, we uncovered that alcohol feeding led to mitochondrial GSH (mGSH) depletion (39). Although alcohol metabolism could affect the TNF signalling at various levels (40), we found that the selective pharmacological depletion of mGSH in control hepatocytes recapitulated the susceptibility to TNF owing to ROS overgeneration (8, 9, 28, 36). Of relevance, ASMase-induced ceramide generation plays an indispensable role in this process, since ASMase−/− hepatocytes failed to overgenerate ROS in response to TNF despite mGSH depletion (28), indicating that ASMase-induced ceramide generation links TNF signalling to mitochondrial ROS stimulation.

In addition, the role of ASMase in ischaemia/reperfusion (I/R) liver injury has been recently examined (41). In a murine model of liver I/R, the levels of ceramide transiently increased after the reperfusion phase because of an early activation of ASMase. In vivo administration of an ASMase inhibitor, imipramine or ASMase knockdown by siRNA decreased ceramide generation during I/R, and attenuated serum transaminases and hepatocellular cell death through a mechanism that involved the blockade of JNK activation and the mitochondrial targeting of the pro-apoptotic Bcl-2 family protein, BIM (41).

Likewise, a role for SLs has also been described in the development of viral hepatitis. In chimeric mice with humanized liver infected with Hepatitis C virus (HCV) genotype 1a or 1b, myriocin an inhibitor of SPT, the first-step enzyme in the SL biosynthetic pathway, suppressed HCV replication (42). Moreover, using an HCV subgenomic replicon cell culture system, a lipophilic long-chain base compound NA255 identified from a secondary fungal metabolite, functioned as a small-molecule HCV replication inhibitor, owing to the blockade of de novo synthesis of SLs through inhibition of SPT that resulted in the disruption of the association among HCV non-structural (NS) viral proteins on the lipid rafts (43). Of note, it was found that NS5B protein exhibits an SL-binding motif in its molecular structure and that this domain was able to directly interact with SM. Thus these studies suggest that the modulation of ceramide generation through ASMase mediating TNF/Fas-induced hepatocellular death may be a promising therapeutic approach in steatohepatitis, hepatic I/R damage or alcohol-induced liver injury, and that inhibition of SL metabolism may provide a new therapeutic strategy for treatment of HCV infection.

SLs and hepatocellular carcinoma

Given the critical role of SLs in cell death/survival regulation, the alteration in SLs metabolism has a significant impact on cancer biology and therapy. In this regard, it has been recently shown that the turnover of SLs in preneoplastic hepatocytes was altered, suggesting that the modulation of this feature may be a potential target for chemoprevention of hepatocellular carcinoma (44). In particular, the balance of ceramide to S1P generation is known to modulate cell fate in response to cancer therapy and alterations in SLs metabolism may thus contribute to multidrug resistance in cancer cells (1, 2). Because CDases modulate the ceramide/S1P ratio through phosphorylation of sphingosine by SKs, these ceramide-metabolizing enzymes promote carcinogenesis and determine the efficacy of cancer therapy. For instance, acid CDase inhibition by a newly developed ceramide analogue, B13, induces apoptosis in cultured human colon cancer cells, and prevents liver metastases in vivo (45). Furthermore, anthracycline therapy (e.g. daunorubicin, DNR) has been shown to activate acid CDase but not neutral CDase by a post-transcriptional-dependent mechanism in established human (HepG2 cells) or mouse (Hepa1c1c7) hepatoma cell lines as well as in primary cells from murine liver tumours, but not in cultured mouse hepatocytes (46). Consequently, acid CDase silencing by small interfering RNA or pharmacological inhibition with n-oleoylethanolamine enhanced the ceramide to S1P balance compared with DNR alone, sensitizing hepatoma cells (HepG2, Hep-3B, SK-Hep and Hepa1c1c7) to DNR-induced cell death, through a mechanism involving mitochondrial targeting. Importantly, in vivo siRNA treatment targeting acid CDase reduced tumour growth in liver tumour xenografts of HepG2 cells and enhanced DNR therapy (46). Moreover, since ganglioside GD3 promotes apoptosis by a dual mechanism (35), we exploited this duality to potentiate the therapeutic effectiveness in hepatoma cells resistant to current cancer therapy (47). The preincubation of HepG2 cells with GD3 blocked the translocation of NF-κB to the nuclei resulting in the sensitization to radiotherapy and daunorubicin-induced cell death owing to the overaccumulation of ROS generated within mitochondria. Thus, SLs arise as a potential target to modulate hepatocarcinogenesis and in the potentiation of current available therapies.

SL accumulation and liver injury

Sphingolipidoses

SLs participate in cell-recognition events and in receptor biology and their metabolic intermediates serve as signalling molecules in apoptotic and proliferative responses. However, despite increasing knowledge of the activity of biosynthetic and degradative intermediates of SLs, little is known about the mechanism by which SLs accumulation contributes to the complex pathology of the human sphingolipidoses when stored in excess. Sphingolipidoses are a subgroup of lysosomal storage diseases (LSDs). LSDs are considered rare biochemical diseases that usually result in progressive deterioration of organ systems including the central nervous system. Although individually these diseases are uncommon, the combined 5000–incidence of over 40 lysosomal disorders approach a birth prevalence of one person per 7000 (48). LSDs include the neuronal ceroid lipofuscinosis, sphingolipidoses, mucopolysaccharidoses, mucolipidoses, oligosaccharidoses, sialidoses, type-II glycogen storage disease (Pompe disease). Among them, sphingolipidoses are the most common subtype, and Gaucher disease (caused by GluCer accumulation) the most prevalent among them.

Sphingolipidoses are defined as disorders caused by a genetic defect in catabolism of sphingosine-containing lipids that result in accumulation of unmetabolized SLs (49) (Table 1). These defects are inherited in an autosomal–recessive fashion, except for Fabry disease that is X-linked, and are very frequent among patients of Ashkenazi Jewish origin. Some SLs stored in the membrane because of degrading enzyme deficiency are not toxic for the cell. However, if those substances are stored in large amounts they interfere with the intracellular transport and other cell activities. This subgroup of LSDs is systematically classified on the basis of the main accumulating lipid (Table 1). Many sphingolipidoses occur in different age-related forms with a clinical onset during infantile, juvenile or adult life. The most severe, the infantile form, occurs with acute brain involvement and patients die within the first years of life. It has been shown for some of them, such as for Gaucher's and Sandhoff diseases, that the infantile forms show a higher degree of enzyme deficiency, apart from the worst prognosis, than the adult forms (50) where residual enzyme activity slows down the symptoms and extends the life span. Genotype differences generally determine the pathogenesis of the disease according to its effect on the enzyme activity level; in some particular cases, the same genotype causes formation of different phenotypes (51). In adults, Gaucher and Niemann-Pick type A diseases are the most common of these disorders. However, the broad spectrum of phenotypes can make recognition of lysosomal storage disorder difficult. The age of onset, severity of symptoms, organ systems affected and central nervous manifestation can vary markedly within a single disorder type or subtype. Neurological symptoms can include seizures, dementia and brainstem dysfunction. Among the peripheral symptoms the most prominent are enlargement of the spleen and liver (hepatosplenomegaly), such as in the Niemann-Pick type C (NPC) disease. Because SLs exhibit a strong affinity towards cholesterol, the accumulation of the former normally results in the upregulation of cholesterol levels, which, independent of SLs accumulation, may contribute to a spectrum of symptoms. This association has been best characterized in mice with NPC disease (51). Mutations in the NPC1 gene, an endosomal/lysosomal protein responsible for the trafficking of cholesterol, results in increased free-cholesterol levels as well as in SM in cellular membranes, and these mice exhibit significant elevation in the plasma levels of alkaline phosphate and aminotransferases, and enhanced hepatocellular apoptosis and susceptibility to TNF (28, 51).

Table 1.   Most common sphingolipidoses
SphingolipidosesMain storage materialEnzyme deficiency
Gaucher diseaseGlucosylceramide (glucocerebroside)Glucosylceramidase (EC 3.2.1.45)
Farber diseaseCeramideCeramidase (EC 3.5.1.23)
Niemann-Pick A and B diseaseSphingomyelin, lysosphingomyelinAcid sphingomyelinase (EC 3.1.4.12)
Fabry diseaseα-Galactosyl-lactosylceramideα-galactosidase (EC 3.2.1.22)
Krabbe diseaseGalactosylceramide (galactocerebroside)Galactosylceramidase (EC 3.2.1.46)
Tay-Sachs diseaseGanglioside GM2β-n-acetylhexosaminidase A (EC 3.2.1.30)
Sandhoff diseaseGanglioside GM2, Globosideβ-n-acetylhexosaminidase A and B (EC 3.2.1.30)
GM1-gangliosidosisGanglioside GM1GM1-β-galactosidase (EC 3.2.1.23)

Sphingolipidoses and the liver

Hepatomegaly is a frequent feature in patients with type 1 Gaucher's disease since the liver is affected by all types of Gaucher disease. The diagnostic features are produced by SL-engorged Kupffer cells and macrophages in sinusoids and portal areas. These cells are large with an eccentric nucleus and eosinophilic, corrugated cytoplasm. Tightly packed lysosomes are filled with tubular glucocerebroside structures. Hepatocytes are not involved in storage. Fibrosis or cirrhosis may evolve in the lobules and portal tracts as a consequence of progressively advancing storage, probably causing pressure atrophy (52). The recent generation of a conditional knockout mouse model for Gaucher disease will help us understand the mechanisms underlying the pathophysiology of the disease (53).

Liver disease is also a common feature of early- and late-onset forms of Niemann-Pick disease. Types A and B are owing to SMase deficiency whereas Type C (NPC) is caused by a defect of intracellular trafficking of cholesterol. Although NPC strictly does not feature SLs accumulation, it shares many characteristics with types A and B, and may be present in early life with a neonatal hepatitis-like picture including giant cell transformation. In Niemann-Pick disease, there is a progressive increase of storage cells within hepatic sinusoids. Another histological feature includes the hepatocellular vacuolation staining as SLs and cholesterol (54). As the disease worsens, hepatic lobules become progressively distorted with portal septal fibrosis and cholestasis.

Infantile GM1 gangliosidosis is an acute neurovisceral disorder that develops in the neonatal period (55). β-galactosidase deficiency leads to GM1 ganglioside accumulation within lysosomes, resulting in enlarged liver Kupffer cells and hepatocytes. The liver in Tay-Sachs or Sandhoff disease in the GM2 group of gangliosidoses is histologically normal, but zebra bodies, cells with abundant lamellar membrane profile owing to storage material accumulation in the lysosomes, may be seen on ultrastructural examination.

Several LSDs are characterized by prominent neurological involvement and minimal peripheral impairment, whereas many sphingolipidoses show peripheral dysfunction with rare brain involvement (e.g. Fabry disease) (56). In some cases such as in NPC disease, the premature death owing to fatal liver failure occurs before the development of the neurological symptoms. For each organ system there is a threshold of enzymatic activity, with clinical manifestations occurring below this threshold, and although specific mutations or types of mutation are sometimes linked to disease outcome, genotype–phenotype correlations are not always strong (57).

Cell biology of sphingolipidoses

The question that arises is how excessive SL storage can cause liver disease? One possible explanation for the pathology of sphingolipidoses would be that this massive storage affects intracellular trafficking by blocking intracellular transport to or from lysosomes. In fact, a common defect in lipid sorting and transport among the different sphingolipidoses has been proposed (58, 59), based on the observation that a derivative of lactosylceramide is targeted to the Golgi apparatus in normal cells, while it accumulates in endosomes and lysosomes in fibroblasts from sphingolipidoses patients. This process is linked to the accumulation of unesterified cholesterol and demonstrates a role for Rab7 and Rab9 in the Golgi targeting of GSLs (60). Recently, Rab8 has arisen as a key regulatory switch for coordinating the cytoskeletal changes and membrane dynamics in cholesterol efflux. In NPC cells, Rab8 overexpression is able to reduce cholesterol content by promoting cholesterol removal from endosomal circuits (61). The fact that cholesterol and SLs levels are interrelated is hardly surprising since both lipids are important components of microdomains/rafts at the cell surface (62), and the accumulation of one will have profound effects on the other. The most prevalent component of the SL fraction in the cell membrane is SM. Tight interactions between the sterol ring of cholesterol and the ceramide moiety of SM promote the lateral association of SLs and cholesterol into ‘distinct microdomains’ called rafts, resulting in their separation from other phospholipids (63, 64). This process results in the transition of these microdomains into a liquid-ordered phase in contrast to the more fluid liquid-disordered phase of the cell membrane (62, 65). For instance, changes in cholesterol caused by defective SL hydrolysis have been observed in Niemann-Pick A and B. Another possible reason for sphingolipidoses pathology, at least for Niemann-Pick A and B diseases, could be the presence of defective signalling as a result of changes in the production of ceramide. On one hand, in Niemann-Pick A and B diseases, which are caused by the defective activity of ASMase, ceramide is not produced in response to various ligands, which could affect the physical–chemical and functional properties of membranes rafts both in plasma membrane and in mitochondria.

Therapies for sphingolipidoses

Treatment of sphingolipidoses is dependent on the restoration of the defective enzymatic activity in lysosomes of abnormal cells. Different approaches have been developed, such as enzyme replacement therapy (ERT), gene therapy and bone marrow transplantation (BMT) that permit the normal enzyme to be taken up by the plasma membrane receptor. These strategies result in delivery of the normal enzyme to the lysosome where it can catabolize stored substrate (66). For example, ERT, the most successful treatment available today, has been proven in the non-neuronopathic forms of Gaucher and Fabry disease (67) using recombinantly produced α-galactosidase A or placental-derived glucocerebrosidase, respectively, ameliorating disease symptoms and quality of life to more than 3000 patients worldwide (68).

In relation to gene therapy, significant progress has been made for gene therapy in cultured cells and animal models (66, 69). For instance, in a recent study (70) using gene therapy to treat Sandhoff mice by stereotaxic intracranial inoculation of recombinant adeno-associated viral vectors encoding the complementing human β-hexosaminidase α and β subunit genes, the onset of the disease was greatly delayed with the preservation of motor function, while inflammation and GM2 ganglioside storage in the brain and spinal cord was significantly reduced. This study highlights the potential of gene therapy to treat human Tay-Sachs-related diseases, and points to the possible use of combined modalities, for example using iminosugars to achieve substrate deprivation, that would show at least an additive therapeutic effect.

Regarding BMT, this approach has been tried on the premise that since lysosomal enzymes undergo a cycle of exocytosis and re-uptake by mannose-6-phosphate receptor, engrafted bone marrow would produce a large number of exocytic cells that could circulate and transfer their normal enzyme to enzyme-deficient cells. Although this procedure improves symptoms to some extent, however taking into consideration the high morbidity and mortality associated with BMT, this procedure has a limited use (71).

Other approaches are based on the prevention of metabolic and cellular defects that occur owing to the accumulation of undegraded substrates, while others deal with the accumulation of substrates by targeting residual enzyme activity that contributes to Sls synthesis. This substrate deprivation has been successfully achieved by using N-butyldeoxynojirimycin (NB-DNJ, OGT-918 or miglustat), an iminosugar inhibitor of GluCer synthase, the first enzyme in the glycosphingolipid synthesis (Fig. 1), in the treatment of disease symptoms in Tay-Sachs and Sandhoff disease mouse models. In addition, this inhibitor has been used in patients with Gaucher's disease, improving clinical signs of adult and juvenile non-neuronopathic forms (72, 73). These strategies however might not be very useful for sphingolipidoses with central nervous system involvement owing to the limited ability of therapeutic agents able to cross the blood–brain barrier. In addition, fatal hepatic fibrosis has been observed in patients with Gaucher's disease despite prolonged ERT therapy (52), indicating the need for urgent alternative strategies or combined therapy.

Concluding remarks

The burst of knowledge in the molecular and cell biology of SLs in the last decade has refined how SLs should be viewed in pathophysiology. The dogma of a predominant structural function in membranes for these lipids has changed dramatically to a new role as signal intermediates in cell metabolism, and in the regulation of cell death. This novel function of SLs in modulating the hepatocellular susceptibility to stress and death ligands-induced cell death may offer new therapeutic strategies for the treatment of liver diseases by preventing hepatocellular loss. Because ASMase-mediated ceramide generation regulates TNF-induced hepatocellular death, the antagonism of this enzyme may be of relevance in steatohepatitis or ischaemia/reperfusion liver injury. Moreover, the upregulation of ceramide along with the downregulation of SLs with anti-apoptotic activity may enhance the effectiveness of current cancer therapy to treat hepatocellular carcinoma. In addition to their role in signal transduction, the accumulation of SLs in sphingolipidoses affects vital cellular function in many key organs, including the liver, which determines in some instances a fatal outcome owing to hepatic failure, suggesting the need for efficient therapy targeting the abnormal metabolism and trafficking of SLs. As our knowledge on the cell biology and regulation of SLs increases, we hope that further progress in this field affecting hepatology may provide life-saving treatments for these patients.

Acknowledgements

The work presented was supported in part by the Research Center for Liver and Pancreatic Diseases, P50 AA 11999, funded by the US National Institute on Alcohol Abuse and Alcoholism; Plan Nacional de I+D grants SAF (2003-04974; 2006-06780); and FIS (04/1039) CIBER-HEPAD supported by Instituto de Salud Carlos III, Spain. M. M. is a Ramon y Cajal investigator.

Ancillary