The Human GM2 Activator Protein - The Journal of Biological Chemistry

7 downloads 0 Views 2MB Size Report
Jan 25, 2016 - Elke M. Meier, Gunter Schwarzmann, Werner Furst, and Konrad Sandhoffs. From the Institute of Organic Chemistry and Biochemistry, University ...
Vol. 266, No. 3, Issue ofJanuary 25, pp. 1879-1887,1991 Printed in U.S.A.

CHEMISTRY THEJOURNALOF BIOLOCXCAL 0 1991 by The American Society for Biochemistry and Molecular Biology, Inc.

The Human GM2Activator Protein A SUBSTRATE SPECIFIC COFACTOR OF /?-HEXOSAMINIDASE A* (Received for publication, August 15,1990)

Elke M. Meier, Gunter Schwarzmann, WernerFurst, and Konrad Sandhoffs From the Institute of Organic Chemistry and Biochemistry, Universityof Bonn, W-5300 Bonn, Federal Republic of Germany

to ceramide by the sequential action of lysosomal exoglycosiGanglioside GD,,-GalNAc was isolated fromTaySachs brain, tritium-labeled in its sphingosine moiety, dases (for a review see Conzelmann and Sandhoff, 1987). In and its enzymic degradation studied in vitro and in the first step, the terminal GalNAc moiety of the GM2 head cultured fibroblasts. When offered as micelles, GI,’,group is removed by the active site on the a-chain of Hex A GalNAc was almost not hydrolyzed by Hex A or Hex (Kytzia and Sanhoff, 1985). B, while after incorporation of the ganglioside into the The hydrolysis of micellar or liposome-incorporated GMZin outer leaflet of liposomes,the terminalGalNAc residue addition requires the presence of a water-soluble, nonenzywas rapidly splitoff by Hex A. In striking contrast to matic protein, the G Mactivator ~ (Sandhoff et al., 1989). The ganglioside GM2,the major glycolipid substrate of Hex human GMzactivator has been purified from kidney (ConzelA, the enzymic hydrolysis ofGD1,-GalNAc was not promoted by the GMz activator protein, although the mann and Sandhoff, 1979) and liver (Li et aL, 1981b). It is , activatorprotein did bind GDl,-GalNAc to form a essential for the in vivo degradation of G M 2 and G A ~as water-soluble complex. Pathobiochemical studies cor- evidenced by the massive cerebral accumulation of these roborate these results. After incorporation of [aH]GD,,- glycosphingolipidscaused by a deficiency of the GM2 activator protein (variant AB of GM2 gangliosidosis, Conzelmann and GalNAc into cultured skin fibroblasts from healthy subjects and from patients with different variants of Sandhoff, 1978; Sonderfeld et al., 1985a). Besides variant AB, GMZ gangliosidosis, its degradation was found to be two other non-allelic forms ofGM2 gangliosidosis are distinstrongly attenuated in mutant cells with Hex A defi- guished variant B (deficiency of a-chain, Okada and O’Brien, ciencies such as variant B (Tay-Sachs disease),variant 1969; Sandhoff, 1969; Sandhoff et al., 1971; Geigeret al., 1975) B’ and variant 0 (Sandhoff disease), while in cells with and variant 0 (deficiency of P-chain, Sandhoff et al., 1968; variant AB(GMz activator deficiency), its catabolism Sandhoff et al., 1971). In addition, a variant allelic to a-chain was blocked only at the level of GM2. In line with these deficiency has been described that due to a point mutation metabolic studies, a normal content ofGD1.-GalNAc does no longer possess an active site on the P-hexosaminidase was found in brains of patients who had succumbed to a-chain(variant B’, Li et al., 1981a; Kytzia et al., 1983; variant ABof GMz gangliosidosis whereas in brains Sonderfeld et al., 1985b;Tanaka et al., 1988; Ohno and Suzuki, from variants B, B’, and 0 , its concentration wasconsiderably elevated(up to 19-fold). Together with stud-1988; Tanaka et al., 1990). The mechanism of action of the GM, activator has been ies on the enzymic degradation of G Mderivatives ~ with studied in some detail. In contrast to the lysosomal activator modifications in the ceramide portion, these results indicate that mainly steric hindranceby adjacent lipid protein SAP-2 (Ho andO’Brien, 1971; Peters et al., 1977; Iyer molecules impedes the access of Hex A to membrane- et al., 1983; Basu and Glew, 1984; Christomanou et al., 1986; bound GMZ (whose degradation therefore depends on Nakano et al., 1989) that has been proposed to directly stimulate the enzymes glucocerebrosidase and acidic sphingomyesolubilization by the GMz activator) and in addition that the interaction between the G M ~ G . Mactivator ~ complex linase, the GM2 activator, like sulfatide activator (Mehl and and theenzyme must be highly specific. Jatzkewitz, 1964; Fischer and Jatzkewitz, 1975; Fischer and Jatzkewitz, 1978; Inui et al., 1983; Li et al., 1985), acts primarily on the glycolipid substrate. It has been demonstrated by ultracentrifugation, isoelectric focusing, and glycolipid Ganglioside GMZ,’ like other glycosphingolipids,is degraded transfer studies that the GM2 activator extracts ganglioside G M from ~ micelles or liposomes and forms a ganglioside*This work was supported by grantsfromtheDeutscheForactivator complex (Conzelmann et al., 1982) which is then schungsgemeinschaft, from the Fonds der Chemischen Industrie, anddegraded by Hex A (Kytzia and Sandhoff, 1985). from the Studienstiftung des deutschen Volkes. The costs of publiEvidently, Hex A is not able to attack ganglioside GM2 cation of this article were defrayed in part by the payment of page directly. For the glycolipid in the micellar state this is easily charges. This article must therefore be hereby marked “aduertisenent” in accordance with 18 U.S.C. Section 1734 solely to indicate explained by tight packing of the head groups. For membranethis fact. 3 TOwhom correspondence should be addressed. 1Cer; GMla-GalNAc,IV‘GalNAc, I13NeuAc-GgOse,Cer, or GalNAcPlI The abbreviations used are: GMZ, I13NeuAc-Ggose3Cer or 4Gal~1-3GalNAc@l-4Ga1(3-2aNeuAc)@l-4Glc~l-1Cer;G~~~-GalNAc, GalNAc~l-4Gal(3-2aNeuAc)@l-4Glcpl-lCer; N-diazirinyl-lyso-GM2, IV4GalNAc, IVqNeuG1-GgOse4Cer, or GalNAc@1-4Ga1(3-2aNeuGI) 5~4[3(trifluoromethyl)diazirinyl]-phenyl)pentanoyl-N-lyso-GM~; f.w., @l-3GalNAc@l-4Galpl-4Glc@l-lCer; Hex A, @-hexosaminidase A; fresh weight;GalNAc: N-acetyl-galactosamine;GA2, GgosesCer or Hex B, @-hexosaminidase B;4-MUGlcNAc, 4-methyl-umhellifery1-2GalNAcpl-4Gal~l-4Glc@l-1Cer; GD1,, IV3NeuAc, II”NeuAc-Ggose, acetamido-2-deoxy-~-~-glucopyranoside; 4-MU-GlcNAc-6-sulfate, 4NeuAcol2-3Gal~l-3GaINAc@l-4Ga1(3-2olNeuAc)@l-4GlcplCer,or methylumbelliferyl-2-acetamido-2-deoxy-6-sulfo-@-~-glucopyrano1Cer; GD,.-GalNAc, IV4GalNAc,IV3NeuAc, I13NeuAc-Ggose4Cer or side; NBD, 7-nitrobenz-2-oxa-l,3-diazol-4-yl; NeuAc, N-acetyl-neuGalNAc~l-4Gal(3-2aNeuAc)~l-3GalNAc~l-4Gal(~-~~NeuAc)p~raminic acid TDC, taurodeoxycholate; TLC, thin layer chromatog4GlcBl-lCer; G M ~II’NeuAc-LacCer , or Gal(3-2~uNeuAc)@l-4GIcpl- raphy.

1879

1880

The GM2Activator: Mode of Action

TABLE 1 or liposome-incorporated G M ~however, , the mechanistic reaDegradation of GD,.-GalNAc and Gm micelles son for the inaccessibility to Hex A is poorly understood. In ~ p ~ were ) incubated for 1.5 h in GDI.-GalNAc(50 p M ) and G M(50 an attempt to solve this problem, we have studied the degradation of the disialoganglioside GDl,-GalNAcby human 0- 10 mM citrate buffer, pH 4.2 (final volume, 50 pl) with the additions hexosaminidases. This ganglioside contains in an identical listed in the table. After TLC in solvent system A, gangliosides were detected with resorcinol/HCl reagent and densitometrically quanticonfiguration the same three terminal sugars a s G Mbut ~ , two fied (at 580 nm). monosaccharide units more remote fromthe ceramide part of Degradation (% of substrate)" the molecule (Fig. 5A). Studies on the enzymic degradation Gn,.-GalNAc GMZ of GDla-GalNAcin vitro and pathobiochemical investigations should give an answer to the question whether G Mactivator ~ 0 0 TDC (2 mM) 0 0 Hex A (150 milliunits) dependence of degradation is a general feature of membrane89 52 Hex A (150 milliunits) + inserted glycolipid substrates of Hex A or is restricted to some TDC (2 mM) lipid substrates (such as G M and ~ GA2) only. In addition, 0 90 Hex A (150 milliunits) + studies on the interaction between GD1,-GalNAc,G Mactiva~ (15 AU) G Mactivator ~ tor, and Hex A should provide further insight into the subwere performed in duplicates; deviations were strate specificity,mechanism of action, and physiological lessa Determinations than 5%unless indicated. relevance of the GM2 activator protein. ~~

TABLE 2 Degradation of GD,,-GalNAc and GM, after incorporation into liposomes Enzymic Degradationof GD1,-GalNAc by Hex A GDI.-GalNAc (50 p M ) or GM2 (50 p M ) were incorporated into the bilayer half of liposomes (2 mol % of total lipid) and incubated Degradation of GDI,-GalNAc micelles-Micellesof GD~,- outer for 12 h in 10 mM citrate buffer, pH 4.2, 1 mgof bovine serum GalNAc or G M(50 ~ p~ each) were incubated at pH 4.2 with albumin/ml with the additions listed in the table (final volume, 50 Hex A. Without any further addition no degradation of the pl). The degradation was determined as described in the legend of ganglioside substrates was measurable (Table 1). In the pres- Table 1. MATERIALS AND METHODS AND RESULTS~

ence of the Gw2 activator protein there was no hydrolysis of GDl.-GalNAc, whereas under identical experimental conditions 90%of G Mwere ~ converted to G M ~After . addition of the detergent TDC (2 mM), GDI,-GalNAc was hydrolyzed faster than G Mby ~ Hex A. Degradation of GDla-GalNAc after Insertion into Liposomes-GD1,-GalNAc and G M were ~ incorporated intothe outer bilayer half of liposomes. As shown in Table 2, GolaGalNAc (50 p ~ was ) degraded by Hex A at pH 4.2 without any further additions. The hydrolysis of GDla-GalNActo GDla was not influenced by the G M activator ~ protein. Inthe presence of 2 mM TDC the degradation ofGDl,-GalNAc increased from 35 to 95%. Liposome-inserted G M ~on , the contrary, was cleaved by HexA only in the presence of either Gw2activator or TDC. The degradation of liposome-inserted GDl.-GalNAc by Hex A was studied in more detail with [3H] GDla-GalNAc by a concentration of 25 pM (incorporated into liposomes at substrate densities between 1 mol % and 5 mol % of total liposomal lipid). Incubation time and amount of enzyme were chosenin such a way that less than 20% of the substrate were degraded. Under the conditions employed, the reaction was linear up to an incubation time of 60 min (Fig. 1)and up to a Hex A concentration of 18.2 milliunits/ZO pl (data not shown).Hydrolysis of 25 PM ['H]GD1,-GalNAc (substrate density, 1 mol %) proceeded at a rate of 48 nmol/ haunitsHex A. In the absence of Hex A there was no measurable hydrolysis of [3H]GDI,-GalNAcat pH 4.0, 37 "C forat least 12 h. The enzymic degradation of [3H]G~1.-GalNAc was not stimulated by the G Mactivator ~ protein (Fig. 1). As shown in Fig. 2a, the reaction displayed a pH optimum between pH 3.5 and 4.2. The degradation of ['H]GD1,-GalNAc was inhibited by an increase of the buffer concentration (Fig. Zb),by bovine serum albumin (Fig. 2c) and by liposomes without substrate (Fig. 2d). As evident from Fig. 3, the reaction rate was strongly dependent on the substrate density (mol % GDIa-GalNAc of total liposomal lipid) on the liposomal Portions of this paper (including "Materials and Methods," Fig. 2, Tables 5and 6) arepresented in miniprintat the end of this paper. Miniprint is easily read with the aid of a standard magnifying glass. Full size photocopies are included in the microfilm edition of the Journal that is available from Waverly Press.

Degradation (% of substrate)" Go,.-GalNAc

GM~

0 35 93

0 33

36

26

35

55

TDC (2 mM) Hex A (150 milliunits) Hex A (150 milliunits) + TDC (2 mM) Hex A (150 milliunits) + GM2 activator (5 AU) Hex A (150 milliunits) + GM2 activator (15 AU) Determinations were performed less than 5% unless indicated.

0

in duplicates; deviations were

FIG. 1. Time dependence of degradation of ['H]CDI.GalNAc. 25 p~ [3H]GDl.-GalNAc,incorporated into liposomes (1 mol % of total lipid), were incubated in 250 pl of 10 mM citrate buffer, pH 4.2, with 15.6 milliunits of Hex A (17.8 units/mg protein) in the (2.5 p ~ G) Mactivator ~ absence (0)and presence (A)of62.5AU protein. After different incubation times, aliquots were analyzed by TLC in solvent system A and fluorography (see inset). [3H]G~~.GalNAc and [%]GD,. were quantified by scraping off of radioactive bands and liquid scintillation counting.

surface. The reduction of the substrate density from 100 mol % (pure micelles) to 30 mol % led to a gradual increaseof the reaction rate from about 0.8 nmol/h. unit to 20 nmol/h. unit. A further decrease of the substrate density caused a steeper increase of the reaction rate, reaching a plateau (80-90 nmol/ h-unit) between 8 mol % and 2 mol % substrate density. A t substrate densities below 2 mol % the reaction velocity de-

The GM2Activator: Mode of Action

1881

porated [3H]G~l,-GalNAcwere degraded to neutral lipids by fibroblasts fromallvariants of GMZ gangliosidosis. In the enzyme-deficient variants, more than 50% of the radioactively labeled monosialoganglioside fraction consisted of [3H]G~l.GalNAc, which could have been formed by the action of a neuraminidase on [3H]GD1.-GalNAc.The radioactive monosialoganglioside fraction of variant AB fibroblasts contained mainly (more than 85%) [3H]GM2.The radioactively labeled disialoganglioside fraction of all cell lines consisted solely of [3H]GDl,-GalNAc. R e l o t l v e o m o m t o f G O ~ ~ - G O I N A ICm a l % of tot01 lipldl Cerebral Storage-The cerebral storageof GD1,-GalNAcwas analyzed by ganglioside extraction from the gray matter of FIG. 3. Substrate density dependence of [3H]Gol.-GalNAc different sets of experiments, [3H]G~l,- human brain, separation by TLC and densitometric quantidegradation. Inthree GalNAc was incorporated at various densities (0.5-50 mol % of total fication. The presumptive GDl,-GalNAc band was shown to lipid) into the outer bilayer half of liposomes. [3H]GD,.-GalNAca t a be resistent to neuraminidase (Vibrio cholerae) and to be final concentration of 25 p~ either in micellar form (substrate dendegraded by human Hex A in the presence of 2 mM TDC sity, 100 mol %) or after incorporation into liposomes was incubated (data not shown). As shown in Table 4, the GD1.-GalNAc for 20 min at 37 "C with 3.6 milliunits of Hex A in 10 mM citrate buffer, pH 4.2 (final volume, 20 p l ) . Preparation of liposomes and concentration in two different brains from patients with GMZ insertion of [3H]GDl.-GalNAcwere performed as described under activator deficiency (variant AB) was not increased over that "Materials and Methods." Degradation of [3H]G~1,-GalNAcwas de- of two control brains (2 nmol GD1,-GalNAc/g f.w.). Cortex termined as described in the legend of Fig. l. tissue from enzyme-deficient variants 0, B, and B', however, contained between 5.7- and 19-fold more GDI,-GalNAc than creased and reached a value of about 50 nmol/h. unit HexA the control brains. at a substrate densityof 1 mol %. the GMP Activator-As Binding of [3H]GDI,-GalNActo shown in Fig. 1 and Tables 1 and 2, the degradation ofGD1,We were not able to determine exact values for the apparent kinetic parameters KM and Vmaxfor the hydrolysis of lipo- GalNAc by Hex A was not stimulated by the G M activator ~ some-incorporated [3H]GDl.-GalNAc. By incubating10 t o 250 protein. To find out if this was due to a lack of complex g M [3H]GD1,-GalNAc(5 mol% in liposomes) with 3.6 milliunformation between ganglioside and GMz activator, the binding its Hex A in 10 mM citrate buffer, p H 4.0, for 10 or 20 min of ['HH]GD,,-G~~NAC to the GM:!activator protein was exam(data not shown),we estimated the KMvalue to be >0.5 mM ined. Complex formation between [3H]GD1,-GalNAcand GMZ and the V,,, value to be >2000 nmol/h .unit, equivalent to activator was demonstrated for [3H]GD1,-GalNAcmicelles by >3 pmol/min.mg (based on the specific activity of 90 units/ cosedimentation of the ganglioside with the G Mactivator ~ in mg, measuredfor highly purified Hex A (Sandhoff et al., a sucrose density gradient by ultracentrifugation (data not 1977)). When we repeated the experiment at substrate con- shown) and for liposome-inserted [3H]G~1,-GalNAc by meascentrations of up to 1 mM employinga 25 mM liposome uring theganglioside transfer activityof the GM2 activator. In preparation(containing some 1.6% of lipid inwater), we the transfer assay, the GM2 activator-mediatedtransfer of observed drastically reduced degradation rates. We assume [3H]GD1,-GalNAcand ['HH]Gm2 from one liposome subpoputhat the preparation of liposomes from such concentrated lation to anotherwas measured. As shown in Fig,4, [%]GD1,lipid solutions yields poorly defined aggregates in which the GalNAc was transferred by the GM2 activator protein with a substrate is hardly degradable by Hex A. rate of 3.6 nmol/AU. h. For ['H]GM2 under the same experiThe degradation rateof [3H]Gol,-GalNAc by Hex B, meas- mental conditions a transfer rate of 10.7 nmol/AU. h was ured under the conditions describedabove, was very low, determined. approximately 5-8%of the rate measured with Hex A and Degradation of G MDerivatives ~ in Micelles and Liposomeswas not influenced by the GMz activator protein (data not The enzymic degradation of several Gm2 derivativeswith shown). alterations in the hydrophobic part of the molecule (Fig. 5b) was studied. When incubated as micelles, all of the testedGM2 Pathobiochemical Studies derivatives were degradedby Hex A without theGM2 activator To find out whether the degradation of GD,,-GalNAc by protein, whereas theenzymic hydrolysis of native GMz under the same experimental conditions was strictly dependent on Hex A is independent of the GM2 activator protein also in uiuo, the metabolism of [3H]G~l.-GalNAc in fibroblast cul- the presence of the GM? activator or of TDC (Table 5). The ~ led to a strong stimulation of tures and the cerebral storage of GD1,-GalNAc were studied addition of the G M activator the degradation of all GM2 derivatives.After incorporation in various variants of GM2 gangliosidosis. Cell Culture Studies-Confluent fibroblasts were fed for 72 into liposomes, the Gw2 derivatives with a hydrophobic fatty were h with 10 p M [3H]G~1.-GalNAc.After harvesting the cells, acyl substituent (NBD-CG-lyso-GM,,diazirinyl-lyso-GM2) lipids were extracted and separatedby ion exchange chroma- no longerdegradable by Hex A without the GM2 activator and G Mderivatives ~ with a short tography intodisialogangliosides, monosialogangliosides, and protein (Table 5). LYSO-GM~ neutral lipids. As shown in Table 3, control fibroblasts me- fatty acyl substituent (N-acetyl-lyso-GM2, N-bromacetyl-lysotabolized about 6.7%of the incorporated [3H]GD,.-GalNAc G M ~were ) still degraded by Hex A without any addition but into monosialogangliosides and about 12% into neutral lipids at a slower rate than in micelles. The cleavage ofall GMZ (mainly sphingomyelin, phosphatidylcholine, lactosylceram- derivatives inserted intoliposomes was stimulated by the GM2 ide, and glucosylceramide, as revealed by TLC). Fibroblasts activator (Table 5). from patientswith GM2 activator deficiency (variantAB) Degradation of GMlb-GalNAc"The degradation ofGMlbdegradedmore than 12%of theincorporated [3H]GD1,- GalNAc by Hex A was studied in the micellar and in the GalNAc into monosialogangliosides, whereas for fibroblasts liposomal system as described above for GM? derivatives. As of the enzyme-deficient variants B,B', and 0 the correspond- shown in Table 6, the hydrolysis of micellar GMlb-GalNAc(50 ing value amounted to 1.9% or less. Less than 0.5% of incor- p M ) was slightly stimulated by the G Mactivator ~ protein (13%

1882

The GM2Activator: Mode

of Action

TABLE3 Distribution of radioactivity between different lipid fractions from normal and GM2 gangliosidosis fibroblasts after incorporation of PH]G~l,-GalNAc Confluent fibroblasts were grown for 72 h in 10 ~ L M[3H]G~l.-GalNAc.Lipids were extracted from the harvested cells and separated into neutral lipids, monosialogangliosides,and disialogangliosides by anion exchange chromatography as described under “Materials and Methods.” The radioactivity in the lipid fractions was determined by liquid scintillation counting. Deviations of the duplicate determinations were less than 5%. Radioactivity (%)

Gol.-GalNAc Cell line Monosialo-gangliosides Neutral incorporation

Disialo-gangliosides lipids

nmol/mg. 72 h

Control

6.6

11.9 11.8 11.9

11.7 12.5 12.2

6.9

81.7 80.7 81.3

13.7 12.1 11.7

0.5 0.4 0.6

12.3 13.7 14.3

87.2 85.9 85.3

B1 B2

11.3 11.4

0.2 0.1

1.7 1.6

98.1 98.3

B’

13.5

0.2

1.8

98.0

01 02

10.0 12.8

0.1 0.1

1.9 1.8

98.0 98.1

6.5 Gw2gangliosidoses AB 1 AB 2 AB 3

TABLE 4 Cerebral storage of GO,.-GalNAc The G”,.-GalNAc content of human brain tissue was determined by densitometric quantification after lipid extraction, partitioning according to Folch’s procedure, enzymic digestion with neuraminidase (from Vibrio cholerae) and TLC, as described under “Materials and Methods.” (brain) Tissue

Control 1 14 Control 2

GDh

GMZ

GDK GalNAc

nmol/g f . w .

nmol/g f.w.

nmol/g f.w.

640 520

24

2 2

Gw2 gangliosidoses

AB 1 AB 2

3325 3688

B’ 1 B’ 2 12

0 B

322

144 218

2 2

4093 28 116

2381 4254

150 2980

2651

of substrate hydrolyzed compared with 88% for GM2 in the presence of 15 AU G Mactivator). ~ After insertion into liposomes GMlb-GalNAc, like GD1.-GalNAc, wasdegraded by Hex A without the Gh.12 activator protein (Table 6). The degradation was slightly stimulated by the GM2 activator (70 and 80% of substrate hydrolyzed in the presence of 5 and 15 AUGM2 activator, respectively, compared with 58% of substrate hydrolyzed without the G Mactivator). ~ DISCUSSION

The degradation of gangliosides GMz and GA2by human lysosomal Hex A is strictly dependent on the presence of the GM.. activator protein (Conzelmann and Sandhoff, 1978; Sonderfeld et al., 1985a; Sandhoff, K., 1984), which has been shown to extract gangliosides from micelles or liposomes (Conzelmann et al., 1982). Other physiological or artificial phexosaminidase substrates such as oligosaccharides or synthetic glycosides, however, are cleaved by Hex A independ-

t 500-

0.01 MZ oa5 t$q2 Actlvator[AL&say]

0.1

FIG. 4. Transfer of gangliosides [SH]Gol.-GalNAcand [3H] G M from ~ donor to acceptor liposomes by the G M activator ~ protein. Donor liposomes (250 nmol of lipid) containing8 mol % of phosphatidic acid and 2 mol % of [’H]G,, (0,A) or 2 mol % of [3H] GD,.-GalNAc (0)were incubated with acceptor liposomes (250 nmol of lipid), containing 2 mol % of phosphatidic acid, in the presence of 4 pg of bovine serum albumin and of the amount of GMZ activator protein, indicated at the abscissa, for 30 min at 37 “C in 40 p1 of 50 mM citrate buffer, pH 4.2. Acceptor liposomes were separated from donor liposomes on DEAE-cellulose columns, as described (Conzelmann et al., 1982). The preparation of liposomes was performed as described (Conzelmann et al., 1982). The transfer of gangliosideswas calculated from the 3H/14Cratio inacceptor liposomes. Controls were run without G,, activator proteinand subtracted from the respective values. Determinations were performed as duplicates; deviations were less than 5% unless indicated by vertical bars.

ently of the GMZ activator protein (Kresse et al., 1981; Bearpark and Stirling, 1985; Thompson et al., 1973; Gloss1 and Kresse, 1982). Evidently, the enzyme cannot directly attack glycolipids GM2and GA2. For liposome-inserted or membranebound glycolipids the inaccessibility for the enzyme is poorly understood. It may be that thehigh molecular weight enzyme Hex A is not able to approach the surface of membranes closely enough to bind to the terminal GalNAc residue of GM, and GA2. Based on the finding of Li et al. (1981b) that the oligosaccharide OfGMZ cannot be cleavedby the enzyme either, however, it may alternatively be assumed that thehead groups of glycolipids G Mand ~ G Aexist ~ in specific conformations that are not recognized as substrates by Hex A. To obtain more

The GM2Activator: Mode

\

$" B

4

no no

HNk

N

N

'0'

N-NBD-C6-Lyro-GM2

insight into theinteraction between Hex A, GMZ activator and glycolipid substrates we have studied the enzymic degradation of GD1,-GalNAc. After incorporation of GDla-GalNAcinto liposomes it was degraded by Hex A without GMz activator or detergent (Table 2, Figs. 1-3). Hex A contains two different active sites being present on polypeptide chains a and 6, respectively (Kytzia and Sandhoff, 1985). GD1,-GalNAc is preferentially cleaved by the active site on the a-chainof Hex A, as can be deduced from the storage of GDI,-GalNAc in the brains of patients with Hex A-deficient variants B or B' of GMz gangliosidosis (see below), fromthe modest degradation by Hex B (less than 8%of hydrolysis by Hex A) and from the pH dependence of the reaction. GDI,-GalNAcdegradation displays an optimum between pH 3.5 and 4.2 (Fig. ea) which resembles those of the hydrolysis of 6-sulfated artificial substrates by the active site on the a-chain of Hex A (Kytzia and Sandhoff, 1985; Kresse et d., 1981). Micellar GD1,-GalNAc was only a poor substrate of Hex A (Table 1, Fig. 3). The incorporation of GDI.-GalNAc into increasing amounts of liposomes led to an increase of the reaction rate reaching a maximum at substrate densities between 8 and 2 mol % (of total lipid). To show that GD,,.-GalNAc is not only in vitro but also in vivo degraded by Hex A independently of the GMZ activator protein, we have studied the metabolism of [3H]GDI,-GalNAc in fibroblast cultures (Table 3) and the cerebral storage of GD1,-GalNAc in various variants ofGMz gangliosidosis. In fibroblasts from variant AB of GMz gangliosidosis, more than 12% of the incorporated ganglioside wereconverted into monosialogangliosides. Control fibroblasts metabolized about 6.7% of the incorporated [3H]GD1,-GalNActo monosialogangliosides and about 12% to neutral lipids. As expected from the strict GM2 activator dependence ofGMz degradation, the degradation of [3H]GDl.-GalNAc in variant AB fibroblasts stopped at thelevel of [3H]GM2, which accumulated 4-fold as compared with control cells. Fibroblasts from patients with

of Action

1883

the enzyme deficient variants B, 0, or B' of GMZ gangliosidosis converted less than 1.9% of the incorporated [3H]G~1,GalNAc into monosialogangliosides. In two variant AB brains no increase ofGD1.-GalNAc content over that of two control brains was found (Table 4). In brains from patients with variants B, 0, and B' a 5.2-19fold accumulation of GD1,-GalNAc wasdetermined. The comparatively small accumulation of GDl,-GalNAc (5.2-fold) and GM2 (17-fold) we found for the variant 0 brain may be explained by the fact that the patienthad already died perinatally. For variant B the cerebral storage of GD1,-GalNAc has already been demonstrated. Iwamori and Nagai (1979) have measured a 9.2-fold accumulation of GDla-GalNAcin a Tay. found a 19-fold (for gray Sachs brain and Yu et ~ l (1983) matter)and 10-fold (for white matter) increase of GD~,GalNAc content in Tay-Sachsbrain over that of normal infant brain. In contrast, Svennerholm and co-workers (Rosengren et d., 1987) found no accumulation of GD1.-GalNAc in a variant 0 brain and a modest increase (less than 2-fold) of GD1,-GalNAc content in avariantBbrain. The GD~,GalNAc content they measured for normal infant brain (54 and 73 nmol/g f.w., Rosengren et ~ l . 1987) , is 27-56-fold higher than thatdetermined by Iwamori and Nagai (1.3 nmol/g f.w., Iwamori and Nagai, 1979) and by us (2 nmol/g f.w.) and at . nmol/g least 10-fold higher than that found by Yu et ~ l (5.7 f.w. for gray substance and4.8 nmol/g f.w. for white substance, Yu et ~ l . , 1983). Previously, however, Svennerholm et ~ l . (1973) reported on the isolation of GDl,-GalNAc fromnormal infant brain with a yield of 1.8 nmol of GD1,-GalNAc/gf.w. which is similar to the low GDI,-GalNAc levelsmeasured by us andby Iwamori and Nagai. In any case, our data, like those from Iwamori and Nagai and those from Yu et ~ l . ,are in accordance with the results presented here for the enzymic degradation of [3H]G~1,-GalNAcinvitro and in fibroblast cultures. GD1,-GalNAc is the first glycolipid substrate of Hex A which unequivocally has been shown to be degraded without the GM2 activator protein. Enzymic hydrolysis of globoside may also be independent of the G Mactivator, ~ as no increase of globoside content was found in brain tissue from variant ABof GMz gangliosidosis (Sandhoff et ~ l . ,1968, 1971). Evidently, GM2 ativator dependence of degradation is nota general feature of glycolipid substrates of Hex A. From our results we conclude that the reason for the strict GMz activator dependence ofGM2 degradation lies in a specific structural feature of ganglioside GM2 that is not shared by GDI,-GalNAc. For this two possibilities exist. 1) The headgroup of GM2 is present in a specific conformation that on principle is not degradable by Hex A. 2) The oligosaccharide chain of GM2is degradable by Hex A without the GM2 activatorprotein; in micelles or membranes, however, it is shielded from the enzymic attack by the headgroups of adjacent lipid molecules. To decide between these alternatives, we have studied the GMPactivator dependence of degradation ofGM2 derivatives that lack the fattyacid or carry a substituted fattyacyl chain (see Fig. 5b). Due to increased polarity or steric hindrance within the hydrophobic part of the molecule such GM2 derivatives are expected to have higher critical micelle concentrations and/or to form micelles of lower stability than GM2. In fact, they were shown to insert spontaneously and rapidly into liposomes and biological membranes (data not shown). Thus, in the micellar system these derivatives should be degraded by Hex A without the GM2 activator protein, if the GMz head group per se is a substrateof the enzyme (possibility 2 as discussesd above). As shown in Table 5, when presented as micelles, all of the tested derivatives were in fact degraded

1884

The GM2Activator: Mode of Action

by Hex A without the GM2 activator protein. After incorpo- significantly lower rate than GDl,-GalNAc (1.4 nmol/AU . h, ration into liposomes the degradation of derivatives carrying Conzelmann et al., 1982). long hydrophobic fatty acyl substituents ( N B D - C ~ - ~ ~ S O - G M We ~ , propose the following model for themechanism of diazirinyl-lyso-Gwp)became strictly dependent on the GM2 action of the G Mactivator ~ protein, explaining its inability to and GMZ derivatives stimulate GDl.-GalNAc degradation. The GMZ activator conactivator protein (Table 5). LYSO-GM~ with a small fatty acyl substituent (acetyl-lyso-GM2,broma- tains an oligosaccharide-binding domain, which recognizes Cetyl-lyso-GM2),however, were still degradable without the sialic acid and GalNAc, and a hydrophobic domain that fixes GM2 activator protein in theliposomal system, but at a lower the ganglioside molecule by interaction with its ceramide rate than in micelles. The difference between the two groups portion. The distance between the domains is such thatGM2 of GM2 derivatives may be explained by their different stability (and GA2) fit exactly into both.For GDl,-GalNAc, in contrast, of association with the lipid bilayer. the binding of the ceramide part to thehydrophobic domain These results indicate that the GM2 activator dependence of GMz activator directs the inner sialicacid and GalNAc of the degradationof ganglioside GM2 by Hex A is not due to residues intothe oligosaccharide-bindingdomain. As this a specific inaccessible conformation of the GM2 head group interaction is somewhat hindered by the terminal sugars, a but may be caused by the steric hindrance of the interaction lower transfer rate, being similar to that ofGDl,, results. between enzyme and substrateby the headgroups of adjacent According to the model, the GM2 activator in addition binds lipid molecules in membranes. For GD,,-GalNAc such a hin- specifically to Hex A. The protein-protein interaction directs drance appears tobe absent or greatly reduced. The terminal the GalNAc residue which is bound to the oligosaccharideGalNAc residue of GDl,-GalNAc is two monosaccharide units binding domain of the activator protein into the active site more remote from the ceramide portion of the molecule than on the a-chain of Hex A. In the case of GDl,-GalNAc, this is theterminal GalNAcresidue of GM2, the difference in a the inner N-acetyl-galactosamine unit, towhich a trisacchastretched Clignment of the head groups being equivalent t o ride is bound. It likely impedes a tight fit between the inner about 11 A. For the lowestenergy conformations of G M ~ GalNAc residue and the active siteso that enzymic cleavage (Wynn and Robson, 1986) and GDI,-GalNAc, predicted by an cannot occur. This model is supported by the following findings. energyminimization technique(WynnandRobson, 1986; 1)The oligosaccharide specificity of the G Mactivator ~ proWynn et al., 1986), the distance between the 0-glycosidic by comparing the transfer rates for bond thatis cleaved byoHex Aand the primary hydro5yl group tein has been shown of sphingosine is 9.2 A for G M and ~ 13.5 or 14.6 A for two various glycosphingolipids (Conzelmann et al., 1982). It was demonstrated that theremoval of sialic acid or GalNAc from GD1,-GalNAc conformer^.^ If gangliosides insert into phospholipid bilayers insuch a way that the double bond of the GMZ molecule significantly reduces the glycolipid binding it reported that sphingosine is adjacent to the sn-Cl-estergroup of phospho- to theGM2 activator protein. In addition, was lipids, it follows from the size of phospholipid head groups after removal of the negative charge from the N-acetyl-neu(Levine et al., 1968; Franks andLieb, 1979; Zaccai et al., 1979) raminic acid residue GM2 was no longer hydrolyzed by Hex A and from the predicted ganglioside conformers that the Hex in the presenceof the GM2 activator (Li et al., 1984). 2) Evidencefor theexistence of ahydrophobic-binding A-susceptible glycosidic bond of the GM2 molecule is only 2.2 domain on the GMz activator protein not only comes from A distant from the membrane surface, whereas ip the GD1,GalNAc molecule itprotrudes for 6.5 or 7.6 Afrom the energetical considerations4 andfrom the binding of the protein to octyl-Sepharose columns (Conzelmann and Sandhoff, membrane surface. This differencemay be crucial for the GM2 activator is accessibility to the high molecular weight enzyme Hex A. In 1979), but also from the finding that the accordance with this conceptG Adegradation ~ isGM2 activator stronglyinhibited by covalentbindingto GM2 derivatives dependent (Conzelmann and Sandhoff, 1978), whereas GMlb- carrying areactive group in the hydrophobic part of the GalNAc, inserted into liposomes, is degraded by Hex A with- molecule (Neuenhofer and Sandhoff,1985).5 In addition, the binding of GMz to theGM2 activator was found tobe inhibited out the GMz activator protein (Table 6). The degradation of GDl,-GalNAc in micelles (Table 1)and only by a large excess (at least 104-fold) of GalNAc, NeuAc, in liposomes (Fig. 1, Table 2) is not stimulated by the G M ~ or the GM2 oligosaccharide,6 and cleavage of the GMz oligosacactivator. This is not due to alack of complex formation charide was not prompted by the GMz activator (Li et al., low GM2 activator-mediated between GD1,-GalNAcand GM2 activator, as bindingof [3H]- 1981b).A furtherhintisthe GD1,-GalNActo the GM2 activator was demonstrated for the transfer of [3H]GDI,-GalNAc between liposomesas compared been discussed above. micellar system by ultracentrifugation (data not shown) and with [3H]GM2 that has A specific binding of the GM2 activator protein to Hex A for the liposomal system by measuring the GMz activatordependent ganglioside transfer rate (Fig. 4). With both meth- has already been deduced from the inhibitory effect of the ods less binding was found for [3H]GD1,-GalNActhan for [3H] ganglioside-free GM2 activator on the degradation of 4-MUGM2 (ultracentrifugation, 28% of [ 3 H ] G ~binding; 2 transfer, GlcNAc-6-sulfate by Hex A (Kytzia and Sandhoff, 1985). In 35% of [%]GM2 binding). [3H]GD1,-GalNAcwas transferred this work further evidence is presented. The apparent K M with a rate (3.6 nmol/AU. h) similar to that determined for value of the GMZ activator-independent degradation of GD1,GalNAc (>0.5 mM) lies in the range of those determined for [“H]GD1,(4.1 nmol/AU.h, Conzelmann et al., 1982). For [3H] synthetic glycoside substrates of the active site on the a-chain GM2 a transferrate of10.7 nmol/AU. hwas determined, compared with 10 nmol/AU. h measured previously (Conzel- (4-MU-GlcNAc-6-sulfate,KM = 0.31 mM (Kytzia and SandKM = 1.25 mM of GM2 hoff, 1985); p-nitrophenyl-GlcNAc-6-sulfate, mann et al., 1982). The difference in the transfer rates (Kresse et al., 1981)). For the enzymic hydrolysis of the G M ~ * and GD1,-GalNAc cannot explain the lack of stimulation of GM2 activator complex and of the G A ~ . Gactivator M~ complex, GDI,-GalNAc degradation by Hex A. For,underidentical experimental conditions, the degradation of GA2 in the micelThe change of free energy for the transfer of ceramide with a CIS lar and liposomal system was induced by the GM2 activator fatty acid from a hydrophobic environment into water is about 151 (data not shown), although this glycolipid is transferred at a kJ/mole (in: Tanford, C., 1980). C. Wynn, personal communication.

E. M. Meier and K. Sandhoff, unpublished results. G. Stephan and K. Sandhoff, unpublished results.

The GM2Activator: Mode of Action on the other hand, KIMvalues of 1.9 and 0.9 PM,respectively, have been measured (Kytzia and Sandhoff, 1985;Conzelmann and Sandhoff, 1979). From the finding that GD1,-GalNAc binds to Hex A with a similar low affinity as water-soluble glycosides, we conclude that the high affinity characteristic of the GMZ-GMz activator and of the G A ~ GM . z activator complex is not a feature of the ganglioside substrates but due to a protein-protein interaction between the GM2 activator and Hex A. It has been reported that, in contrast to GDla-GalNAc, degradation of GMMlb-GalNAc (Itoh et al., 1981), GalNAcB14Ga1(3-2aNeuAc)@1-4GlcNAc@l-3Gal@l-4Glc@l-lcer(DeGasperi et al., 1987), and GalNAc@l-4Gal(3-2aNeuAc)@l4GlcNAc@l-3Ga1(4-1@GalNAc)@l-4Glc@1-lcer(DeGasperi et al., 1987) by Hex A is stimulatedby the GM2 activator protein, but no comparison with the stimulatory effect on GM2 degradation was performed. We found that stimulation of G ~ l b GalNAc hydrolysis is much weaker than that of GM2 degradation (Table 6). These results are in line with our model for the mechanism of action of the GM2 activator protein. All glycolipids with a longer oligosaccharide chain than GMZ, for which some stimulation of degradation by the GM2 activator has been reported, do not contain sialic acid at the inner galactose residue. Thus,it may be assumed that they, in contrast to GDI.-GalNAc, tend to bind to the G Mactivator ~ with their terminal sialic acid and N-acetyl-galactosamine residues, even if this causes some distortion of the glycosphingolipid head group. Consequently, at least some molecules are correctly presented to the active site of Hex A and can be cleaved. In summary, our studies show that the Ghl2 activator displays a so far unknown degree of substrate specificity. The GM* activator is a cofactor that seems to have been “tailored to induce the degradation of glycolipids GM2 and GA2by Hex A. With regard to themechanism of action the GM2 activator differs fundamentally from the sulfatide activator (“dispersin” (Li et al., 1986)) that has been shown to act like a “biodetergent,” exhibiting a very broad enzyme and substrate specificity (Li et al., 1988). Acknowledgment-We thank Dr. Jasna Peter-Katalinic (Bonn) for the FAB-MS spectra, Dr. Robert K.Yu (Richmond, VA), and Dr. Akemi Suzuki (Tokyo) for their generous gifts of GD1.-GalNAc and GMlb-Ga1NAc,respectively, Dr. Kunihiko Suzuki (Chapel Hill) for reading the manuscript and Otto Kiipper for his help with the preparation of G Mactivator ~ and p-hexosaminidases. REFERENCES Basu, A., and Glew, R. H. (1984) Biochem. J. 2 2 4 , 515-524 Bearpark, T. M., and Stirling, J. L. (1978) Biochem. J . 173, 9971000 Beutler, E., and Kuhl, W . (1975) Nature 2 5 8 , 262-264 Bradford, M. M. (1976) Anal. Biochem. 72,24&254 Burg, J., Banerjee, A., Conzelmann, E., and Sandhoff, K. (1983) Hoppe-Seyler’s 2. Physiol. Chem. 364, 821-829 Christomanou, H., Aignesberger, A., and Linke, R. P. (1986) Biol. Chem. Hoppe-Seyler 3 6 6 , 245-256 Conzelmann, E., and Sandhoff, K. (1978) Proc. Nutl. Acad. Sci. U.S. A. 75,3979-3983 Conzelmann, E., and Sandhoff, K. (1979) Hoppe-Seyler’s Z. Physiol. Chem. 3 6 0 , 1837-1847 Conzelmann, E., Burg, J., Stephan,G., and Sandhoff, K. (1982) Eur. J . Biochem. 1 2 3 , 455-464 Conzelmann, E., Kytzia, H.-J., Navon, R., and Sandhoff, K. (1983) Am. J . Hum. Genet. 35,900-913 DeGasperi, R., Koerner, T. A. W., Quarles, R. H., Ilyas, A.A., Ishikawa, Y., Li, S.-C., and Li, Y.-T. (1987) J. Biol. Chem. 262, 17149-17155 Egge, H., and Peter-Katalinic, J. (1987) MassSpectrome. Reu. 6 , 331-393 Egge, H., Peter-Katalinic, J., Reuter, G., Schauer, R., Ghidoni, R.,

1885

Sonnino, S., and Tettamanti, G . (1985) Chem. Phys. Lipids 3 7 , 127-141 Fischer, G., and Jatzkewitz, H. (1975) Hoppe-Seyler’s 2. Physiol. Chem. 356,605-613 Fischer, G., and Jatzkewitz, H.(1978) Biochim. Biophys. Acta 5 2 8 , 69-76 Folch, J., Lees, M., and Sloane Stanley, G. H. (1957) J. Biol. Chem. 226,497-509 Franks, N. P., and Lieb, W . R. (1979) J. Mol. Biol. 133,469-500 Geiger, B., Navon, R., Ben-Joseph, Y., and Arnon, R. (1975) J. Biochem. (Tokyo)56,311-318 Glossl, J., and Kresse, H. (1982) Biochem. J. 203,335-338 Ho, M. W., and O’Brien (1971) Proc. Nutt. Acud. Sci. U. S. A. 68, 2810-2813 Inui, K., Emmett, M., and Wenger, D. A. (1983) Proc. Natl. Acud. Sei. U.S. A. 80,3074-3077 Itoh, T., Li, Y.-T., Li, S.-C., and Yu, R. K. (1981) J. Bioi. Chem. 256, 165-169 Iwamori, M., and Nagai, Y. (1979) J. Neurochem. 32,767-777 Iyer, S. S., Berent, S. L., and Radin, N. S. (1983) Biachim. Biophys. Acta 7 4 8 , l - 7 Kresse, H., Fuchs, W., Glossl, J., Holtfrerich, D., and Gilberg, W. (1981) J. Biol. Chem. 256, 12926-12932 Kytzia, H.-J., and Sandhoff, K. (1985) J. Biol. Chem. 2 6 0 , 75687572 Kytzia, H.-J., Hinrichs, U., Maire, I., Suzuki, K., and Sandhoff, K. (1983) EMBO J. 2, 1201-1205 Levine. Y. K.. Bailev. _ .A. J.. and Wilkins. M. H. F. (1968) Nature 220,577-578 Li. S.-C.. Nakamura. T.. Oeamo. A.. and Li.Y.-T. (1979) . , J. Biol. Chem.254, 10592110595 Li, S.-C., Hirabayashi, Y., and Li, Y.-T. (1981a) Biochem. Biophys. Res. Commun. 101,479-485 Li, S.-C., Hirabayashi, Y., and Li, Y.-T. (1981b) J. Biol. Chem. 2 5 6 , 6234-6240 Li, S.-C., Serizawa, S., Li, Y.-T., Nakamura, K., and Handa, S. (1984) J. Biol. Chem. 259,5409-5410 Li, S.-C., Kihara, H., Serizawa, S., Li, Y.-T., Fluharty, A. L., Mayes, J. S., and Shapiro, L. J. (1985) J. Biol. Chem. 260,1867-1871 Li, S.-C., and Li, Y.-T. (1986) in Enzymes of Lipid Metabolism I1 (Freysz, L., Dreyfus, H., Massarelli, R., and Gatt, S., eds) pp. 307314, Plenum Publishing Co., New York Li, S.-C., Sonnino, S., Tettamanti, G., and Li, Y.-T. (1988) J. Biol. Chem. 263,6588-6591 Lowry, 0. H., Rosebrough, N. J., Farr,A. L., and Randall, R. J. (1951) J. Biol. Chem. 193,265-275 Mehl, E., and Jatzkewitz, H. (1964) Hoppe-Seyler’s Z. Physiol. Chem. 339,260-276 Nakano, T., Sandhoff, K., Stumper, J., Christomanou, H., and Suzuki, K. (1989) J . Biochem. 105,152-154 Neuenhofer, St., and Sandhoff, K. (1985) FEBS Lett. 185,112-114 Ohno, K., and Suzuki, K. (1988) J . Neurochem. 50,316-318 Okada, S., and O’Brien, J. S. (1969) Science 165,698-700 Peters, S. P., Coyle, P., Coffee, C. J., Glew, R. H., Kuhlenschmidt, M. S., Rosenfeld, L., and Lee, Y. C. (1977) J. Biol. Chem. 2 5 2 , 563-573 Randerath, K. (1970) Anal. Biochem. 34, 188-205 Rosengren, B., Mansson, J.-E., and Svennerholm, L. (1987) J. Neurochem. 49,834-840 Sandhoff, K. (1969) FEBS Lett. 4 , 351-354 Sandhoff, K., Andreae, U., and Jatzkewitz, H. (1968) Puthol. Eur. 3 , 278-285 Sandhoff, K., Harzer, K., Wassle, W., and Jatzkewitz, H. (1971) J. Neurochem. 18,2469-2489 Sandhoff, K., Conzelmann, E., andNehrkorn,H. (1977) HoppeSeyler’s Z. Physiol. Chem. 3 5 8 , 779-787 Sandhoff, K., Conzelmann, E., Neufeld, E., Kaback, M.M., and Suzuki, K. (1989) The Metabolic Basis of Inherited Diseuses (Scriver, Ch., Beaudet, A. L., Sly, W . S., and Valle, eds), 6th edition, David McGraw Hill, NY Schwarzmann, G. (1978) Biochim. Bwphys. Acta 5 2 9 , 106-114 Schwarzmann, G., and Sandhoff, K. (1986) Methods Enzymol. 1 3 8 , 319-341 Schwarzmann, G., Hoffmann-Bleihauer, P., Schubert, J., Sandhoff, K., and Marsh, D. (1983) Biochemistry 22,5041-5048 Sonderfeld, S., Conzelmann, E., Schwarzmann, G., Burg, J., Hinrichs, U., and Sandhoff, K. (1985a) Eur. J . Biochem. 1 4 9 , 247-255 Sonderfeld, S., Brendler, S., Sandhoff, K., Galjaard, H., and Hooge-

-

I

,

The G M 2 Activator: Mode of Action

1886

veen, A. T. (1985b) Hum. Gen. 7 1 , 196-200 Svennerholm, L. (1957) Biochim. Biophys. Acta 24,604-611 Svennerholm, L., Mansson, J.-E., and Li, y.-T. (1973) J. Bioi. C k m , 248,740-742 Tanaka, A,, Ohno, K., and Suzuki, K. (1988) Biochem. Biophys. Res. Commun. 156, 1015-1019 Tanaka, A., Ohno, K., Sandhoff, K., Maire, I., Kolodny, E. H., Brown, A., and Suzuki, K. (1990) A m . J. Hum. Genet. 46,329-339 Tanford, C. (1980) The Hydrophobic Effect. Formation of Micelles and Biological Membranes, John Wiley & Sons, New York

-

... .-. . . ., ... , ., . . , '5 N NH1Olil0.4 0 CaCIa"5HaU 160/40/1/5. v / v l v / V l

Thompson, J. N., Stoolmuller, A. C., Matalon, R., and Dorfman, A. (1973) Science 1 8 1,866-867 Weitz, G., Lindl, T., Hinrichs, u., and Sandhoff, K. (1983) HOPPeSeyler's Chem. Z. Physiol. 3 6 4 , 863-871 Wynn7 H'pand B' (1986) J. Bioi' 123' 221-230 Wynn, C. H., Marsden, A,, and Robson, B. (1986) J. Theor. Biol. 119,81-87 yu, R. K., Itoh, T., Yohe, H. c.,and Mac&, L. J. (1983) Brain Res. 275,47-52 Zaccai, G., Buldt, G., Seelig, A., and Seelig, J. (1979) J. Mol. Bid. 134,693-706

The GM2Activator: Mode of Action

1887

r Hex A 1150 mU1

Flqure 2b

lex A 1150 nu TDC 12 nul

0

0

40

87

20

50

37

93

17

38

45

77

16

36

42

56

17

31

61

75

90

100

100

100

0

0

22

31

0

0

62

35

7

I4

61

100

6

12

61

90

8

17

69

85

0

0

70

97