Glycosidase active site mutations in human α-L-iduronidase

Glycobiology vol. 11 no. 9 pp. 741–750, 2001

© 2001 Oxford University Press 741

Glycosidase active site mutations in human α-L-iduronidase

Doug A. Brooks1,3, Sylvie Fabrega2, Leanne K. Hein3, Emma J. Parkinson3, Patrick Durand2,4, Gouri Yogalingam3, Ursula Matte6, Roberto Giugliani6, Ayan Dasvarma3, Jobin Eslahpazire4, Bernard Henrissat7, Jean-Paul Mornon5, John J. Hopwood3, and Pierre Lehn4

3Lysosomal Diseases Research Unit, Department of Chemical Pathology, Women’s and Children’s Hospital, King William Road, North Adelaide, SA 5006, Australia; 4INSERM U458, Hopital Robert Debré, 48 Bd Serurier, 75019 Paris, France; 5Systèmes Moléculaires et Biologie Structurale, Laboratoire de Minéralogie-Cristallographie, CNRS UMR 7590, Universités Paris VI-Paris VII, T16, case 115, 4 place Jussieu, 75252 Paris Cedex 5, France; 6Federal University of Rio Grande do Sul, Porto Alegre, Brazil; and 7Architecture et Fonction des Macromolécules Biologiques, CNRS UMR 6098, Universités d’Aix-Marseille I and II, 31 chemin Joseph Aiguier, 13402 Marseille Cedex 20, France.

Received on March 1, 2001; revised on May 30, 2001; accepted on June 1, 2001.

Mucopolysaccharidosis type I (MPS I; McKusick 25280)results from a deficiency in α-L-iduronidase activity. Usinga bioinformatics approach, we have previously predicted theputative acid/base catalyst and nucleophile residues in theactive site of this human lysosomal glycosidase to be Glu182and Glu299, respectively. To obtain experimental evidencesupporting these predictions, wild-type α-L-iduronidase andsite-directed mutants E182A and E299A were individuallyexpressed in Chinese hamster ovary–K1 cell lines. We havecompared the synthesis, processing, and catalytic propertiesof the two mutant proteins with wild-type human α-L-iduro-nidase. Both E182A and E299A transfected cells producedcatalytically inactive human α-L-iduronidase protein at levelscomparable to the wild-type control. The E182A protein wassynthesized, processed, targeted to the lysosome, and secretedin a similar fashion to wild-type α-L-iduronidase. The E299Amutant protein was also synthesized and secreted similarlyto the wild-type enzyme, but there were alterations in its rateof traffic and proteolytic processing. These data indicate thatthe enzymatic inactivity of the E182A and E299A mutantsis not due to problems of synthesis/folding, but to theremoval of key catalytic residues. In addition, we haveidentified a MPS I patient with an E182K mutant allele.The E182K mutant protein was expressed in CHO-K1 cellsand also found to be enzymatically inactive. Together, theseresults support the predicted role of E182 and E299 in thecatalytic mechanism of α-L-iduronidase and we proposethat the mutation of either of these residues wouldcontribute to a very severe clinical phenotype in a MPS Ipatient.

Key words: active site residues/catalytic machinery/α-L-iduronidase/Hurler syndrome/mucopolysaccharidosis I

Introduction

α-L-iduronidase (EC. 3.2.1.76) is a lysosomal glycosidehydrolase, which cleaves α-linked iduronic acid residues fromthe nonreducing end of the glycosaminoglycans (GAGs),heparan sulfate, and dermatan sulfate. This glycosidase is oneof a series of 10 lysosomal enzymes involved in the sequentialdegradation of these GAGs. α-L-iduronidase is synthesised inthe endoplasmic reticulum (ER) as a 653-amino-acid poly-peptide (following signal peptide cleavage) and is glycosylatedwith six N-linked oligosaccharides to produce a 74-kDaprecursor molecule (Figure 1). The N-linked oligosaccharideson α-L-iduronidase are modified to produce mainly “complextype” oligosaccharides and at least two of these N-linkedoligosaccharides have been shown to be mannose-6-phos-phorylated (Zhao et al., 1997). α-L-iduronidase has beenfound to undergo extensive proteolytic processing to produceat least 10 polypeptides (Mr 74, 69, 65, 60, 49, 44, 25, 16, 9,and 5 kDa; Figure 1; Taylor et al., 1991; Scott et al., 1991;Brooks, 1993). This extensive proteolysis is thought to occurintracellularly, as a result of normal residence in the endosome-lysosome compartments.

The lysosomal storage disorder mucopolysaccharidosistype I (MPS I; McKusick 25280) is an autosomal recessivelyinherited genetic disease, caused by a deficiency in α-L-iduroni-dase (Neufeld and Muenzer, 1995). Failure to remove α-linkediduronic acid residues from the nonreducing end of GAGsresults in the accumulation of these substrates in lysosomalorganelles. This storage causes the progressive deterioration ofcells, tissues, organs, and urinary secretion of the storageproduct.

MPS I patients present with a wide spectrum of clinicalphenotypes, ranging from very severe for Hurler syndrome tonearly normal in Scheie syndrome (Neufeld and Muenzer,1995). About 57 IDUA gene mutations have been reported(Scott et al., 1995; Bunge et al., 1995; Yamagishi et al., 1996;Krawczak and Cooper, 1997). The most common mutantalleles in Caucasians, involve truncation of the α-L-iduroni-dase protein and include the W402X and Q70X “null alleles,”which produce no detectable α-L-iduronidase protein (Scott etal., 1992a,b; Ashton et al., 1992). Together, these nonsensemutations account for approximately 60% of disease alleles inCaucasians and in the homozygous condition are associatedwith the most severe form of MPS I patient clinical phenotype.R89Q (Scott et al., 1993) is one of the most common mutantalleles in Japanese MPS I patients (Yamagishi et al., 1996) andis associated with a clinical phenotype of intermediate severity

1To whom correspondence should be addressed2Present address: Hybrigenics S.A., 180 avenue Daumesnil, 75012 Paris, France

D.A. Brooks et al.

742

(Scott et al., 1993, 1995). To date, no active site mutationshave been reported for α-L-iduronidase, although one muta-tion (D349N) has been shown to result in a high level of cata-lytically inactive protein (Brooks et al., 1992).

Glycoside hydrolases (EC. 3.2.1–3.2.3) are a large group ofenzymes that function by two distinct catalytic mechanisms,involving either overall retention or inversion of configurationat the anomeric carbon of the substrate (Koshland, 1953;Sinnott, 1990; McCarter and Withers, 1994; Davies andHenrissat, 1995). In both mechanisms, catalysis requires a pairof carboxylic acid groups, typically provided by either asparticacid or glutamic acid residues. In the “retaining” enzymes likeα-L-iduronidase, the critical residues are organized on eitherside of the glycosidic bond and are separated by a distance of∼5.5Å (McCarter and Withers, 1994). These two critical aminoacid residues are involved in a two-step catalysis. In the firststep, one of the residues performs a nucleophilic attack at thesugar anomeric carbon, while the other residue functions as anacid/base and assists aglycon departure by protonation of theglycosidic oxygen. The result of this first step is the formationof a covalent glycosyl-enzyme intermediate. In the secondstep, the deprotonated acid/base residue abstracts a protonfrom a water molecule, which attacks the glycosyl-enzyme torelease a sugar with a stereochemistry identical to that of the

substrate (Koshland, 1953; Sinnott, 1990; McCarter andWithers, 1994; Davies and Henrissat, 1995).

A number of strategies, including sequence alignment, 3Dstructure analysis, labeling techniques and combined bioinfor-matics approaches have been used to identify active site residuesin glycosidases (Withers and Aebersold, 1995; Henrissat et al.,1995; Durand et al., 1997, 2000; Callebaut et al., 1997). TheCellulomonas fimi β-1,4-glycanase Cex was the first retainingglycoside hydrolase for which the 3D structure of a catalyticallycompetent glycosyl-enzyme intermediate was determined(White et al., 1996). Based on amino acid similarities, glyco-side hydrolases have been classified into a number of families(these families can be found on a continuously updated serverat http://afmb.cnrs-mrs.fr/∼pedro/CAZY/db.html). A higherhierarchical level of classification was introduced, the “clans,”which groups families displaying the same fold and catalyticmachinery (Henrissat and Bairoch, 1996; Henrissat andDavies, 1997). The largest of these clans, clan GH-A, containsmany of the glycosidases responsible for lysosomal storagedisorders, including β-glucuronidase (Sly syndrome), β-gluco-cerebrosidase (Gaucher disease), β-galactosidase (GM-1gangliosidosis, Morquio type B syndrome), β-mannosidase(mannosidosis), and α-L-iduronidase (Hurler and Scheiesyndromes) (Durand et al., 2000). All clan GH-A membersshare the same retaining mechanism and a common 3D struc-ture, a (β/α)8 barrel, in which the acid/base and the nucleophilicresidues are located at the C-terminal end of strands β4 and β7,respectively. Sequence comparisons and computer modeling havepreviously allowed us to predict that E182 and E299 are thelikely acid/base and nucleophilic residues of α-L-iduronidase(Henrissat et al., 1995; Durand et al., 1997, 2000). In addition,Arg 89 was shown to be located at the C-terminal end of strandβ2 and was hypothesized to play a role in activation of thenucleophile, a prediction in agreement with the involvement ofa R89Q mutation in MPS I disease (Durand et al., 1997).

The aim of the present work was to generate experimentaldata supporting that residues E182 and E299 play a critical rolein the catalytic machinery of human α-L-iduronidase. Thus,we have constructed the E182A and E299A mutants (as a Gluto Ala mutation is isosteric) as well as an E182K mutant (as apatient with a severe MPS I clinical phenotype has beenidentified with this allele). In these mutants, the putativecatalytic residues have been replaced by amino acid residueswhose side chains are unable to participate in the enzymaticreaction. We herein report the results of expression studies ofthese mutant human α-L-iduronidase proteins in Chinesehamster ovary (CHO)–K1 cells, which confirm our bioinfor-matics predictions.

Results and discussion

Epitope mapping of the monoclonal antibody ID1A

A panel of monoclonal antibodies against human α-L-iduroni-dase has previously been generated (Ashton et al., 1992). Themonoclonal antibody ID1A was selected from this panel andused in the current study because of its ability to reactspecifically with human α-L-iduronidase and its capacity todetect both native and denatured protein. Indeed, ID1A hasbeen shown to detect multiple molecular forms of human α-L-iduronidase on immunoblots (denatured protein) and was

Fig. 1. Schematic of α-L-iduronidase protein. The open boxes on the α-L-iduronidase sequence represent the proposed β-sheet elements reported by Durand et al. (1997, 2000) except for the β8-sheet. This β8-sheet has yet to be properly defined in α-L-iduronidase and was only represented here based on the previously reported mutation D349N (Brooks et al., 1992; residue position shown with a star), which caused ablation of α-L-iduronidase catalytic activity and may therefore represent a possible substrate binding region. The proposed nucleophilic residue at E299 (N–) and the acid-base catalyst at E182 (P+) are also shown. The carbohydrate structures shown were as determined by Zhao et al. (1997)), where C represented complex carbohydrate chains, M represented high mannose, and P represented phosphorylated high mannose (N.B. Some of these residues were only partially characterized). The proteolytic processing sites, which are known to give rise to multiple α-L-iduronidase processed forms (Scott et al., 1991), are shown at points with scissors and the residue number. The known polypeptide processed forms are depicted at the base of the figure with arrows showing the proteolytic processing sites. These processing sites are either for signal peptide cleavage, which occurs in the ER, or for other proteolytic events, which appear to be endosome/lysosome (E/L) events. The boxed N and C represent the appropriate termini of the protein.

Glycosidase active site mutations

743

therefore presumed to react with a linear sequence epitope(Brooks, 1993). This was a necessary prerequisite in thecurrent study due to the need to detect expression of mutatedinactive human α-L-iduronidase in CHO-K1 expression cellsand to distinguish it from endogenous CHO-K1 α-L-iduroni-dase. Thus epitope mapping was first performed to ensure thatthe ID1A epitope was distant from the E182 and E299 residuesand that, consequently, mutation of these residues wouldprobably not interfere with the binding capacity of the mono-clonal antibody. ID1A was found to react with a linearsequence peptide, between residues 427 and 438 in the α-L-iduronidase linear sequence (Figure 2), which is flanked bytwo N-linked glycosylation sites at residues 415 and 451(Figure 1). The reactivity of the ID1A antibody was notdependent on glycosylation as shown by the detection of thelinear sequence peptide with high affinity. Thus the ID1Aepitope was spatially distinct from the E182 and E299 residues. Inaddition, as ID1A reacted strongly with active α-L-iduronidase,

this epitope was presumably exposed on the surface of thenative α-L-iduronidase molecule. Of note, as the N-linkedoligosaccharide at residue 451 is mannose-6-phosphorylatedand thereby implicated in lysosomal targeting, it is probablyalso exposed on the surface of the α-L-iduronidase molecule.

Immunoquantification and enzyme activity of wild-type, E182A, and E299A human α-L-iduronidase expressed in CHO-K1 cells

The mutations E182A and E299A were expressed in CHO-K1cells to investigate the consequences of replacing the presumedcritical active site residues E182 and E299 with structurallyconservative residues that prevent enzyme catalysis. Thecatalytic mechanism of glycosidases utilizes two separatecarboxylic acid residues (typically glutamic or aspartic acid) aseither a nucleophile or the acid/base catalyst. In other glyco-sidases, the replacement of these residues with alanine residueshas been shown to ablate glycoside enzyme activity, due toremoval of the critical carboxylic acid side chains (e.g., Fabregaet al., 2000). Here, plasmids carrying wild-type and mutanthuman α-L-iduronidase cDNAs were transferred into CHO-K1cells and stably transfected CHO-K1 cell clones developed.

To characterize the expression of wild-type and mutant α-L-iduronidase protein from CHO-K1 cells, cell extracts andmedia were immunoquantified using a polyclonal capture stepand an ID1A monoclonal antibody detection step (Table I). Inthis assay, no detectable α-L-iduronidase reactivity was observedfor untransfected CHO-K1 cell extracts, demonstrating that ID1Ahad no immunoreactivity with endogenous CHO-K1 α-L-iduroni-dase. In contrast, the CHO-K1 cell line transfected with thewild-type construct had human α-L-iduronidase protein bothin the cell extract and secreted into the culture medium. TheE182A and E299A mutants had similar levels of human α-L-iduronidase protein detected in CHO-K1 cell extracts, whencompared to the wild-type control. Both E182A and E299Acells also showed human α-L-iduronidase protein in the cellculture media. There was an increase in the level of α-L-iduro-nidase protein secreted by the E182A cells (and also E182Kcells), compared to the wild-type cells (Table I), indicating apossible minor alteration in the efficacy of mannose-6-phos-phorylation. Clearly, the ability of the CHO-K1 cell lines trans-fected with the mutant constructs to produce high amounts ofhuman α-L-iduronidase protein and their capacity to alsosecrete α-L-iduronidase protein strongly suggested that the

Fig. 2. Epitope mapping of the monoclonal antibody ID1A. The Y-axis of the figure shows the optical density level of ELISA reactivity for the individual α-L-iduronidase peptide pins. The only peptide pin with significant ELISA reactivity above 2 standard deviations of background (ascertained for second antibody-only control), was peptide pin number 72. This epitope corresponded to the sequence PQGPADAWRAAV, which was located between residues 427 and 438 in the α-L-iduronidase protein. The ID1A epitope is located between two N-linked glycosylation sites at residues 415 and 451 (see Zhao et al., 1997 and Figure 1 for the glycosylation structures present at these sites). This region is probably not part of the postulated (β/α)8-barrel structure for α-L-iduronidase.

Table I. α-L-iduronidase protein and activity in CHO-K1 cells expressing human wild-type, E182A, E299A, and E182K IDUA

ND, not detectable. For cell extracts, all α-L-iduronidase protein and activity results were expressed relative to the total cell protein in the respective CHO-K1 cell extracts. For culture media, results were expressed as the amount of either α-L-iduronidase protein or activity in the culture medium relative to the amount of total cell protein in the CHO-K1 cell extract.

CHO-K1 cell line

IDUA protein cell extract (µg/mg)

IDUA activity cell extract direct assay (nmol/min/mg)

IDUA activity cell extract immunocapture assay (nmol/min/mg)

IDUA protein culture medium (µg/mg)

IDUA activity culture medium direct assay(nmol/min/mg)

Untransfected CHO-K1 ND 0.6 ± 0.0 ND ND ND

Wild-type IDUA 4.1 13.2 ± 0.6 13.4 ± 0.7 0.4 3.1 ± 0.0

E182A 3.3 0.4 ± 0.0 ND 1.8 ND

E182K 4.0 0.30 ND 1.8 ND

E299A 5.9 0.2 ± 0.0 ND 0.1 ND

D.A. Brooks et al.

744

mutant proteins could correctly fold and pass the ER qualitycontrol system. Of note, despite having the highest level ofintracellular α-L-iduronidase protein, the E299A cell line hadthe lowest level of α-L-iduronidase protein in the cell culturemedium.

The wild-type human α-L-iduronidase produced in CHO-K1cells had enzyme activity when either directly assayed usingthe fluorogenic substrate 4-methylumbelliferyl α-L-iduronide(MUI) or when assayed in an immunobinding assay, afterimmune capture by the monoclonal antibody ID1A, to separateaway endogenous CHO-K1 α-L-iduronidase (Table I). Thoughthe E182A and E299A CHO-K1 cell extracts had high levels ofα-L-iduronidase protein, their enzyme activities were onlyconsistent with endogenous CHO-K1 α-L-iduronidaseenzyme. This residual α-L-iduronidase activity was clearly notdue to the mutant proteins, as immune capture (with the humanspecific monoclonal antibody ID1A) prior to enzyme assayresulted in no detectable α-L-iduronidase activity. α-L-iduro-nidase activity was also detected for the secreted wild-typeprotein, but no activity was found in the culture media fromE182A and E299A CHO-K1 cells despite the presence of α-L-iduronidase protein (Table I). The catalytic inactivity of themutant proteins suggested that the glutamic acid residues E182and E299 play a key role in human α-L-iduronidase activityand may indeed be the α-L-iduronidase acid/base and nucleo-phile catalytic residues, as predicted by our bioinformaticsanalysis (Henrissat et al., 1995; Durand et al., 1997, 2000).

Characterization of an MPS I patient with an E182K mutation

An MPS I patient with a severe clinical phenotype was identifiedas having the genotype W402X/E182K and was selected forfurther analysis. W402X has previously been identified as anull allele, because no α-L-iduronidase protein was detected inMPS I patients with a W402X/W402X genotype (Scott et al.,1992a; Ashton et al., 1992). Although a W402X mutantprotein may possibly contain the (β/α)8-barrel catalytic domain(although strand β8 is yet to be identified), the truncatedprotein was presumably unstable and failed to fold correctly,being retained and degraded by the ER quality control system.Here, expression studies with the E182K mutant were under-taken to better understanding how this mutation contributes toa severe clinical phenotype. From a mechanistic point of view,the E182K protein should be catalytically inactive, because alysine residue cannot perform the acid/base catalysis requiredfor glycosidase activity. This was confirmed by the productionof a catalytically inactive E182K α-L-iduronidase protein inCHO-K1 cells (Table I), supporting the hypothesis that theE182K mutation would contribute to a severe clinical outcome.Interestingly, the mutation of the glutamic acid to a lysine atposition 182 would introduce a net positive charge into theactive site of human α-L-iduronidase, but this was not sufficientto reduce the intracellular level of the mutant protein. Mostother α-L-iduronidase mutations appear to cause problemswith protein folding, evoking ER recognition/degradation ofthe mutant protein and resulting in low levels of intracellularα-L-iduronidase protein (reviewed in Brooks, 1997).

Identification of the molecular forms of wild-type and mutant human α-L-iduronidase produced in CHO-K1 cells

Cell extracts from the wild-type, E182A, E182K, and E299Ahuman α-L-iduronidase-expressing CHO-K1 cells were

immunoblotted to define the α-L-iduronidase molecularspecies produced at steady state (Figure 3). Cell extract fromCHO-K1 cells expressing wild-type enzyme demonstratedmolecular species of 74 (minor), 69 (major), and 65 (major)kDa by immunoblot analysis (plus molecular forms of 49 and44 kDa, which were barely visible), while in the culturemedium an 81-kDa molecular species was detected. Theseprocessed forms were consistent with the extensive processingof α-L-iduronidase, which has been previously reported inhuman skin fibroblasts (Taylor et al., 1991; Scott et al., 1991).Despite not being proteolytically processed, the 81 kDa wildtype α-L-iduronidase detected in the culture medium wasactive (see Table I), indicating that proteolytic processing wasnot a necessary prerequisite for α-L-iduronidase enzymeactivity. Importantly, a similar pattern of processed forms wasobserved for E182A, E182K, and wild-type α-L-iduronidase.In contrast, the E299A immunoblot pattern was characterizedby a major 74-kDa species and only a minor 69 kDa molecularform, a finding that suggests delayed traffic or altered proteolyticprocessing. Delayed processing/traffic of E299A α-L-iduronidasewas also supported by the immunoquantification data above,which showed a high level of intracellular compared tosecreted E299A α-L-iduronidase (at steady state). Whereas thetraffic and processing of the E299A mutant appeared to bedelayed or altered, its culture medium contained the same 81-kDahuman α-L-iduronidase species detected in both of the E182mutants and wild-type culture media. This indicated that theE299A mutant protein eventually reached the culture mediumfollowing processing in the Golgi, but was still catalyticallyinactive (Table I).

Synthesis and processing of wild-type and mutant human α-L-iduronidase

As a different steady state-pattern of α-L-iduronidase molecularspecies was observed for the E299A mutant, when compared tothe pattern of the wild-type enzyme and the two E182 mutants, weinvestigated the synthesis and processing of α-L-iduronidase inCHO-K1 expression cells. Wild-type human α-L-iduronidasewas synthesised as a 74-kDa precursor, which after 24 h wasprocessed intracellularly to 69-kDa and 65-kDa molecular

Fig. 3. Characterization of the molecular species of human wild-type and mutant α-L-iduronidase, either residing in CHO-K1 cells or secreted into the culture media. Cell extracts and media samples from either CHO-K1 cells expressing wild-type (lanes 1 and 6), E182A (lanes 2 and 7), E299A (lanes 3 and 8), and E182K (lanes 4 and 9) or control, untransfected CHO-K1 cells (lanes 5 and 10) were analyzed by western blotting. Lanes 1–5 show the pattern obtained for cell extracts, and lanes 6–10 show the pattern obtained for culture media samples. The α-L-iduronidase protein was visualized using a polyclonal antibody to α-L-iduronidase and a peroxidase-labeled second antibody detection system. The molecular mass of α-L-iduronidase molecular species are indicated on the figure (arrows).

Glycosidase active site mutations

745

species (Figure 4a, lanes 4–6). A large proportion of radio-labeled human α-L-iduronidase was secreted into the culturemedium of CHO-K1 cells expressing the wild-type enzymeand had a molecular mass of 81 kDa (Figure 4b lanes 3 and 4).It may be hypothesized that this secreted form presumablyunderwent additional glycosylation (in the Golgi), resulting inits molecular mass being greater than that of the 74-kDaprecursor species. The E182A mutant protein was synthesizedand processed with an identical pattern to that observed for thewild-type α-L-iduronidase control (Figure 4). This wasconsistent with the immunoblot data discussed above (andshown in Figure 3), which demonstrated similar molecularspecies, for both wild-type and E182A α-L-iduronidaseproteins, at steady state. The E299A mutant α-L-iduronidasewas also synthesized as a 74-kDa precursor. However, thisprotein was not intracellularly processed to the 69/65 kDaforms (Figure 4a) and there was a lag phase in secretion(Figure 4b) when compared to either the wild-type or theE182A proteins. This again suggested delayed processing/traf-ficking of the E299A α-L-iduronidase, raising the question ofwhether this protein was capable of reaching the lysosomalcompartment.

Subcellular localisation of wild-type and mutant human α-L-iduronidase in CHO-K1 expression cell lines

Percoll gradient granular fractionation of the different CHO-K1expression cells was performed to determine if the E182A,E182K, and E299A mutant α-L-iduronidase proteins weretrafficked to the lysosomal compartment. Each fraction of thegradient was assayed for either mutant or wild-type protein by

immunoblot analysis, using a specific α-L-iduronidase poly-clonal antibody. The wild-type, E182A, and E182K proteinswere mainly located in higher density fractions containingβ-hexosaminidase and characteristic of lysosomes (Figure 5).The 74 kDa α-L-iduronidase precursor was evident in lower-density fractions, whereas the 69-kDa and 65-kDa specieswere mainly in the higher-density fractions of the gradient(Figure 5). This was consistent with traffic of the E182 mutantsand wild-type α-L-iduronidase to lysosomes. It was also in

Fig. 4. Immunoprecipitation of radiolabeled wild-type and mutant human α-L-iduronidase expressed in CHO-K1 cells. CHO-K1 cells were pulse radiolabeled, then either harvested (0 h) or chased for either 4 h or 24 h (times shown under lane markers). (a) Cell extracts, which were from either untransfected (lanes 1–3), wild-type (lanes 4–6), E182A (lanes 7–9), or E299A (lanes 10–13) CHO-K1 cells. (b) Media samples from either untransfected (lanes 1, 2), wild-type (lanes 3, 4), E182A (lanes 5, 6), or E299A (lanes 7, 8) CHO-K1 cells.

Fig. 5. Subcellular localization of molecular forms of wild-type and mutant α-L-iduronidase in CHO-K1 cells. Postnuclear supernatants were prepared from CHO-K1 cells expressing wild-type and mutant human α-L-iduronidase and then the whole granular fraction, subfractionated on 18% Percoll gradients. Ten 2-ml fractions were collected from the top to the bottom of each gradient, freeze/thawed to release the α-L-iduronidase protein, then 20 µl of each fraction electrophoresed and immunoblotted. The α-L-iduronidase protein was visualized using a polyclonal antibody to α-L-iduronidase and a peroxidase-labeled second antibody detection system. The western blots show the pattern obtained from wild-type (a), E182A (b), E299A (c), and E182K (d) CHO-K1 expression cells. β-hexosaminidase activities were determined on each fraction of the gradients and showed that > 90% of this enzyme activity was recovered in fractions 7–10 of each gradient, indicating appropriate fractionation and minimal organelle breakage.

D.A. Brooks et al.

746

agreement with the processing experiments described abovewhere the 69-kDa and 65-kDa species were only generatedafter long chase times (up to 24 h) as shown in Figure 4. Theappearance of some of these processed forms in the upperfractions of the gradients may be due to impure fractionation.However, disappearance of the 74-kDa precursor speciestoward the higher-density fractions of the gradient suggestedthat its processing to the 69-kDa and 65-kDa molecular formswas an endosome-lysosome event. This was also consistentwith the processing experiments where the processed formswere only observed after 4-h chase times, as trafficking of theprecursor does take some time. Unlike the E182A, E182K, andwild-type α-L-iduronidase, the E299A mutant was detected asa single unprocessed 74-kDa precursor molecule. This findingfurther supports that the proteolytic processing of the E299Aprotein was either inhibited or delayed, as already substantiated bythe synthesis/processing experiments above (Figure 4) and bythe western blot data (Figure 3). Importantly, intracellulartraffic of the E299A mutant was not totally impaired, as thismutant protein was concentrated in the higher density fractionsof the gradient which are characteristic of lysosomes (alsoimplying that the endosome-lysosome proteolytic processingof this protein was impaired).

Conclusions and perspectives

Expression of wild-type human α-L-iduronidase in CHO-K1cells resulted in a normal pattern of protein synthesis (74-kDaprecursor) and intracellular processing (69/65-kDa molecularspecies; major forms), comparable to that previously reportedfor α-L-iduronidase in human skin fibroblasts. Similarly, wild-type α-L-iduronidase was secreted into the CHO-K1 culturemedium, with the same molecular mass (81-kDa species) asthat reported for human skin fibroblasts (Taylor et al., 1991).Expression of E182A α-L-iduronidase resulted in normalsynthesis, processing, traffic to the lysosome, and secretion ofthe mutant protein, when compared with wild-type α-L-iduro-nidase. However, this mutant protein was catalytically inactive.Thus these data strongly support our previous bioinformaticsprediction that E182 may be the acid/base catalyst in α-L-iduronidase (Henrissat et al., 1995; Durand et al., 1997, 2000).In addition, an MPS I patient with the genotype W402X/E182Kwas identified. Expression of E182K α-L-iduronidase alsoresulted in similar levels (and molecular forms) of intracellularand secreted mutant protein, when compared to wild-type α-L-iduronidase. This provides additional support for the criticalrole of the E182 residue in the α-L-iduronidase catalytic mech-anism. The complete abrogation of enzyme activity caused bythe E182K mutation, together with the fact that the W402Xallele is a “null allele,” was consistent with the severe clinicalphenotype observed in this MPS I patient.

The E299A mutation, which involves the residue proposedto act as the nucleophile in the catalytic mechanism of α-L-iduronidase, was also investigated. E299A α-L-iduronidaseprotein was synthesized and secreted, with a molecular masssimilar to that of the wild-type and the two E182 mutantsdescribed above. This indicated that the E299A mutant proteincould pass the ER/Golgi quality control process. The E299Amutant protein was catalytically inactive, a finding againproviding support for our bioinformatics prediction that E299may be the nucleophilic residue in the catalytic site of α-L-iduronidase (Henrissat et al., 1995; Durand et al., 1997, 2000).

However, there was evidence of a delayed rate of traffic andsecretion, and the E299A mutant protein that reached the lyso-somes was not proteolytically processed to the 69/65 species.This may indicate that, unlike other glycosidases, the glutamicacid to alanine change in α-L-iduronidase at position 299 hassome minor structural effect on the mutant protein. Thissuggested that despite folding and subsequent traffic to thecorrect destination, this minor protein structure modificationcould impede interaction with processing enzymes.

The glutamic acid residues studied here in human α-L-iduro-nidase are also present in dog and mouse α-L-iduronidase(Stoltzfus et al., 1992; Clarke et al., 1994). There are regions ofabsolute sequence identity around these critical glutamic acidresidues for both dog and mouse, compared to human α-L-iduronidase, and despite an altered positional location withinthe overall sequence, the two residues are exactly the samenumber of residues apart within all three sequences. Thissuggests that the location of the two glutamic acid residues isimportant and is consistent with the organization of theseglutamic acid residues in other glycosidases, on either side ofthe glycosidic bond, with a precise spatial separation (McCarterand Withers, 1994). Additional studies are planned to investigatethe interaction of substrate with the wild-type, E182A, andE299A α-L-iduronidase to confirm the role of these glutamicacid residues. Studies with E182D and E299D mutants, wherethe key glutamic acid residues are replaced by aspartic acidresidues whose side chains also bear a carboxylic acid group,should allow further insight into the structural constraintsexisting in the catalytic site of human α-L-iduronidase.Finally, it will be important to confirm these conclusions bydirect active site labelling (Withers and Aebersold, 1995) andultimately by 3D structural analysis.

Materials and methods

Materials

The wild-type human α-L-iduronidase cDNA has been previouslydescribed (IDUA; Scott et al., 1991). Plasmid pBS-II-KS+-IDUA,where the wild-type human IDUA cDNA is inserted into theEcoRI site of plasmid pBlue-Script-II-KS+ (Stratagene), waskindly provided by J.M. Heard (Institut Pasteur, Paris).Plasmid pCI-Neo was obtained from Promega. Oligonucleo-tides for mutagenesis were purchased from either Sigma orGenset. The T7 sequencing kit was from Pharmacia. Cationicliposomes bis-guanidinium-tren-cholesterol/dioleoylphosphatidyl-ethanolamine were generously supplied by J.P. Vigneron(Collège de France, Paris).

Polyvinylchloride plates (96-well, enzyme-linked immuno-sorbent assay [ELISA] plates) were obtained from Costar.Nitrocellulose membrane and 2,2′-azino-bis(3-ethylbenz-thiazoline-6-sulfonic acid) (ABTS) substrate were purchasedfrom Bio-Rad. Pansorbin cells (507861), and the α-L-iduroni-dase substrate MUI were from Calbiochem. Ovalbumin andbovine serum albumin (BSA) were from Sigma. G418 waspurchased from either Gibco or Sigma. Sheep anti-mouseimmunoglobulin, horseradish peroxidase–conjugated sheepanti-mouse immunoglobulin (SAM-Ig) and horseradishperoxidase–conjugated sheep anti-rabbit immunoglobulinwere purchased from Silenus Laboratories (a subsidiary ofChemicon). α-L-iduronidase peptide pins were synthesized by

Glycosidase active site mutations

747

Chiron Mimitopes. Fetal calf serum (FCS) for tissue culture,L-glutamine, penicillin, streptomycin, and Dulbecco’sModified Eagle’s Medium/Ham’s Nutrient Mixture F-12Coon’s Modified were from Life Technologies/Gibco BRL.35S-methionine radiolabel was purchased from NEN.

Site-directed mutagenesis for construction of mutants E182A and E299A

The wild-type human IDUA cDNA was isolated fromplasmid pBS-II-KS+-IDUA by EcoRI digestion, subclonedinto pCI-Neo and the resulting plasmid pCI-wt-Neo was usedfor the generation of mutants, by Kunkel’s method (Kunkel,1985). The mutant oligonucleotides (antisense) were asfollows: (1) 5′-GTCAAAGTCGTGGTGGTCTGGTGCATT-CCACGTC-3′ for construction of the E182A mutant bymodifying the GAG triplet coding for Glu182 into a GCA tripletcoding for Ala to create a BsmI restriction site facilitatingscreening; (2) 5′-CGGGTCCGCGGCGTCGTTGTAAATG-3′for construction of the E299A mutant by replacing the GAGtriplet coding for Glu299 with a GCC triplet coding for Ala tocreate a SacII restriction site facilitating screening. Bothmutant cDNAs were verified to exclude undesired mutationsby DNA sequencing by the dideoxy chain termination method.

Generation of CHO-K1 clones expressing wild-type, E182A, and E299A human IDUA proteins

Plasmids pCI-wt-Neo, pCI-E182A-Neo, and pCI-E299A-Neo(where the wild-type and E182A and E299A mutant humanIDUA cDNAs, respectively, are under the transcriptionalcontrol of the cytomeglovirus promoter) were transfected intoCHO cells using BGTC/DOPE cationic liposomes as previouslydescribed (Vigneron et al., 1996). Because these plasmids alsocontain a Neo expression cassette, stably transfected CHO-K1clones could be generated via G418 selection. CHO-K1 clonescharacterized by a high copy number of the different plasmidDNAs were identified by Southern blotting and used foranalysis of the corresponding human IDUA proteinsexpressed. The control cell line expressing wild-type humanIDUA was also selected via determination of the level ofenzyme activity by a MUI assay.

Generation of E182K expression cell line

The E182K missense mutation was engineered into the wild-type cDNA using a quick change site-directed mutagenesis kit(Stratagene). A clone containing the mutagenized IDUAcDNA construct was identified by hybridization with an allele-specific oligonucleatide, sequenced to ensure that no changesother than E182K were introduced and then subcloned into theexpression vector pEFNeo (Unger et al., 1994). Large-scaleplasmid stocks were prepared using a BRESApure plasmid kit(Bresatec). Ten micrograms of pEFNeoE182K was electro-porated into CHO-K1 cells as previously described (Anson et al.,1992) and selected with 0.75 mg/ml G418. G418-resistantmass-cultures were maintained in medium containing 0.5 mg/mlG418 for at least 2 weeks.

Cell culture

CHO-K1 cells were grown in either Dulbecco’s ModifiedEagle’s Medium or Ham’s Nutrient Mixture F-12 Coon’sModified supplemented with 10% FCS and antibiotics. Theculture media from confluent 75 cm2 flasks were collected,

clarified by centrifugation (200 x g for 10 min at 4°C) andstored sterile at 4°C. For harvesting, the cell layers werewashed twice with 10 ml of Dulbecco’s phosphate bufferedsaline (PBS) and the cells released from the culture surface byincubation with 10% (v/v) trypsin versene solution for 2 min at37°C. The harvested cells were centrifuged at 200 × g for 5 minand the cell pellet washed with an additional 10 ml of PBS andthen recentrifuged. The supernatant was removed and the cellpellet used for either organelle fractionation or cell extractsprepared for immunoquantification, immunobinding assay,and western blotting.

Monoclonal antibody epitope mapping

Peptide pin technology (Chiron Mimitopes) was used to determinethe epitope reactivity of the monoclonal antibody ID1A (Ashtonet al., 1992) with linear peptide sequences of α-L-iduronidase, aspreviously described with another lysosomal protein (Turneret al., 1999). Briefly, individual 13 amino acid peptides weresynthesized onto polyethylene pins as previously described(Geyson et al., 1984). A six-amino-acid overlap for consecutivepeptides ensured linear sequence epitopes were not splitbetween peptides. The array of peptide pins (96-well format)were subject to ELISA to quantify reactive linear sequenceepitopes. The peptide pins were slotted into a 96-well platewith each well containing 200 µl of block buffer (1% [w/v]ovalbumin and 0.1% [v/v] Tween 20, in 1 x PBS, pH 7.2), for1 h at 20°C to reduce nonspecific background reactivity withthe peptide pins. All incubations were at 20°C and aided by theuse of a plate shaker (Milenia, Micromix 4). The peptide pinswere washed by submerging in PBS, pH 7.2, for 10 min at20°C on an orbital shaker to remove unbound protein. Themonoclonal antibody was diluted 1:3 in block buffer (1% [w/v]ovalbumin and 0.1% (v/v) Tween 20 in PBS, pH 7.2) and 200 µladded to each well of an ELISA plate. Peptide pins weresubmerged into each well and incubated at 20°C for 1 h, priorto an overnight incubation at 4°C. Plate pins were sequentiallywashed three times in PBS, pH 7.2, to remove unbound mono-clonal antibody. A volume of 200 µl of a 1:1000 dilution ofhorseradish peroxidase–conjugated SAM-Ig in block buffer,was added to each well of a fresh 96-well plate. Peptide pinswere submerged into each well and incubated at 20°C for 1 h,then washed (as above) to remove unbound antibody. Avolume of 200 µl of ABTS peroxidase substrate solution wasadded to each well of a fresh 96-well plate and the peptide pinssubmerged into the wells. Color reaction was recorded after a20-min incubation at 20°C, by determining the optical densityof each well at 414 nm (Ceres 900). All results were comparedto positive and negative control pins, and a second antibody-only control with each individual peptide pin (i.e., performedas a separate assay on the same peptide pin plate).

Preparation of cell extracts and cell protein assays

Confluent CHO cells were harvested as described above and acell pellet produced. For immunoquantification, immuno-binding assay and western blot analysis, each cell pellet wasresuspended in 200 µl of 20 mM Tris–HCl pH 7.0 containing0.5 M NaCl, then freeze/thawed six times. The samples werethen centrifuged at 10,000 × g for 5 min and the supernatantcollected and stored at –20°C. The amount of protein in themedia and each cell lysate was determined by the method ofBradford (1976).

D.A. Brooks et al.

748

Immunoquantification

α-L-iduronidase polypeptide levels were measured with animmunoquantification assay, as previously described (Ashtonet al., 1992), using a monospecific polyclonal antibody tocapture the protein and then ID1A monoclonal antibody and aperoxidase-labeled second antibody to detect and quantify thebound protein.

Immunobinding assay

An immune capture method was developed to specifically bindhuman α-L-iduronidase protein and followed by a MUI assayto determine α-L-iduronidase enzyme activity. Briefly, 100 µlof affinity purified SAM-Ig (20 µg/ml in 0.1 M NaHCO3,pH 8.5) was attached to each well of a 96-well polyvinyl-chloride plate (ELISA plate) by incubation at 37°C for 2 h,then overnight at 4°C. The plate was then washed to removeunbound antibody (cycle involving three successive washes ona plate washer using 0.02 M Tris–HCl, pH 7.0, containing0.25 M NaCl). Each well on the plate was then incubated with200 µl of 0.02 M Tris–HCl, pH 7.0, containing 0.25 M NaCland 1% (w/v) ovalbumin to reduce nonspecific binding andblock any remaining reactive sites on the ELISA plate. Themonoclonal antibody ID1A was then adsorbed to the SAM-Igon the plate, by incubation of each well with 100 µl of ID1Amonoclonal antibody culture supernatant, for 4 h at 20°C. Theplate was then rewashed, as above, to remove unbound anti-body. Dilutions of the α-L-iduronidase to be assayed were thenincubated (50 µl/well) in each well of the plate (e.g., standardcurve of known enzyme activity plus dilutions of cell extractcontaining unknown levels of α-L-iduronidase) for 16 h at4°C. The plate was then rewashed, as above, to removeunbound sample and non-α-L-iduronidase protein. Wells werethen assayed for α-L-iduronidase activity by incubating 30 µlof MUI substrate (0.5 mM), containing 0.05 M sodium dimethyl-glutarate, pH 5.7, and 3.5 mg/ml BSA in 0.9% (w/v) NaCl, for4 h at 37°C. Reactions were stopped by placing the plate onice, then recovering the substrate and adding 2 ml of 0.05 Mglycine-carbonate buffer, pH 10.7. Enzyme activity wasestimated by the release of 4-methylumbelliferone using aPerkin Elmer spectrofluorimeter and an excitation wavelengthof 366 nm and an emission wavelength of 446 nm. Resultswere interpolated through a standard curve and enzymeactivity expressed as nmol/min/ml of sample.

Western blot analysis

Cell lysates (60 µg) and media samples (110 µg) were electro-phoresed on 10% sodium dodecyl sulfate–polyacrylamide gelelectrophoresis (SDS–PAGE) using standard conditions(Laemmli, 1970) and transferred to nitrocellulose and immuno-blotted as previously described (Clements et al., 1989).Briefly, the immunoblot was blocked with 5% (w/v) BSA in0.25 M NaCl/20 mM Tris, pH 7.0, and then incubated in theprimary antibody, which was a monospecific polyclonalagainst human α-L-iduronidase (1:1000 dilution in 1% [w/v]ovalbumin in 0.25 M NaCl/20 mM Tris, pH 7.0). Horseradishperoxidase–conjugated sheep anti-rabbit immunoglobulin wasused as a second antibody (1:1000 dilution) and followed bydetection using 4-chloro-1-naphthol and H2O2 in 0.02 M Tris–HCl,pH 7.0, containing 0.25 M NaCl.

Pulse chase labeling

Confluent 75-cm2 tissue culture flasks of CHO-K1 cells werepreincubated for 45 min in cystine/methionine free Dulbecco’sModified Eagle’s Medium containing 10% (v/v) FCS and2 mM glutamine, before labeling for 30 min at 37°C with0.49 mCi of 35S-methionine in 5 ml of the same culturemedium. The culture medium containing the radiolabel wasthen aspirated, and the tissue culture flasks washed twice with10 ml of PBS before either harvesting (pulse label) or adding8 ml of fresh culture media (for chase, which was in Hams-F12containing 10% [v/v] FCS and 2 mM glutamine). The pulselabel was followed by either 4 h or 24 h chase, then the culturemedium was collected from each flask and centrifuged at200 x g for 5 min and stored sterile at 4°C prior to immuno-precipitation. The CHO-K1 cell layers were harvested atappropriate time points, as described above. The cell pellet wasresuspended in 10 ml of solubilization buffer (PBS containing1% [w/v] Na deoxycholate, 0.1% [w/v] SDS, 0.5% [v/v] TritonX100). The samples were left at 4°C for a minimum of 24 hbefore immunoprecipitation.

Immunoprecipitation

Radiolabeled cell extracts were immunoprecipitated usingantibody attached to pansorbin cells (i.e., source of protein A).The Pansorbin cells were equilibrated by washing four timeswith solubilization buffer, using centrifugation at 10,000 x gfor 2 min at 4°C. Cell extracts and culture media wereprecleared by incubation first, with 60 µl of equilibratedPansorbin cells for 2 h at 4°C, with constant mixing (i.e., minusantibody). Second, 60 µl of SAM-Ig bound Pansorbin cells(15 µg of SAM-Ig per 60 µl of Pansorbin cells) was incubatedfor 2 h at 4°C. Subsequently, 5 ml of Id1A monoclonal culturesupernatant (approximately 10 µg/ml monoclonal antibody)was bound to an additional 500 µl of Pansorbin-SAM-Ig (byincubation at 4°C for 2 h, with mixing), then 60 µl of thisPansorbin mix incubated with each cell extract/culturemedium, overnight at 4°C with mixing. The Pansorbincomplex was pelleted by centrifugation at 10,000 × g for 5 min,prior to three 1-ml washes with solubilization buffer andfinally one wash with 0.5 ml water. The Pansorbin complexwas then resuspended in SDS–PAGE sample buffer, reducedusing β-mercaptoethanol, boiled for 5 min to release the α-L-iduronidase protein then centrifuged at 10,000 g for 5 min. Thesupernatant was loaded onto a 10% polyacrylamide gel, andelectophoresed as described above. The gel was fixed in 40%(v/v) methanol:10% (v/v) acetic acid for 1 h, prior to placing inAmplify solution overnight at 20°C and subsequent autoradio-graphic detection of radiolabeled α-L-iduronidase.

Organelle fractionation of CHO-K1 cells

CHO-K1 cells were harvested and a granular fraction preparedfor organelle subfractionation, on an 18% (v/v) Percollgradient, as previously described (Brooks et al., 1997). Briefly,cells were harvested as described above and the cell pelletresuspended in 10 mM HEPES, pH 7.0, containing 0.25 Msucrose, 1 mM ethylenediamine tetraacetic acid, and proteaseinhibitors. The cell suspension was drawn through a 23-gaugeneedle and then subjected to hypobaric shock, prior to centrif-ugation at 200 x g for 10 min at 4°C to remove cellular debris.The postnuclear supernatant (whole granular fraction)

Glycosidase active site mutations

749

produced was then subfractionated on an 18% (v/v) Percollgradient, by centrifugation at 31,400 × g (average g) for 1 h at4°C. Fractions were collected from the top of the gradient, as10 2-ml samples. Each fraction was assayed for β-hexos-aminidase and acid phosphatase activity as previouslydescribed (Brooks et al., 1992), using the fluorogenicsubstrates 4-methylumbelliferyl-N-acetyl-β-D-glucosaminideand 4-methylumbelliferyl-phosphate respectively (Leabackand Walker, 1961; Kolodny and Mumford, 1976). The saltconcentration of each fraction was adjusted to 0.25 M and thenfreeze/thawed six times before centrifuging at 100,000 × g for1 h at 4°C to remove the Percoll and cell membranes. Thesupernatant of each fraction (20 µl of the total) was run on a10% SDS–PAGE and immunoblotted as described above.

Acknowledgments

This work was supported by a NH&MRC program grant inAustralia and by grants from the Association Vaincre lesMaladies Lysosomales (VML, Evry) in France.

Abbreviations

ABTS, 2,2′-azino-bis(3-ethylbenzthiazoline-6-sulfonic acid);BSA, bovine serum albumin; CHO, Chinese hamster ovary;ELISA, enzyme linked immunosorbent assay; ER, endoplasmicreticulum; FCS, fetal calf serum; GAG, glycosaminoglycans;MPS I, mucopolysaccharidosis type I; MUI, 4-methylum-belliferyl α-L-iduronide; PBS, phosphate buffered saline;SAM-Ig, sheep anti-mouse immunoglobulin; SDS–PAGE,sodium dodecyl sulfate–polyacrylamide gel electrophoresis.

References

Anson, D.S., Taylor, J.A., Bielicki, J., Harper, G.S., Peters, C., Gibson, G.J.,and Hopwood, J.J. (1992) Correction of human MPS type VI fibroblastswith recombinant N-acetylgalactosamine-4-sulphatase. Biochem. J., 284,789–794.

Ashton, L.J., Brooks, D.A., McCourt, P.A.G., Muller, V.J., Clements, P.R.,and Hopwood, J.J. (1992) Immunoquantification and enzyme kinetics ofα-L-iduronidase in fibroblasts from normal controls and mucopolysaccha-ridosis type I patients. Am. J. Hum. Genet., 50, 787–794.

Bradford, M.M. (1976) A rapid and sensitive method for the quantitation ofmicrogram quantities of protein utilising the principle of protein-dyebinding. Anal. Biochem., 72, 248–254.

Brooks, D.A. (1993) Review: the immunochemical analysis of enzyme frommucopolysaccharidoses patients. J. Inher. Metab. Dis., 16, 3–15.

Brooks, D.A. (1997) Minireview: protein processing: a role in the patho-physiology of genetic disease. FEBS Lett., 409, 115–120.

Brooks, D.A., Bradford, T.M., Carlsson, S.R., and Hopwood, J.J. (1997) Amembrane protein primarily associated with the lysosomal compartment.Biochim. Biophys. Acta, 1327, 162–170.

Brooks, D.A., Harper, G.S., Gibson, G.J., Ashton, L.J., Taylor, J.A.,McCourt, P.A.G., Freeman, C., Clements, P.R., Hoffman, J.W., andHopwood, J.J. (1992) Hurler syndrome: a patient with abnormally highlevels of α-L-iduronidase protein. Biochem. Med. Metab. Biol., 47, 211–220.

Bunge, S., Kleijer, W.J., Steglich, C., Beck, M., Schwinger, E., and Gal, A.(1995) Mucopolysaccharidosis type I: identification of 13 novel mutationsof the α-L-iduronidase gene. Hum. Mutat., 6, 91–94.

Callebaut, I., Labesse, G., Durand, P., Poupon, A., Canard, L., Chomilier, J.,Henrissat, B., and Mornon, J.P. (1997) Deciphering protein sequenceinformation through hydrophobic cluster analysis (HCA): current statusand perspectives. Cell Mol. Life Sci., 53, 621–645.

Clarke, L.A., Nasir, J., Zhang, H.F., Mcdonald, H., Applegarth, D.A., Hayden, M.R.,and Toone, J. (1994) Murine alpha-L-iduronidase—cDNA isolation andexpression. Genomics, 24, 311–316.

Clements, P.R., Brooks, D.A., McCourt, P.A.G., and Hopwood, J.J. (1989)Immunopurification and characterisation of human α-L-iduronidase withthe use of monoclonal antibodies. Biochem. J., 259, 199–208.

Davies, G. and Henrissat, B. (1995) Structures and mechanisms of glycosylhydrolases. Structure, 3, 853–859.

Durand, P., Lehn, P., Callebaut, I., Faberega, S., Henrissat, B., and Mornon, J.-P.(1997) Active-site motifs of lysosomal acid hydrolases: invariant featuresof clan GH-A glycosyl hydrolases deduced from hydrophobic clusteranalysis. Glycobiology, 7, 277–284.

Durand, P., Faberega, S., Henrissat, B., Mornon, J.-P., and Lehn, P. (2000)Structural features of normal and mutant human lysosomal glycosidehydrolases deduced from bioinformatics analysis. Hum. Mol. Genet., 9,967–977.

Fabrega, S., Durand, P., Codogno, P., Bauvy, C., Delomenie, C., Henrissat, B.,Martin, B.M., McKinney, C., Ginns, E.I., Mornon, J.P., and Lehn, P.(2000) Human glucocerebrosidase : heterologous expression of active sitemutants in murine null cells. Glycobiology, 10, 1217–1224.

Geyson, H.M., Meloen, R.H., and Barteling, S.J. (1984) Use of peptidesynthesis to probe viral antigens for epitopes to a resolution of a singleamino acid. Proc. Natl Acad. Sci. USA, 81, 3998–4002.

Henrissat, B. and Bairoch, A. (1996) Updating the sequence-based classificationof glycosyl hydrolases. Biochem. J., 316, 695–706.

Henrissat, B. and Davies, G. (1997) Structural and sequence-based classificationof glycoside hydrolases. Curr. Opin. Struct. Biol., 7, 637–644.

Henrissat, B., Callebaut, I., Fabrega, S., Lehn, P., Mornon, J.P., and Davies, G.(1995) Conserved catalytic machinery and the prediction of a common foldfor several families of glycosyl hydrolases. Proc. Natl Acad. Sci. USA, 92,7090–7094.

Kolodny, E.H. and Mumford, R.A. (1976) Human leukocyte acid hydrolases:characterisation of eleven lysosomal enzymes and study of reactionconditions for their automated analysis. Clin. Chim. Acta, 70, 247–257.

Koshland, D.E. (1953) Stereochemistry and the mechanism of enzymaticreactions. Biol. Rev., Cambridge Phil. Soc., 28, 416–436.

Krawczak, M. and Cooper, D.N. (1997) The human gene mutation database.Trends Genet., 13, 121–122.

Kunkel, T.A. (1985) Rapid and efficient site-specific mutagenesis withoutphenotypic selection. Proc. Natl. Acad. Sci. USA, 82, 488–492.

Laemmli, U.K. (1970) Cleavage of structural proteins during the assembly ofthe head of bacteriophage T4. Nature, 227, 680–685.

Leaback, D.H. and Walker, P.G. (1961) Studies on glucosaminidase. IV. Thefluorometric assay of N-acetyl-β-glucosaminidase. Biochem. J., 78, 151–156.

McCarter, J.D., and Withers, S.G. (1994) Mechanisms of enzymatic glycosidehydrolysis. Curr. Opin. Struct. Biol., 4, 885–892.

Neufeld, E.F. and Muenzer, J. (1995) The mucopolysaccharidoses. In Scriver,C.R., Beaudet, A.L., Sly, W.S., and Valle, D., eds., The metabolic basis ofinherited disease, 7th ed. New York: McGraw-Hill, pp. 2465–2494.

Scott, H.S., Anson, D.S., Osborne, A.M., Nelson, P.V., Clements, P.R.,Morris, C.P., and Hopwood, J.J. (1991) Human α-L-iduronidase: cDNAisolation and expression. Proc. Natl Acad. Sci. USA, 88, 9695–9699.

Scott, H.S., Bunge, S., Gal, A., Clarke, L.A., Morris, C.P., and Hopwood, J.J.(1995) Molecular genetics of mucopolysaccharidosis type I: diagnostic,clinical and biological implications. Hum. Mutat., 6, 288–302.

Scott, H.S., Litjens, T., Hopwood, J.J., and Morris, C.P. (1992a) A commonmutation for mucopolysaccharidosis type I associated with a severe Hurlersyndrome phenotype. Hum. Mutat., 1, 103–108.

Scott, H.S., Litjens, T., Nelson, P.V., Brooks, D.A., Hopwood, J.J., andMorris, C.P. (1992b) α-L-iduronidase mutations (Q70X and P533R)associate with severe Hurler syndrome phenotype. Hum. Mutat., 1, 333–339.

Scott, H.S., Litjens, T., Nelson, P.V., Thompson, P.R., Brooks, D.A.,Hopwood, J.J. and Morris, C.P. (1993) Identification of mutations in theα-L-iduronidase gene (IDUA) that cause Hurler and Scheie syndromes.Am. J. Hum. Genet., 53, 973–986.

Sinnott, M.L. (1990) Catalytic mechanisms of enzymic glycosyl transfer.Chem. Rev., 90, 1171–1202.

Stoltzfus, L.J., Sosa-Pineda, B., Moskowitz, S.M., Menon, K.P., Dlott, B.,Hooper, L., Teplow, D.B., Shull, R.M., and Neufeld, E.F. (1992) Architectureof the canine IDUA gene and mutation underlying canine mucopoly-saccharidosis I. Genomics, 14, 763–768.

Taylor, J.A., Gibson, G.J., Brooks, D.A., and Hopwood, J.J. (1991) Biosynthesisand maturation of human α-L-iduronidase in normal and mucopoly-saccharidosis type I human skin fibroblasts. Biochem. J., 274, 263–268.

D.A. Brooks et al.

750

Turner, C.T., Hopwood, J.J., Bond, C.S., and Brooks, D.A. (1999) Immuneresponse to enzyme replacement therapy: 4-sulfatase epitope reactivity ofplasma antibodies from MPS VI cats. Mol. Genet. Metab., 67, 194–205.

Unger, E.G., Durrant, J., Anson, D.S., and Hopwood, J.J. (1994) Recombinantα-L-iduronidase—characterisation of the purified enzyme and correctionof mucopolysaccharidosis type I fibroblasts. Biochem. J., 304, 43–49.

Vigneron, J.P., Oudrhiri, N., Fauquet, M., Vergely, L., Bradley, J.C.,Basseville, M., Lehn, P., and Lehn, J.M. (1996) Guanidinium-cholesterolcationic lipids: efficient vectors for the transfection of eukaryotic cells.Proc. Natl Acad. Sci. USA, 93, 9682–9686.

White, A., Tull, D., Johns, K., Withers, S.G., and Rose, D.R. (1996) Crystalstructure of the catalytic domain of the β-1, 4-glycanase cex fromCellulomonas fimi. Nat. Struct. Biol., 3, 149–154.

Withers, S.G. and Aebersold, R. (1995) Approaches to labeling and identificationof active site residues in glycosidases. Prot. Sci., 4, 361–372.

Yamagishi, A., Tomatsu, S., Fukuda, S., Uchiyama, N., Shimozawa, N.,Suzuki, Y., Kondo, N., Sukegawa, T., and Orii, T. (1996) Mucopoly-saccharidosis type I: identification of common mutations that cause Hurlerand Scheie syndromes in Japanese populations. Hum. Mutat., 7, 23–29.

Zhao, K-W., Faull, K.F., Kakkis, E.D., and Neufeld, E.F. (1997) Carbohydratestructures of recombinant human α-L-iduronidase secreted by chinesehamster ovary cells. J. Biol. Chem., 272, 22758–22765.