Structural and membrane-binding properties of saposin D

Structural and membrane-binding properties of saposin D

Massimo Tatti1, Rosa Salvioli1, Fiorella Ciaffoni1, Piero Pucci2, Annapaola Andolfo2, Angela Amoresano2 andAnna Maria Vaccaro1

1Laboratorio Metabolismo e Biochimica Patologica, Istituto Superiore SanitaÁ, Roma, Italy; 2Centro Internazionale Servizi di Spettrometria di

Massa, CNR-UniversitaÁ di Napoli Federico II and Dipartimento di Chimica Organica e Biologica, UniversitaÁ di Napoli Federico II, Italy

Saposin D is generated together with three similar proteins, saposins A, B and C, from a common precursor,

called prosaposin, in acidic organelles such as late endosomes and lysosomes. Although saposin D has been

reported to stimulate the enzymatic hydrolysis of sphingomyelin and ceramide, its physiological role has not yet

been clearly established. In the present study we examined structural and membrane-binding properties of

saposin D. At acidic pH, saposin D showed a great affinity for phospholipid membranes containing an anionic

phospholipid such as phosphatidylserine or phosphatidic acid. The binding of saposin D caused destabilization of

the lipid surface and, conversely, the association with the membrane markedly affected the fluorescence

properties of saposin D. The presence of phosphatidylserine-containing vesicles greatly enhanced the intrinsic

tyrosine fluorescence of saposin D, which contains tyrosines but not tryptophan residues.

The structural properties of saposin D were investigated in detail using advanced MS analysis. It was found

that the main form of saposin D consists of 80 amino acid residues and that the six cysteine residues are linked in

the following order: Cys5±Cys78, Cys8±Cys72 and Cys36±Cys47. The disulfide pattern of saposin D is

identical with that previously established for two other saposins, B and C, which also exhibit a strong affinity for

lipids. The common disulfide structure probably has an important role in the interaction of these proteins with

membranes.

The analysis of the sugar moiety of saposin D revealed that the single N-glycosylation site present in the

molecule is mainly modified by high-mannose-type structures varying from two to six hexose residues.

Deglycosylation had no effect on the interaction of saposin D with phospholipid membranes, indicating that the

glycosylation site is not related to the lipid-binding site.

The association of saposin D with membranes was highly dependent on the composition of the bilayer. Neither

ceramide nor sphingomyelin, sphingolipids whose hydrolysis is favoured by saposin D, promoted its binding,

while the presence of an acidic phospholipid such as phosphatidylserine or phosphatidic acid greatly favoured the

interaction of saposin D with vesicles at low pH. These results suggest that, in the acidic organelles where

saposins are localized, anionic phospholipids may be determinants of the saposin D topology and, conversely,

saposin D may affect the lipid organization of anionic phospholipid-containing membranes.

Keywords: disulfide bridges; mass spectrometry, membrane interaction; phospholipids; saposin D.

Saposin D is released together with three other similar proteins,saposins A, B and C, from a common precursor calledprosaposin [1±5]. This precursor is proteolytically processedinto mature saposins in late endosomes and lysosomes [6,7].Genetic defects of saposins have provided insights into theirrole in the degradation of sphingolipids. When all saposins aremissing, as seen in a patient carrying a mutation in theprosaposin initiation codon, accumulation of several sphingo-lipids such as ceramide, glucosylceramide, lactosylceramideand ganglioside GM3 has been observed [8,9]. A mutant mouseline in which the prosaposin gene was inactivated exhibited aclinical, pathological and biochemical phenotype closely

resembling that of the human disease [10]. Saposins B and Cappear to be involved in the catabolism of specific sphingo-lipids, as indicated by the storage of sulfatide caused by aselective deficiency of saposin B in an atypical form ofmetachromatic leukodystrophy and the storage of glucosylcer-amide caused by a selective deficiency of saposin C in a variantform of Gaucher's disease [11±13]. An isolated deficiency insaposin A or D has not been reported so far and thus their actualphysiological functions are still not known. In vitro, saposin Ais able to activate the degradation of glucosylceramide [14,15]and galactosylceramide [14], and saposin D that of sphingo-myelin and ceramide [16,17]. The addition of saposin D to theculture medium of fibroblasts from patients with prosaposindeficiency leads to a decrease in the accumulated ceramide[18], and thus the function of this saposin seems to be related toceramide degradation rather than sphingomyelin degradation.

As saposins are involved in the catabolism of membranecomponents such as sphingolipids, a knowledge of theirinteractions with lipid bilayers is of critical importance to ourunderstanding of their physiological behaviour. We haverecently found that at least two saposins, C and D, are able tobind to phospholipid large unilamellar vesicles (LUVs) at low

Eur. J. Biochem. 263, 486±494 (1999) q FEBS 1999

Correspondence to A. M. Vaccaro, Department of Metabolism and

Pathological Biochemistry, Istituto Superiore Sanita', Viale Regina Elena

299, 00161 Roma, Italy. Fax: + 39 06 49387149, Tel.: + 39 06 49902416,

E-mail: [email protected]

Abbreviations: SAPLIP, saposin-like protein; LBPA, lysobisphosphatidic

acid; LUV, large unilamellar vesicle; ES-MS, electrospray mass

spectrometry; MALDI-MS, matrix-assisted laser-desorption ionization

mass spectrometry.

(Received 8 February 1999; accepted 4 May 1999)

q FEBS 1999 Properties of saposin D (Eur. J. Biochem. 263) 487

pH and that the trigger for association with membranes is apH-dependent increase in their superficial hydrophobicity [19].Although saposins C and D share the capacity to interact withphospholipid bilayers, they exhibit distinct properties. Forinstance, after interaction with saposin C, PtdSer-containingLUVs acquire the capacity to bind glucosylceramidase [19,20],the lysosomal enzyme that degrades glucosylceramide, whilethe association of saposin D with these LUVs does not promotebinding of this enzyme [19].

Saposins are of similar size and share a structural motifcharacterized by the conserved location of six cysteineresidues. A further common structural feature is the presence ofan N-glycosylation site, only saposin A having a secondoligosaccharide chain in its molecule [1±4]. The secondarystructure appears to be of great importance for physiologicalfunction [21]; mutations that change cysteines in saposin B or Cto another amino acid impairing the formation of the correctdisulfide bridges result in accumulation of sulfatide orglucosylceramide, respectively. Also in-vitro experimentsindicate that the properties of saposins depend on theirdisulfide structure. In fact, after reduction and alkylation,saposins A [14] and C [22] lose their capacity to stimulateglucosylceramidase activity; moreover the saposin C abilityto perturb PtdSer-containing bilayers is greatly impaired [22].

Further progress in elucidating the molecular mechanisms ofsaposins is critically dependent on characterization of theirstructure and on understanding their interaction with mem-branes at low pH values similar to those found in the acidicorganelles in which saposins are localized. To gain a betterinsight into the properties and functions of saposin D, weinvestigated its structure in relation to its interaction withbilayers. The disulfide structure was assessed and the effectof the oligosaccharide chain on the interaction withmembranes evaluated. Moreover the influence of membranecomposition on the partitioning of saposin D into the bilayerhas been highlighted.

E X P E R I M E N T A L P R O C E D U R E S

Materials

PtdCho from egg yolk and PtdSer from bovine brain were fromAvanti Polar Lipids, Inc. (Alabaster, AL, USA). l-a-Dipal-mitoyl[dipalmitoyl-1-14C]PtdCho (110 mCi´mmol21) was fromNEN Research Products, DuPont de Nemours, Germany.Cholesterol, ceramide, sphingomyelin, trypsin (treated withTos-Phe-CH2Cl) and pepsin were from Sigma. Calcein wasfrom Molecular Probes (Eugene, OR, USA). EndoproteinaseGlu-C, N-glycosidase F and endoglycosidase F/N-glycosidaseF were from Boehringer-Mannheim. CNBr was from Pierce.Sep-Pak cartridges C18 were from Waters. All other reagentswere of the purest available grade.

Preparation of saposins C and D

Saposins C and D were purified from spleens of patientswith type-1 Gaucher's disease following a previously reportedprocedure [19]; it consisted of heat treatment of a waterhomogenate, ion-exchange chromatography on DEAE-Sephacel,gel filtration on Sephadex G-75 and reverse-phase HPLC on aprotein C4 column (Vydac). The purity of the final preparationwas verified by N-terminal sequence analysis, SDS/PAGE andelectrospray MS (ES-MS).

Deglycosylation of native saposin D

In a final volume of 50 ml, 200 mg of saposin D were incubatedin 10 mm sodium phosphate buffer, pH 8.6, containing 0.1 mmEDTA, with 0.25 U of endoglycosidase F/N-glycosidase F at37 8C for 24 h. The deglycosylated protein was purified byreverse-phase HPLC on a protein C4 column. The column wasequilibrated with solvent A [water/acetonitrile (6 : 4, v/v)containing 0.1% trifluoroacetic acid] and eluted with a lineargradient of 0±100% solvent B [water/acetonitrile (2 : 8, v/v)containing 0.1% trifluoroacetic acid) over 70 min at a flow rateof 0.5 ml´min21. The HPLC-purified sample was lyophilizedand redissolved in water. On SDS/PAGE, deglycosylatedsaposin D moved faster than the native saposin.

Vesicle preparation

LUVs were prepared by filter exclusion using a high-pressureextrusion apparatus (Lipex Biomembranes, Vancouver, BC,Canada) as previously described [19,22]. In short, the dry lipidswere dispersed by vortex-mixing in buffer A (2 mm l-histidine,2 mm Tris, 150 mm NaCl, 1 mm EDTA, pH 7.4). Thesuspension was submitted to 10 cycles of freezing and thawingand then extruded 15 times through two stacked polycarbonatefilters (pore-size 0.1 mm, Nucleopore, Pleasanton, CA, USA).All vesicles were supplemented with trace amounts of labelledPtdCho and their concentration was determined by radioactivemeasurement.

Binding of saposins to vesicles

For binding studies, each saposin (20 mg) was incubated withLUVs (50 mg) in 0.2 ml of buffer B (10 mm acetate, 150 mmNaCl, 1 mm EDTA) adjusted to the desired pH, at 37 8C, for10 min. The mixture was then centrifuged with a 42.2 Ti rotor(Beckman), in polycarbonate centrifuge tubes (7 � 20 mm,Beckman), at 100 000 g for 1 h. More than 95% of vesicleswere found in the pellet, as determined by radioactivitymeasurements. Conversely, in control experiments withoutliposomes, the saposins did not sediment during the ultra-centrifugation. After separation of the supernatant, the pelletedvesicles were rinsed once with 0.2 ml of buffer B and finallyresuspended in 0.2 ml of the same buffer. Saposins in the initialsupernatant (free saposins) and in the resuspended vesicles(liposome-bound saposins) were precipitated by the addition of2 ml of cold acetone. After 2 h at 220 8C the precipitatedsaposins were collected by centrifugation, solubilized with30 mL of electrophoresis sample buffer and visualized bySDS/PAGE followed by silver staining (see below).

SDS/PAGE

A discontinuous Tris/glycine SDS/PAGE system was used with15% acrylamide [23]. After electrophoresis the saposins werevisualized with the silver staining kit, Protein (PharmaciaBiotech) according to the manufacturer's instructions.

Leakage assay

Leakage of liposome contents was monitored by the release ofcalcein trapped inside the vesicles [24]. The leakage wascarried out at 37 8C and monitored with a Fluoromax spectro-fluorimeter equipped with a constant-temperature cell holderand stirrer (Spex Industries Inc., Edison, NJ, USA). LUVs forleakage experiments were prepared by hydrating dried films of

488 M. Tatti et al. (Eur. J. Biochem. 263) q FEBS 1999

lipids in 60 mm calcein, pH 7.4, followed by 10 cycles offreeze±thawing. The resulting multilamellar vesicles wereextruded 15 times through two polycarbonate filters (pore-size 0.1 mm). Free calcein was separated from the dye-containing LUVs by chromatography on a Sephadex G-75column. On addition of saposins, leakage of calcein into theexternal medium was followed by the increase in fluorescencecaused by calcein dilution and the consequent relief of self-quenching (excitation 470 nm, emission 520 nm). Complete(100%) leakage was established by lysing the vesicles with0.3% (v/v) Triton X-100.

Tyrosine fluorescence measurements

Tyrosine fluorescence of saposin D was measured at 37 8Cusing a Fluoromax spectrofluorimeter equipped with aconstant-temperature cell holder and stirrer. Emission spectrawere obtained by exciting saposin D samples (4 mm) at 280 nm(3-nm slit width) and scanning emission from 280 to 360 nm(3-nm slit width). The fluorescence spectra of saposin D wererecorded in the presence and absence of LUVs after incubationfor 10 min at 37 8C. The spectra were corrected for thefluorescence associated with LUVs themselves.

Chemical and enzymatic hydrolyses

Native saposin D (150 mg) was incubated with a 15-fold excessof CNBr (w/w) in 70% trifluoroacetic acid, overnight, at roomtemperature in the dark. The protein sample was thendiluted 10-fold with water, evaporated in a Speed-Vaccentrifuge (Savant) and lyophilized. The CNBr-treated saposinwas resuspended in 5% formic acid, digested with pepsin for18 h at 37 8C (enzyme/substrate 1 : 50, w/w) and lyophilized.Deglycosylation of the sample was performed by incubationwith N-glycosidase F (0.1 enzyme unit per 100 mg of saposin)at 37 8C overnight, in 50 mm ammonium bicarbonate, pH 8.5.The nicked protein was subdigested with trypsin and endopro-teinase Glu-C in 50 mm ammonium acetate, pH 8.5, for18 h at 37 8C (enzyme/substrate 1 : 50, w/w). The samplewas eventually lyophilized.

Isolation and characterization of oligosaccharide chains

N-linked glycans released by N-glycosidase F were separatedfrom the peptide material by reverse-phase chromatography onprepacked Sep-Pak cartridges C18. To increase the sensitivity ofthe MS analyses, the oligosaccharides eluted in the hydrophilicphase were permethylated with methyl iodide in dimethylsulfoxide after the NaOH procedure [25]. Derivatized sampleswere diluted with water, extracted in chloroform, dried down ina Speed-Vac centrifuge and directly submitted to analysis.

Mass spectrometry

Electrospray MS (ES-MS) analyses were performed on a Bio-Qtriple-quadrupole instrument (Micromass, Manchester, UK).Protein and peptide samples were dissolved in 1% acetic acid in50% acetronitrile and injected into the ion source at a flow rateof 10 ml´min21. Spectra were acquired and elaborated using themass lynx software provided by the manufacturer. Calibrationwas performed by means of a separate injection of horse heartmyoglobin (16951.5 Da). All mass values are reported asaverage masses.

Matrix-assisted laser-desorption ionization (MALDI) spectrawere recorded using a Voyager mass spectrometer (PerSeptiveBiosystems, Boston, MA, USA). A mixture of analytesolution (1 ml) and a-cyano-4-hydroxycinnamic acid or

Fig. 1. Effect of LUVs on the fluorescence spectrum of saposin D.

Fluorescence emission spectra of saposin D (4 mm) in the absence (dotted

line) and presence of 50 mm (dashed line) or 100 mm (solid line) LUVs

composed of cholesterol/PtdCho/PtdSer (25 : 55 : 20) in buffer B, pH 4.5

(A) or pH 5.0 (B). The excitation wavelength was 280 nm and the

temperature was 37 8C.

Fig. 2. Leakage induced by saposin D as function of saposin concen-

tration and pH. Different amounts of saposin D were injected into a stirred

cuvette thermostatically controlled at 37 8C and containing a 1-ml solution

of LUVs composed of cholesterol/PtdCho/PtdSer (25 : 55 : 20) (75 mm

total lipid) in buffer B, pH 4.5 (X) or pH 5.0 (B), or buffer A, pH 7.4 (O).

After 3 min, the percentage of calcein leakage was measured as described

in Experimental Procedures. The points represent means of at least three

different experiments. The deviation for all samples was less than ^ 5% of

the corresponding mean value.

q FEBS 1999 Properties of saposin D (Eur. J. Biochem. 263) 489

2,5-dihydroxybenzoic acid (Sigma) was applied to the metallicsample plate and air-dried before insertion into the massspectrometer. Mass calibration was performed with the quasi-molecular ions from bovine insulin at 5734.6 Da and a matrixpeak at 379.4 Da as internal standard. Raw data were analysedusing computer software provided by the manufacturer. Allmass values are reported as average masses.

R E S U LT S

Interaction of saposin D with phospholipid membranes

We have previously shown that, at low pH, saposin D has agreater affinity for PtdSer-containing membranes than saposinsA and B, but less than saponin C [19]. To examine further theinteraction of saposin D with the bilayer, protein binding was

followed by measuring the intrinsic fluorescence of thesaposin in the presence and absence of PtdSer-containingLUVs. Saposin D contains three tyrosines at positions 14,43 and 54, four phenylalanines and no tryptophan residues.Considering the small contribution of phenylalanine in theregion of tyrosine excitation and emission, the intrinsicfluorescence spectrum of saposin D is dominated bytyrosines, as indicated by the characteristic maximum at310 nm (Fig. 1). At pH 4.5 addition of increasingconcentrations of LUVs up to a lipid/saposin molar ratioof 25 increased the fluorescence intensity of saposin D (Fig.1A). At pH 5.0 a much smaller increase in fluorescence wasobserved, indicating weaker binding of saposin D to themembrane at this pH (Fig. 1B). At pH 7.4, the presence ofLUVs did not affect the fluorescence of saposin D at all(data not shown).

Fig. 3. Transformed electrospray mass

spectra of deglycosylated (upper panel) and

glycosylated (lower panel) saposin D. The

individual species are indicated by capital letters.

The molecular mass of the components of

deglycosylated saposin D are reported as Da.

490 M. Tatti et al. (Eur. J. Biochem. 263) q FEBS 1999

Table 1. ES-MS analysis of glycosylated saposin D. The components

indicated with a capital letter refer to the ES-MS spectra of glycosylated

saposin D shown in Fig. 3 (lower panel). The molecular masses are reported

as Da. Hex, hexose; HexNAc, N-acetylhexosamine; 1±80 etc is the length

of the polypeptide chain.

Components Molecular mass Structures

A 10130.1 �^ 0.3 1±80 + Hex5HexNAc2

B 9967.1 �^ 0.1 1±80 + Hex4HexNAc2

C 9803.9 �^ 0.7 1±80 + Hex3HexNAc2

D 9640.1 �^ 0.1 1±80 + Hex2HexNAc2

E 9471.5 �^ 0.1 3±80 + Hex2HexNAc2

F 9116.3 �^ 0.2 1±80 + HexNAc2

G 8914.3 �^ 0.1 1±80

H 10014.3 �^ 0.1 2±80 + Hex5HexNAc2

I 9851.9 �^ 0.2 2±80 + Hex4HexNAc2

L 9692.1 �^ 0.3 2±80 + Hex3HexNAc2

M 9530.3 �^ 0.2 2±80 + Hex2HexNAc2

Table 2. MALDI-MS analysis of the oligosaccharide structures present

in the saposin D molecule. The N-linked glycans were released from the

peptide backbone, permethylated and analysed by MALDI-MS as described

in Experimental procedures. The mass signals observed in the MALDI-MS

spectra of the permethylated glycans released from saposin D are reported

as sodium adducts. Hex, hexose; HexNAc, N-acetylhexosamine; Fuc,

fucose.

MNa+ Structures

967�.4 Hex2HexNAc2

1141�.5 FucHex2HexNAc2

1172�.1 Hex3HexNAc2

1345�.6 Fuc Hex3HexNAc2

1375�.9 Hex4HexNAc2

1579�.8 Hex5HexNAc2

1783�.5 Hex6HexNAc2

Fig. 4. Assignment of S±S bridge pattern.

Partial MALDI-MS spectrum of saposin D

treated with CNBr, digested with pepsin,

deglycosylated by endoglycosidase

F/N-glycosidase F and subdigested with trypsin

(upper panel). The peptide mixture was further

subdigested with endoproteinase GluC and

submitted to MALDI-MS analysis (lower panel).

The mass signals and the assignments to the

corresponding disulfide-bridged peptides are

reported.

q FEBS 1999 Properties of saposin D (Eur. J. Biochem. 263) 491

The effect of increasing concentrations of saposin D on thepermeability of LUVs was also examined. The destabilizationof the membrane was assessed by measuring release of calceinincorporated into the vesicles. At pH 4.5, leakage increasedsharply as the saposin concentration increased up to about0.1 mm (lipid/saposin molar ratio 750), beyond which theleakage increase was more gradual (Fig. 2). The membrane-perturbing activity of saposin D decreased at pH 5.0 and wasinhibited at pH 7.4.

Structural characterization of saposin D and assignment ofdisulfide bridges

We have previously found that the membrane-binding proper-ties of saposin C are related to its structural properties. Toobtain more information on the structure of saposin D, its sizeand cysteine coupling were investigated. Microheterogeneity atthe amino acid sequence level was demonstrated by ES-MSanalysis of deglycosylated saposin D. As shown in Fig. 3(upper panel), four components could be recognised in thetransformed ES-MS spectrum. The major component had amolecular mass of 8913.8^0.1 Da corresponding to saposin D1±80 (theoretical mass value 8912.4 Da). Minor componentswith molecular masses of 8741.5^0.2 Da and 8798.6^0.4 Dawere also detected. They were identified as the 3±80(theoretical mass value 8740.2 Da) and 2±80 (theoreticalmass value 8797.3 Da) truncated forms of saposin D inwhich the N-terminal dipeptide Asp-Gly or only the firstamino acid Asp, respectively, were missing. Finally, the signalat 9116.7^0.6 Da corresponds to the 1±80 form of the saposincontaining a single N-acetylglucosamine residue still attachedto the N-glycosylation site because of incomplete cleavage ofthe glycosidic bond (theoretical mass value 9115.8 Da).

A series of components were identified in the spectrum ofglycosylated saposin D (Fig. 3, lower panel and Table 1)demonstrating the occurrence of microheterogeneity also in theglycan chains linked to the single N-glycosylation site atAsn22. The different glycoforms were identified on the basis ofthe theoretical molecular masses of the polypeptide portion andof the oligosaccharide structures expected from the knownbiosynthetic pathway of N-linked glycans. Table 1 reports theassignment of the different molecular masses. High-mannosestructures consisting of two to five hexose residues were linkedto the 1±80 and 2±80 forms of saposin D. A small percentageof the non-glycosylated 1±80 form of saposin D (componentG) was also detected.

The structures of the glycosidic moiety of saposin D wereconfirmed by direct MALDI-MS analysis of the permethylatedoligosaccharide mixture (Table 2). As expected from theES-MS spectrum of the native saposin D, high-mannoseoligosaccharides spanning two to six hexose residues repre-sented the most abundant components. However, minor speciesconsisting of fucosylated truncated complex glycans were alsoobserved.

The complete pattern of S±S bridges in saposin D wasestablished by combining chemical and enzymatic digestions ofthe protein with MS analyses [22]. As saposins are known to beresistant to proteases [1±4], saposin D was first cleaved withCNBr at the single methionine residue at position 66. Thenicked protein was digested with pepsin, deglycosylated byN-glycosidase F treatment and subdigested with trypsin.

Figure 4 (upper panel) shows the MALDI-MS analysis of thecorresponding peptide mixture. The signals at m/z 2795.1 and1896.4 were assigned to the peptide pairs (35±45) + (46±58)and (35±45) + (46±50), respectively, linked by the disulfidebridge between Cys36 and Cys47 and originating fromcleavage at Leu58 and Phe50. Both signals were accompaniedby satellite peaks occurring 17 Da lower because of thecyclization of Gln46 into pyroglutamic acid, thus confirmingthe interpretation of the data. These assignments were furtherverified by both reduction of the peptide mixture withdithiothreitol and submission of the whole mixture to a singlestep of manual Edman degradation followed by MALDI-MSre-examination. A series of related signals was detectedat m/z 2536.3, 2407.5, 2290.5 and 2163.5 and assignedto three-peptide clusters consisting of fragments (1±10)or (1±9) + (67±74) + (75±80) and fragments (4±10) or(4±9) + (67±74) + (75±80) joined by two S±S bridgesinvolving the four cysteine residues Cys5, Cys8, Cys72 andCys78. On the basis of the above data, the possible pairings ofthe four cysteines are either Cys5±Cys72 and Cys8±Cys78 orCys5±Cys78 and Cys8±Cys72.

To determine the correct assignment of the two disulfides,the peptide mixture was further incubated with endoproteaseGlu-C taking advantage of the presence of a glutamic acidresidue at position 6. The MALDI-MS analysis of thefragments obtained (Fig. 4, lower panel) showed theoccurrence of two mass signals at m/z 1383.6 and 1255.2which were attributed to peptides 7±10 and 7±9, respectively,linked to fragment 67±74 by the S±S bridge Cys8±Cys72.Moreover, the peak doublets at m/z 2193.5 and 2177.1 wereassigned to the peptide pairs (35±45) + (46±53) containingeither Gln or pyroglutamic acid at position 46, thusconfirming the occurrence of a disulfide bridge betweenCys36 and Cys47. This S±S bond was further confirmed bythe peak doublets at m/z 2022.6, 2005.8 and 1875.6, 1858.1.The remaining S±S bridge was then inferred by exclusion asCys5±Cys78.

The complete pattern of disulfide bridges in saposin D wasthus Cys5±Cys78, Cys8±Cys72 and Cys36±Cys47 (Fig. 5).

Influence of glycosylation on the interaction of saposin Dwith phopholipid membranes

To investigate the effect of the single sugar chain on thepartitioning of saposin D into phospholipid bilayers, the

Fig. 5. Schematic disulfide structure of

saposin D (Sap D). Black lines linking

cysteines indicate disulfide bonds.

Fig. 6. Partitioning of native and deglycosylated saposin D on LUVs.

Native and deglycosylated saposin D were incubated in buffer B, pH 5.0,

with LUVs of the same composition as in Fig. 1. Free and liposome-bound

saposin were separated and analysed by electrophoresis as described in

Experimental procedures. Lane 1, native saposin D as standard; lane 2, free

saposin D; lane 3, liposome-bound saposin D; lane 4, deglycosylated

saposin D as standard; lane 5, free deglycosylated saposin D; lane 6,

liposome-bound deglycosylated saposin D. Bands were visualized by silver

staining. Each experiment was repeated at least three times with similar

results.

492 M. Tatti et al. (Eur. J. Biochem. 263) q FEBS 1999

saposin was deglycosylated and its association with PtdSer-containing membranes was assessed at pH 5.0. As shown inFig. 6, the binding of deglycosylated saposin D to theliposomes was similar to that of the native form, i.e. only asmall percentage of saposin D was associated with the lipidbilayer. Moreover the leakage induced by glycosylated anddeglycosylated saposin D at pH 4.5 was identical (data notshown). Thus, the removal of the sugar moiety does not appearto affect the membrane-binding properties of saposin D atlow pH.

Effect of membrane lipid composition on the association ofsaposin D

We finally investigated whether the presence of two sphingo-lipids, namely ceramide and sphingomyelin, could influencethe association of the saposin with the bilayer. Thesesphingolipids were chosen because it has been reported thatsaposin D stimulates their degradation [16±18]. Figure 7Ashows that, at pH 4.5, saposin D binds weakly to LUVscomposed of PtdCho and cholesterol. The presence of neither10% ceramide (Fig. 7B) nor 10% sphingomyelin (Fig. 7C) inthe bilayer was able to promote the binding. In contrast, 10%PtdSer markedly increased the association of saposin D (Fig.7D). Similar results were obtained when PtdSer was replacedwith phosphatidic acid (data not shown). Thus, the presence ofacidic phospholipids, but not that of ceramide and sphingo-myelin, appears to be a requirement for the association ofsaposin D with membranes.

D I S C U S S I O N

Saposins are four small cysteine-rich glycoproteins involved inthe degradation of sphingolipids. It has been reported, andwidely accepted for a long time, that, at acidic pH valuesmimicking the interior pH of lysosomes, three saposins (A, Cand D) bind to specific lysosomal hydrolases [1±4], not tophospholipids [26]. Conversely, we have recently shown thatsaposin C reconstitutes the activity of glucosylceramidase, alysosomal hydrolase, by first binding to PtdSer-containingLUVs and then mediating the binding of the enzyme to thesevesicles [19,20].

In the present study we have analysed the interaction ofanother saposin, D, with phospholipids. At low pH, saposin Dexhibits a great affinity for PtdSer-containing LUVs, asindicated by the large increase in tyrosine fluorescence of thesaposin after membrane interaction. Such an increase influorescence intensity is probably associated with the transfer

of the tyrosine phenolic rings to more apolar microenviron-ments like those existing in membrane interiors. The change insaposin D fluorescence can be interpreted either in terms of adirect involvement of tyrosine side chain(s) in the interactionwith the bilayer or as a result of conformational changesoccurring in the protein structure on binding to the membranesurface. The binding of saposin D results in a markedperturbation of the lipid surface. Of the saposins, D and Care the more powerful destabilizing agents of PtdSer-containingmembranes [19].

We have previously shown that the interaction of saposin Cwith phospholipid membranes can be dramatically affected by achange in saposin structure [22]. We therefore thought that adetailed knowledge of the structure of saposin D may be usefulin understanding its membrane-binding properties. Saposin Disolated from spleen from patients with Gaucher's diseaseshowed microheterogeneity at both the protein and oligo-saccharide chain levels. The major component of saposin Dconsisted of 80 amino acid residues; minor fractions lackingone or two amino acids at the N-terminus were also detected.Previously, a single 78-residue form of saposin D, in which thetwo N-terminal amino acids were missing, was found in humankidney [5]. It is conceivable that different proportions ofsaposin D isoforms are present in different tissues.

MS investigation of native saposin D from spleen frompatients with Gaucher's disease showed that the singleN-glycosylation site at Asn22 contained high-mannose-typeoligosaccharides varying from two to six hexose residues andminor species consisting of fucosylated truncated complex-typeglycans. These data are in agreement with previous studies byIto et al. [27]. We also observed a significant amount ofcompletely deglycosylated saposin D.

The question of how glycosylation affects saposin functionshas not yet been clarified. A mutation that abolishes the onlyglycosylation site of saposin B has been reported to cause avariant form of metachromatic leukodystrophy [28,29].Thecarbohydrate moiety does not appear to be critical for the invitro activity of saposins. In fact, deglycosylation ofsaposins A, B and C does not impair their ability to activategalactosylceramidase, sulfatide sulfatase and glucosylcer-amidase, respectively [14,30,31]. In the present study, wefound that removal of the sugar moiety did not even influencethe association of saposin D with phospholipid membranes,suggesting that the glycosylation site is not related to the regioninteracting with the lipid bilayer.

Although deglycosylation does not seem to affect theproperties of saposins, the maintenance of the disulfidestructure is of critical importance for their ability to stimulatesphingolipid degradation [1±4,21]. Mutations that disruptdisulfide linkages in saposin B and C cause metachromaticleukodystrophy and Gaucher-like disease, respectively [32±34].To establish the disulfide pattern of saposin D, we combinedCNBr cleavage with proteolytic digestions followed byMALDI-MS analysis of the resulting peptide mixture. It wasfound that the six cysteines of saposin D are linked in anorderly manner, i.e. the first cysteine is linked to the last, thesecond to the second last one and the third to the fourth, withoutintersection of the three S±S bridges. This unusual arrange-ment of cysteines is identical with that of saposins B and C[22] and of other similar low-molecular-mass cysteine-richproteins, which constitute a family of sequence-related proteins(saposin-like proteins, SAPLIPs) [21]. These proteins performdifferent functions, but the mechanism underlying theiractivities is related to their ability to bind to lipids. In fact, aSAPLIP such as the surfactant protein B interacts preferentially

Fig. 7. Partitioning of saposin D on LUVs of different composition.

Saposin D was incubated in buffer B, pH 4.5, with LUVs of the following

composition: A, cholesterol/PtdCho (25 : 75); B, cholesterol/PtdCho/cera-

mide (25 : 65 : 10); C, cholesterol/PtdCho/sphingomyelin (25 : 65 : 10);

D, cholesterol/PtdCho/PtdSer (25 : 65 : 10). Free and liposome-bound

saposin were separated and analysed by electrophoresis as described in

Experimental procedures. Lanes 1, free saposin D; lanes 2, liposome-

bound saposin D. Bands were visualized by silver staining. Each

experiment was repeated at least three times with similar results.

q FEBS 1999 Properties of saposin D (Eur. J. Biochem. 263) 493

with anionic lipids forming monolayers capable of lowering thenormal surface tension at the alveolar interface [35,36]. Asecond SAPLIP, the highly cytotoxic pore-forming peptide ofEntamoeba histolytica, inserts into membranes causing theformation of ion channels [37]. Also a third SAPLIP, NK-lysin,a tumorolytic and antibacterial peptide of NK and T cells [38],interacts with and destabilizes lipid bilayers [39]. As far assaposins are concerned, saposin B activates cerebrosidesulfatase by interacting with sulfatides [40] and saposin Cstimulates glucosylceramidase by interacting with PtdSer-containing membranes [19,20]. After reduction of disulfidebridges, not only saposins but also SAPLIPs lose their activity[41,42]. It is thus conceivable that the common disulfide patternis necessary to stabilize a structure that enables these proteins tointeract with membranes. A structure, referred to as the`saposin-fold', has recently been defined for NK-lysin [43]. Itconsists of five amphipathic a helices kept together by the threedisulfide bridges and folded into a single globular domain; thisstructure is probably representative of all SAPLIPs because oftheir high degree of homology.

Saposin D is possibly involved in ceramide catabolism[17,18], but its mode of action is still not known. Some authorshave reported that it stimulates the activity of partially purifiedacid ceramidase in the presence of Triton X-100 [17], whileothers did not observe any substantial stimulation of purifiedceramidase activity in a detergent-based assay system [44]. Ourfindings indicate that saposin D associates weakly withceramide- or sphingomyelin-containing vesicles. Only thepresence of an acidic phospholipid such as PtdSer or phosphatidicacid dramatically increases the binding of saposin D to thebilayer at acidic pH. This observation suggests that theformation of anionic phospholipid domains may cause topologychanges in saposin D. A function for anionic phospholipids inthe localization and organization of proteins in membranes hasbeen previously proposed [45,46].

The acidic organelles in which saposins are localized,namely late endosomes and lysosomes [6], contain largeamounts of a unique anionic phospholipid, lysobisphosphatidicacid (LBPA) [47±50]. Recently it has been reported that LBPAregulates the structure and function of late endosomes formingspecialized microdomains in the internal membranes of theseorganelles [50]. Some proteins distribute preferentially withinthese internal membranes and segregate to the LBPA-richdomains, e.g. the multifunctional receptor for mannose-6-phosphate-bearing ligands and insulin growth factor [50]. Onconsideration of the presently observed effect of acidic phos-pholipids on the partition of saposin D into bilayers, one is ledto hypothesize that LBPA-rich domains may also modulate theinteraction of saposin D with late endosomal and lysosomalmembranes.

In conclusion, our study provides new insights into thestructure of saposin D and its interaction with lipids. Definitiveevidence for the interaction of saposin D with phospholipidmembranes has been obtained; moreover the key role of acidicphospholipids in promoting saposin binding has been high-lighted. The analogy of saposin D disulfide structure with thatof SAPLIPs, the activity of which is mediated via membraneinteraction, further supports the notion that lipids are theprimary target for saposin D. Our present and previousfindings indicate that at least two saposins, C and D, havehigh affinity for phospholipid membranes, each having itsspecific mode of interaction. A detailed characterization ofthe interaction of saposins with phospholipids may offer anew approach to the understanding of their physiologicalproperties.

A C K N O W L E D G E M E N T S

This work was partly supported by ISS Research Project `Prevention of risk

factors of maternal and child health'. The authors thank Mr E. Raia for

technical assistance.

R E F E R E N C E S

1. O'Brien, J. & Kishimoto, Y. (1991) Saposin proteins: structure,

function, and role in human lysosomal storage disorders. FASEB J. 5,

301±308.

2. Kishimoto, Y., Hiraiwa, M. & O'Brien, J.S. (1992) Saposins:

structure, function distribution, and molecular genetics. J. Lipid

Res. 33, 1255±1267.

3. FuÈrst, W. & Sandhoff, K. (1992) Activator proteins and topology of

lysosomal sphingolipid catabolism. Biochim. Biophys. Acta 1126,

1±16.

4. Sandhoff, K., Harzer, K., FuÈrst, W. (1995) Sphingolipid activator

proteins. In The Metabolic and Molecular Bases of Inherited

Disease (Scriver, C.R., Beaudet, A.L., Sly, W.S. & Valle, D.,

eds), pp. 2427±2441, McGraw-Hill Book Co., New York.

5. FuÈrst, W., Machleidt, W. & Sandhoff, K. (1988) The precursor of

sulfatide activator protein is processed to three different proteins.

Biol. Chem. Hoppe-Seyler 369, 317±328.

6. Vielhaber, G., Hurwitz, R. & Sandhoff, K. (1996) Biosynthesis,

processing, and targeting of sphingolipid activator protein (Sap)

precursor in cultured human fibroblasts. J. Biol. Chem. 271,

32438±32446.

7. Hiraiwa, M., Martin, B.M., Kishimoto, Y., Conner, G.E., Tsuji, S. &

O'Brien, J.S. (1997) Lysosomal proteolysis of prosaposin, the

precursor of saposins (sphingolipid activator proteins): its mechanism

and inhibition by ganglioside. Arch. Biochem. Biophys. 341, 17±24.

8. Schnabel, D., SchroÈder, M., FuÈrst, W., Klein, A., Hurwitz, R., Zenk, T.,

Weber, J., Harzer, K., Paton, B.C., Poulos, A., Suzuki, K. & Sandhoff,

K. (1992) Simultaneous deficiency of sphingolipid activator proteins

1 and 2 is caused by a mutation in the initiation codon of their

common gene. J. Biol. Chem. 267, 3312±3315.

9. BradovaÂ, V., SmõÂd, F., Ulrich-Bott, B., Roggendorf, W., Paton, B.C. &

Harzer, K. (1993) Prosaposin deficiency: further chacterization of

the sphingolipid activator protein-deficient sibs. Multiple glyco-

lipid elevations (including lactosylceramidosis), partial enzyme

deficiencies and storage disease. Hum. Genet. 92, 143±152.

10. Fujita, N., Suzuki, K., Vanier, M.T., Popko, B., Maeda, N., Klein, A.,

Henseler, M., Sandhoff, K., Nakayasu, H. & Suzuki, K. (1996)

Targeted disruption of the mouse sphingolipid activator protein gene:

a complex phenotype, including severe leukodystrophy and wide-

spread storage of multiple sphingolipids. Hum. Mol. Genet. 5,

711±725.

11. Wenger, D.A., De Gala, G., Williams, C., Taylor, H.A., Stevenson,

R.E., Pruitt, J.R., Miller, J., Garen, P.D. & Balentine, J.D. (1989)

Clinical, pathological, and biochemical studies on an infantile case of

sulfatide/GM1 activator protein deficiency. Am. J. Med. Genet. 33,

255±265.

12. Schlote, W., Harzer, K., Christomanou, H., Paton, B.C., Kustermann-

Kuhn, B., Schmid, B., Seeger, J., Beudt, U., Schuster, I. &

Langenbeck, U. (1991) Sphingolipid activator protein 1 deficiency

in metachromatic leukodystrophy with normal arylsulfatase A

activity. A clinical, morphological, biochemical and immunological

study. Eur. J. Pediatr. 150, 584±591.

13. Christomanou, H., Chabas, A., Pampols, T. & Guardiola, A. (1989)

Activator protein deficient Gaucher's disease. Klin. Wochenschr. 67,

999±1003.

14. Morimoto, S., Martin, B.M., Yamamoto, Y., Kretz, K.A., O'Brien, J.S.

& Kishimoto, Y. (1989) Saposin A: second cerebrosidase activator

protein. Proc. Natl. Acad. Sci. USA 86, 3389±3393.

15. Vaccaro, A.M., Tatti, M., Ciaffoni, F., Salvioli, R., Barca, A. & Scerch,

C. (1997) Effect of saposins A and C on the enzymatic hydrolysis of

liposomal glucosylceramide. J. Biol. Chem. 272, 16862±16867.

16. Morimoto, S., Martin, B.M., Kishimoto, Y. & O'Brien, J.S. (1988)

494 M. Tatti et al. (Eur. J. Biochem. 263) q FEBS 1999

Saposin D: a sphingomyelinase activator. Biochem. Biophys. Res.

Commun. 156, 403±410.

17. Azuma, N., O'Brien, J.S., Moser, H.W. & Kishimoto, Y. (1994)

Stimulation of acid ceramidase activity by saposin D. Arch. Biochem.

Biophys. 311, 354±357.

18. Klein, A., Henseler, M., Klein, C., Suzuki, K., Harzer, K. & Sandhoff,

K. (1994) Sphingolipid activator protein D (sap-D) stimulates the

lysosomal degradation of ceramide `in vivo'. Biochem. Biophys. Res.

Commun. 200, 1440±1448.

19. Vaccaro, A.M., Ciaffoni, F., Tatti, M., Salvioli, R., Barca, A., Tognozzi,

D. & Scerch, C. (1995) pH-dependent conformational properties of

saposins and their interactions with phospholipid membranes. J. Biol.

Chem. 270, 30576±30580.

20. Vaccaro, A.M., Tatti, M., Ciaffoni, F., Salvioli, R., Maras, B. &

Barca, A. (1993) Function of saposin C in the reconstitution of

glucosylceramidase by phosphatidylserine liposomes. FEBS Lett.

336, 159±162.

21. Munford, R.S., Sheppard, P.O. & O'Hara, P.J. (1995) Saposin-like

proteins (SAPLIP) carry out diverse functions on a common

backbone structure. J. Lipid Res. 36, 1653±1663.

22. Vaccaro, A.M., Salvioli, R., Barca, A., Tatti, M., Ciaffoni, F.,

Maras, B., Siciliano. R., Zappacosta, F., Amoresano, A. & Pucci,

P. (1995) Structural analysis of saposin C and B. J. Biol. Chem.

270, 9953±9960.

23. Laemmli, U.K. (1970) Cleavage of structural proteins during the

assembly of the head of bacteriophage T4. Nature (London) 227,

680±685.

24. Straubinger, R.M., Hong, K., Friend, D.S. & Papahadjopoulos, D.

(1983) Endocytosis of liposomes and intracellular fate of encap-

sulated molecules: encounter with a low pH compartment after

internalization in coated vesicles. Cell 32, 1069±1079.

25. Dell, A. (1990) Preparation and desorption mass spectrometry of

permethyl and peracetyl derivatives of oligosaccharides. Methods

Enzymol. 193, 647±660.

26. Morimoto, S., Kishimoto, Y., Tomich, J., Weiler, S., Ohashi, T.,

Barranger, J.A., Kretz, K.A. & O'Brien, J.S. (1990) Interaction of

saposins, acidic lipids, and glucosylceramidase. J. Biol. Chem. 265,

1933±1937.

27. Ito, K., Takahashi, N., Takahashi, A., Shimada, I., Arata, Y., O'Brien,

J.S. & Kishimoto, Y. (1993) Structural study of the oligosaccharide

moieties of sphingolipid activator proteins, saposins A,C and D

obtained from the spleen of a Gaucher patient. Eur. J. Biochem. 215,

171±179.

28. Kretz, K.A., Carson, G.S., Morimoto, S., Kishimoto, Y., Fluharty, A.L.

& O'Brien, J.S. (1990) Characterization of a mutation in a family

with saposin B deficiency: a glycosylation site defect. Proc. Natl

Acad. Sci. USA 87, 2541±2544.

29. Rafi, M.A., Zhang, X.L., de Gala, G. & Wenger, D.A. (1990) Detection

of a point mutation in sphingolipid activator protein-1 mRNA in

patients with a variant form of metachromatic leukodystrophy.

Biochem. Biophys. Res. Commun. 166, 1017±1023.

30. Hiraiwa, M., Soeda, S., Martin, B.M., Fluharty, A.L., Hirabayashi, Y.,

O'Brien, J.S. & Kishimoto, Y. (1993) The effect of carbohydrate

removal on stability and activity of saposin B. Arch. Biochem.

Biophys. 303, 326±331.

31. Sano, A. & Radin, N.S. (1988) The carbohydrate moiety of the

activator protein for glucosylceramide beta-glucosidase. Biochem.

Biophys. Res. Commun. 154, 1197±1203.

32. Holtschmidt, H., Sandhoff, K., Kwon, H.Y., Harzer, K., Nakano, T. &

Suzuki, K. (1991) Sulfatide activator protein. Alternative splicing that

generates three mRNAs and a newly found mutation responsible for a

clinical disease. J. Biol. Chem. 266, 7556±7560.

33. Rafi, M.A., de Gala, G., Zhang, X.L. & Wenger, D.A. (1993)

Mutational analysis in a patient with a variant form of Gaucher

disease caused by Sap-2 deficiency. Som. Cell Mol. Genet. 19, 1±7.

34. Schnabel, D., SchroÈder, M. & Sandhoff, K. (1991) Mutation in the

sphingolipid activator protein 2 in a patient with a variant of Gaucher

disease. FEBS Lett. 284, 57±59.

35. Johansson, J., Curstedt, T. & Jornvall, H. (1991) Surfactant protein B:

disulfide bridges, structural properties, and kringle similarities.

Biochemistry 30, 6917±6921.

36. Longo, M.L., Bisagno, A.M., Zasadzinski, J.A., Bruni, R. & Waring,

A.J. (1993) A function of lung surfactant protein SP-B. Science 261,

453±456.

37. Leippe, M., Ebel, S., Schoenberger, O.L., Horstmann, R.D. & MuÈller-

Eberhard, H.J. (1991) Pore-forming peptide of pathogenic Enta-

moeba histolytica. Proc. Natl Acad. Sci. USA 88, 7659±7663.

38. Andersson, M., Gunne, H., Agerberth, B., Boman, A., Bergman, T.,

Sillard, R., Jornvall, H., Mutt, V., Olsson, B., Wigzell, H., Dagerlind,

A., Boman, H.G. & Gudmundsson, G.H. (1995) NK-lysin, a novel

effector peptide of cytotoxic T and NK cells. Structure and cDNA

cloning of the porcine form, induction by interleukin 2, antibacterial

and antitumor activity. EMBO J. 14, 1615±1625.

39. Ruysschaert, J.M., Goormaghtigh, E., HombleÂ, F., Andersson, M.,

Liepinsh, E. & Otting, G. (1998) Lipid membrane binding of

NK-lysin. FEBS Lett. 425, 341±344.

40. Fischer, G. & Jatzkewitz, H. (1978) The activator of cerebroside

sulfatase: purification from human liver and identification as a

protein. Biochim. Biophys. Acta 528, 69±76.

41. Andersson, M., Holmgren, A. & Spyrou, G. (1996) NK-lysin, a

disulfide-containing effector peptide of T-lymphocytes, is reduced

and inactivated by human thioredoxin reductase. J. Biol. Chem. 271,

10116±10120.

42. Leippe, M., Tannich, E., Nickel, R., van der Goot, G., Pattus, F.,

Horstmann, R.D. & MuÈller-Eberhard, H.J. (1992) Primary and

secondary structure of the pore-forming peptide of pathogenic

Entamoeba histolytica. EMBO J. 11, 3501±3506.

43. Liepinsh, E., Andersson, M., Ruysschaert, J.M. & Otting, G. (1997)

Saposin fold revealed by the NMR structure of NK-lysin. Nat. Struct.

Biol. 4, 793±795.

44. Bernardo, K., . Hurwitz, R., Zenk, T., Desnick, R.J., Ferlinz, K.,

Schuchman, E.H. & Sandhoff, K. (1995) Purification, characteriz-

ation, and biosynthesis of human acid ceramidase. J. Biol. Chem. 270,

11098±11102.

45. de Kruijff, B. (1994) Anionic phospholipids and protein translocation.

FEBS Lett. 346, 78±82.

46. van Klompenburg, W., Nilsson, I., von Heijne, G. & de Kruijff, B.

(1997) Anionic phospholipids are determinants of membrane protein

topology. EMBO J. 16, 4261±4266.

47. Wherrett, J.R. & Huterer, S. (1972) Enrichment of Bis-(monoacyl-

glyceryl) phosphate in lysosomes from rat liver. J. Biol. Chem. 247,

4114±4120.

48. Tjiong, H.B. & Debuch, H. (1978) Lysosomal Bis (monoacylglycero)

phosphate of rat liver, its induction by chloroquine and its structure.

Hoppe-Seyler's Z. Physiol. Chem. 359, 71±79.

49. Frentzen, M. & Debuch, H. (1981) Production of Bis (monoacyl-

glycero) phosphate from phosphatidylglycerol in isolated liver

lysosomes of chloroquine-pretreated rats. Hoppe-Seyler's Z. Physiol.

Chem. 362, 1229±1236.

50. Kobayashi, T., Stang, E., Fang, K.S., de Moerloose, P., Parton, R.G. &

Gruenberg, J. (1998) A lipid associated with the antiphospholipid

syndrome regulates endosome structure and function. Nature

(London) 392, 193±197.