Supporting InformationDeane et al. 10.1073/pnas.1105639108SI Materials and MethodsProtein Expression and Purification. Mouse galactocerebrosidase(mGALC) cDNA was subcloned from the I.M.A.G.E. clone#2395239, kindly donated by the Medical Research CouncilUK Human Genome Mapping Project Resource Centre, intopSP73 (Promega). It was linked to the woodchuck hepatitis virus(WHV) posttranscriptional regulatory element (WPRE) at the 3′end. The WPRE sequence was obtained from the WHV genome(American Type Culture Collection) by PCR. The GALC-WPREsequence was cloned into the pSecTag2B plasmid (Invitrogen) toinclude the Ig k-chain leader sequence for secretion, a His6 tag,and factor Xa cleavage site immediately upstream of the mGALCcoding region. The resultant plasmid, pSecTag2B-mGALC, wasused to transfect HEK 293Tcells by the calcium phosphate trans-fection method. The transfected cells were maintained in Dulbec-co’s modified medium (Gibco) under constant positive selectionby bleomycin at 500 μg∕mL (ZeocinTM, Invitrogen) for over 4 mo.

The supernatant fraction containing secreted GALC was incu-bated at 4 °C with cobalt-chelated agarose resin (Pierce) preequi-librated in wash buffer (PBS pH 7.4, 10 mM imidazole). The resinwas pelleted (700 g, 10 min, 4 °C), supernatant removed, andwashed twice with wash buffer. The washed resin was transferredto a gravity column and protein was eluted with PBS containing150 mM imidazole. Fractions containing purified GALC werepooled, concentrated, and exchanged into 10 mM HepespH 7.4, 150 mM NaCl using Millipore Amicon Ultra 10 k mole-cular weight cutoff centrifugal concentrators. ConcentratedGALC (4 mg∕mL) was immediately subjected to crystallizationtrials.

Crystallization. Crystallization experiments were performed in 96-well nanolitre-scale sitting drops (200 nL of 4 mg∕mL GALCplus 200 nL of precipitant) equilibrated at 20 °C against 80-μLreservoirs of precipitant. Diffraction quality crystals grew againsta reservoir of 0.2 M sodium acetate, 0.1 M sodium cacodylate,pH 6.8, 34% (wt∕vol) polyethylene glycol 8000. Crystals for heavyatom derivatizations were grown by microseeding into nanolitre-scale sitting drops (1). Heavy atom derivatives were prepared bymaking tenfold dilutions of reservoir solution saturated withethylmercuric thiosalicylate (EMTS) or potassium tetrachloro-platinate (K2PtCl4) and soaking crystals overnight at 20 °C. Ga-lactose was introduced into GALC crystals by soaking in 20 mMD-galactose for 45 min. Crystals were cryoprotected by a quicksweep through perfluoropolyether oil (Hampton Research)before being flash-cryocooled by plunging into liquid nitrogen.

Data Collection.Diffraction data were recorded at Diamond LightSource beamline I03 on an Area Detector Systems CorporationQ315 CCD detector. Native and galactose-soaked datasets werecollected at λ ¼ 0.98 Å. Platinum and mercury derivatives werecollected at λ ¼ 0.80 Å. Diffraction data were indexed, inte-grated, and scaled using MOSFLM (2) and SCALA (3) via thexia2 automated data processing pipeline (4).

Structure Determination and Refinement. The structure of GALCwas determined by multiple isomorphous replacement using mer-cury and platinum heavy atom derivatives. The heavy atom siteswere determined using HySS, their positions were refined andinitial phases calculated using SOLVE, and density modificationwas performed using RESOLVE all as implemented in the Auto-Sol wizard (5) of the PHENIX suite (6). An initial model auto-built by ARP/wARP (7) comprised 70% of the final structure.Manual model building was carried out in COOT (8) with refine-ment using phenix.refine (6). Determination of carbohydratestructures was informed by knowledge of correct core carbohy-drate linkages (9). For the native and galactose-bound struc-tures of GALC 96.7% and 95.8% of residues, respectively, are inthe favored region of the Ramachandran plot as determined byMolprobity (10).

Structure Analysis. The analysis of buried surface area in theGALC structure was carried out using the PISA service at theEuropean Bioinformatics Institute (EBI, http://www.ebi.ac.uk/msd-srv/prot_int/pistart.html) (11). Protein structure comparisonwas made using the Protein Data Bank (PDB) eFold service atthe EBI (http://www.ebi.ac.uk/msd-srv/ssm) (12). For the electro-static potential calculations partial charges were assigned usingPDB2PQR (13), which uses PROPKA (14) to determine proteinpKa values. Electrostatic surfaces were calculated using APBS(15) and molecular graphics were prepared using PyMOL (De-Lano Scientific). The schematic diagram illustrating bonding in-teractions between GALC and galactose was generated usingLIGPLOT (16). Sequence alignments were prepared using Aline(17).

The atomic coordinates and structure factors (PDB ID codes3zr5 and 3zr6) have been deposited in the Protein Data Bank(http://www.rcsb.org/).

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

2. Leslie AGW (1992) Recent changes to the MOSFLM package for processing film andimage plate data Joint CCP4 +ESF-EAMCB Newsletter on Protein Crystallography, 26.

3. Evans P (2006) Scaling and assessment of data quality. Acta Crystallogr D Biol Crystal-logr 62:72–82.

4. Winter G (2009) xia2: An expert system for macromolecular crystallography datareduction. J Appl Cryst 43:186–190.

5. 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.

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

7. 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.

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

9. Lutteke T (2009) Analysis and validation of carbohydrate three-dimensional structures.Acta Crystallogr D Biol Crystallogr 65:156–168.

10. Chen VB, et al. (2010) MolProbity: All-atom structure validation for macromolecularcrystallography. Acta Crystallogr D Biol Crystallogr 66:12–21.

11. Krissinel E, Henrick K (2007) Inference of macromolecular assemblies from crystallinestate. J Mol Biol 372:774–797.

12. Krissinel E, Henrick K (2004) Secondary-structure matching (SSM), a new tool for fastprotein structure alignment in three dimensions. Acta Crystallogr D Biol Crystallogr60:2256–2268.

13. Dolinsky TJ, Nielsen JE, McCammon JA, Baker NA (2004) PDB2PQR: An automatedpipeline for the setup of Poisson-Boltzmann electrostatics calculations. Nucleic AcidsRes 32:W665–667.

14. Li H, Robertson AD, Jensen JH (2005) Very fast empirical prediction and rationalizationof protein pKa values. Proteins 61:704–721.

15. Baker NA, Sept D, Joseph S, Holst MJ, McCammon JA (2001) Electrostatics of nanosys-tems: Application to microtubules and the ribosome. Proc Natl Acad Sci USA98:10037–10041.

16. Wallace AC, Laskowski RA, Thornton JM (1995) LIGPLOT: A program to generateschematic diagrams of protein-ligand interactions. Protein Eng 8:127–134.

17. Bond CS, Schuttelkopf AW (2009) ALINE: A WYSIWYG protein-sequence alignmenteditor for publication-quality alignments. Acta Crystallogr D Biol Crystallogr65:510–512.

Deane et al. www.pnas.org/cgi/doi/10.1073/pnas.1105639108 1 of 7

Fig. S1. Sequence alignment of mGALC with human GALC (hGALC). Secondary structure of GALC is shown above the sequence, β-sheets (arrows) and α-helices(cylinders) are colored as for Fig. 1. Residues not modeled due to poorly defined electron density (416–418) are in lowercase. Highlighted on the sequence arecysteines that form the disulfide bond (yellow circles), asparagines where N-linked glycosylation was observed in electron density (pink diamonds), catalyticresidues at the active site (open stars), and disease-causing mutations discussed in the text (purple triangles).

Fig. S2. Calcium binding site in the lectin domain. The seven oxygen ligands that form the pentagonal bipyramidal coordination of calcium are shown: theside chains of D660 (bidentate) and N479; the backbone carbonyls of D477 and F511; and two water molecules.

Fig. S3. Schematic representation of GALC amino acid side chain interactions with the galactose ligand. Hydrogen bonds (green dashed lines) and hydro-phobic interactions (red fans) are illustrated.

Deane et al. www.pnas.org/cgi/doi/10.1073/pnas.1105639108 2 of 7

Fig. S4. Comparison of galactose binding. (A) GALC with galactose bound in the active site. (B) Acid α-galactosidase with galactose bound in the active site(PDB ID code 3GXP) shown in the same orientation as GALC in A. (C) Overlay of these two structures showing the different binding sites for the galactosemolecules.

Fig. S5. Substrate specificity of lysosomal glycosyl hydrolases. Structures of GALC (blue), human acid β-glucosidase (PDB ID code 1OGS, wheat), and β-glucur-onidase (PDB ID code 3HN3, green) were overlaid using the catalytic glutamate residues. The relative orientations of the substrate-specificity determiningtryptophan residues are shown (sticks).

Deane et al. www.pnas.org/cgi/doi/10.1073/pnas.1105639108 3 of 7

Fig. S6. Comparison of human acid β-glucosidase (PDB ID code 1OGS) and GALC active site loops. (A, Top) Loop 1 (red, residues 311–319, implicated in con-formational changes to accommodate substrate) and catalytic side chains are shown for human acid β-glucosidase. (Bottom) Main chain temperature factorsare shown for residues 250–400 with those of loop 1 shown in red. (B, Top) The equivalent region of GALC is shown highlighting the disulfide bond (yellow),R380 (orange), and H237 (purple). (Bottom) Main chain temperature factors are shown for residues 200–450 with R380 and H237 highlighted. The temperaturefactors support the observation that loop 1 of acid β-glucosidase is more flexible than the surrounding parts of the structure, whereas the equivalent regions ofGALC do not display a similar increase in temperature factors.

Deane et al. www.pnas.org/cgi/doi/10.1073/pnas.1105639108 4 of 7

Fig. S7. Electrostatic surface of GALC and acid β-glucosidase. (A, Left) The molecular surface of GALC is shown colored by electrostatic potential at the solventaccessible surface from red (negative, −3 kT∕e) to blue (positive, þ3 kT∕e). Electrostatic potential was calculated using a pH value of 4.8 when assigning sidechain protonation. Galactose is shown in the binding site as pink sticks. (A, Right) Ribbon diagram illustrating the orientation of GALC (Left). (B) Rotation ofGALC by 180° about the horizontal axis relative to (A). (C) Electrostatic surface (calculated as inA) and ribbon diagrams of acid β-glucosidase (PDB ID code 1OGS)in the same orientation as for GALC in A.

Deane et al. www.pnas.org/cgi/doi/10.1073/pnas.1105639108 5 of 7

Table S1. Data collection and refinement statistics

Native Galactose-bound EMTS K2PtCl4

Data collectionSpace group H32 H32 H32 H32Cell dimensions

a, b, c, Å a ¼ b ¼ 250.4, c ¼ 78.1 a ¼ b ¼ 250.0, c ¼ 77.5 a ¼ b ¼ 249.4, c ¼ 77.7 a ¼ b ¼ 249.4, c ¼ 77.5Resolution, Å* 73–2.1 (2.15–2.10) 73–2.4 (2.50–2.44) 38–2.8 (2.82–2.75) 42–3.0 (3.10–3.02)Rmerge 0.12 (0.70) 0.12 (0.75) 0.18 (0.88) 0.14 (0.78)hI∕σIi 7.0 (1.7) 8.9 (1.7) 14.2 (3.0) 20.7 (4.7)Completeness, % 99.6 (99.8) 100.0 (100.0) 99.9 (99.8) 99.9 (99.8)Redundancy 3.7 (3.7) 5.6 (4.9) 11.2 (10.6) 18.3 (18.7)RefinementResolution, Å 63–2.1 (2.12–2.10) 63–2.4 (2.47–2.44)No. reflections 99,641 (2,983) 66,636 (2,603)Rwork∕Rfree 18.1∕21.4 17.8∕22.3Ramachandran favored region, % 96.7 95.8Ramachandran outliers, % 0.2 0.3No. atoms

Protein 5,163 5,136Carbohydrate 98 95Water 258 139Galactose 0 12Ca2þ ions 1 1

B factorsProtein 32.3 50.4Carbohydrate 71.0 94.9Water 34.0 40.6Galactose — 52.7Ca2þ ions 31.3 98.5

rms deviationsBond lengths, Å 0.010 0.010Bond angles, ° 1.245 1.330

*Values in parentheses are for highest-resolution shell.

Table S2. Missense mutations of GALC that cause Krabbe disease

Mutation*,†

Alternatenumbering‡ ASA, %§ B/E¶ Structural features∥ Ref.

TIM barrelG41S G57S 0.0 B introduction of polar s/c (1)G43R G59R 0.0 B introduction of large, basic s/c into barrel (2)A44T A60T 0.0 B hydrophobic s/c to polar s/c at barrel (1)S52F S68F 0.4 B polar s/c to large hydrophobic s/c (2)R63H R79H 20.8 basic s/c to polar s/c, loss of stabilizing hydrogen bond (3)I66M I82M 0.0 B replacement with larger hydrophobic s/c, steric hindrance (4)G95S G111S 0.0 B introduction of polar s/c (3)T96A T112A 0.4 B polar s/c to hydrophobic s/c, loss of stabilizing hydrogen bond (5)M101L M117L 6.4 B replacement with smaller hydrophobic s/c (3)E114K E130K 9.6 B severe misfolding: buried acidic s/c to large, basic s/c (1)D171V D187V 83.9 E acidic s/c to small hydrophobic s/c, change local surface charge (5)G178A G194A 0.2 B introduction of hydrophobic s/c near SBS (6)E215K E231K 42.9 E acidic s/c to large, basic s/c, change local surface charge (3)L223P L239P 86.4 E backbone torsion angles not compatible with proline geometry (1)I234T I250T 0.0 B hydrophobic s/c to polar s/c (3)A247T A263T 0.0 B hydrophobic s/c to polar s/c (6)S257F S273F 0.0 B severe misfolding: polar s/c near SBS to large hydrophobic s/c (7)T262I T278I 7.3 B polar s/c to hydrophobic s/c, near SBS (2)G268S G284S 0.0 B introduction of polar s/c, near loop from β-sandwich (3)G270D G286D 0.0 B introduction of acidic s/c, near loop from β-sandwich (4, 8)N279T N295T 0.9 B replacement with smaller polar s/c, loss of stabilizing hydrogen bonds (6)S287F S303F 0.0 B polar s/c to large hydrophobic s/c in barrel (6)I289V I305V 0.0 B replacement with smaller hydrophobic s/c near SBS, polymorphism associated

with I66M; these residues are not close structurally(4)

Y298C Y314C 4.2 B large to small polar s/c, loss of stabilizing hydrogen bonds (3)P302A/R P318A/R 7.6 B hydrophobic s/c near SBS to large basic s/c (9, 10)Y319C Y335C 1.0 B large to small polar s/c, loss of stabilizing hydrogen bond (2)β-sandwichL364R L380R 0.0 B severe misfolding: hydrophobic s/c to large basic s/c (7)R380W/L R396W/L 52.4 E directly binds substrate, mutation will severely affect substrate binding (6, 11)P384L P400L 97.8 E introduces branched chain hydrophobic s/c on the loop from β-sandwich (6)

Deane et al. www.pnas.org/cgi/doi/10.1073/pnas.1105639108 6 of 7

Mutation*,†

Alternatenumbering‡ ASA, %§ B/E¶ Structural features∥ Ref.

W410G W426G 4.9 B severe misfolding: loss of large, hydrophobic s/c (2)Lectin domainF498S F514S 0.0 B large hydrophobic s/c to small polar s/c (6)T513M T529M 0.0 B polar s/c to hydrophobic s/c, loss of stabilizing hydrogen bond and steric

hindrance(6)

R515C/H R531C/H 18.3 basic s/c to polar s/c, loss of stabilizing hydrogen bonds (2, 6)D528N D544N 0.1 B introduction of new glycosylation site and loss of stabilizing hydrogen bonds (12, 13)G537R G553R 2.2 B severe misfolding: introduction of large, basic s/c (8, 10)V550G V566G 0.0 B loss of hydrophobic s/c (9)Y551S Y567S 22.0 loss of stabilizing pi-stacking interactions (6)A563E A579E 0.0 B small hydrophobic s/c to large acidic s/c (11)I583S I599S 0.0 B hydrophobic s/c to polar s/c (12)T617M T633M 3.2 B polar s/c to hydrophobic s/c (11)L618S L634S 0.0 B hydrophobic s/c to polar s/c (4)L629R L645R 12.7 severe misfolding: hydrophobic s/c to large basic s/c (6, 14)T652P T668P 0.0 B backbone torsion angles not compatible with proline geometry (2)

*Missense mutations resulting in early termination codons are not included in this table. Missense mutations that are known to occur on the same allele as aframeshift mutation are not listed as in these cases it is unlikely that the missense mutation will be the disease-determining feature.

†Mutations in the human sequence are listed. Because of the insertion of a residue in the human sequence at residue 507, the equivalent residue in the mousestructure beyond this point is n − 1.

‡The first start site in exon 1 is used for numbering in Uniprot entry P54803 and publications in the journal Human Mutation. For clarity, these designations arealso provided.

§ASA: accessible surface area: percentage of the residue surface area that is accessible to solvent as calculated by GetArea (15).¶B: buried, residues with <10% ASA are considered buried; E: exposed, residues with >40% ASA are considered exposed.∥Abbreviations: s/c, side chain; SBS, substrate binding site.

1. 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 28:742.2. Fu L, et al. (1999) Molecular heterogeneity of Krabbe disease. J Inherit Metab Dis 22:155–162.3. De Gasperi R, et al. (1996) Molecular heterogeneity of late-onset forms of globoid-cell leukodystrophy. Am J Hum Genet 59:1233–1242.4. Furuya H, et al. (1997) Adult onset globoid cell leukodystrophy (Krabbe disease): Analysis of galactosylceramidase cDNA from four Japanese patients. Hum Genet 100:450–456.5. Luzi P, Rafi MA, Wenger DA (1996) Multiple mutations in the GALC gene in a patient with adult-onset Krabbe disease. Ann Neurol 40:116–119.6.Wenger DA, Rafi MA, Luzi P (1997) Molecular genetics of Krabbe disease (globoid cell leukodystrophy): Diagnostic and clinical implications. Hum Mutat 10:268–279.7. Xu C, Sakai N, TaniikeM, Inui K, Ozono K (2006) Six novel mutations detected in the GALC gene in 17 Japanese patients with Krabbe disease, and new genotype-phenotype correlation.J Hum Genet 51:548–554.

8. De Gasperi R, et al. (1999) Molecular basis of late-life globoid cell leukodystrophy. Hum Mutat 14:256–262.9. Tatsumi N, et al. (1995) Molecular defects in Krabbe disease. Hum Mol Genet 4:1865–1868.10. Tappino B, et al. (2010) Identification and characterization of 15 novel GALC gene mutations causing Krabbe disease. Hum Mutat 31:E1894–1914.11. Selleri S, et al. (2000) Deletion of exons 11–17 and novel mutations of the galactocerebrosidase gene in adult- and early-onset patients with Krabbe disease. J Neurol 247:875–877.12. Rafi MA, Luzi P, Zlotogora J, Wenger DA (1996) Two different mutations are responsible for Krabbe disease in the Druze and Moslem Arab populations in Israel. Hum Genet

97:304–308.13. Lee WC, et al. (2010) Molecular characterization of mutations that cause globoid cell leukodystrophy and pharmacological rescue using small molecule chemical

chaperones. J Neurosci 30:5489–5497.14. Jardim LB, et al. (1999) Protracted course of Krabbe disease in an adult patient bearing a novel mutation. Arch Neurol 56:1014–1017.15. Fraczkiewicz R, Braun W (1998) Exact and efficient analytical calculation of the accessible surface areas and their gradients for macromolecules. J Comp Chem 19:319–333.

Deane et al. www.pnas.org/cgi/doi/10.1073/pnas.1105639108 7 of 7