- Email: [email protected]
Biochemistry of glycosphingolipid degradation
ELSEVIER
Clinica Chimica Acta 266 (1997) 51-61
Biochemistry of glycosphingolipid degradation Konrad Sandhoff*, Thomas Kolter Kekuld-lnstitut fiir Organisehe Chemie und Biochemie der Universitiit, Gerhard-Domagk-Str. 1, 53121 Bonn, Germany
Received 1 September 1996; accepted 23 July 1997
Abstract
Glycosphingolipids (GSLs) form cell-type-specific patterns on the surface of eukaryotic cells. Degradation of GSLs requires endocytotic membrane flow of plasma membrane-derived GSLs into the lysosomes as the digesting organelles. Recent research focused on the mechanisms leading to selective membrane degradation in the lysosomes and on the mechanism and physiological function of sphingolipid activator proteins, which are needed for degradation of GSLs with short oligosaccharide chains in addition to hydrolysing enzymes. Both, the inherited deficiency of lysosomal hydrolases and of sphingolipid activator proteins give rise to sphingolipid storage diseases. In some cases it was possible to correlate residual enzyme activities with the onset and the course of the disease. © 1997 Elsevier Science B.V. Keywords: Sphingolipid activator proteins; Glycosphingolipids; Lysosomes; Sphingolipidoses; Storage diseases
1. Introduction
Glycosphingolipids (GSLs) are components of the plasma membrane of eukaryotic cells [ 1]. Each GSL contains a hydrophobic ceramide moiety that acts as membrane anchor and a hydrophilic, extracellular oligosaccharide chain. Variations in the type, number and linkage of sugar residues in the oligosaccharide chain give rise to the wide range of naturally occurring GSLs. These Corresponding author. Tel./fax: + 49 228 737778; e-mail: [email protected] 0009-8981/97/$17.00 © 1997 Elsevier Science B.V. All rights reserved. P l l S0009-8981 (97)00166-6
52
K. Sandhoff, T. Kolter / Clinica Chimica Acta 266 (1997) 5 1 - 6 1
form cell-type-specific patterns at the cell surface which change with cell growth, differentiation, viral transformation and oncogenesis [2]. GSLs interact at the cell surface with toxins, viruses, and bacteria as well as with membrane bound receptors and enzymes [3]. They are also involved in cell-type-specific adhesion processes and, in addition, lipophilic products of GSL metabolism such as sphingosine and ceramide, play a role in signal transduction events [4]. Finally, GSLs form a protective layer on biological membranes protecting them from inappropriate degradation. Degradation of GSLs occurs in lysosomal compartments within the cell [5]. The plasma membranes containing GSLs destined for degradation are endocytosed and travel through the endosomal compartments to the lysosome. Here, hydrolysing enzymes cleave sugar residues sequentially to produce ceramide which is deacylated to sphingosine. This can leave the lysosome, reenter the biosynthetic pathway or be further degraded. To date, all the inherited diseases of GSL metabolism that have been described affect GSL degradation, and none have been described that affect GSL biosynthesis. Presumably, severe disorders affecting GSL biosynthesis would result in incorrect GSL patterns on the cell surface, and these in turn would affect morphogenesis and embryogenesis. A study of the inherited disorders in degradation, the lysosomal storage diseases, has provided insights into the processes involved in GSL degradation.
2. Topology of endocytosis and lysosomal digestion Catabolism of complex GSLs derived from the plasma membrane occurs in the lysosomes after endocytosis. In the conventional model, components of plasma membranes reach the lysosomal compartment by endocytic membrane flow via the early and late endocytic reticulum [6] (Fig.). During this vesicular flow, molecules in the membrane are subjected to a sorting process which directs some of the molecules to the lysosomal compartment, some others to the Golgi apparatus and yet others even back to the plasma membrane. After a series of vesicle budding and fusion events through the endosomal compartment, the membrane fragments reach the lysosomal compartment. As a consequence of endocytic membrane flow, the plasma membrane fragments are incorporated into the lysosomal membrane. Lysosomal degradation of former components of the plasma membrane must then proceed selectively within the lysosomal membrane without destroying it. It is difficult to see how this could occur, especially as the inner leaflet of the lysosomal membrane is covered with a thick glycocalix composed of glycoproteins. In an alternative model for the topology of endocytosis [5,7] (Fig. 1), components of the plasma membrane pass through the endosomal compartment as intraendosomal vesicles that become intralysosomal vesicles on reaching the
K. Sandhoff, T. K o l t e r / Clinica C h i m i c a A c t a 2 6 6 ( 1 9 9 7 ) 5 1 - 6 1
A
B 'l l t
-
PM
\
/ Go!glapparatus ~
Late endosorr~e~
~
Transient fusion
G~ccallx
53
Transient fusionand transfer
I r'~rr'~ar~,
~rr~ rn~'r~ane
Fig. 1. Two models for the topology of endocytosis and lysosomal digestion of glycosphingolipids (GSLs) derived from the plasma membrane. Conventional model (A): degradation of GSLs derived from the plasma membrane occurs selectively within the lysosomal membrane. Alternative model for the topology of endocytosis and digestion of GSLs (B): during endocytosis glycolipids of the plasma membrane become incorporated into the membranes of intraendosomal vesicles (multivesicular bodies). The vesicles are transferred into the lysosomal compartment when the late endosome fuses with primary lysosomes. PM, plasma membrane; ~, glycosphingolipid [5].
lysosomes. The vesicles are initially formed in the early endosome by selective invagination (budding-in) of endosomal membranes enriched for components of the plasma membrane. The surrounding endosome then passes along the endocytic pathway by the normal, successive events of membrane fission and fusion. The intraendosomal vesicles, however, are carried along as passengers. When the vesicle reaches the lysosome, glycoconjugates originating from the outer leaflet of the plasma membrane face the lysosol on the outer leaflet of intralysosomal vesicles and are thus in the correct place for degradation. In this model there is no glycocalix blocking the action of hydrolases. This hypothesis is supported by a series of observations: 1. multivesicular bodies are observed at the level of the early and late endosomal reticulum (reviewed in [5]); 2. the epidermal growth factor receptor derived from the PM and internalized into lysosomes of hepatocytes is not integrated into the lysosomal membrane [8];
54
K. Sandhoff, T. Kolter / Clinica Chimica Acta 266 (1997) 5 1 - 6 1
3. spherical multivesicular bodies are the predominant endocytic compartments in HEp-2 cells. Moreover, these multivesicular bodies mature within 60-90 min into lysosomes which still contain internal vesicles [9]; 4. multivesicular storage bodies accumulate in cells from patients with sphingolipid storage diseases. They also occur in cells, e.g. Kupffer cells and fibroblasts, of a patient with another sphingolipid storage disease, a combined activator protein deficiency [10,11]. Analysis of cultivated fibroblasts of this patient shows that the storage vesicles refer to late endosomal or lysosomal compartments. These compartments are still functionally active, except their inability to degrade sphingolipids with short oligosaccharide head groups. After complementation of the medium of these cells with the missing SAP-precursor, not only the degradation block is abolished but also the morphological shape of the multivesicular bodies is reversed [12]. To date, budding-in of vesicles on the stage of the early endosome is not observed by microscopy. This membrane fission event is postulated to take place since this would explain both the occurrence of intralysosomal vesicles and the selective degradation of glycosphingolipids derived from the plasma membrane in contrast to those of the lysosomal membrane. Within the lysosome, degradation of GSLs occurs by the step-wise action of specific acid exohydrolases [5]. More than 10 different exohydrolases are involved in this process. If any of these enzymes is deficient, the corresponding lipid substrate accumulates and is stored in the lysosomal compartment. For glycosphingolipids with long carbohydrate chains of more than four sugar residues the presence of an enzymatically active exohydrolase is sufficient for degradation in vivo. Detailed biochemical analysis of sphingolipidoses such as metachromatic leukodystrophy and Tay-Sachs disease demonstrated that the in vivo degradation of some glycolipids involves not only exohydrolases but also a second type of effector molecule [13]. In these cases lipid storage is observed in patients without any enzyme deficiency, but with deficiency of a protein cofactor. The degradation of membrane bound GSLs with short oligosaccharide chains requires the cooperation of an exohydrolase and a sphingolipid activator protein. Several sphingolipid activator proteins are now known including the GM2 activator and the saposins. A detailed study of these proteins has illuminated some of their main features and indicated the mechanisms by which they facilitate digestion of GSLs in the lysosome.
3. The GM2 activator and its role in lysosomal digestion The enzymatic degradation of ganglioside GM2, the main storage material in Tay-Sachs disease, requires [~-hexosaminidase A and a lysosomal ganglioside
K. Sandhoff, T. Kolter / Clinica Chimica Acta 266 (1997) 5 1 - 6 1
55
binding protein, the GM2 activator [14,15]. The GM2 activator binds ganglioside GM2 as well as related gangliosides forming water-soluble complexes (mostly in a 1:1 molar ratio). In vitro it functions as a ganglioside transfer protein transferring gangliosides from donor membranes to acceptor membranes. The GM2 activator behaves as a 'liftase'. It recognizes GM2 within the membrane and by binding to it lifts the lipid out of the bilayer and presents it to the water soluble [3-hexosaminidase A for degradation. However, it is also possible that the activator-lipid complex leaves the membrane and the enzymatic reaction takes place in free solution. Formation of the ternary complex presumably involves also a protein-protein interaction between the GM2 activator and [3-hexosaminidase A [16]. Point mutations within the structural gene of the GM2 activator have been identified in four patients with the AB-variant of GM2 gangliosidosis, e.g. [17]. The mutated proteins are proteolytically labile so that some of them are degraded in the ER, others in the Golgi compartment. The resulting loss of the GM2 activator causes a block within the GM2 degradation in the lysosomes as demonstrated in metabolic studies in cultured fibroblasts of the patients. The block can be bypassed by feeding native GM2 activator to the culture medium of the mutant fibroblasts [18].
4. Other activator proteins The first activator protein was identified in 1964 as a protein which is necessary for the hydrolytic degradation of glycosphingolipids carrying a sulphuric ester group (sulfatides) by lysosomal arylsulfatase A [19]. This sulfatide activator, SAP-B (saposin B), is a small lysosomal glycoprotein consisting of 80 amino acids, with an N-linked carbohydrate chain and three disulfide bridges [20]. Like the GM2 activator it binds GSLs but has a broader specificity. In vitro it behaves similarly to the GM2 activator in some aspects, i.e. it can recognize and bind several different GSLs on the surface of micelles by forming a stoichiometric complex and is then able to transfer the GSLs to the membranes of acceptor liposomes [13]. SAP-B can also present GSLs to water-soluble enzymes as substrates [21]. The inherited deficiency of the sulfatide activator leads to a lysosomal storage disease which resembles metachromatic leukodystrophy. However, unlike typical metachromatic leukodystrophy not only sulfatide but also additional glycolipids, e.g. globotriaosyl ceramide, accumulate owing to a blockage of degradation at several points in the catabolic pathway [5]. The sulfatide activator has sequence homology with three further activator proteins: the Gaucher factor (SAP-C), SAP-A and SAP-D. Molecular biological analysis has revealed that these four activator proteins are formed by proteolytic
56
K. Sandhoff, T. Kolter / Clinica Chimica Acta 266 (1997) 5 1 - 6 1
processing of a common precursor protein, the SAP-precursor (reviewed in [13]). The four activator proteins, SAP-A, B, C and D show homology to each other and have similar properties, but differ in their function and their mechanism of action [13]. The physiological function of these sphingolipid activator proteins is only partially clarified to date; most of our knowledge on this has emerged from studies on patients with atypical lipid storage diseases. As mentioned above, an inherited deficiency of SAP-B leads to an atypical form of metachromatic leukodystrophy. Similarly, SAP-C deficiency causes an atypical form of Gaucher disease, where glucosyl ceramide accumulates [22,23]. In one patient with a complete deficiency of the whole SAP-precursor protein due to a homoallelic mutation within the start codon (AUG --> UUG) [11] there is simultaneous storage of many sphingolipids, including ceramide, glucosyl ceramide, lactosyl ceramide, ganglioside GM3, galactosyl ceramide, sulfatides, digalactosyl ceramide and globotriaosyl ceramide [24]. The storage of GSLs can be demonstrated in cultured mutant cells by pulse-chase experiments [25]. Storage substances in cells from different patients can be visualized by biosynthetic labelling of glycolipids with [14C]serine followed by long chase periods. As previously mentioned, the glycolipids that accumulate reflect the types of GSLs synthesized in each cell type. In the case of fibroblasts, glucosylceramide accumulates in cells from patients with SAP-C deficiency, ganglioside GM2 accumulates when there is deficiency of the GM2 activator and ceramide, glucosylceramide, lactosylceramide and ganglioside GM3 all accumulate when the SAP precursor is deficient. Cell culture models such as these are not only suitable for diagnostic purposes, but also allow the elucidation of the function of the different activator proteins. Although, the four SAPs show great homology, they act very differently in these experiments. If purified SAP-B is fed to the cells with SAP-precursor deficiency, the storage of lactosylceramide is abolished but ceramide and glucosylceramide still accumulate. In contrast, when SAP-D is fed to the cells ceramide accumulation is prevented. These experiments give the first hint to the in vivo function of this activator protein [25]. As expected, if the mutant cells are fed with the SAP-precursor in nanomolar concentrations storage of all these glycolipids is abolished and the aberrant accumulation of intralysosomal membrane structures prevented (see above). These studies on cells in culture from patients with different storage diseases support our hypothesis concerning the function of sphingolipid activator proteins. In their absence, membrane bound GSLs with short oligosaccharide chains are not readily accessible for digestion by water soluble exohydrolases. By analogy to a razor blade, the hydrolases can only attack oligosaccharide chains of GSLs that protrude far enough from the membrane into the aqueous phase. This is no problem for long chain oligosaccharides but for the degradation of GSLs with short sugar chains the assistance of activator proteins is
K. Sandhoff, T, Kolter I Clinica Chimica Acta 266 (1997) 51-61
57
required. Some of them act as binding proteins or liftases, forming a watersoluble complex with the lipid and hence raising the lipid out of the membrane. However, the function of activator proteins is not necessarily restricted to binding lipids in this way. For example, SAP-C can directly activate glucosylceramide [3-glucosidase [26] and the GM2 activator specifically interacts with [3-hexosaminidase A [ 16]. The simultaneous deficiency of the four sphingolipid activator proteins, as seen in the patient described earlier, causes an intraendosomal and intralysosomal accumulation of vesicles and other membrane fragments (multivesicular bodies) and supports our view of the topology of endocytosis and lysosomal digestion. Furthermore, the occurrence of storage bodies in the combined activator protein deficiency suggests a possible function of glycolipids. The increased concentrations of glycolipids within distinct membranes appears to protect the vesicles from degradation by phospholipases, proteases and other hydrolases that occur in high concentration in the lysosomes.
4.1. Mechanisms of the pathogenesis of lipidoses The pathogenesis of inherited disorders is of a complex nature and not fully understood for any disease. This is also valid for lipidoses although they are guided by simple biochemical principles [27]. On the one hand, mutations in genes coding for different enzymes can lead to diseases with similar or even identical phenotype. For example, deficiencies of the a-chains in 13-hexosaminidase A and S or of the [3-chains in [3-hexosaminidase B and A or a deficiency in the GM2 activator protein give rise to similar clinical phenotypes that were characterized before as amaurotic idiocy. On the other hand mutations in the same structural gene can cause different clinical phenotypes which in some cases are known under different eponyms. For example, severe forms of c~-iduronidase deficiency are known as Hurler disease whereas less severe forms are called Scheie disease. Even patients with identical mutations within the same structural gene e.g. that one coding for arylsulfatase A can show different clinical forms, obviously due to a differing genetic background. Despite this enormous heterogeneity and only indirect correlation between genotype and phenotype of a disease it is possible to state some factors that connect genotype and phenotype of sphingolipidoses. 1. Due to the cell-type-specific expression of GSLs, lipid storage in lipidosis patients is observed especially in those cells and organs, in which the lipid substrates of the corresponding deficient enzyme are prevalently synthesized (e.g. complex gangliosides in neuronal cells) or by which they are taken up by phagocytosis (glucosylceramide and its precursors in macrophages in case of Gaucher disease).
58
K. Sandhoff, T. Kolter / Clinica Chimica Acta 266 (1997) 51-61
2. Mutations in proteins of GSL degradation give rise to disorders in GSL turnover. Although having different effects on the corresponding proteins, t h e y can be treated as Vmax mutations that finally reduce the catabolic turnover of substrates [28]. 3. Impeded degradation can cause the accumulation of morphologically active or toxic compounds. 4.2. The extent of degradation disorders causes different clinical courses of disease
In normal ceils the maximum turnover rate of a catabolic system (Vmax) is always higher than the influx rate (vi) into the lysosomal compartment of the substrates to be degraded. Only when the maximum turnover rate becomes smaller than the influx rate, an irreversible substrate storage occurs and it becomes a disease. Due to homoallelic or heteroallelic mutations, the degrading system can approach different residual activities which should correspond to the pathogenesis of the disease. The mutated proteins and the resulting maximum enzyme activities Vmax constitute a decisive connection between genotype and phenotype of a disease. Starting with the Michaelis-Menten equation, the steady state substrate concentration within the lysosome can be calculated as a function of the residual enzyme activity [28]. In normal cells, substrate concentrations are usually far below the Michaelis-Menten constant, i.e. the substrate concentration at which a half maximal turnover rate is achieved. A decrease of the maximum turnover rate of the degrading system on values of e.g. 50 to 20% of normal cells, a typical range for heterozygote carriers of inherited diseases, does not influence the turnover rate v. The ratio v/v i remains constant, since lowering of maximal enzyme activity is compensated by a higher substrate saturation of the degrading enzyme (Fig. 2 [27]). This compensatory mechanism works until the maximal turnover rate decreased to the value of substrate influx into the lysosome. At this critical threshold value all enzyme molecules occur in the form of enzyme substrate complexes. The decrease of residual enzyme activities to this value should not lead to an irreversible pathological substrate storage. This explains that probands with pseudodeficiencies, e.g. a mean loss of enzyme activity of up to 90% of [3-hexosaminidase A or arylsulfatase A do not fall ill. Only the decrease of the maximal catabolic enzyme activity (Vmax) below the critical threshold value vi causes storage of the corresponding lipid substrate. So the model predicts that the decrease of the maximal catabolic enzyme activity to the calculated threshold value does not influence the flux rate of the substrate and that pathological storage occurs only below this level. We examined this model by measuring flux rates and enzyme activities in cultured skin fibroblasts from patients with GM2 gangliosidoses [29]. The experimentally observed values support our hypotheses since the flux rate decreases linearly with
59
K. Sandhoff, T. Kolter / Clinica Chimica Acta 266 (1997) 51-61
x
]
xx
x
~
x
K
xX~
aa
RH
a
a
o
o
Q
a
a ma
"
~-.=_ o
200
400
Hexosamlnldase
600
A activity
800 (pmol
/ h x mg
1000
1200
x AU)
i
Z
i
-~ls-
i
i i
0
Critical
threshold
activity
...
er,J
~,
......................................................
10-
• i
:>
,.
.~
1 .0
I
_ __
Hypothetical
solubility
of s u b s t r a t e
0.5
W
,
~
,
~
,.
M
I--
i
I
1'0 Enzyme
activity
1'5
2'0
Vm=x/V,
Fig. 2. Theoretical correlation between the residual activity of an enzyme and the steady state concentration and turnover rate of its substrate in a defined compartment [28]. The substrate concentration is expressed as multiples of K m turnover rate and enzyme activity (v,.ax) as multiples of the influx rate vi. Above the critical threshold activity (Vm,x > Vi), the turnover rate is limited by the influx rate; below this threshold (Vmax < Vi) it is limited by the remaining capacity of the deficient enzyme; in the latter case, the substrate accumulates at a rate of va¢c = vi - Vma~. - - , steady state substrate concentration; . . . . . . . . , turnover (flux) rate; . . . . . . . . . . critical threshold activity; . . . . . . , critical threshold activity, taking the limited solubility of the substrate into account. Above: residual activity of [3-hexosaminidase A and turnover of its substrate ganglioside GM2 in cell culture. Skin fibroblasts from normal probands, from patients with various forms of GM2 gangliosidosis and from heterozygotes were fed with [3H]ganglioside GMZ for 3 days in culture. Cells were then harvested, homogenized in water and assayed for the following three parameters: incorporation of the substrate, percentage of degraded substrate, activity of J3-hexosaminidase A towards ganglioside GM2 in the presence of the GM2 activator protein [29]. Filled circles, a-chain deficiency (B-variant of GM2 gangliosidosis, Tay Sachs disease), infantile type; empty circles, a-chain deficiency, juvenile type; filled squares, a-chain deficiency, adult type; filled triangles, activator deficiency (AB-variant of GM2 gangliosidosis); healthy probands, x, obligate heterozygotes from GM2 gangliosidosis; empty squares, normal controls.
60
K. Sandhoff, T. Kolter / Clinica Chimica Acta 266 (1997) 51-61
decreasing e n z y m e activity only below a threshold value. Patients with the adult course of the disease show a significantly higher turnover than patients with the juvenile course and these a higher turnover than patients with the infantile course. Smaller differences in the residual e n z y m e activities correspond to significant differences in the respective flux rates.
4.3. A n i m a l
models
In the meantime, animal models o f the G M 2 gangliosidoses have b e c o m e available and permit further studies o f the pathogenesis o f this disease. In contrast to the human diseases, the mouse models of B-Variant and 0-Variant show very different neurological phenotypes. The Tay-Sachs model showed no neurological abnormalities while the Sandhoff model was severely affected. This phenotypic difference between the two mouse models is due to differences in the ganglioside degradation pathway between mice and humans [30].
Acknowledgements The work performed in the laboratory of the authors was supported by the Deutsche Forschungsgemeinschaft (SFB 284).
References [1] Wiegandt, H. In: Neuberger A, van Deenen LLM, editors. New comprehensive biochemistry, vol. 10. Amsterdam: Elsevier, 1985:199-260. [2] Hakomori S. Annu Rev Biochem 1981;50:733-64. [3] Karlsson K-A. Annu Rev Biochem 1989;58:309-50. [4] Hannun YA, Obeid LM. Trends Biochem Sci 1995;20:73-7. [5] Sandhoff K, Kolter T. Trends Cell Biol 1996;6:98-103. [6] Griffiths GW, Hoflack B, Simons K, Mellman IS, Kornfeld S. Cell 1988;52:329-41. [7] FiJrst W, Sandhoff K. Biochim Biophys Acta 1992;1126:1-16. [8] Renfrew CA, Hubbard AL. J Biol Chem 199l;266:21265-73. [9] van Deurs B, Holm PK, Kayser L, Sandvig K, Hansen SH. Eur J Cell Biol 1993;61:208-24. [10] Harzer K, Paton BC, Poulos A, Kustermann-Kuhn B, Roggendorf W, Grisar T, Popp M. Eur J Pediatr 1989;149:31-9. [ 11] Schnabel D, Schr6der M, Fiirst W, Klein A, Hurwitz R, Zenk T, Weber J, Harzer K, Paton BC, Poulos A, Suzuki K, Sandhoff K. J Biol Chem 1992;267:3312-5. [12] Burkhardt JK, Hiittler S, Klein A, M6bius W, Habermann A, Griffiths G, Sandhoff K. Eur J Biochem 1997;73:10-8. [13] Sandhoff K, Harzer K, Fiirst W. In: Scriver C, Beaudet AL, Sly WS, Valle D, editors. The metabolic and molecular basis of inherited disease. New York: McGraw Hill, 1995:2427-41.
K. Sandhoff, T. Kolter / Clinica Chimica Acta 266 (1997) 51-61
61
[14] Gravel RA, Clarke JTR, Kaback MM, Mahuran D, Sandhoff K, Suzuki K. In: Scriver C, Beaudet AL, Sly WS, Valle D, editors. The metabolic and molecular basis of inherited disease. New York: McGraw Hill, 1995:2839-79. [15] Meier EM, Schwarzmann G, Fiirst W, Sandhoff K. J Biol Chem 1991;266:1879-87. [16] Kytzia H-J, Sandhoff K. J Biol Chem 1985;260:7568-72. [17] Schepers U, Glombitza G, Lemm T, Hoffmann A, Chabas A, Ozand P, Sandhoff K. Am J Hum Genet 1996;59:1048-56. [18] Klima H, Klein A, van Echten G, Schwarzmann G, Suzuki K, Sandhoff K. Biochem J 1993;292:571-6. [19] Mehl E, Jatzkewitz H. Hoppe-Seyler's Z Physiol Chem 1964;339:260-76. [20] FiJrst W, Schubert J, Machleidt W, Meyer HE, Sandhoff K. Eur J Biochem 1990;192:709-14. [21] Li S-C, Sonnino S, Tettamanti G, Li Y-T. J Biol Chem 1988;263:6588-91. [22] Christomanou H, Aignesberger A, Linke RP. Biol Chem Hoppe Seyler 1986;367:879-90. [23] Schnabel D, Schr6der M, Sandhoff K. FEBS Lett 1991;284:57-9. [24] Bradova V, Smid F, Ulrich-Bott B, Roggendorf W, Paton BC, Harzer K. Hum Genet 1993;92:143-52. [25] Klein A, Henseler M, Klein C, Suzuki K, Harzer K, Sandhoff K. Biochem Biophys Res Commun 1994;200:1440-8. [26] Ho MW, O'Brien JS. Proc Natl Acad Sci USA 1971;68:2810-3. [27] Sandhoff K, Kolter T. Naturwissenschaften 1995;82:403-13. [28] Conzelmann E, Sandhoff K. Dev Neurosci 1983/84;6:58-71. [29] Leinekugel P, Michel S, Conzelmann E, Sandhoff K. Hum Genet 1992;88:513-23. [30] Sango K, Yamanaka S, Hoffmann A, Okuda Y, Grinberg A, Westphal H, McDonald MP, Crawley JN, Sandhoff K, Suzuki K, Proia RL. Nature Genet 1995;11:170-6.
Clinica Chimica Acta 266 (1997) 51-61
Biochemistry of glycosphingolipid degradation Konrad Sandhoff*, Thomas Kolter Kekuld-lnstitut fiir Organisehe Chemie und Biochemie der Universitiit, Gerhard-Domagk-Str. 1, 53121 Bonn, Germany
Received 1 September 1996; accepted 23 July 1997
Abstract
Glycosphingolipids (GSLs) form cell-type-specific patterns on the surface of eukaryotic cells. Degradation of GSLs requires endocytotic membrane flow of plasma membrane-derived GSLs into the lysosomes as the digesting organelles. Recent research focused on the mechanisms leading to selective membrane degradation in the lysosomes and on the mechanism and physiological function of sphingolipid activator proteins, which are needed for degradation of GSLs with short oligosaccharide chains in addition to hydrolysing enzymes. Both, the inherited deficiency of lysosomal hydrolases and of sphingolipid activator proteins give rise to sphingolipid storage diseases. In some cases it was possible to correlate residual enzyme activities with the onset and the course of the disease. © 1997 Elsevier Science B.V. Keywords: Sphingolipid activator proteins; Glycosphingolipids; Lysosomes; Sphingolipidoses; Storage diseases
1. Introduction
Glycosphingolipids (GSLs) are components of the plasma membrane of eukaryotic cells [ 1]. Each GSL contains a hydrophobic ceramide moiety that acts as membrane anchor and a hydrophilic, extracellular oligosaccharide chain. Variations in the type, number and linkage of sugar residues in the oligosaccharide chain give rise to the wide range of naturally occurring GSLs. These Corresponding author. Tel./fax: + 49 228 737778; e-mail: [email protected] 0009-8981/97/$17.00 © 1997 Elsevier Science B.V. All rights reserved. P l l S0009-8981 (97)00166-6
52
K. Sandhoff, T. Kolter / Clinica Chimica Acta 266 (1997) 5 1 - 6 1
form cell-type-specific patterns at the cell surface which change with cell growth, differentiation, viral transformation and oncogenesis [2]. GSLs interact at the cell surface with toxins, viruses, and bacteria as well as with membrane bound receptors and enzymes [3]. They are also involved in cell-type-specific adhesion processes and, in addition, lipophilic products of GSL metabolism such as sphingosine and ceramide, play a role in signal transduction events [4]. Finally, GSLs form a protective layer on biological membranes protecting them from inappropriate degradation. Degradation of GSLs occurs in lysosomal compartments within the cell [5]. The plasma membranes containing GSLs destined for degradation are endocytosed and travel through the endosomal compartments to the lysosome. Here, hydrolysing enzymes cleave sugar residues sequentially to produce ceramide which is deacylated to sphingosine. This can leave the lysosome, reenter the biosynthetic pathway or be further degraded. To date, all the inherited diseases of GSL metabolism that have been described affect GSL degradation, and none have been described that affect GSL biosynthesis. Presumably, severe disorders affecting GSL biosynthesis would result in incorrect GSL patterns on the cell surface, and these in turn would affect morphogenesis and embryogenesis. A study of the inherited disorders in degradation, the lysosomal storage diseases, has provided insights into the processes involved in GSL degradation.
2. Topology of endocytosis and lysosomal digestion Catabolism of complex GSLs derived from the plasma membrane occurs in the lysosomes after endocytosis. In the conventional model, components of plasma membranes reach the lysosomal compartment by endocytic membrane flow via the early and late endocytic reticulum [6] (Fig.). During this vesicular flow, molecules in the membrane are subjected to a sorting process which directs some of the molecules to the lysosomal compartment, some others to the Golgi apparatus and yet others even back to the plasma membrane. After a series of vesicle budding and fusion events through the endosomal compartment, the membrane fragments reach the lysosomal compartment. As a consequence of endocytic membrane flow, the plasma membrane fragments are incorporated into the lysosomal membrane. Lysosomal degradation of former components of the plasma membrane must then proceed selectively within the lysosomal membrane without destroying it. It is difficult to see how this could occur, especially as the inner leaflet of the lysosomal membrane is covered with a thick glycocalix composed of glycoproteins. In an alternative model for the topology of endocytosis [5,7] (Fig. 1), components of the plasma membrane pass through the endosomal compartment as intraendosomal vesicles that become intralysosomal vesicles on reaching the
K. Sandhoff, T. K o l t e r / Clinica C h i m i c a A c t a 2 6 6 ( 1 9 9 7 ) 5 1 - 6 1
A
B 'l l t
-
PM
\
/ Go!glapparatus ~
Late endosorr~e~
~
Transient fusion
G~ccallx
53
Transient fusionand transfer
I r'~rr'~ar~,
~rr~ rn~'r~ane
Fig. 1. Two models for the topology of endocytosis and lysosomal digestion of glycosphingolipids (GSLs) derived from the plasma membrane. Conventional model (A): degradation of GSLs derived from the plasma membrane occurs selectively within the lysosomal membrane. Alternative model for the topology of endocytosis and digestion of GSLs (B): during endocytosis glycolipids of the plasma membrane become incorporated into the membranes of intraendosomal vesicles (multivesicular bodies). The vesicles are transferred into the lysosomal compartment when the late endosome fuses with primary lysosomes. PM, plasma membrane; ~, glycosphingolipid [5].
lysosomes. The vesicles are initially formed in the early endosome by selective invagination (budding-in) of endosomal membranes enriched for components of the plasma membrane. The surrounding endosome then passes along the endocytic pathway by the normal, successive events of membrane fission and fusion. The intraendosomal vesicles, however, are carried along as passengers. When the vesicle reaches the lysosome, glycoconjugates originating from the outer leaflet of the plasma membrane face the lysosol on the outer leaflet of intralysosomal vesicles and are thus in the correct place for degradation. In this model there is no glycocalix blocking the action of hydrolases. This hypothesis is supported by a series of observations: 1. multivesicular bodies are observed at the level of the early and late endosomal reticulum (reviewed in [5]); 2. the epidermal growth factor receptor derived from the PM and internalized into lysosomes of hepatocytes is not integrated into the lysosomal membrane [8];
54
K. Sandhoff, T. Kolter / Clinica Chimica Acta 266 (1997) 5 1 - 6 1
3. spherical multivesicular bodies are the predominant endocytic compartments in HEp-2 cells. Moreover, these multivesicular bodies mature within 60-90 min into lysosomes which still contain internal vesicles [9]; 4. multivesicular storage bodies accumulate in cells from patients with sphingolipid storage diseases. They also occur in cells, e.g. Kupffer cells and fibroblasts, of a patient with another sphingolipid storage disease, a combined activator protein deficiency [10,11]. Analysis of cultivated fibroblasts of this patient shows that the storage vesicles refer to late endosomal or lysosomal compartments. These compartments are still functionally active, except their inability to degrade sphingolipids with short oligosaccharide head groups. After complementation of the medium of these cells with the missing SAP-precursor, not only the degradation block is abolished but also the morphological shape of the multivesicular bodies is reversed [12]. To date, budding-in of vesicles on the stage of the early endosome is not observed by microscopy. This membrane fission event is postulated to take place since this would explain both the occurrence of intralysosomal vesicles and the selective degradation of glycosphingolipids derived from the plasma membrane in contrast to those of the lysosomal membrane. Within the lysosome, degradation of GSLs occurs by the step-wise action of specific acid exohydrolases [5]. More than 10 different exohydrolases are involved in this process. If any of these enzymes is deficient, the corresponding lipid substrate accumulates and is stored in the lysosomal compartment. For glycosphingolipids with long carbohydrate chains of more than four sugar residues the presence of an enzymatically active exohydrolase is sufficient for degradation in vivo. Detailed biochemical analysis of sphingolipidoses such as metachromatic leukodystrophy and Tay-Sachs disease demonstrated that the in vivo degradation of some glycolipids involves not only exohydrolases but also a second type of effector molecule [13]. In these cases lipid storage is observed in patients without any enzyme deficiency, but with deficiency of a protein cofactor. The degradation of membrane bound GSLs with short oligosaccharide chains requires the cooperation of an exohydrolase and a sphingolipid activator protein. Several sphingolipid activator proteins are now known including the GM2 activator and the saposins. A detailed study of these proteins has illuminated some of their main features and indicated the mechanisms by which they facilitate digestion of GSLs in the lysosome.
3. The GM2 activator and its role in lysosomal digestion The enzymatic degradation of ganglioside GM2, the main storage material in Tay-Sachs disease, requires [~-hexosaminidase A and a lysosomal ganglioside
K. Sandhoff, T. Kolter / Clinica Chimica Acta 266 (1997) 5 1 - 6 1
55
binding protein, the GM2 activator [14,15]. The GM2 activator binds ganglioside GM2 as well as related gangliosides forming water-soluble complexes (mostly in a 1:1 molar ratio). In vitro it functions as a ganglioside transfer protein transferring gangliosides from donor membranes to acceptor membranes. The GM2 activator behaves as a 'liftase'. It recognizes GM2 within the membrane and by binding to it lifts the lipid out of the bilayer and presents it to the water soluble [3-hexosaminidase A for degradation. However, it is also possible that the activator-lipid complex leaves the membrane and the enzymatic reaction takes place in free solution. Formation of the ternary complex presumably involves also a protein-protein interaction between the GM2 activator and [3-hexosaminidase A [16]. Point mutations within the structural gene of the GM2 activator have been identified in four patients with the AB-variant of GM2 gangliosidosis, e.g. [17]. The mutated proteins are proteolytically labile so that some of them are degraded in the ER, others in the Golgi compartment. The resulting loss of the GM2 activator causes a block within the GM2 degradation in the lysosomes as demonstrated in metabolic studies in cultured fibroblasts of the patients. The block can be bypassed by feeding native GM2 activator to the culture medium of the mutant fibroblasts [18].
4. Other activator proteins The first activator protein was identified in 1964 as a protein which is necessary for the hydrolytic degradation of glycosphingolipids carrying a sulphuric ester group (sulfatides) by lysosomal arylsulfatase A [19]. This sulfatide activator, SAP-B (saposin B), is a small lysosomal glycoprotein consisting of 80 amino acids, with an N-linked carbohydrate chain and three disulfide bridges [20]. Like the GM2 activator it binds GSLs but has a broader specificity. In vitro it behaves similarly to the GM2 activator in some aspects, i.e. it can recognize and bind several different GSLs on the surface of micelles by forming a stoichiometric complex and is then able to transfer the GSLs to the membranes of acceptor liposomes [13]. SAP-B can also present GSLs to water-soluble enzymes as substrates [21]. The inherited deficiency of the sulfatide activator leads to a lysosomal storage disease which resembles metachromatic leukodystrophy. However, unlike typical metachromatic leukodystrophy not only sulfatide but also additional glycolipids, e.g. globotriaosyl ceramide, accumulate owing to a blockage of degradation at several points in the catabolic pathway [5]. The sulfatide activator has sequence homology with three further activator proteins: the Gaucher factor (SAP-C), SAP-A and SAP-D. Molecular biological analysis has revealed that these four activator proteins are formed by proteolytic
56
K. Sandhoff, T. Kolter / Clinica Chimica Acta 266 (1997) 5 1 - 6 1
processing of a common precursor protein, the SAP-precursor (reviewed in [13]). The four activator proteins, SAP-A, B, C and D show homology to each other and have similar properties, but differ in their function and their mechanism of action [13]. The physiological function of these sphingolipid activator proteins is only partially clarified to date; most of our knowledge on this has emerged from studies on patients with atypical lipid storage diseases. As mentioned above, an inherited deficiency of SAP-B leads to an atypical form of metachromatic leukodystrophy. Similarly, SAP-C deficiency causes an atypical form of Gaucher disease, where glucosyl ceramide accumulates [22,23]. In one patient with a complete deficiency of the whole SAP-precursor protein due to a homoallelic mutation within the start codon (AUG --> UUG) [11] there is simultaneous storage of many sphingolipids, including ceramide, glucosyl ceramide, lactosyl ceramide, ganglioside GM3, galactosyl ceramide, sulfatides, digalactosyl ceramide and globotriaosyl ceramide [24]. The storage of GSLs can be demonstrated in cultured mutant cells by pulse-chase experiments [25]. Storage substances in cells from different patients can be visualized by biosynthetic labelling of glycolipids with [14C]serine followed by long chase periods. As previously mentioned, the glycolipids that accumulate reflect the types of GSLs synthesized in each cell type. In the case of fibroblasts, glucosylceramide accumulates in cells from patients with SAP-C deficiency, ganglioside GM2 accumulates when there is deficiency of the GM2 activator and ceramide, glucosylceramide, lactosylceramide and ganglioside GM3 all accumulate when the SAP precursor is deficient. Cell culture models such as these are not only suitable for diagnostic purposes, but also allow the elucidation of the function of the different activator proteins. Although, the four SAPs show great homology, they act very differently in these experiments. If purified SAP-B is fed to the cells with SAP-precursor deficiency, the storage of lactosylceramide is abolished but ceramide and glucosylceramide still accumulate. In contrast, when SAP-D is fed to the cells ceramide accumulation is prevented. These experiments give the first hint to the in vivo function of this activator protein [25]. As expected, if the mutant cells are fed with the SAP-precursor in nanomolar concentrations storage of all these glycolipids is abolished and the aberrant accumulation of intralysosomal membrane structures prevented (see above). These studies on cells in culture from patients with different storage diseases support our hypothesis concerning the function of sphingolipid activator proteins. In their absence, membrane bound GSLs with short oligosaccharide chains are not readily accessible for digestion by water soluble exohydrolases. By analogy to a razor blade, the hydrolases can only attack oligosaccharide chains of GSLs that protrude far enough from the membrane into the aqueous phase. This is no problem for long chain oligosaccharides but for the degradation of GSLs with short sugar chains the assistance of activator proteins is
K. Sandhoff, T, Kolter I Clinica Chimica Acta 266 (1997) 51-61
57
required. Some of them act as binding proteins or liftases, forming a watersoluble complex with the lipid and hence raising the lipid out of the membrane. However, the function of activator proteins is not necessarily restricted to binding lipids in this way. For example, SAP-C can directly activate glucosylceramide [3-glucosidase [26] and the GM2 activator specifically interacts with [3-hexosaminidase A [ 16]. The simultaneous deficiency of the four sphingolipid activator proteins, as seen in the patient described earlier, causes an intraendosomal and intralysosomal accumulation of vesicles and other membrane fragments (multivesicular bodies) and supports our view of the topology of endocytosis and lysosomal digestion. Furthermore, the occurrence of storage bodies in the combined activator protein deficiency suggests a possible function of glycolipids. The increased concentrations of glycolipids within distinct membranes appears to protect the vesicles from degradation by phospholipases, proteases and other hydrolases that occur in high concentration in the lysosomes.
4.1. Mechanisms of the pathogenesis of lipidoses The pathogenesis of inherited disorders is of a complex nature and not fully understood for any disease. This is also valid for lipidoses although they are guided by simple biochemical principles [27]. On the one hand, mutations in genes coding for different enzymes can lead to diseases with similar or even identical phenotype. For example, deficiencies of the a-chains in 13-hexosaminidase A and S or of the [3-chains in [3-hexosaminidase B and A or a deficiency in the GM2 activator protein give rise to similar clinical phenotypes that were characterized before as amaurotic idiocy. On the other hand mutations in the same structural gene can cause different clinical phenotypes which in some cases are known under different eponyms. For example, severe forms of c~-iduronidase deficiency are known as Hurler disease whereas less severe forms are called Scheie disease. Even patients with identical mutations within the same structural gene e.g. that one coding for arylsulfatase A can show different clinical forms, obviously due to a differing genetic background. Despite this enormous heterogeneity and only indirect correlation between genotype and phenotype of a disease it is possible to state some factors that connect genotype and phenotype of sphingolipidoses. 1. Due to the cell-type-specific expression of GSLs, lipid storage in lipidosis patients is observed especially in those cells and organs, in which the lipid substrates of the corresponding deficient enzyme are prevalently synthesized (e.g. complex gangliosides in neuronal cells) or by which they are taken up by phagocytosis (glucosylceramide and its precursors in macrophages in case of Gaucher disease).
58
K. Sandhoff, T. Kolter / Clinica Chimica Acta 266 (1997) 51-61
2. Mutations in proteins of GSL degradation give rise to disorders in GSL turnover. Although having different effects on the corresponding proteins, t h e y can be treated as Vmax mutations that finally reduce the catabolic turnover of substrates [28]. 3. Impeded degradation can cause the accumulation of morphologically active or toxic compounds. 4.2. The extent of degradation disorders causes different clinical courses of disease
In normal ceils the maximum turnover rate of a catabolic system (Vmax) is always higher than the influx rate (vi) into the lysosomal compartment of the substrates to be degraded. Only when the maximum turnover rate becomes smaller than the influx rate, an irreversible substrate storage occurs and it becomes a disease. Due to homoallelic or heteroallelic mutations, the degrading system can approach different residual activities which should correspond to the pathogenesis of the disease. The mutated proteins and the resulting maximum enzyme activities Vmax constitute a decisive connection between genotype and phenotype of a disease. Starting with the Michaelis-Menten equation, the steady state substrate concentration within the lysosome can be calculated as a function of the residual enzyme activity [28]. In normal cells, substrate concentrations are usually far below the Michaelis-Menten constant, i.e. the substrate concentration at which a half maximal turnover rate is achieved. A decrease of the maximum turnover rate of the degrading system on values of e.g. 50 to 20% of normal cells, a typical range for heterozygote carriers of inherited diseases, does not influence the turnover rate v. The ratio v/v i remains constant, since lowering of maximal enzyme activity is compensated by a higher substrate saturation of the degrading enzyme (Fig. 2 [27]). This compensatory mechanism works until the maximal turnover rate decreased to the value of substrate influx into the lysosome. At this critical threshold value all enzyme molecules occur in the form of enzyme substrate complexes. The decrease of residual enzyme activities to this value should not lead to an irreversible pathological substrate storage. This explains that probands with pseudodeficiencies, e.g. a mean loss of enzyme activity of up to 90% of [3-hexosaminidase A or arylsulfatase A do not fall ill. Only the decrease of the maximal catabolic enzyme activity (Vmax) below the critical threshold value vi causes storage of the corresponding lipid substrate. So the model predicts that the decrease of the maximal catabolic enzyme activity to the calculated threshold value does not influence the flux rate of the substrate and that pathological storage occurs only below this level. We examined this model by measuring flux rates and enzyme activities in cultured skin fibroblasts from patients with GM2 gangliosidoses [29]. The experimentally observed values support our hypotheses since the flux rate decreases linearly with
59
K. Sandhoff, T. Kolter / Clinica Chimica Acta 266 (1997) 51-61
x
]
xx
x
~
x
K
xX~
aa
RH
a
a
o
o
Q
a
a ma
"
~-.=_ o
200
400
Hexosamlnldase
600
A activity
800 (pmol
/ h x mg
1000
1200
x AU)
i
Z
i
-~ls-
i
i i
0
Critical
threshold
activity
...
er,J
~,
......................................................
10-
• i
:>
,.
.~
1 .0
I
_ __
Hypothetical
solubility
of s u b s t r a t e
0.5
W
,
~
,
~
,.
M
I--
i
I
1'0 Enzyme
activity
1'5
2'0
Vm=x/V,
Fig. 2. Theoretical correlation between the residual activity of an enzyme and the steady state concentration and turnover rate of its substrate in a defined compartment [28]. The substrate concentration is expressed as multiples of K m turnover rate and enzyme activity (v,.ax) as multiples of the influx rate vi. Above the critical threshold activity (Vm,x > Vi), the turnover rate is limited by the influx rate; below this threshold (Vmax < Vi) it is limited by the remaining capacity of the deficient enzyme; in the latter case, the substrate accumulates at a rate of va¢c = vi - Vma~. - - , steady state substrate concentration; . . . . . . . . , turnover (flux) rate; . . . . . . . . . . critical threshold activity; . . . . . . , critical threshold activity, taking the limited solubility of the substrate into account. Above: residual activity of [3-hexosaminidase A and turnover of its substrate ganglioside GM2 in cell culture. Skin fibroblasts from normal probands, from patients with various forms of GM2 gangliosidosis and from heterozygotes were fed with [3H]ganglioside GMZ for 3 days in culture. Cells were then harvested, homogenized in water and assayed for the following three parameters: incorporation of the substrate, percentage of degraded substrate, activity of J3-hexosaminidase A towards ganglioside GM2 in the presence of the GM2 activator protein [29]. Filled circles, a-chain deficiency (B-variant of GM2 gangliosidosis, Tay Sachs disease), infantile type; empty circles, a-chain deficiency, juvenile type; filled squares, a-chain deficiency, adult type; filled triangles, activator deficiency (AB-variant of GM2 gangliosidosis); healthy probands, x, obligate heterozygotes from GM2 gangliosidosis; empty squares, normal controls.
60
K. Sandhoff, T. Kolter / Clinica Chimica Acta 266 (1997) 51-61
decreasing e n z y m e activity only below a threshold value. Patients with the adult course of the disease show a significantly higher turnover than patients with the juvenile course and these a higher turnover than patients with the infantile course. Smaller differences in the residual e n z y m e activities correspond to significant differences in the respective flux rates.
4.3. A n i m a l
models
In the meantime, animal models o f the G M 2 gangliosidoses have b e c o m e available and permit further studies o f the pathogenesis o f this disease. In contrast to the human diseases, the mouse models of B-Variant and 0-Variant show very different neurological phenotypes. The Tay-Sachs model showed no neurological abnormalities while the Sandhoff model was severely affected. This phenotypic difference between the two mouse models is due to differences in the ganglioside degradation pathway between mice and humans [30].
Acknowledgements The work performed in the laboratory of the authors was supported by the Deutsche Forschungsgemeinschaft (SFB 284).
References [1] Wiegandt, H. In: Neuberger A, van Deenen LLM, editors. New comprehensive biochemistry, vol. 10. Amsterdam: Elsevier, 1985:199-260. [2] Hakomori S. Annu Rev Biochem 1981;50:733-64. [3] Karlsson K-A. Annu Rev Biochem 1989;58:309-50. [4] Hannun YA, Obeid LM. Trends Biochem Sci 1995;20:73-7. [5] Sandhoff K, Kolter T. Trends Cell Biol 1996;6:98-103. [6] Griffiths GW, Hoflack B, Simons K, Mellman IS, Kornfeld S. Cell 1988;52:329-41. [7] FiJrst W, Sandhoff K. Biochim Biophys Acta 1992;1126:1-16. [8] Renfrew CA, Hubbard AL. J Biol Chem 199l;266:21265-73. [9] van Deurs B, Holm PK, Kayser L, Sandvig K, Hansen SH. Eur J Cell Biol 1993;61:208-24. [10] Harzer K, Paton BC, Poulos A, Kustermann-Kuhn B, Roggendorf W, Grisar T, Popp M. Eur J Pediatr 1989;149:31-9. [ 11] Schnabel D, Schr6der M, Fiirst W, Klein A, Hurwitz R, Zenk T, Weber J, Harzer K, Paton BC, Poulos A, Suzuki K, Sandhoff K. J Biol Chem 1992;267:3312-5. [12] Burkhardt JK, Hiittler S, Klein A, M6bius W, Habermann A, Griffiths G, Sandhoff K. Eur J Biochem 1997;73:10-8. [13] Sandhoff K, Harzer K, Fiirst W. In: Scriver C, Beaudet AL, Sly WS, Valle D, editors. The metabolic and molecular basis of inherited disease. New York: McGraw Hill, 1995:2427-41.
K. Sandhoff, T. Kolter / Clinica Chimica Acta 266 (1997) 51-61
61
[14] Gravel RA, Clarke JTR, Kaback MM, Mahuran D, Sandhoff K, Suzuki K. In: Scriver C, Beaudet AL, Sly WS, Valle D, editors. The metabolic and molecular basis of inherited disease. New York: McGraw Hill, 1995:2839-79. [15] Meier EM, Schwarzmann G, Fiirst W, Sandhoff K. J Biol Chem 1991;266:1879-87. [16] Kytzia H-J, Sandhoff K. J Biol Chem 1985;260:7568-72. [17] Schepers U, Glombitza G, Lemm T, Hoffmann A, Chabas A, Ozand P, Sandhoff K. Am J Hum Genet 1996;59:1048-56. [18] Klima H, Klein A, van Echten G, Schwarzmann G, Suzuki K, Sandhoff K. Biochem J 1993;292:571-6. [19] Mehl E, Jatzkewitz H. Hoppe-Seyler's Z Physiol Chem 1964;339:260-76. [20] FiJrst W, Schubert J, Machleidt W, Meyer HE, Sandhoff K. Eur J Biochem 1990;192:709-14. [21] Li S-C, Sonnino S, Tettamanti G, Li Y-T. J Biol Chem 1988;263:6588-91. [22] Christomanou H, Aignesberger A, Linke RP. Biol Chem Hoppe Seyler 1986;367:879-90. [23] Schnabel D, Schr6der M, Sandhoff K. FEBS Lett 1991;284:57-9. [24] Bradova V, Smid F, Ulrich-Bott B, Roggendorf W, Paton BC, Harzer K. Hum Genet 1993;92:143-52. [25] Klein A, Henseler M, Klein C, Suzuki K, Harzer K, Sandhoff K. Biochem Biophys Res Commun 1994;200:1440-8. [26] Ho MW, O'Brien JS. Proc Natl Acad Sci USA 1971;68:2810-3. [27] Sandhoff K, Kolter T. Naturwissenschaften 1995;82:403-13. [28] Conzelmann E, Sandhoff K. Dev Neurosci 1983/84;6:58-71. [29] Leinekugel P, Michel S, Conzelmann E, Sandhoff K. Hum Genet 1992;88:513-23. [30] Sango K, Yamanaka S, Hoffmann A, Okuda Y, Grinberg A, Westphal H, McDonald MP, Crawley JN, Sandhoff K, Suzuki K, Proia RL. Nature Genet 1995;11:170-6.