Eur. J. Biochem. 200, 591 -597 (1991) 0 FEBS 1991
001429569100573D
Glycosphingolipid specifity of the human sulfatide activator protein A. VOGEL, G. SCHWARZMANN and K. SANDHOFF Institut fur Organische Chemie und Biochemie der Universitat Bonn, Federal Republic of Germany Received January 28/April 3, 1991) - EJB 91 0145
The interaction of the sulfatide activator protein with different glycosphingolipids has been studied in detail. The following findings were made. 1. The sulfatide activator protein forms water-soluble complexes with sulfatides [Fischer, G. and Jatzkewitz, H. (1977) Hoppe-Seyler’s Z. Physiol. Chem. 356, 6588 - 65911 and various other glycosphingolipids. 2. In the absence of degrading enzymes the activator protein acts in vitro as a glycosphingolipid transfer protein,transporting glycosphingolipids from donor to acceptor liposomes. Lipids having less than three hexoses, e.g. galactosylceramide, sulfatide and ganglioside GM3were transferred at very slow rates, whereas complex lipids such as gangliosides GMZ, GMland GDla were transferred much faster than the former. The transfer rate increased with increasing length of the carbohydrate chain of the lipid molecules. 3. Both the acyl residue in the ceramide moiety and the nature of the carbohydrate chain are significant for recognition of the glycosphingolipids by the sulfatide activator protein. Apparently, both residues serve as an anchor and the longer they are the better they are recognized by the protein. 4. In the absence of activator protein, degradation rates of sulfatide derivatives by arylsulfatase A, and of ganglioside GM1 derivatives by P-galactosidase, increase with decreasing length of acyl residues in their hydrophobic ceramide moiety. Addition of activator protein stimulates the degradation of only those GM1 and sulfatide derivatives that have long-chain fatty acids in their hydrophobic ceramide anchor.
Glycosphingolipids are degraded to ceramide by the sequential action of lysosomal exohydrolases [2]. In addition, the physiological degradation of glycosphingolipids with oligosaccharide chains up to a length of three hexoses requires the assistance of water-soluble, nonenzymatic glycoproteins, so-called sphingolipid activator proteins [3]. Genetic defects Correspondence to K. Sandhoff, Institut fur Organische Chemie of these lysosomal activator proteins result in sphingolipid und Biochemie der Universitat Bonn, Gerhard-Domagk-Strasse 1, storage diseases [4- lo]. While the human GM2 activator W-5300 Bonn 1, Federal Republic of Germany (SAP-3) specifically stimulates the hydrolysis of ganglioside Abbreviations. Cer, ceramide or N-acylsphingoid; Sph, D-erythroand GA2by hexosaminidase A [I 11, the sulfatide activator sphingosine; NeuAc, N-acetylneuraminic acid; GalNAc, 2-acet- GM2 amido-2-deoxy-~-galactose; GalCer, galactosylceramide; GlcCer, (SAP-I) furthers unspecifically the degradation of different glucosylceramide; LacCer, lactosylceramide; sulfatide or 3’-O-sulfo- glycosphingolipids in vitro [12,13]. In the test tube the sulfatide galactosyl-ceramide; globotriaosylceramide or GbOse3Cer or Galcll activator can stimulate the hydrolysis of sulfatides by + 4Galpl 4Glc/l1+ 1Cer; lysosulfatide or 3’-O-SUlfO-galaCtOSy1-Darylsulfatase A, ganglioside GMlby f3-galactosidase,GbOse3erythro-sphingosine; C,-sulfatide, derivative of sulfatide containing Cer by a-galactosidase A [12,14] and some gangliosides (Gola, acyl chains with x C atoms (x = 2, 6, 18); G M ~I13NeuAc-LacCer , or GTlb,GM3)by lysosomal sialidase [15]. Gal(3+2clNeuAc)~l+4Glc~I+lCer; G M Z , I13NeuAc-GgOse3Cer Fischer and Jatzkewitz [16] observed that the sulfatide or GalNAcBl + 4Ga1(3 + 2clNeuAc)pl + 4Glcj31 + ICer; GMl, I13NeuAc - GgOse4Cer or GalBl + 3GalNAcpl + 4Ga1(3 + activator extracts sulfatides from micelles and forms water2aNeuAc)pl 4Glc/l1 + 1Cer; GDla, IV3NeuAc, I13NeuAc - soluble sulfatide - activator protein complexes which are then GgOse4Cer or Gal(3 + 2clNeuAc)jll + 3GalNAcbl + 4Ga1(3 c degraded by arylsulfatase A. To gain further insight into the 2aNeuAc)pl + 4GlcpI + 1Cer; GTlb, IV3NeuAc,I13(NeuAc)2- specificity of the sulfatide activator and its interaction with GgOse4Cer or Gal(3 + 2clNeuAc)pl + 3GalNAcB1 + 4Ga1(3 + glycosphingolipids, we studied the ability of the protein to 2clNeuAc8 + 2clNeuAc)pl + 4Glc/lI + Cer; GA1, GgOse4Cer or act as a glycosphingolipid transfer protein, stimulating the Gal~l+3GalNAc/31+4Gal/lI +GlcpI+ 1 Cer; G A Z , GgOse3 Cer or GalNAcj?l+4Galfll+4Glcfll +1Cer, GMl alcohol, ganglioside with transfer rates of various glycosphingolipids from donor to the carboxyl group reduced to a hydroxymethyl group; lyso-GMl, acceptor liposomes. Since there is evidence that the hydroI13NeuAc-GgOse,Sph; deacylated GM1, lyso-GM1which lacks the N- phobic ceramide moiety of sulfatides, especially the acyl chain, is important for the protein-lipid interaction [17], we also acetyl group in the sialic acid moiety or I13Neu-GgOse4Sph, C,-GM1 derivative of GMl containing acyl chains with x C atoms (x = 2, 8, analyzed the enzymic degradation of synthetically modified 18). glycosphingolipids containing oligosaccharide and acyl chains Enzymes. 8-Galactosidase (EC 3.2.1.23); arylsulfatase of different lengths.
Glycosphingolipids are components of the outer leaflet of plasma membranes. They form cell-type-specific patterns on the surfaces of mammalian cells. These patterns change characteristically with differentiation and oncogenic transformation [l].
--f
--f
(EC 3.1.6.1).
592 MATERIALS AND METHODS Protein purification The sulfatide activator protein was purified from human kidney according to procedures described previously [3, 51 using acid precipitation, heat precipitation, anion-exchange chromatography on DEAE-cellulose, gel filtration, isoelectric focussing, hydrophobic chromatography on octyl-Sepharose and anion-exchange chromatography and gel filtration on HPLC columns. The preparation was pure according to electrophoresis and Western blot analysis. The purified protein gave one major single band on SDSjPAGE (not shown) and its molecular mass was determined as 10 kDa by extrapolation from the mobilities of molecular mass markers. Protein was determined according to Lowry et al. [18] using bovine serum albumin as standard. In the presence of other proteins, activator proteins are generally quantified by measuring their capability to accelerate in vivo the hydrolysis of glycosphingolipid substrates by the respective hydrolases. Thus, one unit of sulfatide activator (AU) is defined as the amount of activator that stimulates the enzymatic hydrolysis of ganghoside GMl by 1 nmol x h-' x (U P-galactosidase)-' under assay conditions where the reaction depends almost linearly on the activator concentration [3]. The sulfatide activator used in this work had a specific activity of 208 500 4500 AU/mg protein. P-Galactosidase was purified from human liver by the method of Miller et al. [19], arylsulfatase A from human liver was purified to electrophoretic homogeneity by the method of Fischer and Jatzkewitz [20].P-Galactosidase and arylsulfatase A had specific activities of 1.1 and 3.85 units/mg protein, respectively. One unit of P-galactosidase ciitalyzes the release of P-galactose from 1 pmol 4-methylumbelliferyl-P-~galactopyranoside/min under standard conditions [21]. One unit of arylsulfatase A catalyzes the release of sulfate from 1 pmol p-nitrocatecholsulfate/min under conditions described [161. Lipid.?
silica-gel-coated thin layer plates and anisaldehyde were obtained from Merck (Darmstadt, FRG). All other chemicals were of analytical grade or highest purity available. Activator assays The stimulation of the enzymatic degradation of ganglioside GM1 was measured as described previously [3]. The assays contained lipid ( 5 nmol), P-gdlactosidase (1.9 mu), bovine serum albumin (30 pg), and detergent (up to 100 nmol) or activator protein (up to 0.1 nmol) in a total volume 50 pl of 50 mM citrate, pH 4.0. After incubation for 1 h at 37"C, the vials were transferred to an ice bath and 1 ml 1 mM galactose solution was added. For the separation of released [3H]galactose from GMl and GM2, the sample was loaded onto small DEAE-cellulose columns (1 ml), and the columns were washed twice with 1 ml 1 mM galactose solution. The combined effluents containing the liberated [3H]galactose were collected in scintillation vials and their radioactivity quantified by liquid scintillation counting. For LacCer and the other GM1 derivatives the assay was modified as follows. For C Z - G Mand ~ Cs-GMl degradation, the glycosphingolipids were purified using small (0.4 ml) Lichroprep RP-18 columns. Prior to loading the columns were washed twice with chloroform/methanol (2/1, by vol.), once with methanol and once with synthetic upper phase (chloroform/methanol/O.l M KCl, 3/48/47, by vol.). The lyophilized lipid mixture ( 5 nmol) of the assay was redissolved in 4.9 ml synthetic upper phase and passed over an RP-18 column (0.4 ml). After washing with water (2.5 ml), the lipids were eluted with 0.5 ml methanol and 2.5 ml chloroform/methanol ( l / l , by vol.). Since LacCer and GlcCer separation by thinlayer chromatography was impaired by taurodeoxycholate, these glycosphingolipids were freed of this detergent by passing the incubation mixture through a DEAE-cellulose column (0.5 ml) using 4 ml methanol. Glycosphingolipids were separated by thin-layer chromatography (solvents: chloroform/methanol/water, 60/40/9, by vol., for C2-GMl,GAl;n-butanol/ethanol/water 2/1/1, by vol., for lyso-GMl; chloroform/methanol/water, 60/25/4, by vol., for LacCer; chloroform/methanol/l5 mM CaCl,, 60/35/8, by vol., for GM1 alcohol, C8-GMl). After staining with anisaldehyde/sulfuric acid/acetic acid [29], the glycosphingolipid bands were quantified by scanning the thin-layer plates at 580 nm with a densitometer (Shimadzu CS-910). For the determination of stimulation of degradation of sulfatide derivatives lipid (5 nmol), arylsulfatase A (1 mu), bovine serum albumin (10 pg), and detergent (100 nmol) or activator protein (0.1 nmol), were incubated in a total volume of 50 pl of 10 mM sodium acetate, 100 mM NaC1, pH 5.0, for 1 h at 37°C. Assays containing 3H-labeled CI8- or C6sulfatide were quantified as described previously [3, 51. The assays containing [3H]C2-sulfatide or lysosulfatide were lyophilized and products and educts separated by thin-layer chromatography (solvents: for lysosulfatide; chloroform/ methanol/water/25% NH3, 70/30/4/1, by vol. ; for C2sulfatide, chloroform/methanol/lS mM CaC12, 60/35/8, by vol.). Lysosulfatide was quantified at 580 nm as described above, [3H]C2-sulfatide was quantified using a radioscanner (Berthold, W-7547, Wildbad 1).
Ganglioside GM1 was a gift from Fidia (Abano, Italy) and tritiated in the galactose moiety by the galactose oxidase/ NaB3H4 method according to Leskawa et al. [22]; its specific activity was 77.7 TBq/mol (2120 Ci/mol). Ganglioside GM2 was isolated from Tay-Sachs brain by the method of Svennerholm [23], and tritiated in the N-acetylgalactosamine moiety by the galactose oxidase/NaB3H4 method according to Suzuki and Suzuki [24]; its specific radioactivity was 440 GBq/mol (12 Ci/mol). GalCer and sulfatide from postmortem human brain, as well as GlcCer and GM3from postmortem human spleen, were tritiated as described [25]. The specific radioactivities for GalCer, sulfatide, GlcCer and GM3 were 2 (54), 7 (190), 3.7 (100) and 12.1 (330) TBq (Ci)/mol, respectively. Ganglioside GDla was isolated from a mixture of bovine brain gangliosides and tritiated in the sphingosine moiety [25] to a specific radioactivity of 25 TBq/mol(675 Ci/ mol). GMM,-alcohol was synthesized as described [26]. LysoGM1and C2-GMlwere synthesized as described [27, 281. C8GMl,[3H]C2-sulfatideand C6-sulfatide, [3H]lysosulfatide and LacCer were available in our laboratory. Egg phosphatidylcholine, a-tocopherol and cholesterol were purchased from Sigma (Miinchen, FRG), ['4C]dipalmitoylglycerophosCentrifugation studies phocholine was procured from Amersham-Buchler (Braunschweig, FGR). Taurocholate and taurodeoxycholate Centrifugation studies were carried out as described prewere from Sigma (Taufiirchen, FRG). Lkhroprep RP-18, viously [30]. For studying complex formation, lipids and acti-
593 vator protein were mixed and incubated for 30 min at 37 C prior to centrifugation. Samples containing either radioactive lipids or activator protein or incubation mixtures of both were loaded on top of a discontinuous gradient (5 - 30%, by mass, sucrose in steps of 5%) prepared in 50 mM citrate, pH 4.2 in 13-ml nitrocellulose tubes. After centrifugation at 210000 x g (35000 rpm in a Beckman swinging bucket rotor SW 41) for 36 h at 4 C, the bottoms of the tubes were pierced and their contents were collected in fractions of about 0.55 -0.65 ml. [3H]Glycosphingolipids were determined by their radioactivity. The activator content of the fractions was determined by its stimulation of ['H]GM1 degradation by /3-galactosidase as follows: 80 ~1 of the fractions obtained after centrifugation were mixed with 20 p1 50 mM citrate, pH 4.2 containing 5 nmol [3H]GMMI, 100 pg bovine serum albumin and 1.9 mU /I-galactosidase. Incubation of this mixture and determination of released [3H]galactose were as described above. Under these assay conditions, pure sulfatide activator protein showed a specific activity of 187000 5500 AU/mg protein. The difference from 208 500 4500 AU/mg protein (see above) is probably due to the presence of sucrose in this assay. Trunsfer ~ f [ ~ H ] l i p i d s 'H-labeled liposomes were prepared and transfer experiments were performed according to Conzelmann et al. [30]. Acceptor liposomes were prepared from egg phosphatidylcholine (17.5 pmol) labeled with a trace amount of [I4C]dipalmitoylglycerophosphocholine (about 4400 dpm), cholesterol (6.5 pmol), dicetyl phosphate (0.5 pmol) and a-tocopherol (0.5 pmol). Donor liposomes were prepared from egg phosphatidylcholine (16.1 pmol), cholesterol (5.9 pmol), dicetylphosphate (2 pmol); a-tocopherol (0.5 pmol) and 0.5 pmol of the respective [3H]glycosphingolipid. The lipids, dissolved in organic solvents, were mixed in the above ratios and the solvent evaporated under a stream of nitrogen. The lipid mixtures were dried overnight under reduced pressure. Then 1 ml 50 mM sodium citrate pH 4.0 was added and, after suspending the lipids by vigorous shaking, the suspension was sonified under nitrogen for 30 min (Branson sonifier B 12, Branson, Connecticut, equipped with a 3-mm microtip) at 50 W. The preparation was then centrifuged at 50000 x g (40000 rpm) for 30 min to sediment titanium particles and any remaining larger aggregates. The clear, slightly opalescent supernatant was used for the experiments. Donor liposomes (250 nmol lipids; 10 pl) were incubated with the same amount of acceptor liposomes in the presence of 1Opg cytochrome c (used to prevent adsorption of glycosphingolipids to the wall of the reaction vials) and various amounts of activator protein in a total volume of 50 pl in 50 mM citrate pH 4.0 at 37°C. After incubation, the samples were transferred to an ice bath and then loaded onto 1.2-ml DEAE-cellulose columns which had been equilibrated with 50 mM phosphate pH 6.0. The columns were immediately eluted twice with 1 ml of the same phosphate buffer containing 12.5 mM NaCl. The combined eluates were collected and the 3H and 14C radioactivity of the lipids used was measured by liquid scintillation counting (Tri-CubR 1900 CA: Packard, Frankfurt).
RESULTS Glycosphingolipid transfer f r o m donor to acceptor liposomes in the presence ojactivator protein The model proposed by Fisher and Jatzkewitz [16] postulates that sulfatides are extracted from micelles or membranes
m
. -
ow 0.0
,I
I
0.1
I
0.2
>
Activator protein [nrnol/50p11
Activator protein Inrnol/50pl1 Fig. 1. Transferof glycosphin~olipidsjrornacceptor to donor liposornes by the sulfatide activator protein. Donor liposomes (250 nmol lipid) containing 2 mol% of the respective glycosphingolipids werc incubated with an equal amount of acceptor liposomes, cytochrome c (10 pg) and increasing amounts of activator protein in a total volume of 50 p1 50 mM citrate pH 4.0, at 37°C for 1 h. Acceptor liposomes were separated from donor liposomes on small DEAE-cellulose columns as described in Materials and Methods; from their 3H/'4C ratio, the transfer of lipids was calculated. Controls were run without activator proteins and the values subtracted from the respective experimental values. Each value is the mean of duplicate determination. (A) Transfer of (0)GDla,( x ) GMl and (0)GM2; (B) transfer of ( 0 ) GalCer, ( x ) sulfatide and (0) GM3
by the sulfatide activator to give freely water-soluble activatorlipid complexes. This interaction should be reversible. If such a complex had diffused some distance before the lipid was reinserted into the membrane, the sulfatide is most likely to be inserted into another membrane different from the one from which it was extracted, i.e. in the absence of enzyme, the activator should work as a sulfatide transfer protein. To test this hypothesis, we studied the transfer of sulfatides and other glycosphingolipids from donor to acceptor liposomes in the presence of the activator protein. A mixture of donor liposomes containing ['Hlglycosphingolipids (2 mol%) and acceptor liposomes devoid of [3H]glycosphingolipids but labeled with [ '4C]dipalmitoylglycophosphocholine was incubated with activator protein (0.1 nmol) for different periods of time at pH 4.0 and 37°C. Then donor and acceptor liposomes were separated by anionexchange chromatography. The yield of acceptor liposomes in the eluate was monitored by following their I4C radioactivity, the yield of the transferred lipids by assaying their 3H
594 Table 1. Transfer qf diferent (3H]glycosphingolipids by the sulfatide activator protein f r o m acceptor to donor liposomes The values were calculated from the slopes of the plots in Fig. 1
-0
Z6
I
Lipid transferred
Transfer rates nmol glycosphingolipid x h x (nmol activator protein)-
'
Ganglioside GDla Ganglioside GM1 Ganglioside GM2 Sulfatide Galactosylceramide Ganglioside GM3 Glucosylceramide
c
'
9.6 4.5 3.0 1 .o 0.5 0.4 < 0.1
1
c3
a, z m 0 ._
T33 C 0
-? OL. x0 I
10
20
Fraction number
radioactivity. Acceptor liposomes, which had a low content of dicetyl phosphate, were eluted by more than 90%. The donor liposomes, owing to their high content of negatively charged lipid (dicetylphosphate), were nearly completely retained on the ion-exchange resin. Without incubation, only up to 2.5% (0.01 nmol) of the glycosphingolipids used were found with the acceptor liposomes. Without addition of the activator protein, the amount of [3H]gangliosides in the acceptor liposomes increased to 0.3 nmol after 15 min of incubation, staying then nearly constant. The rate of transfer of the gangliosides GD1,,GMland GM2 in the presence of the activator protein was almost linear for the first 30 min, approaching a plateau after 90 rnin (not shown) ; the transfer rates of the gangliosides increased with increasing concentrations of sulfatide activator protein (Fig. 1A, Table 1). The transfer rates of sulfatide, GalCer and GlcCer were much lower. Therefore, their time dependencies were assayed in the presence of increased activator protein concentrations (0.5 nmol/assay; not shown). After 1 h of incubation, small and increasing transfer rates were observed for GalCer, sulfatide and GM3 with increasing concentration of the activator protein (Fig. 1B, Table 1).
Complex formation between the suljiitide activator protein and different glycosphingolipids
Complex formation between the GM,-activator protein and different glycosphingolipids could be demonstrated by ultracentrifugation studies [30]. Similar experiments were performed with the sulfatide activator protein and different glycosphingolipids. Micellar ganglioside or activator protein or a mixture of both after incubation for 30 min at 37 "C were layered on top of a discontinous sucrose gradient (5-30% mass/vol.) and centrifuged at 210000 x g for 40 h. When centrifuged alone, the ganglioside micelles sedimented considerably faster than the activator protein (Fig. 2A). When the activator was mixed and incubated for 30 rnin at 37°C with a tenfold excess of glycosphingolipid, e.g. ganglioside GMl,and then centrifuged, some of the lipid was bound to and moved with the activator protein (Fig. 2 B). Complex formation could also be demonstrated for the gangliosides Gola,GMZ, GM3and for sulfatides. The mixtures of the respective glycosphingolipids and activator protein were incubated for 30 rnin at 37 "C followed by ultracentrifugation and the molar ratios of activator protein to glycosphingolipid bound were calculated (Table 2).
x .c _
> ._ c U
0,
._ c 3
2; 9 1 a0 E
b C
c-
0 5 ._ c
Y "
0
5
10
15
2 0-
Fraction number Fig. 2. Demonstration of complex formation between ganglioside GMl and the sulfatide activator by ultracentrifugation. Activator protein (2.4 nmol, molecular mass 10 kDa) and [3H]GMl (12 nmol) were mixed and incubated as described for Table 2 prior to loading on top of a discontinuous sucrose gradient ( 5 - 30%) in 50 mM citrate pH 4.2 containing cytochrome c (1.5 mgiml). After centrifugation at 210000 x g (35000 rpm) in a Beckman Ti41 swinging-bucket rotor) for 36 h, the content of the tubes were fractionated and assayed for [3H]gangliosideand for activator protein as described in Materials and Methods. (A) Activator protein and ganglioside GMl were centrifuged separately. (B) Mixture of activator and ganglioside GMl.(0)Activator protein activity; ( x ) [3H]GM1
Degradation of glycosphingolipid derivatives in the presence of the sulfatide activator protein
Sulfatide derivatives were degraded by arylsulfatase A, ganglioside GM1 derivatives by j-galactosidase in the absence of activator proteins or detergents (Tables 3 and 4). Degradation rates increased up to 19-fold for sulfatide derivatives and up to 64-fold for GMlderivatives, when the acyl chains of their hydrophobic ceramide moieties decreased from C1 (over C8, C6, C,) to the respective lysolipids with no acyl chain (Tables 3 and 4). Addition of activator protein stimulated the degradation of only those glycosphingolipid derivatives which had acyl chains of 8 C atoms or longer in the case of GMl derivatives, and acyl chains of 6 C atoms or longer in the case of sulfatide derivatives. The highest stimulation was observed for glycosphingolipids containing the naturally occurring stearoyl residues in their ceramide moieties with an activation of almost 13-fold in the case of sulfatide and 9-fold in the case of ganglioside GMlhydrolysis.
595 Table 2. Comples,formation between the sulfatide activator protein and dijyerent micellar lipids Activator protein (1.2 nmol) and the respective [3H]glycosphingolipid (12 nmol) were incubated for 1 h at 37°C in 90 ~1 50 mM citrate, pH 4.0 prior to loading on top of a discontinous sucrose gradient ( 5 30%) in 50 mM citrate pH 4.2 containing cytochrome c (1.5 mgiml). After centrifugation (210000 x g , 36 h) the content of the tubes was fractionated and asssayed for [3H]glycosphingolipid and for activator protein activity as described in Materials and Methods. Fractions containing both activator protein and lipid were collected and their molar ratio determined. The values given are the means of four independent measurements and were calculated for the dimeric form of the sulfatide activator with a molecular mass of 20 kDa
Lipids
mol/mol
GD1, GMl GM2 GM3
Degradation rate without activator
Glycosphingolipid/sulfatide activator
Lipids
Ganglioside Ganglioside Ganglioside Ganglioside Sulfatide
Table 4. Degradation of dcferent GMl derivates by a-gulactosidase in the presence of activator protein Ganglioside GMlderivatives were degraded by human a-galactosidase either in the absence or in the presence of sulfatide activator or in the presence of detergent (taurodeoxycholate) as described under Materials and Methods. The given values are the means from four measurements (without or with detergent) or six measurements (with activator protein)
with detergent (2 mM)
nmol x h-' x m u - ' 0.020 0.006 0.40 k0.02 c2-GM1 0.85 f 0 . 0 2 Lyso-GMla 1.29 fO.01 Ganglioside GM1 0.020 k 0.006 GM,-alCOhoI 0.020 kO.008 GA 1 0.025 & 0.006 LacCer 0.020 f 0.005 c1 8 - G M l
C8-GM1
1.32 f 0.02 1.00 f 0.02 0.89 f 0.01 0.80 f 0.04 0.66 f 0.02
with activator (0.1 nmol)
0.18 & 0.01 0.52 f0.02 0.90 50.04 1.28 fO.05 0.18 f 0.01 0.18 fO.01 0.06 f 0.001 0.020 f 0.004
0.037 f 0.001 1.08 fO.001 1.11 f 0.02 1.020 fO.001 0.037 f 0.001 0.70 fO.01 0.353 f 0.020 0.293 f 0.010
a In the case of Iyso-GMlthe assays contained 0.9 m u fl-galactosidase and were incubated for 1.5 h
Table 3. Degradation of sulfatide derivates by the arylsulfatase A in the presence of the sulfatide activator protein Sulfatide derivatives were degraded by human arylsulfatase A either in the absence or in the presence of sulfatide activator or in the presence of detergent (taurodeoxycholate) as described under Materials and Methods. The values given are means from four measurements (without or with detergent) or six measurements (with activator protein) Lipids
Degradation rate without activator
with activator (0.1 nmol)
with detergent (2 mM)
nmol x h - ' x m u - ' C18-sulfatide C6-sulfatide C2-sulfatide Lyso-sulfatide
0.03 fO.001 0.05 0.01 0.13 f 0.01 0.58 f 0.03
+
0.38 f 0 . 0 7 0.18 f 0.02 0.140 f 0.001 0.57 f 0.05
0.17 f 0 . 0 3 0.620 f 0.001 1.25 k 0.06 0.85 f 0.03"
a The assay contained 20nmol taurocholate instead of taurodeoxycholate
Addition of detergent (sodium taurodeoxycholate) instead of activator protein to the enzymic assay systems resulted in less stimulation for glycosphingolipids containing the natural stearoyl residues but in significantly higher stimulation for glycosphingolipids containing acyl chains of medium chain length (C, and C,). Modifications in the oligosaccharide chain of ganglioside GMl also affected the hydrolysis by /I-galactosidase and its stimulation by the activator protein (Table 4). Reduction of the carboxyl group of the sialic acid residue to a primary hydroxyl group (GMl alcohol) or its removal (yielding GA1) increased the stimulatory effect of the detergent. However, removal of the sialic acid residue (yielding GA1) reduced the stirnulatory effect of the activator protein. Shortening the oligosaccharide chain by three sugar residues (yielding LacCer) abolished the stimulatory effect of the activator protein but increased that of the detergent.
DISCUSSION The sulfatide activator protein (SAP 1) stimulates the degradation of various glycosphingolipids by their respective hydrolases [2, 12, 131. According to the model of Fischer and Jatzkewitz [16], the protein does not activate the enzyme itself but extracts sulfatide monomers from micelles or membranes to give water-soluble complexes which are then degraded by arylsulfatase A. Such a model was proved for the GM2 activator protein by transfer experiments and by analysis of ganglioside-protein complex formation.This activator protein forms a stoichiometric complex with ganglioside GMZ which is then specifically recognized by human hexosaminidase A as substrate [II, 301. Binding Studies
Formation of water-soluble complexes could also be demonstrated between various glycosphingolipids and the sulfatide activator protein in ultracentrifugation studies (Fig. 2). One mole of ganglioside GM1 was bound to two moles of sulfatide activator as calculated on the basis of a molecular mass of 10 kDa. The molecular mass of the sulfatide activator without carbohydrate as calculated from the amino acid sequence is 8973 Da [31] which corresponds well with the molecular mass found on SDSjPAGE in this report. It seems likely, however, that the molecule in its physiological state is a dimer with a molecular mass in the range of 20-22 kDa as determined by gel filtration [12, 14, 201 depending on the amount of glycosylation. If the calculation is done for the dimeric form of the sulfatide activator then a binding ratio of 1 : 1 for GM1 and sulfatide activator is obtained. However, this stoichiometry was higher for the more complex ganglioside GDlaand lower for less complex gangliosides and sulfatides (Table 2). These data indicate that the sulfatide activator protein is an unspecific glycosphingolipid binding protein, and support the model proposed by Wynn [32] for the interaction of sulfatide activator protein with dissimilar glycosphingolipids. With increasing extension and complexity of their oligosaccharide
596 small fatty acyl substituents) became degradable by the enzyme even in the absence of the GM2activator protein after incorporation into liposomes. In studies employing micelles a similar principle could be demonstrated for the degradation of sulfatide derivatives by aryl sulfatase A (Table 3) and ganglioside G M i derivatives by P-galactosidase (Table 4). The sulfatide activator protein stimulated only the hydrolysis of glycosphingolipids with long-chain fatty acyl residues. Those with no or a short-chain fatty acyl residue were hydrolyzed quite readily by the enzyme in the absence of activator protein. Glycosphingolipids with truncated hydrophobic moieties probably have an increased critical micellar concentration and form micelles with decreased stability giving rise to an 0.0 increased monomer concentration which makes them directly available to the enzyme. Incubation time [minl In contrast to the glycosphingolipids with long-chain fatty Fig. 3. Complex formation between uctivator protein and ganglioside acyl residues, the enzymic degradation of the truncated GM1as u function qf time. Activator protein (2.4 nmol) and [3H]GW1 (12 nmol) were mixed in 90 pI 50 mM citrate pH 4.0, incubated glycosphingolipids was not stimulated by the addition of the for the indicated periods of timcs at 37°C and loaded on top of activator protein, indicating that it does not activate either a discontinous sucrose gradient (5 - 30%). After centrifugation at arylsulfatase A or fi-galactosidase directly. 21 0000 x g (35 000 rpm) for 36 h, the content of the tubes was fractionated and assayed for [3H]ganglioside and for activator protein as described in Materials and Methods. Fractions containing both lipid and activator protein were collected and their molar ratio quantified.
REFERENCES
1. Hakomori, S.-I. (1984) Trends Biochem. Sci. 9, 453-458. 2. Conzelmann, E. & Sandhoff, K . (1987) Adv. Enzymol. 60, 89216. chains glycosphingolipids are apparently extracted more ef3. Conzelmann, E. & Sandhoff, K. (1987) Methods Enzymol. 138, ficiently from their micelles by the activator protein. Kinetic 792- 815. measurements, however, showed that complex formation with 4. Conzelmann, E. & Sandhoff, K. (1978) Proc. Natl Acud. Sci. USA ganglioside GMl is rather slow (Fig. 3), saturating the sulfatide 75, 3979-3983. activator only after 30 min at 37°C. 5. Stevens, R. L., Fluharty, A. L., Kihara, H., Kaback, M. M., Shapiro, L. J., Marsh, B., Sandhoff, K., & Fischer, G. (1981) Am. J . Human Genet. 33,900-906. Transff r experiments 6. Zhang, X., Mohammad, A. R, DeGala, G., & Wenger, D. A. (1990) Proc. Natl Acud. Sci. USA 87, 1426- 1430. In transfer experiments similar to those performed before 7. Rafi, M. A., Zhang, X.-L., DeGala, G. & Wenger, D. A. (1990) with the GMZ activator [30] the protein was shown to be Biochem. Biophys. Res. Cornmun. 166,1017- 1023. an unspecific glycosphingolipid transfer protein : transfer of 8. Christomanou, H., Erzberger A. & Linke, R. P. (1986) Bid. various 3H-labeled glycosphingolipids from donor to acceptor Chem. Hoppe-Seyler 367, 879 - 890. liposomes was catalyzed by the sulfatide activator protein 9. Kretz, A. Keith, Carson, G. S., Morimoto, S., Kishimoto, Y., Fluharty, L. A. & O’Brien, J. S. (1990) Proc. Natl Acad. Sci. (Fig. 1 and Table 1). The highest transfer rates were observed USA 87,2541 -2544. for glycosphingolipids with extended oligosaccharide chains like ganglioside GD1,.As expected from the binding studies the 10. Holtschmidt, H., Sandhoff, K., Kwon, H., Harzer, K., Nakano, T. & Suzuki, K. (1991) J . Biol. Chem. 266, in the press. transfer rates dropped with shortening of the oligosaccharide 11. Meier, E. M., Schwarzmann, G., Fiirst, W. & Sandhoff, K. (1 991) chain with the exception of GalCer. The latter was transferred J . Bid. Chem. 266,1879- 1887. more efficiently than GlcCer and about as efficiently as 12. Li, S.-C. & Li, Y.-T. (1976) J . Biol. Chem. 251, 1159-1163. ganglioside GM3. 13. Vogel, A., Fiirst, W., Abo-Hashish, M. A,, Lee-Vaupel, M., Glycosphingolipids containing less than four sugars (GM3, Conzelmann, E. & Sandhoff, K. (1987) Arch. Biochem. Biophys. sulfatide, GalCer) probably extend less than a nanometer from 259,627 - 638. the surface of the membrane into the aqueous space [32, 331, 14. Gartner, S., Conzelmann, E. & Sandhoff, K. (1983) J . Biol. Chem. 255, 12378 - 12 385. so that they are poorly recognized and transferred slowly by the activator protein. According to a model of Brown et al. 15. Fingerhut, R. (1989) Diploniurbeit Universitat, Bonn. [34], the activator can extract them only by changing their 16. Fischer, G . & Jatzkewitz, H. (1977) Biochim. Biophys. Actu 481, 561 - 572. ‘off rate’ from the lipid bilayer. For GlcCer no transfer was 17. Mitsuyama, T., Gasa, S., Nozima, T., Taniguchi, N. & Makita, observed in the presence of the activator protein. A. (1985) J . Biochem. (Tokyo) 98, 605-613. 18. Lowry, 0. H., Rosebrough, N. Y., Farr, A. L. & Randall, R. J. (1951) J . Biol. Chem. 193.265-275. Enzyme studies 19. Miller, A. L., Frost, R. G. & OBrien, J. S. (1977) Biochem. J . 165, 591 -594. Studies in a liposomal system on the enzymic degradation of ganglioside GM2indicate that mainly steric hindrance by 20. Fischer, G. & Jatzkewitz, H. (1975) Hoppe-Seyler’sZ. Physiol. Chem. 356,606-613. adjacent lipid molecules impedes the access of hexosaminidase 21. Ho, M. W. & O’Brien, J. S. (1971) Clin. Chim. Actu 32, 443A to the membrane-bound GMz whose degradation therefore 450. depends on solubilization by the GM2activator protein [I 11. 22. Leskawa, C. K., Dasgupta, S., Chien, J.-L. & Hogan, E. L. (1984) Derivatives of ganglioside GMZ which are less tightly bound Anal. Biochem. 140, 172 - 177. by the lipid bilayer and have an increased ‘off rate’ [34] (such 23. Svennerholm, L. (1972) Methods Curbohydr. Chem. 6,464-474. as lyso-GM2,having no acyl residue, or GM2derivatives with 24. Suzuki, Y. & Suzuki, K. (1972) J . Lipid Res. 13, 687-690.
597 25. Schwarzmann, G. (1978) Biochim. Biophys. Acta 529, 106-114. 26. Handa, S. & Nakamura, K. (1984) J . Biochem. (Tokyo) 95, 1323- 1329. 27. Neuenhofer, S., Schwarzmann, G., Egge, H. & Sandhoff, K. (1985) Biochemistry 24, 525-533. 28. Schwarzmann, G. & Sandhoff, K. (1987) Methods Enzymol. 138, 319 - 341. 29. Stahl, E. & Kaltenbach, U. (1961) J . Chromatogr. 5, 351 -355.
30. Conzelmann, E., Burg, J., Stephan, G. & Sandhoff, K. (1982) Eur. J . Biochem. 123, 455 -464. 31. Fiirst, W., Schubert, J., Machleidt, W., Meyer, H. E. & Sandhoff, K. (1990) Eur. J . Biochem. 192,709-714. 32. Wynn, C. H. (1986) Biochem. J . 240, 921 -924. 33. Wynn, C. H. & Robson, B. (1986) J . Theor. Biol. 123,221 -230. 34. Brown, R. E. Stephenson, F. A,, Markello, T., Barenholz, Y. & Thompson, T. E. (1985) Chem. Phys. Lipids 38, 79 -93.
001429569100573D
Glycosphingolipid specifity of the human sulfatide activator protein A. VOGEL, G. SCHWARZMANN and K. SANDHOFF Institut fur Organische Chemie und Biochemie der Universitat Bonn, Federal Republic of Germany Received January 28/April 3, 1991) - EJB 91 0145
The interaction of the sulfatide activator protein with different glycosphingolipids has been studied in detail. The following findings were made. 1. The sulfatide activator protein forms water-soluble complexes with sulfatides [Fischer, G. and Jatzkewitz, H. (1977) Hoppe-Seyler’s Z. Physiol. Chem. 356, 6588 - 65911 and various other glycosphingolipids. 2. In the absence of degrading enzymes the activator protein acts in vitro as a glycosphingolipid transfer protein,transporting glycosphingolipids from donor to acceptor liposomes. Lipids having less than three hexoses, e.g. galactosylceramide, sulfatide and ganglioside GM3were transferred at very slow rates, whereas complex lipids such as gangliosides GMZ, GMland GDla were transferred much faster than the former. The transfer rate increased with increasing length of the carbohydrate chain of the lipid molecules. 3. Both the acyl residue in the ceramide moiety and the nature of the carbohydrate chain are significant for recognition of the glycosphingolipids by the sulfatide activator protein. Apparently, both residues serve as an anchor and the longer they are the better they are recognized by the protein. 4. In the absence of activator protein, degradation rates of sulfatide derivatives by arylsulfatase A, and of ganglioside GM1 derivatives by P-galactosidase, increase with decreasing length of acyl residues in their hydrophobic ceramide moiety. Addition of activator protein stimulates the degradation of only those GM1 and sulfatide derivatives that have long-chain fatty acids in their hydrophobic ceramide anchor.
Glycosphingolipids are degraded to ceramide by the sequential action of lysosomal exohydrolases [2]. In addition, the physiological degradation of glycosphingolipids with oligosaccharide chains up to a length of three hexoses requires the assistance of water-soluble, nonenzymatic glycoproteins, so-called sphingolipid activator proteins [3]. Genetic defects Correspondence to K. Sandhoff, Institut fur Organische Chemie of these lysosomal activator proteins result in sphingolipid und Biochemie der Universitat Bonn, Gerhard-Domagk-Strasse 1, storage diseases [4- lo]. While the human GM2 activator W-5300 Bonn 1, Federal Republic of Germany (SAP-3) specifically stimulates the hydrolysis of ganglioside Abbreviations. Cer, ceramide or N-acylsphingoid; Sph, D-erythroand GA2by hexosaminidase A [I 11, the sulfatide activator sphingosine; NeuAc, N-acetylneuraminic acid; GalNAc, 2-acet- GM2 amido-2-deoxy-~-galactose; GalCer, galactosylceramide; GlcCer, (SAP-I) furthers unspecifically the degradation of different glucosylceramide; LacCer, lactosylceramide; sulfatide or 3’-O-sulfo- glycosphingolipids in vitro [12,13]. In the test tube the sulfatide galactosyl-ceramide; globotriaosylceramide or GbOse3Cer or Galcll activator can stimulate the hydrolysis of sulfatides by + 4Galpl 4Glc/l1+ 1Cer; lysosulfatide or 3’-O-SUlfO-galaCtOSy1-Darylsulfatase A, ganglioside GMlby f3-galactosidase,GbOse3erythro-sphingosine; C,-sulfatide, derivative of sulfatide containing Cer by a-galactosidase A [12,14] and some gangliosides (Gola, acyl chains with x C atoms (x = 2, 6, 18); G M ~I13NeuAc-LacCer , or GTlb,GM3)by lysosomal sialidase [15]. Gal(3+2clNeuAc)~l+4Glc~I+lCer; G M Z , I13NeuAc-GgOse3Cer Fischer and Jatzkewitz [16] observed that the sulfatide or GalNAcBl + 4Ga1(3 + 2clNeuAc)pl + 4Glcj31 + ICer; GMl, I13NeuAc - GgOse4Cer or GalBl + 3GalNAcpl + 4Ga1(3 + activator extracts sulfatides from micelles and forms water2aNeuAc)pl 4Glc/l1 + 1Cer; GDla, IV3NeuAc, I13NeuAc - soluble sulfatide - activator protein complexes which are then GgOse4Cer or Gal(3 + 2clNeuAc)jll + 3GalNAcbl + 4Ga1(3 c degraded by arylsulfatase A. To gain further insight into the 2aNeuAc)pl + 4GlcpI + 1Cer; GTlb, IV3NeuAc,I13(NeuAc)2- specificity of the sulfatide activator and its interaction with GgOse4Cer or Gal(3 + 2clNeuAc)pl + 3GalNAcB1 + 4Ga1(3 + glycosphingolipids, we studied the ability of the protein to 2clNeuAc8 + 2clNeuAc)pl + 4Glc/lI + Cer; GA1, GgOse4Cer or act as a glycosphingolipid transfer protein, stimulating the Gal~l+3GalNAc/31+4Gal/lI +GlcpI+ 1 Cer; G A Z , GgOse3 Cer or GalNAcj?l+4Galfll+4Glcfll +1Cer, GMl alcohol, ganglioside with transfer rates of various glycosphingolipids from donor to the carboxyl group reduced to a hydroxymethyl group; lyso-GMl, acceptor liposomes. Since there is evidence that the hydroI13NeuAc-GgOse,Sph; deacylated GM1, lyso-GM1which lacks the N- phobic ceramide moiety of sulfatides, especially the acyl chain, is important for the protein-lipid interaction [17], we also acetyl group in the sialic acid moiety or I13Neu-GgOse4Sph, C,-GM1 derivative of GMl containing acyl chains with x C atoms (x = 2, 8, analyzed the enzymic degradation of synthetically modified 18). glycosphingolipids containing oligosaccharide and acyl chains Enzymes. 8-Galactosidase (EC 3.2.1.23); arylsulfatase of different lengths.
Glycosphingolipids are components of the outer leaflet of plasma membranes. They form cell-type-specific patterns on the surfaces of mammalian cells. These patterns change characteristically with differentiation and oncogenic transformation [l].
--f
--f
(EC 3.1.6.1).
592 MATERIALS AND METHODS Protein purification The sulfatide activator protein was purified from human kidney according to procedures described previously [3, 51 using acid precipitation, heat precipitation, anion-exchange chromatography on DEAE-cellulose, gel filtration, isoelectric focussing, hydrophobic chromatography on octyl-Sepharose and anion-exchange chromatography and gel filtration on HPLC columns. The preparation was pure according to electrophoresis and Western blot analysis. The purified protein gave one major single band on SDSjPAGE (not shown) and its molecular mass was determined as 10 kDa by extrapolation from the mobilities of molecular mass markers. Protein was determined according to Lowry et al. [18] using bovine serum albumin as standard. In the presence of other proteins, activator proteins are generally quantified by measuring their capability to accelerate in vivo the hydrolysis of glycosphingolipid substrates by the respective hydrolases. Thus, one unit of sulfatide activator (AU) is defined as the amount of activator that stimulates the enzymatic hydrolysis of ganghoside GMl by 1 nmol x h-' x (U P-galactosidase)-' under assay conditions where the reaction depends almost linearly on the activator concentration [3]. The sulfatide activator used in this work had a specific activity of 208 500 4500 AU/mg protein. P-Galactosidase was purified from human liver by the method of Miller et al. [19], arylsulfatase A from human liver was purified to electrophoretic homogeneity by the method of Fischer and Jatzkewitz [20].P-Galactosidase and arylsulfatase A had specific activities of 1.1 and 3.85 units/mg protein, respectively. One unit of P-galactosidase ciitalyzes the release of P-galactose from 1 pmol 4-methylumbelliferyl-P-~galactopyranoside/min under standard conditions [21]. One unit of arylsulfatase A catalyzes the release of sulfate from 1 pmol p-nitrocatecholsulfate/min under conditions described [161. Lipid.?
silica-gel-coated thin layer plates and anisaldehyde were obtained from Merck (Darmstadt, FRG). All other chemicals were of analytical grade or highest purity available. Activator assays The stimulation of the enzymatic degradation of ganglioside GM1 was measured as described previously [3]. The assays contained lipid ( 5 nmol), P-gdlactosidase (1.9 mu), bovine serum albumin (30 pg), and detergent (up to 100 nmol) or activator protein (up to 0.1 nmol) in a total volume 50 pl of 50 mM citrate, pH 4.0. After incubation for 1 h at 37"C, the vials were transferred to an ice bath and 1 ml 1 mM galactose solution was added. For the separation of released [3H]galactose from GMl and GM2, the sample was loaded onto small DEAE-cellulose columns (1 ml), and the columns were washed twice with 1 ml 1 mM galactose solution. The combined effluents containing the liberated [3H]galactose were collected in scintillation vials and their radioactivity quantified by liquid scintillation counting. For LacCer and the other GM1 derivatives the assay was modified as follows. For C Z - G Mand ~ Cs-GMl degradation, the glycosphingolipids were purified using small (0.4 ml) Lichroprep RP-18 columns. Prior to loading the columns were washed twice with chloroform/methanol (2/1, by vol.), once with methanol and once with synthetic upper phase (chloroform/methanol/O.l M KCl, 3/48/47, by vol.). The lyophilized lipid mixture ( 5 nmol) of the assay was redissolved in 4.9 ml synthetic upper phase and passed over an RP-18 column (0.4 ml). After washing with water (2.5 ml), the lipids were eluted with 0.5 ml methanol and 2.5 ml chloroform/methanol ( l / l , by vol.). Since LacCer and GlcCer separation by thinlayer chromatography was impaired by taurodeoxycholate, these glycosphingolipids were freed of this detergent by passing the incubation mixture through a DEAE-cellulose column (0.5 ml) using 4 ml methanol. Glycosphingolipids were separated by thin-layer chromatography (solvents: chloroform/methanol/water, 60/40/9, by vol., for C2-GMl,GAl;n-butanol/ethanol/water 2/1/1, by vol., for lyso-GMl; chloroform/methanol/water, 60/25/4, by vol., for LacCer; chloroform/methanol/l5 mM CaCl,, 60/35/8, by vol., for GM1 alcohol, C8-GMl). After staining with anisaldehyde/sulfuric acid/acetic acid [29], the glycosphingolipid bands were quantified by scanning the thin-layer plates at 580 nm with a densitometer (Shimadzu CS-910). For the determination of stimulation of degradation of sulfatide derivatives lipid (5 nmol), arylsulfatase A (1 mu), bovine serum albumin (10 pg), and detergent (100 nmol) or activator protein (0.1 nmol), were incubated in a total volume of 50 pl of 10 mM sodium acetate, 100 mM NaC1, pH 5.0, for 1 h at 37°C. Assays containing 3H-labeled CI8- or C6sulfatide were quantified as described previously [3, 51. The assays containing [3H]C2-sulfatide or lysosulfatide were lyophilized and products and educts separated by thin-layer chromatography (solvents: for lysosulfatide; chloroform/ methanol/water/25% NH3, 70/30/4/1, by vol. ; for C2sulfatide, chloroform/methanol/lS mM CaC12, 60/35/8, by vol.). Lysosulfatide was quantified at 580 nm as described above, [3H]C2-sulfatide was quantified using a radioscanner (Berthold, W-7547, Wildbad 1).
Ganglioside GM1 was a gift from Fidia (Abano, Italy) and tritiated in the galactose moiety by the galactose oxidase/ NaB3H4 method according to Leskawa et al. [22]; its specific activity was 77.7 TBq/mol (2120 Ci/mol). Ganglioside GM2 was isolated from Tay-Sachs brain by the method of Svennerholm [23], and tritiated in the N-acetylgalactosamine moiety by the galactose oxidase/NaB3H4 method according to Suzuki and Suzuki [24]; its specific radioactivity was 440 GBq/mol (12 Ci/mol). GalCer and sulfatide from postmortem human brain, as well as GlcCer and GM3from postmortem human spleen, were tritiated as described [25]. The specific radioactivities for GalCer, sulfatide, GlcCer and GM3 were 2 (54), 7 (190), 3.7 (100) and 12.1 (330) TBq (Ci)/mol, respectively. Ganglioside GDla was isolated from a mixture of bovine brain gangliosides and tritiated in the sphingosine moiety [25] to a specific radioactivity of 25 TBq/mol(675 Ci/ mol). GMM,-alcohol was synthesized as described [26]. LysoGM1and C2-GMlwere synthesized as described [27, 281. C8GMl,[3H]C2-sulfatideand C6-sulfatide, [3H]lysosulfatide and LacCer were available in our laboratory. Egg phosphatidylcholine, a-tocopherol and cholesterol were purchased from Sigma (Miinchen, FRG), ['4C]dipalmitoylglycerophosCentrifugation studies phocholine was procured from Amersham-Buchler (Braunschweig, FGR). Taurocholate and taurodeoxycholate Centrifugation studies were carried out as described prewere from Sigma (Taufiirchen, FRG). Lkhroprep RP-18, viously [30]. For studying complex formation, lipids and acti-
593 vator protein were mixed and incubated for 30 min at 37 C prior to centrifugation. Samples containing either radioactive lipids or activator protein or incubation mixtures of both were loaded on top of a discontinuous gradient (5 - 30%, by mass, sucrose in steps of 5%) prepared in 50 mM citrate, pH 4.2 in 13-ml nitrocellulose tubes. After centrifugation at 210000 x g (35000 rpm in a Beckman swinging bucket rotor SW 41) for 36 h at 4 C, the bottoms of the tubes were pierced and their contents were collected in fractions of about 0.55 -0.65 ml. [3H]Glycosphingolipids were determined by their radioactivity. The activator content of the fractions was determined by its stimulation of ['H]GM1 degradation by /3-galactosidase as follows: 80 ~1 of the fractions obtained after centrifugation were mixed with 20 p1 50 mM citrate, pH 4.2 containing 5 nmol [3H]GMMI, 100 pg bovine serum albumin and 1.9 mU /I-galactosidase. Incubation of this mixture and determination of released [3H]galactose were as described above. Under these assay conditions, pure sulfatide activator protein showed a specific activity of 187000 5500 AU/mg protein. The difference from 208 500 4500 AU/mg protein (see above) is probably due to the presence of sucrose in this assay. Trunsfer ~ f [ ~ H ] l i p i d s 'H-labeled liposomes were prepared and transfer experiments were performed according to Conzelmann et al. [30]. Acceptor liposomes were prepared from egg phosphatidylcholine (17.5 pmol) labeled with a trace amount of [I4C]dipalmitoylglycerophosphocholine (about 4400 dpm), cholesterol (6.5 pmol), dicetyl phosphate (0.5 pmol) and a-tocopherol (0.5 pmol). Donor liposomes were prepared from egg phosphatidylcholine (16.1 pmol), cholesterol (5.9 pmol), dicetylphosphate (2 pmol); a-tocopherol (0.5 pmol) and 0.5 pmol of the respective [3H]glycosphingolipid. The lipids, dissolved in organic solvents, were mixed in the above ratios and the solvent evaporated under a stream of nitrogen. The lipid mixtures were dried overnight under reduced pressure. Then 1 ml 50 mM sodium citrate pH 4.0 was added and, after suspending the lipids by vigorous shaking, the suspension was sonified under nitrogen for 30 min (Branson sonifier B 12, Branson, Connecticut, equipped with a 3-mm microtip) at 50 W. The preparation was then centrifuged at 50000 x g (40000 rpm) for 30 min to sediment titanium particles and any remaining larger aggregates. The clear, slightly opalescent supernatant was used for the experiments. Donor liposomes (250 nmol lipids; 10 pl) were incubated with the same amount of acceptor liposomes in the presence of 1Opg cytochrome c (used to prevent adsorption of glycosphingolipids to the wall of the reaction vials) and various amounts of activator protein in a total volume of 50 pl in 50 mM citrate pH 4.0 at 37°C. After incubation, the samples were transferred to an ice bath and then loaded onto 1.2-ml DEAE-cellulose columns which had been equilibrated with 50 mM phosphate pH 6.0. The columns were immediately eluted twice with 1 ml of the same phosphate buffer containing 12.5 mM NaCl. The combined eluates were collected and the 3H and 14C radioactivity of the lipids used was measured by liquid scintillation counting (Tri-CubR 1900 CA: Packard, Frankfurt).
RESULTS Glycosphingolipid transfer f r o m donor to acceptor liposomes in the presence ojactivator protein The model proposed by Fisher and Jatzkewitz [16] postulates that sulfatides are extracted from micelles or membranes
m
. -
ow 0.0
,I
I
0.1
I
0.2
>
Activator protein [nrnol/50p11
Activator protein Inrnol/50pl1 Fig. 1. Transferof glycosphin~olipidsjrornacceptor to donor liposornes by the sulfatide activator protein. Donor liposomes (250 nmol lipid) containing 2 mol% of the respective glycosphingolipids werc incubated with an equal amount of acceptor liposomes, cytochrome c (10 pg) and increasing amounts of activator protein in a total volume of 50 p1 50 mM citrate pH 4.0, at 37°C for 1 h. Acceptor liposomes were separated from donor liposomes on small DEAE-cellulose columns as described in Materials and Methods; from their 3H/'4C ratio, the transfer of lipids was calculated. Controls were run without activator proteins and the values subtracted from the respective experimental values. Each value is the mean of duplicate determination. (A) Transfer of (0)GDla,( x ) GMl and (0)GM2; (B) transfer of ( 0 ) GalCer, ( x ) sulfatide and (0) GM3
by the sulfatide activator to give freely water-soluble activatorlipid complexes. This interaction should be reversible. If such a complex had diffused some distance before the lipid was reinserted into the membrane, the sulfatide is most likely to be inserted into another membrane different from the one from which it was extracted, i.e. in the absence of enzyme, the activator should work as a sulfatide transfer protein. To test this hypothesis, we studied the transfer of sulfatides and other glycosphingolipids from donor to acceptor liposomes in the presence of the activator protein. A mixture of donor liposomes containing ['Hlglycosphingolipids (2 mol%) and acceptor liposomes devoid of [3H]glycosphingolipids but labeled with [ '4C]dipalmitoylglycophosphocholine was incubated with activator protein (0.1 nmol) for different periods of time at pH 4.0 and 37°C. Then donor and acceptor liposomes were separated by anionexchange chromatography. The yield of acceptor liposomes in the eluate was monitored by following their I4C radioactivity, the yield of the transferred lipids by assaying their 3H
594 Table 1. Transfer qf diferent (3H]glycosphingolipids by the sulfatide activator protein f r o m acceptor to donor liposomes The values were calculated from the slopes of the plots in Fig. 1
-0
Z6
I
Lipid transferred
Transfer rates nmol glycosphingolipid x h x (nmol activator protein)-
'
Ganglioside GDla Ganglioside GM1 Ganglioside GM2 Sulfatide Galactosylceramide Ganglioside GM3 Glucosylceramide
c
'
9.6 4.5 3.0 1 .o 0.5 0.4 < 0.1
1
c3
a, z m 0 ._
T33 C 0
-? OL. x0 I
10
20
Fraction number
radioactivity. Acceptor liposomes, which had a low content of dicetyl phosphate, were eluted by more than 90%. The donor liposomes, owing to their high content of negatively charged lipid (dicetylphosphate), were nearly completely retained on the ion-exchange resin. Without incubation, only up to 2.5% (0.01 nmol) of the glycosphingolipids used were found with the acceptor liposomes. Without addition of the activator protein, the amount of [3H]gangliosides in the acceptor liposomes increased to 0.3 nmol after 15 min of incubation, staying then nearly constant. The rate of transfer of the gangliosides GD1,,GMland GM2 in the presence of the activator protein was almost linear for the first 30 min, approaching a plateau after 90 rnin (not shown) ; the transfer rates of the gangliosides increased with increasing concentrations of sulfatide activator protein (Fig. 1A, Table 1). The transfer rates of sulfatide, GalCer and GlcCer were much lower. Therefore, their time dependencies were assayed in the presence of increased activator protein concentrations (0.5 nmol/assay; not shown). After 1 h of incubation, small and increasing transfer rates were observed for GalCer, sulfatide and GM3 with increasing concentration of the activator protein (Fig. 1B, Table 1).
Complex formation between the suljiitide activator protein and different glycosphingolipids
Complex formation between the GM,-activator protein and different glycosphingolipids could be demonstrated by ultracentrifugation studies [30]. Similar experiments were performed with the sulfatide activator protein and different glycosphingolipids. Micellar ganglioside or activator protein or a mixture of both after incubation for 30 min at 37 "C were layered on top of a discontinous sucrose gradient (5-30% mass/vol.) and centrifuged at 210000 x g for 40 h. When centrifuged alone, the ganglioside micelles sedimented considerably faster than the activator protein (Fig. 2A). When the activator was mixed and incubated for 30 rnin at 37°C with a tenfold excess of glycosphingolipid, e.g. ganglioside GMl,and then centrifuged, some of the lipid was bound to and moved with the activator protein (Fig. 2 B). Complex formation could also be demonstrated for the gangliosides Gola,GMZ, GM3and for sulfatides. The mixtures of the respective glycosphingolipids and activator protein were incubated for 30 rnin at 37 "C followed by ultracentrifugation and the molar ratios of activator protein to glycosphingolipid bound were calculated (Table 2).
x .c _
> ._ c U
0,
._ c 3
2; 9 1 a0 E
b C
c-
0 5 ._ c
Y "
0
5
10
15
2 0-
Fraction number Fig. 2. Demonstration of complex formation between ganglioside GMl and the sulfatide activator by ultracentrifugation. Activator protein (2.4 nmol, molecular mass 10 kDa) and [3H]GMl (12 nmol) were mixed and incubated as described for Table 2 prior to loading on top of a discontinuous sucrose gradient ( 5 - 30%) in 50 mM citrate pH 4.2 containing cytochrome c (1.5 mgiml). After centrifugation at 210000 x g (35000 rpm) in a Beckman Ti41 swinging-bucket rotor) for 36 h, the content of the tubes were fractionated and assayed for [3H]gangliosideand for activator protein as described in Materials and Methods. (A) Activator protein and ganglioside GMl were centrifuged separately. (B) Mixture of activator and ganglioside GMl.(0)Activator protein activity; ( x ) [3H]GM1
Degradation of glycosphingolipid derivatives in the presence of the sulfatide activator protein
Sulfatide derivatives were degraded by arylsulfatase A, ganglioside GM1 derivatives by j-galactosidase in the absence of activator proteins or detergents (Tables 3 and 4). Degradation rates increased up to 19-fold for sulfatide derivatives and up to 64-fold for GMlderivatives, when the acyl chains of their hydrophobic ceramide moieties decreased from C1 (over C8, C6, C,) to the respective lysolipids with no acyl chain (Tables 3 and 4). Addition of activator protein stimulated the degradation of only those glycosphingolipid derivatives which had acyl chains of 8 C atoms or longer in the case of GMl derivatives, and acyl chains of 6 C atoms or longer in the case of sulfatide derivatives. The highest stimulation was observed for glycosphingolipids containing the naturally occurring stearoyl residues in their ceramide moieties with an activation of almost 13-fold in the case of sulfatide and 9-fold in the case of ganglioside GMlhydrolysis.
595 Table 2. Comples,formation between the sulfatide activator protein and dijyerent micellar lipids Activator protein (1.2 nmol) and the respective [3H]glycosphingolipid (12 nmol) were incubated for 1 h at 37°C in 90 ~1 50 mM citrate, pH 4.0 prior to loading on top of a discontinous sucrose gradient ( 5 30%) in 50 mM citrate pH 4.2 containing cytochrome c (1.5 mgiml). After centrifugation (210000 x g , 36 h) the content of the tubes was fractionated and asssayed for [3H]glycosphingolipid and for activator protein activity as described in Materials and Methods. Fractions containing both activator protein and lipid were collected and their molar ratio determined. The values given are the means of four independent measurements and were calculated for the dimeric form of the sulfatide activator with a molecular mass of 20 kDa
Lipids
mol/mol
GD1, GMl GM2 GM3
Degradation rate without activator
Glycosphingolipid/sulfatide activator
Lipids
Ganglioside Ganglioside Ganglioside Ganglioside Sulfatide
Table 4. Degradation of dcferent GMl derivates by a-gulactosidase in the presence of activator protein Ganglioside GMlderivatives were degraded by human a-galactosidase either in the absence or in the presence of sulfatide activator or in the presence of detergent (taurodeoxycholate) as described under Materials and Methods. The given values are the means from four measurements (without or with detergent) or six measurements (with activator protein)
with detergent (2 mM)
nmol x h-' x m u - ' 0.020 0.006 0.40 k0.02 c2-GM1 0.85 f 0 . 0 2 Lyso-GMla 1.29 fO.01 Ganglioside GM1 0.020 k 0.006 GM,-alCOhoI 0.020 kO.008 GA 1 0.025 & 0.006 LacCer 0.020 f 0.005 c1 8 - G M l
C8-GM1
1.32 f 0.02 1.00 f 0.02 0.89 f 0.01 0.80 f 0.04 0.66 f 0.02
with activator (0.1 nmol)
0.18 & 0.01 0.52 f0.02 0.90 50.04 1.28 fO.05 0.18 f 0.01 0.18 fO.01 0.06 f 0.001 0.020 f 0.004
0.037 f 0.001 1.08 fO.001 1.11 f 0.02 1.020 fO.001 0.037 f 0.001 0.70 fO.01 0.353 f 0.020 0.293 f 0.010
a In the case of Iyso-GMlthe assays contained 0.9 m u fl-galactosidase and were incubated for 1.5 h
Table 3. Degradation of sulfatide derivates by the arylsulfatase A in the presence of the sulfatide activator protein Sulfatide derivatives were degraded by human arylsulfatase A either in the absence or in the presence of sulfatide activator or in the presence of detergent (taurodeoxycholate) as described under Materials and Methods. The values given are means from four measurements (without or with detergent) or six measurements (with activator protein) Lipids
Degradation rate without activator
with activator (0.1 nmol)
with detergent (2 mM)
nmol x h - ' x m u - ' C18-sulfatide C6-sulfatide C2-sulfatide Lyso-sulfatide
0.03 fO.001 0.05 0.01 0.13 f 0.01 0.58 f 0.03
+
0.38 f 0 . 0 7 0.18 f 0.02 0.140 f 0.001 0.57 f 0.05
0.17 f 0 . 0 3 0.620 f 0.001 1.25 k 0.06 0.85 f 0.03"
a The assay contained 20nmol taurocholate instead of taurodeoxycholate
Addition of detergent (sodium taurodeoxycholate) instead of activator protein to the enzymic assay systems resulted in less stimulation for glycosphingolipids containing the natural stearoyl residues but in significantly higher stimulation for glycosphingolipids containing acyl chains of medium chain length (C, and C,). Modifications in the oligosaccharide chain of ganglioside GMl also affected the hydrolysis by /I-galactosidase and its stimulation by the activator protein (Table 4). Reduction of the carboxyl group of the sialic acid residue to a primary hydroxyl group (GMl alcohol) or its removal (yielding GA1) increased the stimulatory effect of the detergent. However, removal of the sialic acid residue (yielding GA1) reduced the stirnulatory effect of the activator protein. Shortening the oligosaccharide chain by three sugar residues (yielding LacCer) abolished the stimulatory effect of the activator protein but increased that of the detergent.
DISCUSSION The sulfatide activator protein (SAP 1) stimulates the degradation of various glycosphingolipids by their respective hydrolases [2, 12, 131. According to the model of Fischer and Jatzkewitz [16], the protein does not activate the enzyme itself but extracts sulfatide monomers from micelles or membranes to give water-soluble complexes which are then degraded by arylsulfatase A. Such a model was proved for the GM2 activator protein by transfer experiments and by analysis of ganglioside-protein complex formation.This activator protein forms a stoichiometric complex with ganglioside GMZ which is then specifically recognized by human hexosaminidase A as substrate [II, 301. Binding Studies
Formation of water-soluble complexes could also be demonstrated between various glycosphingolipids and the sulfatide activator protein in ultracentrifugation studies (Fig. 2). One mole of ganglioside GM1 was bound to two moles of sulfatide activator as calculated on the basis of a molecular mass of 10 kDa. The molecular mass of the sulfatide activator without carbohydrate as calculated from the amino acid sequence is 8973 Da [31] which corresponds well with the molecular mass found on SDSjPAGE in this report. It seems likely, however, that the molecule in its physiological state is a dimer with a molecular mass in the range of 20-22 kDa as determined by gel filtration [12, 14, 201 depending on the amount of glycosylation. If the calculation is done for the dimeric form of the sulfatide activator then a binding ratio of 1 : 1 for GM1 and sulfatide activator is obtained. However, this stoichiometry was higher for the more complex ganglioside GDlaand lower for less complex gangliosides and sulfatides (Table 2). These data indicate that the sulfatide activator protein is an unspecific glycosphingolipid binding protein, and support the model proposed by Wynn [32] for the interaction of sulfatide activator protein with dissimilar glycosphingolipids. With increasing extension and complexity of their oligosaccharide
596 small fatty acyl substituents) became degradable by the enzyme even in the absence of the GM2activator protein after incorporation into liposomes. In studies employing micelles a similar principle could be demonstrated for the degradation of sulfatide derivatives by aryl sulfatase A (Table 3) and ganglioside G M i derivatives by P-galactosidase (Table 4). The sulfatide activator protein stimulated only the hydrolysis of glycosphingolipids with long-chain fatty acyl residues. Those with no or a short-chain fatty acyl residue were hydrolyzed quite readily by the enzyme in the absence of activator protein. Glycosphingolipids with truncated hydrophobic moieties probably have an increased critical micellar concentration and form micelles with decreased stability giving rise to an 0.0 increased monomer concentration which makes them directly available to the enzyme. Incubation time [minl In contrast to the glycosphingolipids with long-chain fatty Fig. 3. Complex formation between uctivator protein and ganglioside acyl residues, the enzymic degradation of the truncated GM1as u function qf time. Activator protein (2.4 nmol) and [3H]GW1 (12 nmol) were mixed in 90 pI 50 mM citrate pH 4.0, incubated glycosphingolipids was not stimulated by the addition of the for the indicated periods of timcs at 37°C and loaded on top of activator protein, indicating that it does not activate either a discontinous sucrose gradient (5 - 30%). After centrifugation at arylsulfatase A or fi-galactosidase directly. 21 0000 x g (35 000 rpm) for 36 h, the content of the tubes was fractionated and assayed for [3H]ganglioside and for activator protein as described in Materials and Methods. Fractions containing both lipid and activator protein were collected and their molar ratio quantified.
REFERENCES
1. Hakomori, S.-I. (1984) Trends Biochem. Sci. 9, 453-458. 2. Conzelmann, E. & Sandhoff, K . (1987) Adv. Enzymol. 60, 89216. chains glycosphingolipids are apparently extracted more ef3. Conzelmann, E. & Sandhoff, K. (1987) Methods Enzymol. 138, ficiently from their micelles by the activator protein. Kinetic 792- 815. measurements, however, showed that complex formation with 4. Conzelmann, E. & Sandhoff, K. (1978) Proc. Natl Acud. Sci. USA ganglioside GMl is rather slow (Fig. 3), saturating the sulfatide 75, 3979-3983. activator only after 30 min at 37°C. 5. Stevens, R. L., Fluharty, A. L., Kihara, H., Kaback, M. M., Shapiro, L. J., Marsh, B., Sandhoff, K., & Fischer, G. (1981) Am. J . Human Genet. 33,900-906. Transff r experiments 6. Zhang, X., Mohammad, A. R, DeGala, G., & Wenger, D. A. (1990) Proc. Natl Acud. Sci. USA 87, 1426- 1430. In transfer experiments similar to those performed before 7. Rafi, M. A., Zhang, X.-L., DeGala, G. & Wenger, D. A. (1990) with the GMZ activator [30] the protein was shown to be Biochem. Biophys. Res. Cornmun. 166,1017- 1023. an unspecific glycosphingolipid transfer protein : transfer of 8. Christomanou, H., Erzberger A. & Linke, R. P. (1986) Bid. various 3H-labeled glycosphingolipids from donor to acceptor Chem. Hoppe-Seyler 367, 879 - 890. liposomes was catalyzed by the sulfatide activator protein 9. Kretz, A. Keith, Carson, G. S., Morimoto, S., Kishimoto, Y., Fluharty, L. A. & O’Brien, J. S. (1990) Proc. Natl Acad. Sci. (Fig. 1 and Table 1). The highest transfer rates were observed USA 87,2541 -2544. for glycosphingolipids with extended oligosaccharide chains like ganglioside GD1,.As expected from the binding studies the 10. Holtschmidt, H., Sandhoff, K., Kwon, H., Harzer, K., Nakano, T. & Suzuki, K. (1991) J . Biol. Chem. 266, in the press. transfer rates dropped with shortening of the oligosaccharide 11. Meier, E. M., Schwarzmann, G., Fiirst, W. & Sandhoff, K. (1 991) chain with the exception of GalCer. The latter was transferred J . Bid. Chem. 266,1879- 1887. more efficiently than GlcCer and about as efficiently as 12. Li, S.-C. & Li, Y.-T. (1976) J . Biol. Chem. 251, 1159-1163. ganglioside GM3. 13. Vogel, A., Fiirst, W., Abo-Hashish, M. A,, Lee-Vaupel, M., Glycosphingolipids containing less than four sugars (GM3, Conzelmann, E. & Sandhoff, K. (1987) Arch. Biochem. Biophys. sulfatide, GalCer) probably extend less than a nanometer from 259,627 - 638. the surface of the membrane into the aqueous space [32, 331, 14. Gartner, S., Conzelmann, E. & Sandhoff, K. (1983) J . Biol. Chem. 255, 12378 - 12 385. so that they are poorly recognized and transferred slowly by the activator protein. According to a model of Brown et al. 15. Fingerhut, R. (1989) Diploniurbeit Universitat, Bonn. [34], the activator can extract them only by changing their 16. Fischer, G . & Jatzkewitz, H. (1977) Biochim. Biophys. Actu 481, 561 - 572. ‘off rate’ from the lipid bilayer. For GlcCer no transfer was 17. Mitsuyama, T., Gasa, S., Nozima, T., Taniguchi, N. & Makita, observed in the presence of the activator protein. A. (1985) J . Biochem. (Tokyo) 98, 605-613. 18. Lowry, 0. H., Rosebrough, N. Y., Farr, A. L. & Randall, R. J. (1951) J . Biol. Chem. 193.265-275. Enzyme studies 19. Miller, A. L., Frost, R. G. & OBrien, J. S. (1977) Biochem. J . 165, 591 -594. Studies in a liposomal system on the enzymic degradation of ganglioside GM2indicate that mainly steric hindrance by 20. Fischer, G. & Jatzkewitz, H. (1975) Hoppe-Seyler’sZ. Physiol. Chem. 356,606-613. adjacent lipid molecules impedes the access of hexosaminidase 21. Ho, M. W. & O’Brien, J. S. (1971) Clin. Chim. Actu 32, 443A to the membrane-bound GMz whose degradation therefore 450. depends on solubilization by the GM2activator protein [I 11. 22. Leskawa, C. K., Dasgupta, S., Chien, J.-L. & Hogan, E. L. (1984) Derivatives of ganglioside GMZ which are less tightly bound Anal. Biochem. 140, 172 - 177. by the lipid bilayer and have an increased ‘off rate’ [34] (such 23. Svennerholm, L. (1972) Methods Curbohydr. Chem. 6,464-474. as lyso-GM2,having no acyl residue, or GM2derivatives with 24. Suzuki, Y. & Suzuki, K. (1972) J . Lipid Res. 13, 687-690.
597 25. Schwarzmann, G. (1978) Biochim. Biophys. Acta 529, 106-114. 26. Handa, S. & Nakamura, K. (1984) J . Biochem. (Tokyo) 95, 1323- 1329. 27. Neuenhofer, S., Schwarzmann, G., Egge, H. & Sandhoff, K. (1985) Biochemistry 24, 525-533. 28. Schwarzmann, G. & Sandhoff, K. (1987) Methods Enzymol. 138, 319 - 341. 29. Stahl, E. & Kaltenbach, U. (1961) J . Chromatogr. 5, 351 -355.
30. Conzelmann, E., Burg, J., Stephan, G. & Sandhoff, K. (1982) Eur. J . Biochem. 123, 455 -464. 31. Fiirst, W., Schubert, J., Machleidt, W., Meyer, H. E. & Sandhoff, K. (1990) Eur. J . Biochem. 192,709-714. 32. Wynn, C. H. (1986) Biochem. J . 240, 921 -924. 33. Wynn, C. H. & Robson, B. (1986) J . Theor. Biol. 123,221 -230. 34. Brown, R. E. Stephenson, F. A,, Markello, T., Barenholz, Y. & Thompson, T. E. (1985) Chem. Phys. Lipids 38, 79 -93.
