Insights into Krabbe disease fromstructures of galactocerebrosidaseJanet E. Deanea,1, Stephen C. Grahamb, Nee Na Kimc, Penelope E. Steinc, Rosamund McNairc,M. Begoña Cachón-Gonzálezc, Timothy M. Coxc, and Randy J. Reada,1

aDepartment of Haematology and bClinical Biochemistry, Cambridge Institute for Medical Research, University of Cambridge, Addenbrooke’s Hospital,Hills Road, Cambridge CB2 0XY, United Kingdom; and cDepartment of Medicine, University of Cambridge, Addenbrooke’s Hospital, Hills Road,Cambridge CB2 2QQ, United Kingdom

Edited by Gregory A. Petsko, Brandeis University, Waltham, MA, and approved July 22, 2011 (received for review April 8, 2011)

Krabbe disease is a devastating neurodegenerative disease charac-terized by widespread demyelination that is caused by defects inthe enzyme galactocerebrosidase (GALC). Disease-causing muta-tions have been identified throughout the GALC gene. However,a molecular understanding of the effect of these mutations hasbeen hampered by the lack of structural data for this enzyme. Herewe present the crystal structures of GALC and the GALC-productcomplex, revealing a novel domain architecture with a previouslyuncharacterized lectin domain not observed in other hydrolases.All three domains of GALC contribute residues to the substrate-binding pocket, and disease-causing mutations are widely distrib-uted throughout the protein. Our structures provide an essentialinsight into the diverse effects of pathogenic mutations on GALCfunction in human Krabbe variants and a compelling explanationfor the severity of many mutations associated with fatal infantiledisease. The localization of disease-associated mutations in thestructure of GALCwill facilitate identification of those patients thatwould be responsive to pharmacological chaperone therapies.Furthermore, our structure provides the atomic framework forthe design of such drugs.

glycoside hydrolase family 59 ∣ glycosyl hydrolase

Krabbe disease, also known as globoid-cell leukodystrophy, isan autosomal recessive neurodegenerative disorder caused

by a deficiency of the lysosomal enzyme galactocerebrosidase(GALC, EC 3.2.1.46) (1). Defects in GALC that result in reduc-tion of enzyme activity lead to the accumulation of the cytotoxicmetabolite psychosine, resulting in demyelination due to theapoptosis of myelin-forming cells (2, 3). Most patients withKrabbe disease suffer from rapidly progressive leukoencephalo-pathy with onset before 6 mo of age and death occurring beforethe age of 2 y (4). Rare attenuated variants of Krabbe diseaseinclude late-infantile, juvenile, and adult presentations with dis-ease severity and progression being highly variable. The preva-lence of Krabbe disease is approximately 1 in 100,000 births, butthere is wide variation among countries. Two genetically homo-genous communities have an extremely high prevalence of infan-tile Krabbe disease with about 1 in 100–150 live births (5).

Sequence characterization of the human GALC gene and thediscovery of the naturally occurring Twitcher mouse model ofthe disease have contributed to the molecular understanding ofKrabbe disease (6–9). GALC is produced and glycosylated inthe endoplasmic reticulum (ER)-Golgi complex after which itis trafficked, via the mannose-6-phosphate (M6P) pathway, tothe lysosome (10). GALC is essential for normal catabolism ofgalactosphingolipids, including the principal lipid componentof myelin, galactocerebroside (10). Sphingolipid degradationrequires the combined action of water-soluble hydrolases andnonenzymatic sphingolipid activator proteins known as saposins(11, 12).

More than 70 mutations in the GALC gene have been asso-ciated with severe clinical phenotypes and these mutations aredistributed throughout the sequence of GALC. The mutational

profile of patients is highly heterogeneous, often occurring incompound heterozygote patterns, such that the course of the dis-ease cannot easily be predicted from the genotype at the GALClocus (13–15). As with many other lysosomal storage diseases,Krabbe patients with similar or identical genotypes can have var-ied clinical presentations and course of their disease (4).

Here we present the structure of mouse GALC (83% identitywith human GALC, Fig. S1) alone and in complex with D-galac-tose. These structures reveal the presence of a unique domainarrangement, identify the nature of substrate binding and cataly-sis, and provide insight into the differing roles of mutationsoccurring in human Krabbe disease variants.

ResultsCrystal Structure of GALC.The crystal structure of GALC refined to2.1-Å resolution contains one molecule per asymmetric unit com-prising residues 25 to 668 (Fig. 1 and Table S1); GALC is thearchetype of the carbohydrate-active enzymes (CAZy) glycosidehydrolase family 59 (GH59). The overall fold comprises threedomains: a central triosephosphate isomerase (TIM) barrel, aβ-sandwich domain, and a lectin domain (Fig. 1). The first 24residues of GALC encode the signal peptide for targeting to theER that in our construct is replaced with the sequence requiredfor secretion by HEK293 cells and is cleaved during secretion(as occurs with the wild-type signal peptide in vivo). The structureof GALC reveals that residues 25–40 contribute two β-strandsto the β-sandwich domain before forming the ðβ∕αÞ8 TIM barrel(residues 41–337) that lies at the center of the structure. Residues338–452 then form the remainder of the β-sandwich domain.Residues 453–471 lie across one face of the structure stretchingfrom the β-sandwich to the lectin domain (residues 472–668). Theinterfaces formed between each of the domains of GALC are verylarge: 22% (1;770 Å2) and 15% (1;480 Å2) of the solvent acces-sible surface areas of the β-sandwich domain and the lectindomain, respectively, are buried in their interfaces with the TIMbarrel. Thus, the domains of GALC are not likely to move relativeto each other.

The central TIM barrel is composed of eight parallel β-strandssurrounded by α-helices. The connecting loops on the C-terminal

Author contributions: J.E.D., P.E.S., M.B.C.-G., T.M.C., and R.J.R. designed research; J.E.D.,S.C.G., N.N.K., R.M., and M.B.C.-G. performed research; J.E.D., S.C.G., T.M.C., and R.J.R.analyzed data; and J.E.D., S.C.G., N.N.K., P.E.S., M.B.C.-G., T.M.C., and R.J.R. wrotethe paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Freely available online through the PNAS open access option.

Data deposition: The atomic coordinates and structure factors have been deposited in theProtein Data Bank, www.pdb.org (PDB ID codes 3zr5 and 3zr6).

See Commentary on page 15017.1To whom correspondence may be addressed. E-mail: jed55@cam.ac.uk or rjr27@cam.ac.uk.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1105639108/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1105639108 PNAS ∣ September 13, 2011 ∣ vol. 108 ∣ no. 37 ∣ 15169–15173

BIOCH

EMISTR

YSE

ECO

MMEN

TARY

Dow

nloa

ded

by g

uest

on

Oct

ober

14,

202

0

side of the barrel (proximal to the lectin domain) are generallylonger than those on the opposite side of the barrel. The β-sand-wich domain comprises two twisted β-sheets with a similar topol-ogy to that seen in other glycosyl hydrolases except for a very longloop that wraps over the top of the TIM barrel. A cysteine residuein this loop, C378, forms a disulfide bridge with C271 in the TIMbarrel. A calcium ion is bound in the lectin domain of GALC in apentagonal bipyramidal configuration (Fig. S2). The lectin do-main of GALC possesses a similar fold and calcium-binding siteto that of β-glucanase (PDB ID code 3d6e, rmsd ¼ 2.38 Å over155 Cα atoms). However, the lectin domain of GALC does notpossess the catalytic residues present in the β-glucanase family.The lectin domain of GALC also possesses structural similarityto galectin proteins (PDB ID code 2yv8, rmsd ¼ 2.45 Å over129 Cα atoms) and other carbohydrate-binding lectins. ResiduesN284, N363, N387, and N542 displayed electron density for gly-cosylation moieties, and these have been modeled in the GALCstructure (Fig. 1). No ordered electron density was observed for

glycosylation at additional predicted sites in the lectin domain(N585 and N629).

Substrate Specificity. GALC catalyses the hydrolysis of the galac-tosyl moiety from glycosphingolipids such as galactocerebrosideand psychosine (Fig. 2A). To better understand the substratebinding of GALC, we solved the structure of the enzyme in com-plex with D-galactose to 2.4-Å resolution (Fig. 2B and Table S1).

Fig. 1. Structure of GALC shown in two orthogonal views. Ribbon diagramof GALC colored by domain: β-sandwich (red), TIM barrel (blue), linker(orange), and lectin domain (green). The disulfide bond (yellow spheres),calcium ion (gray sphere), and glycosylation moieties (violet sticks) areshown. The N and C termini are marked by labeled circles (Top).

Fig. 2. Substrate-binding site of GALC. (A) Schematic diagram of theGALC substrate galactocerebroside. (B) Ribbon diagram of GALC illustratingbound galactose (pink sticks) with domains colored as in Fig. 1. (C) Surfacerepresentation of GALC zoomed in on the galactose-binding site, coloredand oriented as in B. (D) Unbiased difference (FO-FC) electron density (greenmesh) corresponding to galactose bound to GALC. (E) Active site of GALCillustrating the relevant atomic distances (purple dashed lines) betweenthe bound galactose (pink sticks) and the proposed catalytic residues: thenucleophile (E258, blue sticks) and the proton donor (E182, blue sticks).(F) Substrate specificity for galactose- (pink sticks) rather than glucose-(yellow sticks) containing glycolipids is conferred by W291 and T93 (bluesticks).

15170 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1105639108 Deane et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

14,

202

0

The overall fold of GALC is unchanged upon galactose binding(rmsd ¼ 0.17 Å over 641 Cα atoms), the core of the bindingpocket being formed by the long loops on the C-terminal faceof the TIM barrel as has been observed for other enzymes con-taining TIM barrels. However, in our structure we see that loopsfrom both the β-sandwich and lectin domains also contribute tothe substrate-binding pocket (Fig. 2C). The position and orienta-tion of galactose in the active site is unambiguous as shown by thecharacteristic shape of the difference density calculated beforeinclusion of galactose in the model (Fig. 2D). The interactionsand atomic distances between atoms of galactose and GALCidentify E258 as the active site nucleophile and E182 as the pro-ton donor (Fig. 2E). The average distance between the carboxyloxygens of these catalytic residues is 5.0 Å, consistent with theretaining mechanism of enzymatic glycosidic bond hydrolysisin which the product retains the same stereochemistry as thesubstrate (16).

In addition to the catalytic glutamates E182 and E258, severalresidues contribute to the formation of the substrate-bindingpocket and form hydrogen bonds with the bound galactosemolecule (Fig. S3). The position and orientation of galactose inthe GALC active site differs from that seen in other galactose-containing hydrolase structures, such as acid α-galactosidase(Fig. S4), but is similar to the position of glucose analogues inglucocerebrosidase (GlcCerase, also known as acid β-glucosi-dase), the hydrolase responsible for cleaving glucose-containingsphingolipids (17, 18). Our structure reveals how GALC specifi-cally recognizes galactose-containing lipids. Galactose and glu-cose differ only in the position of a hydroxyl group at onecarbon (C4, Fig. 2F). In the galactose-bound structure of GALC,this hydroxyl forms a hydrogen bond with T93: a residue withunusual backbone dihedral angles (φ ¼ 139°, ψ ¼ −48°) thatare maintained in the absence of galactose, indicating this residueconfers specificity rather than undergoing an induced fit. Glucoseis not compatible with the substrate-binding pocket as the posi-tion of its hydroxyl group would clash with residue W291. Theconformation of W291 is stabilized by a hydrogen bond withthe backbone carbonyl of G47. Interestingly, other lysosomal gly-cosyl hydrolases, including GlcCerase and β-glucuronidase, con-serve a tryptophan residue at this position but in a differentconformation that is incompatible with the structure of GALC(Fig. S5). The binding of galactose is further stabilized by hydro-gen bonding to R380 at the tip of the long loop that stretchesfrom the β-sandwich domain. This loop is secured in positionby the disulfide bond formed between C271 in the TIM barreland C378 in the loop. The position of the side chain of R380 al-ters slightly upon galactose binding to accommodate the ligand.

DiscussionSubstrate Binding. The galactose-bound structure of GALC re-veals that the active site pocket accommodates only the galactosemoiety and does not have space for the lipid tails that are presentin the natural substrates (Fig. 2 A and C). Analysis of the loopssurrounding the active site suggests that GALC is unlikely toundergo a conformational change similar to that describedfor GlcCerase in order to accommodate substrate (Fig. S6)(18, 19). The scissile bond points out toward the surface ofthe protein but in order to gain access to the membrane-em-bedded substrate, glycolipid hydrolases require saposins (saps)(20). GALC-mediated degradation of galactolipids requires theactivity of sapA (21–23) and possibly also sapC (21, 22, 24).Two mechanisms of action for saposins have been postulated.The first suggests saposins extract the lipids from bilayers to formwater-soluble lipid–protein complexes that present the substrateto the appropriate enzyme (11). The second involves the bindingof enzyme at the bilayer surface where saposin molecules facil-itate the access to substrate by distorting the bilayer (12, 25). Dur-ing GALC activation it is possible that both sapA and sapC will

be involved via these different mechanisms. Indeed, it has beenshown for GlcCerase that both sapA and sapC can activate theenzyme and that they mediate their activating effects bybinding to distinct sites on the enzyme (26). It has been notedpreviously that saposins possess a highly negatively charged sur-face (27) and that in the case of sapC this needs to be partiallyneutralized to trigger membrane binding (28). Our structurereveals that the surface charge of GALC at pH 4.8, equivalentto the acidified lysosome, is þ10 e− with a very large, highly po-sitively charged patch surrounding the substrate-binding site(Fig. S7). As this patch is near to the substrate-binding pocketand possesses a surface charge that would be compatible withsapC or negatively charged lipid binding, it is likely to be the siteof their association. Interestingly, human GlcCerase, which alsobinds sapC, possesses a similar surface charge distribution butwithout conserving the residues that confer this charge (Fig. S7).

Disease-Associated Mutations of GALC. Mutations have been iden-tified in the GALC gene that affect gene splicing and mRNAstability or cause deletions, frameshifts, and missense mutations(5, 13–15, 29–34). A large proportion of the missense mutationsin GALC that cause Krabbe disease are likely to result in proteinmistargeting or premature degradation (33). Mistargeting couldoccur at several stages in the GALC processing pathway, includ-ing failure to transit from the ER to Golgi due to misfolding, andblockage of trafficking to the lysosome due to altered binding tothe M6P receptor. In addition, if mutations affect the binding ofGALC to essential activating factors in the lysosome (includingsaposins) then, despite being correctly processed and localized,GALC may not be able to access substrate for efficient glyco-sphingolipid cleavage.

The structure of GALC provides detailed insight into the roleof many disease-associated mutations. Nearly 70% of the mis-sense mutations that result in Krabbe disease involve the modi-fication of residues that are buried within the structure of theenzyme (Table S2). These mutations will lead to instability or mis-folding of GALC resulting in the premature degradation of theenzyme. Examples of such mutations are not restricted to thecore TIM barrel domain but can be found in each domain ofGALC. Mutations that are likely to result in severe misfoldinginclude E114K and S257F in the TIM barrel, L364R andW410G in the β-sandwich domain, and G537R and L629R inthe lectin domain (Fig. 3 and Table S2). Previous work has shownthat GALC possessing this last mutation, L629R, does indeed ac-cumulate in the ER and is not secreted by cells or detected in thelysosome (33). Misfolding or destabilizing mutations of GALC donot necessarily result in an enzyme that is catalytically inactive.Several disease-associated mutations that are completely buriedin the structure have been shown to retain residual enzyme activ-ity (Table S2) (14, 15, 33, 35, 36). In those cases where the mu-tated form of GALC is trapped in the ER but still capable ofcleaving substrate the use of pharmacological chaperone therapy(PCT) would be clinically relevant. PCT involves the use of smallmolecules that stabilize the enzyme structure allowing it to escapethe ER, thus correcting the trafficking defect. Pharmacologicalchaperones have been identified for similar lysosomal storage dis-eases such as Fabry, Gaucher, and Pompe diseases and are nowbeing translated into clinical applications (37).

Within the ER-Golgi complex GALC undergoes modificationby the addition of glycans to specific asparagine residues. ThisN-linked glycosylation is essential for the correct trafficking ofGALC including binding to the M6P receptor (36). A commonmutation found in Krabbe disease patients of Arab ancestry isD528N (5), which has been shown to introduce a new glycosyla-tion site into GALC (33). The mutant GALC protein is hypergly-cosylated, not efficiently taken up by cells and not present in thelysosome (33). D528 lies on a loop of the lectin domain with theside chain buried in the interface with the TIM barrel (Fig. 3).

Deane et al. PNAS ∣ September 13, 2011 ∣ vol. 108 ∣ no. 37 ∣ 15171

BIOCH

EMISTR

YSE

ECO

MMEN

TARY

Dow

nloa

ded

by g

uest

on

Oct

ober

14,

202

0

Mutation to asparagine will not only alter the glycosylation stateof GALC but will destabilize this part of the structure.

Some disease-causing mutations may not affect the residualenzyme activity, processing, or localization of GALC, but mayinterfere with binding to activating factors. Consistent with thishypothesis, the disease-associated mutation E215K has beenshown to have a very mild effect on GALC enzymatic activity(29), and our structure reveals that it is exposed on the surfaceof the TIM barrel (Fig. 3). This mutation confers an oppositecharge on the same face as the substrate-binding pocket suggest-ing that the mechanism of disease for this mutation will involvethe perturbation of a binding face for an activating factor, such asa saposin, that is essential for efficient in vivo glycolipid metabo-lism. Similarly, residue P302 that is found on the surface ofGALC very close to the substrate-binding pocket is mutated toarginine in Krabbe disease, a mutation that would significantlyalter the surface properties in this region (Fig. 3).

Perhaps the most unexpected observation from the structure ofGALC is the contribution from the β-sandwich domain of a longloop that forms an integral part of the substrate-binding site.R380 at the tip of this loop directly binds the galactose moleculein the active site (Fig. 3). The importance of this residue forGALC activity is confirmed by its mutation to tryptophan or leu-cine leading to severe infantile Krabbe disease (13, 38).

Role of the Lectin Domain.A striking feature of the GALC structureis the presence of the lectin domain sitting over one face of theTIM barrel. Although several lysosomal enzymes possess TIMbarrel and β-sandwich domains similar to that seen in GALC,the presence of a lectin domain is unique. This domain contri-butes directly to the formation of the substrate-binding cleft(Fig. 2C) and the presence of several disease-causing mutationsin this domain highlights its importance in the normal processingand activity of GALC (Fig. 3 and Table S2). It is likely that thelectin domain will be involved in processes that are not found inrelated enzymes lacking this domain (18). For example, the lectindomain is unlikely to be involved in saposin binding as GlcCerase,which does not possess the lectin domain, retains both the binding

of sapA and sapC as activators (26, 39) and the positively chargedpatch on the TIM barrel near the active site (Fig. S7). The lectinfold is almost universally a carbohydrate recognition domain (40),suggesting that this region of GALC may play a role via carbo-hydrate binding. This activity may involve binding of glycosylatedproteins during the processing and trafficking of GALC. In sup-port of this, one family of lectins known as galectins, to which theGALC lectin domain possesses structural homology, has identi-fied roles in protein trafficking and sorting into specific vesiclepopulations (41). It has been noted previously that GALC maybe taken up by neighboring cells via M6P-independent mechan-isms (13). In cultured skin fibroblasts the uptake of GALC by un-transduced cells is inhibited less than 50% byM6P, indicating thatthe uptake is not solely via the M6P receptor-mediated pathway.It is possible that the lectin domain of GALC is responsible forconferring this additional uptake mechanism. However, we notethat other lysosomal enzymes also undergo M6P-independenttrafficking via mechanisms that do not require the presence ofa lectin domain (42).

Cleavage of GALC. Many studies of GALC have focused on theprocessing of GALC into 50- and 30-kDa fragments following up-take into the lysosome (6, 10, 43, 44). The cleavage site lies withina loop of the β-sandwich domain, far from the active site. Follow-ing cleavage it is extremely unlikely that the two fragments woulddissociate given the very large buried surface area between them.We therefore conclude that this cleavage event has no relevancefor enzyme activity, consistent with the previous identification ofactivity of the 80-kDa “precursor” (33). Thus, it is incorrecttargeting, with the side effect of failure to encounter processingenzymes, rather than lack of cleavage per se that is likely to be thecritical deficit in mutant forms of GALC that are not cleaved.

The structure of GALC has provided essential insights intothe role of disease-associated mutations in the pathogenesis ofKrabbe disease. GALC is now seen to possess a unique domainarrangement, including a previously uncharacterized lectin do-main and a substrate-binding site that involves each of the threedomains of the protein. This structure provides a rich framework

Fig. 3. Disease-associated mutations of GALC. A selection of residues that are mutated in Krabbe disease are shown as sticks (beige) on the structure of GALC.For each residue a relevant view is provided (Insets) illustrating the surrounding region of the structure that would be perturbed by the mutation. Because ofthe insertion of a residue in the human sequence at position 507, residues 528 and 629 correspond to 527 and 628, respectively, in the mouse structure.

15172 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1105639108 Deane et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

14,

202

0

for further understanding of Krabbe disease and the developmentof potential therapeutics.

Materials and MethodsHis-tagged mouse GALC was produced recombinantly in HEK293T cells andpurified from the supernatant using cobalt-affinity chromatography. Theprotein was crystallized at 4 mg∕mL using sitting-drop vapor diffusion withmicroseeding (45) against a reservoir of 0.2 M sodium acetate, 0.1 M sodiumcacodylate, 34% PEG 8000. Diffraction data were recorded at Diamond LightSource beamline I03. The structure was solved by MIRAS using AutoSol (46)with data from Mercury and Platinum derivatives. For the product complex

crystals were soaked with 20 mM D-galactose for 45 min before data collec-tion. The structure was built using ARP/wARP (47) and COOT (48) and refinedusing phenix.refine (49). Detailed methods, data collection, and refinementstatistics are provided in SI Materials and Methods.

ACKNOWLEDGMENTS. We thank the staff of beamline I03 at the DiamondLight Source. J.E.D. is funded by a Wellcome Trust grant (to R.J.R.). S.C.G.is an 1851 Research Fellow. N.N.K. was supported by a grant from the JeanShanks Foundation and the Cambridge Overseas Trust. M.B.C.-G. was fundedby the Hunter’s Hope Foundation and the Biomedical Research Centre of theNational Institute of Health Research.

1. Suzuki K (2003) Globoid cell leukodystrophy (Krabbe’s disease): Update. J Child Neurol18:595–603.

2. Nagara H, Ogawa H, Sato Y, Kobayashi T, Suzuki K (1986) The twitcher mouse: Degen-eration of oligodendrocytes in vitro. Brain Res 391:79–84.

3. Tanaka K, Nagara H, Kobayashi T, Goto I (1988) The twitcher mouse: Accumulation ofgalactosylsphingosine and pathology of the sciatic nerve. Brain Res 454:340–346.

4. Wenger DA, Rafi MA, Luzi P, Datto J, Costantino-Ceccarini E (2000) Krabbe disease:Genetic aspects and progress toward therapy. Mol Genet Metab 70:1–9.

5. Rafi MA, Luzi P, Zlotogora J, Wenger DA (1996) Two different mutations are respon-sible for Krabbe disease in the Druze and Moslem Arab populations in Israel. HumGenet 97:304–308.

6. Chen YQ, Rafi MA, de Gala G, Wenger DA (1993) Cloning and expression of cDNAencoding human galactocerebrosidase, the enzyme deficient in globoid cell leukody-strophy. Hum Mol Genet 2:1841–1845.

7. Sakai N, et al. (1994) Krabbe disease: Isolation and characterization of a full-lengthcDNA for human galactocerebrosidase. Biochem Biophys Res Commun 198:485–491.

8. Duchen LW, Eicher EM, Jacobs JM, Scaravilli F, Teixeira F (1980) Hereditary leucodystro-phy in the mouse: The new mutant twitcher. Brain 103:695–710.

9. Sakai N, et al. (1996) Molecular cloning and expression of cDNA for murine galacto-cerebrosidase and mutation analysis of the twitcher mouse, a model of Krabbe’sdisease. J Neurochem 66:1118–1124.

10. Nagano S, et al. (1998) Expression and processing of recombinant human galactosyl-ceramidase. Clin Chim Acta 276:53–61.

11. Sandhoff K, Kolter T (1996) Topology of glycosphingolipid degradation. Trends CellBiol 6:98–103.

12. Kolter T, Sandhoff K (2005) Principles of lysosomal membrane digestion: Stimulationof sphingolipid degradation by sphingolipid activator proteins and anionic lysosomallipids. Annu Rev Cell Dev Biol 21:81–103.

13. Wenger DA, Rafi MA, Luzi P (1997) Molecular genetics of Krabbe disease (globoid cellleukodystrophy): Diagnostic and clinical implications. Hum Mutat 10:268–279.

14. Xu C, Sakai N, Taniike M, Inui K, Ozono K (2006) Six novel mutations detected in theGALC gene in 17 Japanese patients with Krabbe disease, and new genotype-pheno-type correlation. J Hum Genet 51:548–554.

15. Tappino B, et al. (2010) Identification and characterization of 15 novel GALC genemutations causing Krabbe disease. Hum Mutat 31:E1894–1914.

16. Davies G, Henrissat B (1995) Structures and mechanisms of glycosyl hydrolases. Struc-ture 3:853–859.

17. Lieberman RL, D'Aquino JA, Ringe D, Petsko GA (2009) Effects of pH and imino-sugar pharmacological chaperones on lysosomal glycosidase structure and stability.Biochemistry 48:4816–4827.

18. Lieberman RL, et al. (2007) Structure of acid beta-glucosidase with pharmacologicalchaperone provides insight into Gaucher disease. Nat Chem Biol 3:101–107.

19. Wei RR, et al. (2011) X-ray and biochemical analysis of N370S mutant human acid beta-glucosidase. J Biol Chem 286:299–308.

20. Sandhoff K, Kolter T (2003) Biosynthesis and degradation of mammalian glycosphin-golipids. Philos Trans R Soc Lond B Biol Sci 358:847–861.

21. Morimoto S, et al. (1989) Saposin A: Second cerebrosidase activator protein. Proc NatlAcad Sci USA 86:3389–3393.

22. Harzer K, et al. (1997) Saposins (sap) A and C activate the degradation of galactosyl-ceramide in living cells. FEBS Lett 417:270–274.

23. Matsuda J, Vanier MT, Saito Y, Tohyama J, Suzuki K (2001) A mutation in the saposin Adomain of the sphingolipid activator protein (prosaposin) gene results in a late-onset,chronic form of globoid cell leukodystrophy in the mouse. Hum Mol Genet10:1191–1199.

24. Zschoche A, Furst W, Schwarzmann G, Sanhoff K (1994) Hydrolysis of lactosylceramideby human galactosylceramidase and GM1-beta-galactosidase in a detergent-free sys-tem and its stimulation by sphingolipid activator proteins, sap-B and sap-C Activatorproteins stimulate lactosylceramide hydrolysis. Eur J Biochem 222:83–90.

25. Wilkening G, Linke T, Sandhoff K (1998) Lysosomal degradation on vesicular mem-brane surfaces. Enhanced glucosylceramide degradation by lysosomal anionic lipidsand activators. J Biol Chem 273:30271–30278.

26. Fabbro D, Grabowski GA (1991) Human acid beta-glucosidase. Use of inhibitory andactivatingmonoclonal antibodies to investigate the enzyme’s catalytic mechanism andsaposin A and C binding sites. J Biol Chem 266:15021–15027.

27. Ahn VE, Leyko P, Alattia JR, Chen L, Prive GG (2006) Crystal structures of saposins A andC. Protein Sci 15:1849–1857.

28. de Alba E, Weiler S, Tjandra N (2003) Solution structure of human saposin C: pH-dependent interaction with phospholipid vesicles. Biochemistry 42:14729–14740.

29. De Gasperi R, et al. (1996) Molecular heterogeneity of late-onset forms of globoid-cellleukodystrophy. Am J Hum Genet 59:1233–1242.

30. Furuya H, et al. (1997) Adult onset globoid cell leukodystrophy (Krabbe disease):Analysis of galactosylceramidase cDNA from four Japanese patients. Hum Genet100:450–456.

31. De Gasperi R, et al. (1999)Molecular basis of late-life globoid cell leukodystrophy.HumMutat 14:256–262.

32. Fu L, et al. (1999) Molecular heterogeneity of Krabbe disease. J Inherit Metab Dis22:155–162.

33. Lee WC, et al. (2010) Molecular characterization of mutations that cause globoid cellleukodystrophy and pharmacological rescue using small molecule chemical chaper-ones. J Neurosci 30:5489–5497.

34. Lee WC, et al. (2006) Suppression of galactosylceramidase (GALC) expression in thetwitcher mouse model of globoid cell leukodystrophy (GLD) is caused by nonsense-mediated mRNA decay (NMD). Neurobiol Dis 23:273–280.

35. Harzer K, Knoblich R, Rolfs A, Bauer P, Eggers J (2002) Residual galactosylsphingosine(psychosine) beta-galactosidase activities and associated GALC mutations in late andvery late onset Krabbe disease. Clin Chim Acta 317:77–84.

36. Kornfeld S (1986) Trafficking of lysosomal enzymes in normal and disease states. J ClinInvest 77:1–6.

37. Parenti G (2009) Treating lysosomal storage diseases with pharmacological chaper-ones: From concept to clinics. EMBO Mol Med 1:268–279.

38. Selleri S, et al. (2000) Deletion of exons 11–17 and novel mutations of the galactocer-ebrosidase gene in adult- and early-onset patients with Krabbe disease. J Neurol247:875–877.

39. Alattia JR, Shaw JE, Yip CM, Prive GG (2007) Molecular imaging of membrane inter-faces reveals mode of beta-glucosidase activation by saposin C. Proc Natl Acad Sci USA104:17394–17399.

40. Weis WI, Drickamer K (1996) Structural basis of lectin-carbohydrate recognition. AnnuRev Biochem 65:441–473.

41. Delacour D, Koch A, Jacob R (2009) The role of galectins in protein trafficking. Traffic10:1405–1413.

42. Reczek D, et al. (2007) LIMP-2 is a receptor for lysosomal mannose-6-phosphate-independent targeting of beta-glucocerebrosidase. Cell 131:770–783.

43. Chen YQ,Wenger DA (1993) Galactocerebrosidase from human urine: Purification andpartial characterization. Biochim Biophys Acta 1170:53–61.

44. Sakai N, et al. (1994) Purification and characterization of galactocerebrosidase fromhuman lymphocytes. J Biochem 116:615–620.

45. Walter TS, et al. (2008) Semi-automated microseeding of nanolitre crystallizationexperiments. Acta Crystallogr Sect F Struct Biol Cryst Commun 64:14–18.

46. Terwilliger TC, et al. (2009) Decision-making in structure solution using Bayesianestimates of map quality: the PHENIX AutoSol wizard. Acta Crystallogr D Biol Crystal-logr 65:582–601.

47. Langer G, Cohen SX, Lamzin VS, Perrakis A (2008) Automated macromolecular modelbuilding for X-ray crystallography using ARP/wARP version 7. Nat Protoc 3:1171–1179.

48. Emsley P, Lohkamp B, Scott WG, Cowtan K (2010) Features and development of Coot.Acta Crystallogr D Biol Crystallogr 66:486–501.

49. Adams PD, et al. (2010) PHENIX: A comprehensive Python-based system for macromo-lecular structure solution. Acta Crystallogr D Biol Crystallogr 66:213–221.

Deane et al. PNAS ∣ September 13, 2011 ∣ vol. 108 ∣ no. 37 ∣ 15173

BIOCH

EMISTR

YSE

ECO

MMEN

TARY

Dow

nloa

ded

by g

uest

on

Oct

ober

14,

202

0