Molecular imaging of membrane interfaces revealsmode of �-glucosidase activation by saposin CJean-Rene Alattia*, James E. Shaw†‡, Christopher M. Yip†‡§, and Gilbert G. Prive*†¶�

*Division of Cancer Genomics and Proteomics, Ontario Cancer Institute, 101 College Street, Toronto, ON, Canada M5G 1L7; and Departments of†Biochemistry, §Chemical Engineering and Applied Chemistry, and ¶Medical Biophysics, and ‡Institute of Biomaterials and Biomedical Engineering,University of Toronto, Toronto, ON, Canada M5S 3G9

Edited by K. Sandhoff, Institut fur Organische Chemie und Biochemie, Bonn, Germany, and accepted by the Editorial Board September 7, 2007 (received forreview May 29, 2007)

Acid �-glucosidase (GCase) is a soluble lysosomal enzyme respon-sible for the hydrolysis of glucose from glucosylceramide andrequires activation by the small nonenzymatic protein saposin C(sapC) to gain access to the membrane-embedded glycosphingo-lipid substrate. We have used in situ atomic force microscopy (AFM)with simultaneous confocal and epifluorescence microscopies toinvestigate the interactions of GCase and sapC with lipid bilayers.GCase binds to sites on membranes transformed by sapC, andenzyme activity occurs at loci containing both GCase and sapC.Using FRET, we establish the presence of GCase/sapC and GCase/product contacts in the bilayer. These data support a mechanism inwhich sapC locally alters regions of bilayer for subsequent attackby the enzyme in stably bound protein complexes.

atomic force microscopy � confocal microscopy � FRET � interfacialcatalysis � lipid storage disease

Gaucher disease is a common lysosomal storage diseasecharacterized by the presence of engorged macrophages in

the liver, spleen, and bone marrow (1, 2). The disorder is causedby abnormal accumulations of glucosylceramide (GlcCer) inthese tissues because of the inability to catabolize this lipidwithin lysosomes. Typically, the breakdown of membrane gly-cosphingolipids requires the combined action of a hydrolyticenzyme and a nonenzymatic ‘‘activator’’ protein, which, in thecase of GlcCer, are acid �-glucosidase (GCase, EC 3.2.1.45) andsaposin C (sapC). In the absence of activator and acidic phos-pholipids, the enzyme does not have direct access to the glyco-sphingolipid, which is tightly packed within the lipid bilayer.Imiglucerase [Cerezyme (Genzyme, Cambridge, MA), recom-binant GCase expressed in mammalian cell cultures] is used inenzyme replacement therapy of Gaucher patients (3).

The activator proteins saposin A, B, C, and D are homologous,soluble, nonenzymatic proteins that interact with lysosomalmembranes and facilitate the breakdown of glycosphingolipidsby specific hydrolases (2, 4, 5). Recent findings have alsoimplicated saposins in glycolipid antigen presentation by CD1molecules, where saposins are thought to provide the means bywhich glycolipids are extracted from membranes and loadedonto CD1 molecules (6–8). Current models for enzyme activa-tion address the location of saposin-mediated lipid–hydrolaseinteractions and define the ‘‘solubilizer’’ and ‘‘liftase’’ modes ofaction. In the former model, target lipid molecules are extractedfrom bilayers by saposins and presented to cognate enzymes assoluble protein–lipid complexes. In contrast, the ‘‘liftase’’ modelinvolves the binding of enzyme to the bilayer surface wheresaposin molecules facilitate the access to the glycosphingolipidsubstrates. Different saposins and enzymes are hypothesized tofall into a particular category of enzyme activation. For example,saposin B is thought to be a detergent-like lipid solubilizer (9,10), whereas sapC may act as a liftase at the bilayer surface (2,11, 12). Strong binding of sapC to lipid bilayers was shown byusing coprecipitation assays (13) and NMR spectroscopic titra-tions (14). Simultaneous atomic force microscopy (AFM)/fluorescence microscopy allowed the visualization of saposin

binding to planar bilayers and the resulting membrane transfor-mation (15–18). Although sapC has no enzymatic activity,sapC-transformed areas contain narrow channels surroundingislets of the original bilayer in a manner strikingly similar to theeffects of secreted phospholipase A2 (19) on lipid bilayers. LikeGCase, phospholipases are water-soluble enzymes that act onmembrane-embedded lipids, some using a membrane-dockingC2 domain distinct from the catalytic domain (20, 21). Pancreaticlipase is another example of a soluble enzyme that hydrolizeslipids in aggregated forms. The enzyme requires the action of asmall activator protein, colipase, which anchors the enzyme tolipid particles and stabilizes its active conformation (22). Con-ceptually, some aspects of the lipase/colipase pair may be similarin the GCase/sapC association acting according to the ‘‘liftase’’model.

Despite extensive in vivo and in vitro studies, it remains unclearwhether the activation of GCase by sapC can be exclusivelyassigned to either model. An understanding of saposin-mediatedglycosphingolipid hydrolysis requires the characterization ofGCase interactions with saposin- and substrate-containing lipidbilayers. In this study, we present the simultaneous visualizationof sapC and GCase action on model membranes by AFM andconfocal f luorescence imaging. Using a membrane-bound flu-orogenic substrate in the combined microscope, we directlyobserve enzyme activation at the membrane interface. Finally,we establish GCase/sapC and GCase/product contacts by FRETwithin the bilayer. Our results are fully consistent with amembrane-bound reaction involving a form of interfacialcatalysis.

ResultsGCase Membrane Localization. The addition of sapC to a supportedbilayer resulted in the lowering of the membrane from a smallnumber of nucleation sites that are preferentially initiated fromthe edges of membrane defects, as described (15, 16) (Fig. 1). Atconcentrations �1 �M, these areas were lowered by �1.5 nm,probably because of partial lipid removal, and contained stablybound sapC (15). After the addition of GCase to a final molarratio of 1:3 (GCase:sapC), material accumulated exclusively onareas that were previously remodeled by sapC to levels that were

Author contributions: J.-R.A., C.M.Y., and G.G.P. designed research; J.-R.A. and J.E.S.performed research; C.M.Y. contributed new reagents/analytic tools; J.-R.A. and G.G.P.analyzed data; and J.-R.A. and G.G.P. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission. K.S. is a guest editor invited by the Editorial Board.

Abbreviations: sapC, saposin C; A488, Alexa Fluor 488; A546-sapC, A546-labeled sapC;AFM, atomic force microscopy; GCase, acid �-glucosidase; GlcCer, glucosylceramide;DFUG, 6,8-difluoro-4-heptadecylumbelliferyl �-D-glucopyranoside; DFHU, 6,8-difluoro-4-heptadecylumbelliferyl �-D-glucopyranoside; MUG, 4-methylumbelliferyl-�-D-glucopyranoside; PI, 1,2-diacyl-sn-3-phosphoinositol.

�To whom correspondence should be addressed. E-mail: prive@uhnres.utoronto.ca.

This article contains supporting information online at www.pnas.org/cgi/content/full/0704998104/DC1.

© 2007 by The National Academy of Sciences of the USA

17394–17399 � PNAS � October 30, 2007 � vol. 104 � no. 44 www.pnas.org�cgi�doi�10.1073�pnas.0704998104

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

8, 2

020

2–8 nm higher than the unaffected bilayer (Fig. 1). We alsoobserved the occasional presence of larger aggregates in theseareas, which are seen as white features in Fig. 1. Although sapCalone has the potential to form raised plaques and aggregatesatop remodeled areas, these appear only at sapC concentrations�5 �M (15). We therefore presume that the material observedupon the addition of the enzyme is composed of GCase or amixture of GCase and sapC. Because the diameter of GCase is�5 nm (23), the change in height is consistent with the presenceof one to two molecules of the enzyme. The features that arosefrom the addition of GCase did not appear in areas unaffectedby sapC action.

To confirm the identity of the proteins bound to the supportedlipid bilayer, we used combined in situ AFM/multiprobe confocalf luorescence imaging (Fig. 2). sapC was labeled with the AlexaFluor 546 (A546) fluorophore at residue 22. This position is ina short loop between helices �1 and �2 of the protein and is thesite normally glycosylated in the natural protein (24). GCase waschemically labeled using an amine-reactive Alexa Fluor 488(A488) succinimidyl conjugate. Images of A546-labeled sapC(A546-sapC) and A488-labeled GCase (A488-GCase) fluores-cence were collected in separate channels (red and green,respectively). Similar to the experiment in Fig. 1, the supportedplanar bilayer had a model lysosomal lipid composition withsome initial defects. As shown (15), f luorescent sapC localizedto the lipid surface in saposin-induced lowered areas [Fig. 2 andsupporting information (SI) Fig. 7]. After the addition of A488-GCase, granular material similar to that observed in high-resolution AFM scans (Fig. 1) was detected in areas of saposin-induced bilayer lowering (Fig. 2B) along with weak A488-GCasefluorescence in the GCase channel (Fig. 2 A).

The low intensity of the A488-GCase signal made it difficultto correlate the sapC and GCase fluorescence distributions.Because the ‘‘liftase’’ mechanism implies a direct sapC/GCaseinteraction, we expected that the fluorophores on the twoproteins would be within their Forster radius (25, 26), resultingin FRET between the A488-GCase donor and the A546-sapCacceptor. In this scenario, FRET would reduce the A488-GCasefluorescence. We tested this hypothesis by photobleaching theA546-sapC with the confocal microscope laser and observed astrong increase of A488-GCase fluorescence (Fig. 2 A). Thus,sapC and GCase are in close contact on the membrane surface.The enhanced A488-GCase intensity allowed clear correlationbetween the A546-sapC and A488-GCase fluorescence distri-butions (Fig. 2B), confirming the assignment of the AFMfeatures.

GCase Enzyme Activity. We used a membrane-bound fluorogenicsubstrate analog to image and assay GCase activity in bilayers.The compound 6,8-dif luoro-4-heptadecylumbelliferyl �-D-glucopyranoside (DFUG) is similar to soluble umbelliferyl-based glucosides used as GCase substrates in previous studies(12, 27, 28) but contains a C17 acyl chain that anchors themolecule to the membrane. DFUG is an excellent analogue ofthe natural GlcCer substrate for measuring membrane-boundGCase activity, because hydrolysis of the glucoside headgroupyields the f luorescent compound 6,8-dif luoro-7-hydroxy-4-heptadecylcoumarin (DFHU), which also remains in the lipidbilayer.

In the absence of activator protein, GCase had a low butdetectable level of enzyme activity when incubated with lipo-somes containing DFUG (Fig. 3). Preincubation of the lipo-somes with varying amounts of sapC before the addition of fixedamounts of enzyme resulted in a dose-dependent increase in theGCase activity. The time curves resemble typical activity rateswhen increasing amounts of substrate are mixed with enzyme,but in this case, the total amount of DFUG substrate is constant,and only the activator concentrations are changing. Thus, sapCis effectively increasing the amount of substrate that is availablefor hydrolysis in this assay. There is a 10-fold increase in theinitial rate of hydrolysis at 2.5 �M sapC, which is similar to the�7-fold increase in GCase activity when incubated with the sameconcentration of sapC and liposomes containing radiolabeledGlcCer (12). No substrate hydrolysis was observed in the absenceof GCase, regardless of the saposin concentration.

These observations were reproduced with supported bilayersin the confocal microscope. A488-GCase was added to DFUG-containing membranes in the absence of sapC. Isolated spots ofintense A488-GCase fluorescence (white in Fig. 4 and SI Fig. 8)correspond to protein bound directly to exposed mica surfacesin areas of large membrane defects and serve as an internalnegative control. Weak but measurable A488-GCase fluores-cence was observed in areas of intact membrane, indicating lowconcentrations of enzyme at the membrane surface. Upon theaddition of A546-sapC, there was a significant increase in theamount of GCase bound at the membrane, giving rise to newlocalized areas of enzyme (Fig. 4A, ‘‘GCase,’’ green patches). Inthe DFHU channel, f luorescence was seen only in areas thatcontained both sapC and GCase (Fig. 4). Thus, sapC resulted inGCase translocation to concentrated patches in the bilayer, andenzyme activation by sapC was confirmed by the colocalizationof the A488-GCase, A546-sapC, and DFHU signals.

- sapC

1 µm 0.1 µm

+ sapC + GCase

05

0 3.5µm-5

0 3.5µm 0 3.5µm 0 0.5µm

nm

Fig. 1. AFM imaging of sapC and GCase interactions with a supported planar bilayer. The series of height images were acquired after the sequential additionof sapC (0.2 �M) and A488-GCase (0.07 �M), as labeled. The images depict the equilibrium state of the bilayer at the end of an evolution period after proteinaddition. The rightmost frame is a higher-resolution image of the GCase-affected area marked by the cyan box in the lower-resolution frame. The lateral XYscale of the images is given by the thick white bars. Pseudocolors are used to represent the height data, with darker areas corresponding to lower features. Theheight profiles corresponding to the dashed lines are all plotted with the same vertical scale.

Alattia et al. PNAS � October 30, 2007 � vol. 104 � no. 44 � 17395

BIO

PHYS

ICS

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

8, 2

020

GCase-sapC–DFHU Interactions. The results described above dem-onstrate the translocation of soluble GCase to the bilayer inareas remodeled by sapC, with the concurrent accumulation ofthe reaction product DFHU at sapC/GCase sites. Next, we usedFRET between A546-sapC and A488-GCase and between A488-GCase and DFHU, as a sensitive tool to determine the presenceof associations between sapC, GCase and membrane-boundDFHU. Fig. 5 shows that the fluorescence from membrane-bound A488-GCase increases when the FRET acceptor A546-sapC is photobleached by the confocal microscope laser (Fig. 5Middle, ‘‘Gcase’’ image). The FRET efficiency, E, for the pairA546-sapC/A488-GCase is 0.65, which indicates very close prox-imity. Similarly, DFHU fluorescence experiences a net increasewhen A488-GCase is photobleached in turn (Fig. 5 Bottom,‘‘DFHU’’ image). This confirms the close proximity of GCase,sapC and the product of the hydrolase reaction, DFHU, in thebilayer. Because there is no spectral overlap between the ab-sorption spectrum of A546 and the emission spectrum of thecoumarin derivative, there is no direct FRET between thelabeled sapC and DFHU. However, we have described FRETbetween sapC and labeled lipid molecules in sapC transformedbilayers (15).

DiscussionGCase is a soluble enzyme that requires a facilitator to access itsmembrane-embedded substrate in the hydrolysis of glucoseheadgroups. Purified GCase was previously shown to interact

546-nm HeNe laser illumination (Bottom Right), resulting in an increase inFRET donor A488-GCase fluorescence in the area corresponding to photo-bleached A546-sapC (Bottom Left). (B) The confocal images labeled 1 and 2 aremagnifications of the respective areas enclosed in dashed boxes in A. Tappingmode AFM height images of the corresponding area are given in Right. TheAFM height image before protein addition is given in Top. Pseudocolors areused to represent the AFM height data, with darker areas corresponding tolower features. Image scales are given by white bars in each image.

10 µm

- sapC

+ sapC

+ GCase

- sapC

+ sapC

+ GCase

A sapC

1

GCase

2

B AFM1 µm

1

2

Fig. 2. Binding of labeled GCase to lipid bilayers remodeled by sapC. (A) Thefirst three rows of confocal microscopy images correspond to the sequentialaddition to a supported planar bilayer (�sapC) of 0.5 �M A546-sapC (�sapC)and 0.05 �M A488-GCase (�Gcase). Left and Right images are, respectively,A488-GCase (green) and A546-sapC (red) channel images of the same bilayerarea. (Bottom) After A546-sapC and A488-GCase addition, A546-sapC (FRETacceptor) was photobleached within a 25 � 25-�m square using intense

0 0.5 1.0 1.5 2.0 2.50

0.5

1.0

[sapC] (µM)

No

rm. I

nit

. rat

e

B

0 1 2 3 4 5

DF

HU

Flu

o. I

nt.

100

200

300

Time (min)

+ G

Cas

e

A

00.10.250.5

1.252.5

[sapC] (µM)

Fig. 3. GCase activity assay using the liposome-anchored DFUG substrate. (A)Kinetics of DFHU accumulation after addition of 0.05 �M GCase to liposomespreincubated with 0, 0.1, 0.25, 0.5, 1.25, and 2.5 �M sapC, as indicated. Thetotal lipid concentration in each reaction was 100 �M, including 10 mol %DFUG. Reaction product DFHU fluorescence intensities are given in arbitraryunits. The arrow indicates the time at which the enzyme was added to eachsample, before mixing and the resumption of fluorescence measurements. (B)Dependence of the GCase initial reaction rates on sapC concentration. Therates were determined from the kinetics in A and normalized relative to thehighest rate. Error bars correspond to �3 standard deviations.

17396 � www.pnas.org�cgi�doi�10.1073�pnas.0704998104 Alattia et al.

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

8, 2

020

only weakly with liposomes and to exhibit a low level of liposomalGlcCer hydrolysis (29). Unlike the pancreatic lipase/colipasesystem (22), there is no indication of an interaction betweensapC and purified delipidated GCase in solution in the absenceof membrane lipids (30–33). The association of the proteins andhydrolase activation is recovered when lipids, especially nega-tively charged lipids, are added (30–33). [Earlier GCase prep-arations were shown to bind Sepharose-coupled sapC (34, 35),and sapC induced the lowering of the GCase Km value for4-methylumbelliferyl-�-D-glucopyranoside (MUG) hydrolysis(35). However, the same preparations were shown to containconsiderable amounts of lipids (35).] Acidic pH and the presenceof the negatively charged lipids 1,2-diacyl-sn-3-phosphoinositol(PI) or bis(monoacylglycero)phosphate (BMP), which are abun-dant in intralysosomal compartment (36), are required for theactivation of GCase activity by sapC (12, 27). In our DFUGassay, the activation of GCase by sapC was maximal in the pHrange 4.8–5.0 and required PI or BMP (SI Fig. 9), as expected.

The sapC preparations used in this study were not glycosylated,and the consequences of covalent modifications at position Asn-22(the single glycosylation site in sapC) deserve comment. Two recentreports have shown that glycosylated recombinant saposins A (37)and B (38) expressed in yeast differ from the unglycosylated formsin their ability to solubilize lipids from immobilized liposomes.However, earlier reports have shown that nonglycosylated saposinsretain their lipid-binding and enzyme activation effects (39), andrecombinant sapC expressed in Escherichia coli was shown to havesimilar properties to sapC purified from natural tissues (40).Notably, E. coli-expressed saposin B (nonglycosylated) was as activeas glycosylated human saposin B purified from urine in an in vitroactivity assay (10) and could functionally complement saposinB-deficient human fibroblast cells (41). We have shown in aprevious report (15) that both unglycosylated sapC and Alexa-labeled N22C sapC have similar lipid extraction activities onsupported bilayers. Moreover, our previous (15) and current results(Figs. 1 and 2 and SI Fig. 7) show that both unlabeled sapC and theAlexa-labeled N22C mutant have similar bilayer transformationeffects when observed by AFM imaging. As an additional test, wefound that the GCase activation properties of Alexa-labeled sapCwere similar to that of unlabeled sapC in the liposomal assay (SI Fig.10). We conclude that the covalent addition of a 1,034-Da fluoro-phore at position Asn-22 has negligible effects on sapC activity inthese assays. Further studies will be required to directly demon-strate any effects of sapC glycosylation in these assays.

Although sapC has the potential to solubilize membrane lipids(12, 15), solution studies with liposomes have led to conflictingconclusions in assigning the GCase/sapC pair to either the‘‘solubilizer’’ or the ‘‘liftase’’ mode of action (12, 33). Our directvisualization of the events occurring at the membrane interfacesuggests that GCase hydrolyzes its substrate at the bilayer levelwith the help of sapC within a complex at the membrane surface.Although we cannot formally exclude the possibility that all orpart of the product may be originating in solution and transferredback to the bilayer, that sapC brings GCase to concentrated areasat the bilayer with a subsequent increase in GCase activity in thesame areas is strong support for an interfacial reaction. Theincreased enzyme activity may be due to several mechanisms,including increased levels of membrane-bound enzyme, a higherspecific activity of the enzyme within an activator complex,and/or better access to the substrate. GlcCer, a membrane-embedded glycosphingolipid with a glucosyl headgroup, is thenatural substrate for GCase. Assuming that the enzyme alonecannot penetrate the bilayer, the lipid substrate must be ‘‘lifted’’for proper docking of the headgroup to the active site (42). Wehave shown that sapC induces a nucleated spread of membraneremodeling that is characterized by a reduced membrane thick-ness (15). sapC is present in the remodeled areas, and aparticular model suggests that sapC removes the top leaflet,whereas its amphiphilic nature allows it to shield the hydropho-

A GCase sapC DFHU

5 µm

GCase DFHU 1.0

0

Flu

ore

scen

ce

- + - +

0.5

B

Fig. 4. sapC-induced GCase binding and activation. (A) Pseudocolor fluorescence images of a lipid bilayer after consecutive additions of 0.1 �M A488-GCaseand 5 �M sapC containing 10% A546-sapC. (Left) Overlay of A488-GCase confocal images before and after sapC addition to the bilayer (SI Fig. 8), showing newlyaccumulated A488-GCase after sapC addition (green areas) and the original areas of GCase accumulation before adding sapC (white patches). The A546-sapCconfocal image (Center) and reaction product DFHU epifluorescence image (Right) of the same area show fluorescence localizations matching the newlyaccumulated A488-GCase fluorescence (Left). (B) Localized A488-GCase and enzyme product (DFHU) fluorescence intensities before (�) and after (�) sapCaddition to the bilayer. Fluorescence intensities were quantified from the dashed box areas and normalized relative to the � intensity in each channel. The areascorrespond to the position of a sapC-induced GCase spot. The images from which the pre-sapC intensities were quantified are not shown.

Fig. 5. Sequential A546-sapC/A488-GCase and A488-GCase/DFHU FRET.(Top) Epifluorescence DFHU (Left) and confocal A488-GCase (Center) andA546-sapC (Right) images of a lipid bilayer after simultaneous exposure to 0.1�M A488-GCase and 1 �M sapC containing 50% A546-sapC. (Top) Imagesbefore any photobleaching. (Middle) A546-sapC was photobleached within a25 �m x 25 �m square using intense 543 nm HeNe laser illumination (Right).(Bottom) A488-GCase was then photobleached within a 25 � 25-�m squareusing intense Ar-ion 488 nm laser illumination (Center). Conditions werechosen to achieve extensive bilayer coverage with sapC and GCase. At thisresolution, the sapC and GCase coverage appears relatively uniform. (Thebright spots correspond to aggregates of sapC and GCase as seen in Fig. 1 andwere confirmed by AFM scans of the same area.) DFUG hydrolysis occurspredominantly in the areas of uniform GCase/sapC coverage.

Alattia et al. PNAS � October 30, 2007 � vol. 104 � no. 44 � 17397

BIO

PHYS

ICS

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

8, 2

020

bic tails of the lower acyl chains (15). The organization of lipidsat the interface between intact bilayers and transformed areaslikely exposes lipid molecules, including the GCase substrate(Fig. 6). The lipid fraction removed by sapC is presumably in asolubilized form and may potentially serve other functions,including loading lipid antigens onto CD1 molecules (6–8).

The creation of perturbed bilayer edges could explain saposin-mediated GCase activation, but it is unclear whether the partiallyexposed substrate is sufficient for enzyme translocation. Elec-trostatic interactions may additionally be involved in securingGCase binding to the membrane. Crystal structures of GCasehave consistently shown the presence of sulfate ion clustersbound in the vicinity of the active site (23, 28, 42, 43). The ionsare thought to mimic negatively charged phospholipids that arealso known to enhance enzyme activity (42). Close proximity ofsapC and GCase molecules is confirmed by our results showingcolocalization of labeled sapC and GCase fluorescences in thebilayer, as well as sapC/GCase FRET. However, because puri-fied and delipidated GCase does not seem to interact with sapCin solution (30–33), GCase binding may depend on the specificrecognition of membrane-bound conformations of sapC.

The sapC-induced increase in GCase activity may also be relatedto a higher intrinsic activity of the enzyme within an activatorcomplex, possibly involving a conformational change in the enzyme(44), as seen in the case of pancreatic lipase. This is supported bythe fact that sapC can increase the hydrolysis of soluble substrateanalogs by up to 17-fold in the presence of large unilamellar vesicles(12, 33). However, this activation cannot be explained by anincrease in the accessibility of catalytic pocket, because the crystalstructure of isolated GCase showed that the active site is easilyaccessible to solution compounds (23). Alternatively, interactionswith sapC and the bilayer leading to an altered configuration of theactive site may account for the enhanced activity. Overall, thesedata provide a view of hydrolytically active sapC/GCase complexes

at membrane surfaces and clarify the role of sapC in the treatmentof Gaucher’s disease by enzyme replacement therapy.

Materials and MethodsFor descriptions of materials, cloning, expression, protein puri-fication, f luorescence labeling, liposome preparation, AFM, andconfocal epif luorescence imaging, refer to SI Text. Otherprocedures are described below.

Lipid Compositions. A mixture of 1,2-diacyl-sn-3-phosphocholine(PC)/cholesterol/PI/1,2-diacyl-sn-3-phosphoethanolamine (PE)/GlcCer (50 mol %/20 mol %/10 mol %/10 mol %/10 mol %) wasused to mimic lysosomal lipid composition in the experimentshown in Fig. 2. In the experiments shown in Fig. 1, a mixture of1,2-dipalmitoyl-sn-3-phosphocholine/phosphatidylcholine/cholesterol/PI/phosphatidylethanolamin/GlcCer (25 mol %/25mol %/20 mol %/10 mol %/10 mol %/10 mol %) was used. Amixture consisting of PC/cholesterol/PI/PE/DFUG (50 mol %/20mol %/10 mol %/10 mol %/10 mol %) was used in theexperiments shown in Figs. 3–5.

Liposome-Based GCase Activity Assay. In this assay, GCase hydrol-izes membrane-embedded DFUG to yield DFHU, which has afluorescence emission peak at �max � 450 nm. DFUG-containingliposomes were prepared as described above in 50 mM sodiumacetate, pH 4.8/150 mM NaCl for use in the assay. Reactionmixtures were 150 �l in volume and contained 100 �M total lipids.The liposomes in each reaction were preincubated for 15 min with0–2.5 �M sapC, as indicated. The reactions were performed at20°C. The DFHU fluorophore was exited at � � 330 nm, andmaximum emission was sampled at 0.5-s intervals over a 10-minperiod using a Photon Technology International (Birmingham, NJ)QM-1 fluorescence spectrophotometer. Baseline fluorescence val-ues were recorded for 1 min before the addition of 0.05 �M GCaseto each reaction sample. Initial rate values were calculated by linearregression of the first 15 points, after GCase addition and mixing.Standard deviations were calculated from residual �2 values.

FRET. FRET was detected as an increase in donor fluorescenceintensity after acceptor photobleaching (26, 45). A488-GCase/A546-sapC and DFHU/A488-GCase were used as FRET donor/acceptor pairs in the experiment shown in Fig. 5. The lipidbilayer containing 10 mol % DFUG (membrane-anchored,f luorogenic GCase substrate). To confirm FRET, a controlexperiment was performed with unlabeled sapC (SI Fig. 11).Consecutive A488-GCase/A546-sapC and DFHU/A488-GCaseFRET experiments were performed by photobleaching within a25 � 25-�m area A546-sapC followed by A488-GCase usingintense HeNe 543- and Ar-ion 488-nm lasers, respectively.Images in the DFHU, A488-Gcase, and A546-sapC channels ofa 70 � 70-�m area centered around the photobleached squarewere acquired before and after each photobleaching event.FRET efficiency, E, was calculated by using the equation E �1 � Db/Da, where Db and Da are the donor fluorescenceintensities before and after acceptor photobleaching, respec-tively (26). Db and Da were quantified from raw images in a 20 �20-�m square within the photobleached area.

We thank Tim Edmunds for support and Genzyme for a gift ofCerezyme. This work was supported by a grant from the CanadianInstitutes of Health Research (to G.G.P. and C.M.Y.).

1. Futerman AH, van Meer G (2004) Nat Rev Mol Cell Biol 5:554–565.2. Kolter T, Sandhoff K (2005) Annu Rev Cell Dev Biol 21:81–103.3. Grabowski GA, Barton NW, Pastores G, Dambrosia JM, Banerjee TK, McKee

MA, Parker C, Schiffmann R, Hill SC, Brady RO (1995) Ann Intern Med122:33–39.

4. Sandhoff K, Kolter T (1996) Trends Cell Biol 6:98–103.

5. Vaccaro AM, Salvioli R, Tatti M, Ciaffoni F (1999) Neurochem Res 24:307–314.6. Winau F, Schwierzeck V, Hurwitz R, Remmel N, Sieling PA, Modlin RL,

Porcelli SA, Brinkmann V, Sugita M, Sandhoff K, et al. (2004) Nat Immunol5:169–174.

7. Zhou D, Cantu C, III, Sagiv Y, Schrantz N, Kulkarni AB, Qi X, Mahuran DJ,Morales CR, Grabowski GA, Benlagha K, et al. (2004) Science 303:523–527.

GCase

Fig. 6. Schematic model for GCase binding to the membrane and saposin-mediated activation. GCase alone cannot extract membrane-embedded Glc-Cer from the bilayer. Saposin can help expose embedded lipids, includingGlcCer (red), to soluble GCase molecules by creating perturbed edges betweenlowered and intact areas of the bilayer. Saposin molecules are colored inpurple; lipid polar groups are blue. Acyl chains belonging to upper and lowerleaflets are colored in light and dark gray, respectively.

17398 � www.pnas.org�cgi�doi�10.1073�pnas.0704998104 Alattia et al.

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

8, 2

020

8. Kang SJ, Cresswell P (2004) Nat Immunol 5:175–181.9. Yuan W, Qi X, Tsang P, Kang SJ, Illarionov PA, Besra GS, Gumperz J,

Cresswell P (2007) Proc Natl Acad Sci USA 104:5551–5556.10. Ahn VE, Faull KF, Whitelegge JP, Higginson J, Fluharty AL, Prive GG (2003)

Protein Expr Purif 27:186–193.11. Sandhoff K, Kolter T (2003) Philos Trans R Soc London Ser B 358:847–861.12. Wilkening G, Linke T, Sandhoff K (1998) J Biol Chem 273:30271–30278.13. Vaccaro AM, Ciaffoni F, Tatti M, Salvioli R, Barca A, Tognozzi D, Scerch C

(1995) J Biol Chem 270:30576–30580.14. de Alba E, Weiler S, Tjandra N (2003) Biochemistry 42:14729–14740.15. Alattia JR, Shaw JE, Yip CM, Prive GG (2006) J Mol Biol 362:943–953.16. You HX, Yu L, Qi X (2001) FEBS Lett 503:97–102.17. You HX, Qi X, Grabowski GA, Yu L (2003) Biophys J 84:2043–2057.18. You HX, Qi X, Yu L (2004) Chem Phys Lipids 132:15–22.19. Grandbois M, Clausen-Schaumann H, Gaub H (1998) Biophys J 74:2398–2404.20. Six DA, Dennis EA (2000) Biochim Biophys Acta 1488:1–19.21. Dessen A, Tang J, Schmidt H, Stahl M, Clark JD, Seehra J, Somers WS (1999)

Cell 97:349–360.22. van Tilbeurgh H, Bezzine S, Cambillau C, Verger R, Carriere F (1999) Biochim

Biophys Acta 1441:173–184.23. Dvir H, Harel M, McCarthy AA, Toker L, Silman I, Futerman AH, Sussman

JL (2003) EMBO Rep 4:704–709.24. Ahn VE, Leyko P, Alattia JR, Chen L, Prive GG (2006) Protein Sci 15:1849–1857.25. Forster T (1948) Annal Phys 2:55–75.26. Berney C, Danuser G (2003) Biophys J 84:3992–4010.27. Salvioli R, Tatti M, Scarpa S, Moavero SM, Ciaffoni F, Felicetti F, Kaneski CR,

Brady RO, Vaccaro AM (2005) Biochem J 390:95–103.28. Liou B, Kazimierczuk A, Zhang M, Scott CR, Hegde RS, Grabowski GA (2006)

J Biol Chem 281:4242–4253.29. Vaccaro AM, Tatti M, Salvioli R, Ciaffoni F, Gallozzi E (1990) Biochim

Biophys Acta 1033:73–79.

30. Sa Miranda MC, Aerts JM, Pinto RA, Magalhaes JA, Barranger JA, Tager JM,Schram AW (1988) Biochim Biophys Acta 965:163–168.

31. Aerts JM, Sa Miranda MC, Brouwer-Kelder EM, Van Weely S, Barranger JA,Tager JM (1990) Biochim Biophys Acta 1041:55–63.

32. Qi X, Grabowski GA (1998) Biochemistry 37:11544–11554.33. Vaccaro AM, Tatti M, Ciaffoni F, Salvioli R, Maras B, Barca A (1993) FEBS

Lett 336:159–162.34. Ho MW (1975) FEBS Lett 53:243–247.35. Berent SL, Radin NS (1981) Biochim Biophys Acta 664:572–582.36. Kobayashi T, Beuchat MH, Chevallier J, Makino A, Mayran N, Escola

JM, Lebrand C, Cosson P, Gruenberg J (2002) J Biol Chem 277:32157–32164.

37. Locatelli-Hoops S, Remmel N, Klingenstein R, Breiden B, Rossocha M,Schoeniger M, Koenigs C, Saenger W, Sandhoff K (2006) J Biol Chem281:32451–32460.

38. Remmel N, Locatelli-Hoops S, Breiden B, Schwarzmann G, Sandhoff K (2007)FEBS J 274:3405–3420.

39. Hiraiwa M, Soeda S, Martin BM, Fluharty AL, Hirabayashi Y, O’Brien JS,Kishimoto Y (1993) Arch Biochem Biophys 303:326–331.

40. Qi X, Leonova T, Grabowski GA (1994) J Biol Chem 269:16746–16753.41. Whitelegge JP, Ahn V, Norris AJ, Sung H, Waring A, Stevens RL, Fluharty CB,

Prive G, Faull KF, Fluharty AL (2003) Cell Mol Biol 49:799–807.42. Brumshtein B, Wormald MR, Silman I, Futerman AH, Sussman JL (2006) Acta

Crystallogr D 62:1458–1465.43. Premkumar L, Sawkar AR, Boldin-Adamsky S, Toker L, Silman I, Kelly JW,

Futerman AH, Sussman JL (2005) J Biol Chem 280:23815–23819.44. Qi X, Kondoh K, Krusling D, Kelso GJ, Leonova T, Grabowski GA (1999)

Biochemistry 38:6284–6291.45. McLean PJ, Kawamata H, Ribich S, Hyman BT (2000) J Biol Chem 275:8812–

8816.

Alattia et al. PNAS � October 30, 2007 � vol. 104 � no. 44 � 17399

BIO

PHYS

ICS

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

8, 2

020