madam professor

Structural snapshots illustrate the catalytic cycle of

β -galactocerebrosidase, the defective enzyme in

Krabbe disease

Chris H. Hill a, Stephen C. Graham b, Randy J. Read a, and Janet E. Deane a,1

aDepartment of Haematology, Cambridge Institute for Medical Research, University of Cambridge, Cambridge CB2 0XY, United Kingdom; and bDepartment of Pathology, University of Cambridge, Cambridge CB2 1QP, United Kingdom

Edited by Gregory A. Petsko, Brandeis University, Waltham, MA, and approved November 8, 2013 (received for review June 27, 2013)

Glycosphingolipids are ubiquitous components of mammalian cell membranes, and defects in their catabolism by lysosomal enzymescauseadiversearrayofdiseases.De ﬁciencies in the enzyme β -galactocerebrosidase (GALC) cause Krabbe disease, a devastating genetic disorder characterized by widespread demyelination andrapid, fatal neurodegeneration. Here, we present a series of high-resolution crystal structures that illustrate key steps in the catalytic cycle of GALC. We have captured a snapshot of the short-lived enzyme –substrate complex illustrating how wild-type GALC binds a bona ﬁde substrate. We have extensively characterized the en- zyme kinetics of GALC with this substrate and shown that the enzyme is active in crystallo by determining the structure of the enzyme –product complex following extended soaking of the crys- tals with this same substrate. We have also determined the struc- ture of a covalent intermediate that, together with the enzyme – substrate and enzyme –product complexes, reveals conformational changes accompanying the catalytic steps and provides key mech-anistic insights, laying the foundation for future design of phar- macological chaperones.

β -galactosylceramidase |lysosomal storage disease |glycosyl hydrolase | pharmacological chaperone therapy

T he recycling and degradation of eukaryotic membrane com- ponents occurs in the lysosome and is essential for cellular

maintenance. The molecular mechanisms of lysosomal lipid deg-

radation are primarily informed by the study of a class of human

diseases, sphingolipidoses, which are caused by inherited defects

in glycosphingolipid catabolism. Krabbe disease is a devastating

neurodegenerative disorder that is caused by de ﬁciencies in the

lysosomal enzyme β-galactocerebrosidase (GALC) (enzyme com-

mission 3.2.1.46). It is essential for the catabolism of galacto-

sphingolipids, including the principal lipid component of myelin,

β -D-galactocerebroside (Fig. 1 A) (1). GALC function has also

been implicated in cancer cell metabolism, primary open-angle

glaucoma and the maintenance of a hematopoietic stem cell

niche (2 –4). GALC catalyzes the hydrolysis of β-D-galactocerebroside to

β -D-galactose and ceramide, as well as the breakdown of psy-

chosine to β-D-galactose and sphingosine. In both cases, removal

of the galactosyl moiety is thought to occur via a retaining two-

step glycosidic bond hydrolysis reaction (5, 6). Our recent

structure of murine GALC identi ﬁed two active site glutamate

residues geometrically consistent with this mechanism (7). In the

ﬁ rst step, the carboxylate group of E258 is hypothesized to per-

form a nucleophilic attack at ring position C 1, forming an enzyme –

substrate intermediate, releasing the ﬁrst product (ceramide or

sphingosine) as the leaving group. In the second step, E182 is

thought to act as a general acid/base to deprotonate a water mol-

ecule, which then attacks the ring, releasing the enzyme and the

second product (galactose). Defects in GALC lead to the accumulation of cytotoxic

metabolites that elicit complex, and still only partially under-

stood, cellular events resulting in apoptosis of myelin-forming

oligodendrocytes and Schwann cells of the central and peripheral

nervous system (8 –10). This leads to the characteristic demy-

elination and neurodegeneration observed in Krabbe disease

patients (11, 12). A wide range of disease-causing missense

mutations have been identi ﬁed throughout the GALC gene (13 –

21). Some of these mutations will result in catalytically inactive

protein, whereas other mutations may destabilize the protein,

compromising lysosomal delivery by causing retention in the

endoplasmic reticulum (ER). For such mutants with residual

catalytic activity, pharmacological chaperone therapy (PCT) may

be an appropriate treatment strategy (22 –24). This approach

involves the use of small molecules that bind to and stabilize the

defective enzyme in the ER and facilitate its delivery to the site

of action, such as the lysosome. PCT is currently under in-

vestigation for the related lysosomal storage disorders Gaucher

disease (25 –27) and Fabry disease (28, 29), with a number of

candidate ligands now in clinical trials (23). Here, we present a comprehensive enzymatic characterization

of GALC accompanied by crystal structures of wild-type GALC

in complex with substrate, product, and covalent intermediate,

illustrating key stable species along the catalytic pathway. These

structures not only reveal the short-lived enzyme –substrate

complex captured using hydrolysable substrate and wild-type

enzyme but also highlight unexpected conformational changes of

active site residues. We con ﬁrm the hypothesized mechanism,

including the involvement of E258 and E182, but also identify

key roles for additional active site residues.

Signi ﬁcance

Defects in the enzyme β-galactocerebrosidase (GALC) result in the devastating neurodegenerative disorder Krabbe disease. GALC is responsible for the degradation and recycling of gly-cosphingolipids that form the primary lipid component of themyelin sheath that insulates nerve cells. A detailed under- standing of how GALC processes substrate will facilitate the development of new drug therapies for Krabbe disease. Thisstudy reveals a series of structural snapshots of GALC captured during different steps of the catalytic cycle. These structures identify speci ﬁc residues within the active site that undergo signi ﬁcant movements during substrate cleavage, providing key insight into the catalytic mechanism of GALC.

Author contributions: C.H.H., S.C.G., and J.E.D. designed research; C.H.H. and J.E.D. per- formed research; C.H.H. and J.E.D. analyzed data; and C.H.H., S.C.G., R.J.R., and J.E.D. wrote the paper.

The authors declare no con ﬂict of interest.

This article is a PNAS Direct Submission.

Data deposition: The atomic coordinates and structure factors have been deposited in the Protein Data Bank, www.pdb.org [PDB ID codes 4CCC (GALC-4N βDG), 4CCD (GALC- D- galactal), and 4CCE (GALC-galactose)]. 1To whom correspondence should be addressed. E-mail: [email protected].

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. 1073/pnas.1311990110/-/DCSupplemental .

www.pnas.org/cgi/doi/10.1073/pnas.1311990110 PNAS |December 17, 2013 |vol. 110 |no. 51 |20479 –20484

BIOCHEMISTRY Results

Enzyme Kinetics. Accurate measurement of the GALC kinetic

parameters has been hampered by the low expression levels of

the enzyme in tissues and the dif ﬁculty in purifying the enzyme

because of its hydrophobicity (30, 31). We have previously de-

termined the structure of mouse GALC by purifying the enzyme

from stably expressing HEK 293T cells: the murine enzyme

shares 83% sequence identity with the human enzyme, and all

active site residues are conserved (7). To characterize the en-

zyme kinetics of GALC, we used the chromogenic substrate 4-

nitrophenyl- β-D-galactopyranoside (4N βDG) (Fig. 1 B) because

of its high solubility at a range of pH values. We con ﬁrm that the

optimum pH of GALC activity is between pH 4.5 and 5.0 con-

sistent with its lysosomal localization (Fig. 1 D). Michaelis –

Menten plots were used to determine Kmand Vmax parameters at

several enzyme concentrations (Fig. 1 Eand Fig. S1 ). At pH 4.6,

the Kmfor 4N βDG is 5.1 ±0.3 mM and the Vmax is 46.6 ±1.6

nmol ·min −1·μg−1, giving a kcat of 57.8 s −1. These values are

comparable to those of another lysosomal hydrolase, human

α -galactosidase, determined using a similar substrate ( Km, 8.3

mM; kcat, 63.5 s −1) (32) and the hydrolysis of lactose and allo-

lactose by Escherichia coli β-galactosidase ( Km,∼1 mM; kcat,

∼ 60 s −1) (33, 34).

Enzyme –Substrate Complex. Typically enzyme –substrate com-

plexes can only be formed by using catalytically inactive mutant

forms of an enzyme or nonhydrolysable substrate analogs. The

combination of small crystal size (5.4 ×10−5mm 3), high solvent

content (61.9%), and reduced catalytic activity of GALC at pH

6.8, the pH at which crystals were grown, made GALC crystals

excellent candidates for diffusion-trapping experiments ( Fig. S2 )

(35, 36). Rapid soaking (20 –40 s) and cryocooling of crystals

allowed us to capture the substrate 4N βDG in the active site of

the wild-type enzyme (Fig. 2 A). Electron-density maps revealed

unambiguously the presence of the unhydrolysed substrate in the

active site, and the enzyme –substrate complex structure was re-

ﬁ ned at 2.1-Å resolution (Fig. 2 B,Fig. S3 A, and Table 1).

The galactosyl hydroxyl groups make multiple speci ﬁc contacts

with the enzyme ( Fig. S4 A) but do not require signi ﬁcant con-

formational change of active site loops, indicating that the sub-

strate binding pocket of GALC is highly preorganized. This is in

contrast to the considerable conformational changes that occur

in loops around the active site of the related enzyme β-gluco-

cerebrosidase (25, 37). The scissile bond and 4-nitrophenyl agly-

cone project from the surface of the enzyme and contribute very

little to substrate speci ﬁcity or recognition (Fig. 2 C). This is con-

sistent with the requirement for lipid-binding saposin proteins in

order for GALC to cleave the natural, lipidated substrates, and it

is likely that the speci ﬁcity of binding is conferred by the saposin

proteins rather than the hydrophobic lipid tails of the substrate

molecules (38, 39). Although there is not substantial movement of active site

loops upon substrate binding, to accommodate substrate in the

active site pocket, the side chain conformations of R380 and the

catalytic acid/base residue E182 are signi ﬁcantly altered (Figs. 3

A and B). The R380 side chain changes conformation to form

hydrogen bonds with the galactosyl 6-hydroxyl, whereas the

movement of the E182 side chain is necessary to accommodate

the scissile bond of the substrate. Despite this movement, E182

maintains the hydrogen bond with H237 seen in the absence of

substrate. In the enzyme –substrate complex, the O e2atom of the

E182 side chain is ∼3.1 Å from the leaving group oxygen, ready

to donate a proton during departure of 4-nitrophenol. Unlike

several other retaining β-glycosidases where alignment of the

anomeric carbon with the syn lone pair of the nucleophile is

achieved by distorting the substrate into 1S3geometry, in the

GALC enzyme –substrate complex, the substrate binds in an undis-

torted low-energy 4C1conformation similar to that seen for E. coli

β -galactosidase (33, 40).

Enzyme –Intermediate Complex. To capture a structural snapshot of

the covalent enzyme –intermediate complex, we incubated GALC

crystals with 1,2-dideoxy- D- lyxo -hex-1-enopyranose ( D-galactal)

(Fig. 1 C).D-galactal is a slow-binding inhibitor of E. coli β-galac-

tosidase and can form an intermediate with a half-life suf ﬁciently

Fig. 1. Enzyme kinetic analysis of GALC. ( A–C) Schematic diagram of the natural substrate β-D-galactocerebroside ( A), the chromogenic substrate 4N βDG ( B), and the inhibitor D-galactal ( C). (D)pHpro ﬁle of enzyme activity for GALC. SEM error bars are shown. ( E) Michaelis –Menten plots of initial velocity vs. substrate concentration at GALC concentrations of 3.18 nM ( ○), 4.24 nM, ( ◇), 5.66 nM, ( ▼), 7.54 nM, ( ▲), 10.1 nM, ( □), and 13.4 nM ( ●). (Inset ) Plot of Vmax obs vs. GALC concentration showing kcat(57.8 s −1) as the gradient. SEM error bars are shown.

20480 |www.pnas.org/cgi/doi/10.1073/pnas.1311990110 Hill et al. long to allow soaking and cryocooling of crystals (33, 41). We

showed that D-galactal is a potent inhibitor of GALC ( Ki=32 ±

1.8 μM; Fig. S5 ) and were able to trap the intermediate complex

for structure determination. The enzyme acts as a catalyst for the

hydration of the double bond of D-galactal to form 2-deoxy-

galactose via a 2-deoxygalactosyl –galactocerebrosidase interme-

diate (Fig. 3 C). The covalent linkage ( ∼1.3 Å) between the C1

atom of D-galactal and the O e 2atom of E258 is unambiguous and

only requires a minor conformational change of E258 (Fig. 3 D

and Fig. S3 B). To form the covalent linkage, there is a signi ﬁcant

Fig. 2. Structure of the wild-type GALC enzyme in complex with substrate. ( A) Ribbon diagram showing the overall structure of GALC with the substrate 4N βDG bound in the active site. The electron density (2 FO-FCcontoured at 0.25 e −/Å3, blue) is shown for uncleaved substrate bound in the active site pocket of GALC. Surface glycans (pink sticks) are shown. ( B) Detail of the GALC active site with bound, uncleaved substrate showing active site residues (sticks) and electron density (as above). ( C) Surface representation of the GALC active site (gray) with the substrate and electron density shown (as above).

Table 1. Data collection and re ﬁnement statistics

Statistic GALC-4N βDG GALC- D-galactal GALC-galactose

Data collection Space group R32 R32 R32 Cell dimensionsa ,b,c(Å) 250.0, 250.0, 77.8 249.5, 249.5, 77.7 248.8, 248.8, 77.6 α ,β,γ(°) 90, 90, 120 90, 90, 120 90, 90, 120 Resolution (Å) 56.38 –2.09 (2.15 –2.09)* 47.46 –1.97 (2.02 –1.97)* 47.01 –2.06 (2.12 –2.06)* Rmerge 0.153 (1.287) 0.138 (1.040) 0.143 (1.402) < I/σI> 13.5 (2.5) 10.6 (2.2) 13.9 (2.4) CC1/2 0.999 (0.579) 0.998 (0.598) 0.998 (0.618) Completeness (%) 99.9 (99.4) 99.8 (100.0) 99.9 (100.0)Redundancy 19.0 (14.8) 9.7 (9.2) 13.5 (13.8) Re ﬁnement Resolution (Å) 56.38 –2.09 (2.12 –2.09) 47.45 –1.97 (1.99 –1.97) 44.25 –2.06 (2.09 –2.06) No. re ﬂections 54646 (2561) 64813 (2683) 56378 (2673) Rwork / Rfree 0.187/0.224 0.170/0.198 0.172/0.206 Ramachandran favored region (%) 95.29 95.76 95.44 Ramachandran outliers (%) 0.47 0.47 0.47No. of atoms Protein 5119 5127 5119 Carbohydrate 98 98 98Ligand 21 10 12Ca2 +ion 111 Water 272 267 206 B-factorsProtein 55.1 36.1 45.3Carbohydrate 88.1 70.2 85.4 Ligand 54.9 22.3 43.5 Ca2 +ion 72.2 36.4 59.7 Water 50.5 35.7 40.0r.m.s. deviations Bond lengths (Å) 0.008 0.007 0.007 Bond angles (°) 1.12 1.07 1.05

*Values in parentheses are for highest-resolution shell. One crystal was used per structure.

Hill et al. PNAS |December 17, 2013 |vol. 110 |no. 51 |20481

BIOCHEMISTRY movement of the C1 atom of the pyranose ring compared with

the enzyme –substrate complex, about 1.3 Å deeper into the active

site pocket (Fig. 3 D). The pyranose ring retains an undistorted 4C1chair conformation and is stabilized by two new hydrogen

bonds to Y238 ( Fig. S4 B). This movement upon intermediate

formation is much more subtle than that seen for E. coli

β -galactosidase (33) and human α-galactosidase (32), more closely

resembling that seen for the covalent intermediate of hen egg

white lysozyme captured using a catalytically inactivated mutant

enzyme (42). The crucial role of E258 in the catalytic mechanism was tested

by generating a mutant of GALC, where this residue was changed

to glutamine (E258Q), resulting in the change of a single non-

hydrogen atom (O e 2to N). At pH 4.6, this mutant enzyme has a

melting temperature ( Tm) similar to the wild-type enzyme and is

correctly folded, and yet all activity is lost, con ﬁrming the role of

E258 as the catalytic nucleophile ( Fig. S6 ).

Enzyme –Product Complex. Extended soaking (10 min) of GALC

crystals with substrate revealed only the product galactose in the

active site, demonstrating that GALC is active in crystallo (Fig. 3 E

and Figs. S3 Cand S4C). This structure con ﬁrms that we have

crystallized a catalytically competent conformation of the enzyme

and, in combination with the enzyme –substrate and covalent in-

termediate structures, provides a series of snapshots illustrating

key stages of the GALC catalytic cycle (Fig. 3 G and Movie S1 ).

Discussion

Movement of Active Site Residues and the Role of R380. The struc-

tural snapshots captured here reveal that during catalysis the side

chain of the catalytic acid/base residue E182 can adopt two

distinct conformations (Fig. 3 Aand C). The reversible move-

ment of catalytic side chains has been observed before for other

enzymes (43, 44). In addition, we observe that the noncatalytic

substrate-binding residue R380 also alternates between two dif-

ferent conformations throughout the catalytic cycle (Fig. 3 Band

F ). Upon binding of substrate, the side chain of R380 moves to

accommodate the substrate and forms a hydrogen bond with the

galactosyl 6-hydroxyl. Upon formation of the covalent interme-

diate, R380 moves back to a conformation similar to that seen in

the unliganded enzyme. However, in this complex, two ordered

water molecules that were not present in the unliganded struc-

ture appear in the active site, hydrogen bonded to both R380 and

E182 ( Figs. S4 Band S7). To complete the catalytic cycle, E182

activates one of these waters to attack the enzyme-bound in-

termediate. In the enzyme –product complex, the side chain of

R380 is again hydrogen bonded to the galactose 6-hydroxyl, as

seen in the enzyme –substrate complex. The importance of R380

for GALC activity is highlighted by the severe infantile Krabbe

disease that arises from its mutation to tryptophan or leucine

(17, 45).

Substrate Distortion. Based on studies with hen egg white lyso-

zyme, a general catalytic mechanism has been proposed for all

retaining β-glycosidases involving initial substrate distortion and

the electrophilic migration of the anomeric carbon along the

reaction coordinate to form a covalent intermediate, typically

from 1S3to 4C1via a 4H3oxocarbenium ion-like transition state

(40, 42). In this mechanism, accumulating positive charge at the

anomeric center is stabilized by the p-like lone pair of the

Fig. 3. Structures of the GALC active site illustrating conformational changes along the reaction coordinate. ( A) Wild-type enzyme in complex with substrate 4N βDG (green). ( B) Overlay of the GALC active site residues in the absence (gray) and presence (green) of substrate. ( C) Covalent intermediate structure illustrating D-galactal (orange) covalently attached to the catalytic nucleophile. ( D) Movement of E258 and the pyranose ring between substrate binding (green) and covalent linkage with the inhibitor D-galactal (orange). ( E) Enzyme –product complex formed following extended incubation with substrate il- lustrating catalytic activity of GALC in crystallo .(F) The two different conformations of the R380 side chain in the covalent intermediate (orange) and product (pink) complexes. ( G) Schematic representation of the proposed retaining two-step glycosidic bond hydrolysis reaction.

20482 |www.pnas.org/cgi/doi/10.1073/pnas.1311990110 Hill et al. endocyclic oxygen when the C1, C2, C5, and O atoms are co- planar. Although this has been validated in several subsequent Michaelis complexes (46, 47), here we observe that in all stable species along the reaction coordinate, the galactosyl moietyremains in the lowest-energy 4C1conformation. This does not preclude distortion as part of the reaction coordinate but is in agreement with the observed lack of substrate distortion seen for the E. coli β-galactosidase (33). It has been proposed that for polymeric substrates the distortion of one sugar is compensated for by the binding interactions of others (46). The shallow sub- strate-binding pocket present in GALC means that there are veryfew speci ﬁc interactions other than those to the galactosyl moiety, suggesting that, for this substrate, distortion may not be ener- getically stable.

Toward PCT. Currently, the only available treatment for early in- fantile Krabbe disease is hematopoietic stem cell transplantation.Although other related lysosomal storage diseases (LSDs) can be treated by enzyme-replacement therapy, for Krabbe disease this strategy is unsuitable because it is primarily a disease of thecentral nervous system: the administered enzyme will not crossthe blood brain barrier and thus will not reach the site of action. For LSDs where the mutant enzyme possesses some residual catalytic activity, PCT is a feasible alternative treatment approachbecause even partial restoration of traf ﬁcking to the lysosome can provide suf ﬁcient enzyme activity to prevent disease (48). High- throughput screening of small-molecule libraries for PCT candi- dates for Krabbe disease has so far been unable to identify anysmall molecules that signi ﬁcantly increase GALC activity (49). One potential PCT candidate for Krabbe disease is α-lobeline, which was shown in tissue culture to increase the activity of the hyperglycosylated mutant D528N form of GALC (50). However,this molecule was not effective with other mutated forms of GALC and, because of its chemical structure, is not likely to bind the active site of GALC, potentially limiting its speci ﬁcity. The elucidation of glycosyl hydrolase structures and mecha-nisms has driven the recent development of candidate PCT molecules that speci ﬁcally bind the enzyme active site to stabilize partially defective enzymes implicated in a range of human dis-eases (26, 51, 52). Glycan-analog PCT candidates have beenidenti ﬁed for other LSDs. One example is the iminosugar 1-deoxynojirimycin, which has been shown to stabilize α-galac- tosidase A by binding to the active site (28) and is currently inphase III clinical trials for the treatment of Fabry disease (29). Three related molecules, isofagomine (25), N-butyl-, and N-nonyl- deoxynojirimycin (26) have been shown to bind the acid β -glucosidase active site and increase enzymatic activity in cell lines and patient ﬁbroblasts expressing clinically relevant muta- tions responsible for Gaucher disease (27, 53). The structures of GALC described here provide the atomic detail necessary to aidthe design of small molecules that speci ﬁcally bind the active site of GALC, facilitating the development of pharmacological chap- erone therapies for Krabbe disease.

To summarize, we have shown how a bona ﬁde substrate binds in the active site pocket of the wild-type GALC enzyme. Struc-tural snapshots of the covalent intermediate and product com-plexes of GALC provide a comprehensive illustration of the

catalytic cycle of this medically important enzyme. The insights gained from this series of structures into enzyme –ligand inter- actions and active site conformational dynamics provide anatomic framework for the rational design of small moleculeinhibitors and pharmacological chaperones.

Materials and Methods

Protein Expression and Puri ﬁcation. His6-tagged murine wild-type and E258Q GALC was recombinantly expressed by stably transfected HEK 293T cell lines and puri ﬁed from conditioned medium using nickel-af ﬁnity chromatogra- phy. When stored at 4 °C, puri ﬁed GALC was stable and retained full en- zymatic activity for at least 2 wk. Detailed methods are provided in SI Materials and Methods .

Crystallization and Small-Molecule Soaks. GALC protein was concentrated to 2.5 mg/mL in 150 mM NaCl, 10 mM Hepes (pH 7.4). Crystals were grown by sitting-drop vapor diffusion with microseeding (54) against a reservoir of 0.2M sodium acetate, 0.1 M sodium cacodylate (pH 6.8), and 34% wt/vol polyethylene glycol 8000. For enzyme –substrate and enzyme –product com- plexes, crystals were soaked with 20 mM 4N βDG for 20 –40 s and 10 min, respectively. For the covalent intermediate complex, crystals were soaked with 20 mM D-galactal for 2 h. Immediately after soaking, crystals were cryoprotected with per ﬂuoropolyether oil before ﬂash-cooling in liquid nitrogen. Diffraction data were recorded at the Diamond Light Source beamline I04-1. Final re ﬁnement statistics for all structures are shown in Table 1, and the re ﬁned models plus structure factors have been deposited in the Protein Data Bank. Detailed data collection and re ﬁnement methods are provided in SI Materials and Methods .

Enzyme Activity Assays. Endpoint assays were conducted with chromogenic substrate 4-nitrophenyl- β-D-galactopyranoside (4N βDG). In all experiments, formation of product 4-nitrophenol was monitored spectrophotometrically by terminating reactions with stopping buffer (360 mM NaOH, 280 mM glycine, pH 10.6) and measuring A410. All experiments were performed at 37 °C with shaking. pH-pro ﬁle experiments were conducted with 10 mM 4N βDG in a range of citrate/phosphate buffers (pH values: 4.0, 4.5, 5.0, 5.5, 6.0, 6.5,and 7.0) supplemented with 50 mM NaCl. Steady-state kinetic experiments were performed in 20 mM sodium acetate, 150 mM NaCl, 0.1% vol/vol Nonidet P-40 (pH 4.6) with 4N βDG concentrations of 50, 25, 12.5, 6.25, 3.13, 1.56, and 0.78 mM and GALC concentrations of 13.4, 10.1, 7.54, 5.66, 4.24, and 3.18 nM. Experiments were performed in triplicate. Kinetic parameterswere obtained by curve- ﬁtting using Prism 5 (GraphPad). Details of proce- dures and analyses are provided in SI Materials and Methods . Validation of steady-state conditions is provided in Fig. S1 .

ACKNOWLEDGMENTS. We thank the staff of beamline I04-1 at the Diamond Light Source. We thank Begoña Cachón-González for the murine β-galacto- cerebrosidase expression construct. C.H.H. is funded by a Wellcome Trust PhD studentship; S.C.G. is supported by a Sir Henry Dale fellowship, jointly funded by the Wellcome Trust and The Royal Society (Grant 098406/Z/12/Z);R.J.R. is supported by a Principal Research Fellowship funded by the Well- come Trust (Grant 082961/Z/07/Z); and J.E.D. is supported by a Royal Society University Research Fellowship (UF100371). The Cambridge Institute for Medical Research is supported by Wellcome Trust Strategic Award 100140.

1. Nagano S, et al. (1998) Expression and processing of recombinant human gal- actosylceramidase. Clin Chim Acta 276(1):53 –61. 2. Beier UH, Görögh T (2005) Implications of galactocerebrosidase and galactosylcere- broside metabolism in cancer cells. Int J Cancer 115(1):6 –10. 3. Liu Y, et al. (2011) GALC deletions increase the risk of primary open-angle glaucoma: The role of Mendelian variants in complex disease. PLoS ONE 6(11):e27134. 4. Visigalli I, et al. (2010) The galactocerebrosidase enzyme contributes to the mainte- nance of a functional hematopoietic stem cell niche. Blood 116(11):1857 –1866. 5. Sandhoff K, Kolter T (1996) Topology of glycosphingolipid degradation. Trends Cell Biol 6(3):98 –103. 6. Davies G, Henrissat B (1995) Structures and mechanisms of glycosyl hydrolases. Structure 3(9):853 –859. 7. Deane JE, et al. (2011) Insights into Krabbe disease from structures of galactocere- brosidase. Proc Natl Acad Sci USA 108(37):15169 –15173. 8. Kanazawa T, et al. (2000) Inhibition of cytokinesis by a lipid metabolite, psychosine. J Cell Biol 149(4):943 –950.

9. Giri S, Khan M, Rattan R, Singh I, Singh AK (2 006) Krabbe disease: Psychosine-mediated activation of phospholipase A2 in o ligodendrocyte cell death. J Lipid Res 47(7):1478 –1492. 10. Formichi P, et al. (2007) Psychosine-induced apoptosis and cytokine activation in im- mune peripheral cells of Krabbe patients. J Cell Physiol 212(3):737 –743. 11. D ’Agostino AN, Sayre GP, Hayles AB (1963) Krabbe ’s disease. Globoid cell type of leukodystrophy. Arch Neurol 8:82 –96. 12. Tanaka K, Nagara H, Kobayashi T, Goto I (1988) The twitcher mouse: Accumulation of galactosylsphingosine and pathology of the sciatic nerve. Brain Res 454(1-2):340 –346. 13. Ra ﬁMA, Luzi P, Zlotogora J, Wenger DA (1996) Two different mutations are re- sponsible for Krabbe disease in the Druze and Moslem Arab populations in Israel.Hum Genet 97(3):304 –308. 14. Tappino B, et al. (2010) Identi ﬁcation and characterization of 15 novel GALC gene mutations causing Krabbe disease. Hum Mutat 31(12):E1894 –E1914. 15. Fiumara A, et al. (2011) Krabbe leukodystrophy in a selected population with high rate of late onset forms: Longer survival linked to c.121G >A (p.Gly41Ser) mutation. Clin Genet 80(5):452 –458.

Hill et al. PNAS |December 17, 2013 |vol. 110 |no. 51 |20483

BIOCHEMISTRY 16. De Gasperi R, et al. (1996) Molecular heterogeneity of late-onset forms of globoid-cell leukodystrophy. Am J Hum Genet 59(6):1233 –1242. 17. Wenger DA, Ra ﬁMA, Luzi P (1997) Molecular genetics of Krabbe disease (globoid cell leukodystrophy): Diagnostic and clinical implications. Hum Mutat 10(4):268 –279. 18. Xu C, Sakai N, Taniike M, Inui K, Ozono K (2006) Six novel mutations detected in the GALC gene in 17 Japanese patients with Krabbe disease, and new genotype-pheno-type correlation. J Hum Genet 51(6):548 –554. 19. Lissens W, et al. (2007) A single mutation in the GALC gene is responsible for the majority of late onset Krabbe disease patients in the Catania (Sicily, Italy) region. Hum Mutat 28(7):742. 20. Fu L, et al. (1999) Molecular heterogeneity of Krabbe disease. J Inherit Metab Dis 22(2):155 –162. 21. Furuya H, et al. (1997) Adult onset globoid cell leukodystrophy (Krabbe disease): Analysis of galactosylceramidase cDNA from four Japanese patients. Hum Genet 100(3-4):450 –456. 22. Cohen FE, Kelly JW (2003) Therapeutic approaches to protein-misfolding diseases. Nature 426(6968):905 –909. 23. Parenti G (2009) Treating lysosomal storage diseases with pharmacological chaper- ones: From concept to clinics. EMBO Mol Med 1(5):268 –279. 24. Fan JQ (2003) A contradictory treatment for lysosomal storage disorders: Inhibitors enhance mutant enzyme activity. Trends Pharmacol Sci 24(7):355 –360. 25. Lieberman RL, et al. (2007) Structure of acid beta-glucosidase with pharmacological chaperone provides insight into Gaucher disease. Nat Chem Biol 3(2):101 –107. 26. Brumshtein B, et al. (2007) Crystal structures of complexes of N-butyl- and N-nonyl- deoxynojirimycin bound to acid beta-glucosidase: Insights into the mechanism of chemical chaperone action in Gaucher disease. J Biol Chem 282(39):29052 –29058. 27. Sun Y, et al. (2012) Ex vivo and in vivo effects of isofagomine on acid β-glucosidase variants and substrate levels in Gaucher disease. J Biol Chem 287(6):4275 –4287. 28. Guce AI, Clark NE, Rogich JJ, Garman SC (2011) The molecular basis of pharmaco- logical chaperoning in human α-galactosidase. Chem Biol 18(12):1521 –1526. 29. Asano N, et al. (2000) In vitro inhibition and intracellular enhancement of lysosomal alpha-galactosidase A activity in Fabry lymphoblasts by 1-deoxygalactonojirimycin and its derivatives. Eur J Biochem 267(13):4179 –4186. 30. Chen YQ, Wenger DA (1993) Galactocerebrosidase from human urine: Puri ﬁcation and partial characterization. Biochim Biophys Acta 1170(1):53 –61. 31. Sakai N, et al. (1994) Puri ﬁcation and characterization of galactocerebrosidase from human lymphocytes. J Biochem 116(3):615 –620. 32. Guce AI, et al. (2010) Catalytic mechanism of human alpha-galactosidase. J Biol Chem 285(6):3625 –3632. 33. Juers DH, et al. (2001) A structural view of the action of Escherichia coli (lacZ) beta- galactosidase. Biochemistry 40(49):14781 –14794. 34. Huber RE, Wallenfels K, Kurz G (1975) The action of beta-galactosidase (Escherichia coli) on allolactose. Can J Biochem 53(9):1035 –1038. 35. Hajdu J, et al. (2000) Analyzing protein functions in four dimensions. Nat Struct Biol 7(11):1006 –1012.

36. Hajdu J (1993) Fast crystallography and time-resolved structures. Annu Rev Biophys Biomol Struct 22:467 –498. 37. Wei RR, et al. (2011) X-ray and biochemical analysis of N370S mutant human acid β -glucosidase. J Biol Chem 286(1):299 –308. 38. Sandhoff K, Kolter T (2003) Biosynthesis and degradation of mammalian glyco- sphingolipids. Philos Trans R Soc Lond B Biol Sci 358(1433):847 –861. 39. Harzer K, et al. (1997) Saposins (sap) A and C activate the degradation of gal- actosylceramide in living cells. FEBS Lett 417(3):270 –274. 40. Vocadlo DJ, Davies GJ (2008) Mechanistic insights into glycosidase chemistry. Curr Opin Chem Biol 12(5):539 –555. 41. Wentworth DF, Wolfenden R (1974) Slow binding of D-galactal, a “reversible ”in- hibitor of bacterial beta-galactosidase. Biochemistry 13(23):4715 –4720. 42. Vocadlo DJ, Davies GJ, Laine R, Withers SG (2001) Catalysis by hen egg-white lysozyme proceeds via a covalent intermediate. Nature 412(6849):835 –838. 43. Millard CB, et al. (1999) Reaction products of acetylcholinesterase and VX reveal a mobile histidine in the catalytic triad. J Am Chem Soc 121(42):9883 –9884. 44. Barabas O, et al. (2013) Catalytic mechanism of alpha-phosphate attack in dUTPase is revealed by X-ray crystallographic snapshots of distinct intermediates, 31P-NMRspectroscopy and reaction path modelling. Nucleic Acids Res , 10.1093/nar/gkt756. 45. Selleri S, et al. (2000) Deletion of exons 11-17 and novel mutations of the gal- actocerebrosidase gene in adult- and early-onset patients with Krabbe disease. J Neurol 247(11):875 –877. 46. Sulzenbacher G, Driguez H, Henrissat B, Schülein M, Davies GJ (1996) Structure of the Fusarium oxysporum endoglucanase I with a nonhydrolyzable substrate analogue: Substrate distortion gives rise to the preferred axial orientation for the leaving group. Biochemistry 35(48):15280 –15287. 47. Tews I, et al. (1996) Bacterial chitobiase structure provides insight into catalytic mechanism and the basis of Tay-Sachs disease. Nat Struct Biol 3(7):638 –648. 48. Schueler UH, et al. (2004) Correlation between enzyme activity and substrate storage in a cell culture model system for Gaucher disease. J Inherit Metab Dis 27(5):649 –658. 49. Ribbens J, et al. (2013) A high-throughput screening assay using Krabbe disease pa- tient cells. Anal Biochem 434(1):15 –25. 50. Lee WC, et al. (2010) Molecular characterization of mutations that cause globoid cell leukodystrophy and pharmacological rescue using small molecule chemical chaper- ones. J Neurosci 30(16):5489 –5497. 51. Yuzwa SA, et al. (2008) A potent mechanism-inspired O-GlcNAcase inhibitor that blocks phosphorylation of tau in vivo. Nat Chem Biol 4(8):483 –490. 52. Ficko-Blean E, Stubbs KA, Nemirovsky O, Vocadlo DJ, Boraston AB (2008) Structural and mechanistic insight into the basis of mucopolysaccharidosis IIIB. Proc Natl Acad Sci USA 105(18):6560 –6565. 53. Sánchez-Ollé G, et al. (2009) Promising results of the chaperone effect caused by imino sugars and aminocyclitol derivatives on mutant glucocerebrosidases causingGaucher disease. Blood Cells Mol Dis 42(2):159 –166. 54. Walter TS, et al. (2008) Semi-automated microseeding of nanolitre crystallization experiments. Acta Crystallogr Sect F Struct Biol Cryst Commun 64(Pt 1):14 –18.

20484 |www.pnas.org/cgi/doi/10.1073/pnas.1311990110 Hill et al.