Structural and Functional Characterization of Streptomyces plicatus β-N-Acetylhexosaminidase by Comparative Molecular Modeling and Site-directed Mutagenesis*

We have sequenced the Streptomyces plicatus β-N-acetylhexosaminidase (SpHex) gene and identified the encoded protein as a member of family 20 glycosyl hydrolases. This family includes human β-N-acetylhexosaminidases whose deficiency results in various forms of GM2 gangliosidosis. Based upon the x-ray structure of Serratia marcescens chitobiase (SmChb), we generated a three-dimensional model ofSpHex by comparative molecular modeling. The overall structure of the enzyme is very similar to homology modeling-derived structures of human β-N-acetylhexosaminidases, with differences being confined mainly to loop regions. From previous studies of the human enzymes, sequence alignments of family 20 enzymes, and analysis of the SmChb x-ray structure, we selected and mutated putative SpHex active site residues. Arg162 → His mutation increased K m 40-fold and reduced V max 5-fold, providing the first biochemical evidence for this conserved Arg residue (Arg178 in human β-N-acetylhexosaminidase A (HexA) and Arg349 in SmChb) as a substrate-binding residue in a family 20 enzyme, a finding consistent with our three-dimensional model of SpHex. Glu314 → Gln reduced V max296-fold, reduced K m 7-fold, and altered the pH profile, consistent with it being the catalytic acid residue as suggested by our model and other studies. Asp246 → Asn reduced V max 2-fold and increasedK m only 1.2-fold, suggesting that Asp246 may play a lesser role in the catalytic mechanism of this enzyme. Taken together with the x-ray structure ofSmChb, these studies suggest a common catalytic mechanism for family 20 glycosyl hydrolases.

a glycosyl hydrolase that removes N-acetylglucosamine or Nacetylgalactosamine residues from the nonreducing end of oligosaccharides and their conjugates. This enzyme may be important for the efficient degradation of polysaccharides by Streptomyces species (1). In humans, the importance of ␤-Nacetylhexosaminidase is illustrated by the fatal neurodegenerative disorders that result from its deficiency.
Human ␤-N-acetylhexosaminidase A (HexA), a heterodimer of subunits ␣ (encoded by HEXA) and ␤ (encoded by HEXB), is essential for the degradation of ganglioside G M2 (2). Mutations in HEXA or HEXB cause Tay-Sachs and Sandhoff disease, respectively. Structure/function studies of HexA have been limited by the difficulty in producing levels of recombinant HexA in mammalian expression systems that are sufficient for kinetic analysis (3,4).
The classification of glycosyl hydrolases into families based on amino acid sequence similarity has greatly facilitated the identification of evolutionary and mechanistic relationships between these enzymes (5)(6)(7). To date, 63 glycosyl hydrolase families have been identified, 24 of which have a three-dimensional structure determined for at least one member.
Family 20 contains ␤-N-acetylhexosaminidases and chitobiases (EC 3.2.1.52) and are thought to use an acid/base mechanism that involves a proton donor and a nucleophile (8). Serratia marcescens chitobiase (SmChb) is the only family 20 member for which a three-dimensional structure has been determined (9). According to the 1.9-Å resolution crystallographic model, SmChb appears to lack the nucleophilic residue necessary for catalysis. Alternatively, Tews et al. (9) suggest that the N-acetyl group on the nonreducing N-acetylglucosamine residue of the chitobiose substrate may act as the nucleophile (substrate-assisted catalysis) (9). The structure also predicts that Glu 540 is the proton donor and that Arg 349 is directly involved in substrate binding. Both residues are conserved among family 20 enzymes. In human HexA, mutagenesis studies of the ␣and ␤-subunit residues homologous to SmChb Glu 540 , ␣-Glu 323 and ␤-Glu 355 , respectively, have been identified as likely proton donors (10,11). Fernandes et al. (10) suggested that ␣-Asp 258 , previously predicted to be a proton donor in HexA, fulfills a lesser role in the catalytic mechanism of this enzyme. These studies on HexA provide biochemical support for Glu 540 of SmChb as the proton donor in family 20 enzymes; however, there is no biochemical evidence in the literature that supports the suggestion that Arg 349 of SmChb or its homologue in other family 20 enzymes is a substrate binding residue.
We have sequenced the gene encoding SpHex, classified it as a new member of family 20 glycosyl hydrolases, and constructed a three-dimensional model of the SpHex enzyme by comparative molecular modeling. In conjunction with the molecular modeling, site-directed mutagenesis was used to demonstrate a role for Arg 162 (human HexA, ␣-Arg 178 ; SmChb, Arg 349 ) in substrate binding. Our biochemical analysis of Glu 314 (human HexA, ␣-Glu 323 ; SmChb, Glu 540 ) and Asp 246 (human HexA, ␣-Asp 258 ; SmChb, Asp 448 ) mutations further substantiate the role of these residues in family 20 enzymes. These studies indicate that the catalytic mechanism of these enzymes may be conserved. Finally, multiple sequence alignments of the enzymes indicate that SpHex is significantly more related in structure to human HexA and -B than SmChb.

EXPERIMENTAL PROCEDURES
Recombinant DNA Methods and Strains-Escherichia coli strain JM109 was used for all plasmid manipulations and fusion protein production. Plasmid DNA was purified using the Nucleobond AX purification system (Macherey-Nagel). Restriction enzymes and Klenow fragment were from New England Biolabs. Restriction enzyme digests and filling-in 3Ј-recessed ends of DNA were according to New England Biolabs instructions. T4 DNA ligase was from Boehringer Mannheim, and DNA ligations were performed according to their instructions. Liquid cultures of S. plicatus (American Type Culture Collection number 27800) were grown at 25°C in International Streptomyces Project tryptone yeast extract broth (ISP medium 1) (Difco). Stocks of S. plicatus were maintained on ISP agar plates at 4°C.
Sequencing the SpHex Clone-Clone 36, expressing SpHex (1), was previously identified by screening a -ZAP library using 4-methylumbelliferyl (4-MU) glycosides of N-acetylglucosamine oligosaccharides as a substrate for an in plate assay (12). Clone 36 was found to have a 5.0-kb insert and expressed a 60-kDa fusion protein (1). Based on this, the coding region for SpHex was predicted to be within a 1.8-kb region of the insert adjacent to the ␤-galactosidase gene in the vector. The 1.8-kb region, bounded by EcoRI (5Ј) and SmaI (3Ј) sites, was restriction enzyme-mapped and subcloned into pBluescript (pBS ϩ ) (Stratagene) as two smaller fragments. The subclones were sequenced at the National Centers of Excellence core sequencing facility (Toronto, Canada) by the dideoxy chain termination method (13). Sequencing was done on double-stranded DNA and covered both strands. To clarify the sequence in some areas, we used sequence-specific primers to sequence singlestranded DNA produced from a 1.8-kb EcoRI/SmaI clone (psHEX-1.8).
To obtain the 5Ј-end of the SpHex gene, genomic DNA was isolated from S. plicatus according to Nisen et al. (14), except that the chloroform-extracted DNA was not dialyzed prior to use. The genomic DNA was digested with combinations of XhoI and various restriction enzymes, and fragments that hybridized to a 600-bp AscI/XhoI 32 P-labeled probe, complementary to the 5Ј-end of clone psHEX-1.8, were identified by Southern blotting (15). A 1.7-kb EcoRV/XhoI fragment was predicted to contain approximately 1.1 kb of sequence beyond the known 5Ј-end of the gene. An area of an agarose gel containing this fragment was excised, purified using a Geneclean II kit (BIO 101), ligated into pBS ϩ (EcoRV/XhoI), and electroporated into E. coli. The resulting colonies were screened by hybridization (15) using the same DNA probe as described above. Ten positive colonies were identified, one of which (ps5HEX-1.7) was used to determine the sequence of the 5Ј-end of the SpHex gene. A 700-bp fragment was removed from the 5Ј-end of the ps5HEX-1.7 insert by digestion with SmaI and EcoRV. The larger vector (4.2 kb) was purified from the 700-bp insert using a Geneclean II kit and religated to create ps5HEX-1.0. Upon sequencing 284 bp of the 5Ј-end of the psHEX-1.0 insert (both strands), the complete coding sequence of the SpHex gene could be determined.
The predicted amino acid sequence had one open reading frame (starting at position 269 of GenBank TM accession number AF063001) and showed homology to other family 20 ␤-hexosaminidases in a Clust-alW1.7 (16,17) multiple sequence alignment. This sequence was compared with protein sequences available in the EMBL/SWISS-PROT data base (release 34; 59,021 sequence entries) using MaxHom, a weighted dynamic multiple sequence alignment algorithm (18).
Comparative Molecular Modeling-Atomic coordinates for SmChb were obtained from the Brookhaven National Laboratory Protein Data Bank (Protein Data Bank code 1qbb) and used as a template for molec-ular modeling. The best alignment between the amino acid sequences of SpHex, SmChb, and the 13 other known family 20 glycosyl hydrolases was obtained using ClustalW1.7. To improve the alignment, sequences that increased the frequency of insertions and deletions were sequentially removed, and the remaining sequences were realigned. This minimized the number of insertions and deletions within the SpHex sequence and maximized alignments of the secondary structural elements as visualized using the graphics program O (19). The final sequence alignment between SpHex and SmChb was optimized manually by comparing initial models of SpHex to the x-ray structure of SmChb using O (20).
Using Homology (Insight software, BioSym Technologies, Inc.), the aligned amino acid sequence of SpHex was substituted onto the SmChb backbone atomic coordinates. Initial side chain steric clashes were relieved using a rotamer library containing the most favorable residue side chain conformations. No amino acid insertions into the SpHex model were required, and gaps were manually joined using the program O. Serious steric clashes were relieved manually.
The model was placed in a primitive lattice (P1 space group) and subjected to energy minimization using a conjugate gradient target function implemented in X-PLOR (21,22). The cell parameters were set sufficiently large to avoid intermolecular contacts during minimization. Minimization was performed with the van der Waals and electrostatic terms turned on. The model was minimized until convergence was reached, as judged by the root mean square of the energy gradient (average derivative Ͻ 0.1 kcal/mol/Å). The overall quality of the model was assessed with the program PROCHECK (23), using comparison values typical for a 2.0-Å resolution x-ray structure.
Construction of the pMAL-c2-SpHex Fusion Protein Vector-A construct pMAL-c2-SpHex (a gift from New England Biolabs) was made by subcloning the insert from clone 36 (BamHI/Hind III) into pMAL-c2 (XmnI/HindIII). The BamHI site was filled in to facilitate subcloning and to allow for correct translation of the maltose-binding protein (MBP)-SpHex fusion protein. 2 To remove 3.2 kb of extraneous DNA, pMAL-c2-SpHex was digested with SmaI and HindIII, the HindIII site was blunt-ended, and both the larger vector (7 kb) and a 1.8-kb (SmaI/ SmaI) fragment containing the coding region, were gel-purified and ligated to generate pmHEX-1.8. A plasmid with the fragment ligated in the correct orientation was chosen using 4-methylumbelliferyl-␤-Nacetylglucosaminide (4-MUG) (Toronto Research Chemicals Inc.) as substrate for an in plate assay (12).
Purification of SpHex-MBP Fusion and Cleavage with Factor Xa Protease-SpHex-MBP fusion protein was prepared in batches based on instructions from New England Biolabs except that fusion protein expression was induced at 25°C, and E. coli suspensions were supplemented with 1 mM phenylmethylsulfonyl fluoride, 5 g/ml aprotinin, 5 g/ml leupeptin, and 2 g/ml pepstatin A prior to lysis. Incubation of the purified fusion protein with 2% (w/w) factor Xa protease (New England Biolabs) in 20 mM Tris-Cl, pH 7.4, 200 mM NaCl, 1 mM EDTA, pH 7.4, containing 2 mM CaCl 2 for 18 -24 h at room temperature facilitated cleavage of the MBP from SpHex. The fusion protein and cleavage products were visualized using SDS-PAGE as described elsewhere (27).
Protein and Enzyme Assays-Protein concentrations were determined by the Bradford method (28) using ␥-globulin as the standard. To 2 S. Wong-Madden, personal communication.
determine the pH optima for mutant and wild type SpHex, activities for each were measured using 4-MUG as a substrate in 10 mM Na 2 HPO 4 , 6 mM citric acid, 0.3% bovine serum albumin ranging from pH 2.0 to 9.0. After a 45-min incubation at 25°C, each reaction was quenched with 0.1 M glycine-carbonate buffer, pH 10. The fluorescent product, 4-MU, was measured (excitation, 364 nm; emission, 448 nm) using a Hitatchi F-2000 fluorescence spectrophotometer. Kinetic data for mutant and wild type SpHex were obtained in an identical manner at the pH optimum of the wild type enzyme using 4-MUG concentrations ranging from 0.033 to 5 mM. K m and V max values were determined using the direct linear plot method (29).
Generation of Antibodies against SpHex-MBP Fusion Protein-Rabbit anti-SpHex-MBP fusion protein antibodies were raised at the National Biological Laboratories (Oakbank, Canada). Briefly, purified SpHex-MBP fusion protein emulsified with Freund's complete adjuvant was injected subcutaneously at multiple sites. Two weeks later, the antigen was emulsified with Freund's incomplete adjuvant and injected subcutaneously at multiple sites. This step was repeated 3 weeks later. Serum was collected 1 week after the final injection.
Western Blot Analysis-To obtain a crude S. plicatus cytosolic fraction for immunoblotting experiments, the cells were collected by centrifugation at 5000 ϫ g for 10 min and then resuspended in 4 ml of TNE (20 mM Tris-Cl, pH 7.4, 200 mM NaCl, 1 mM EDTA, pH 7.4) containing 1 mM phenylmethylsulfonyl fluoride, 5 g/ml aprotinin, 5 g/ml leupeptin, and 2 g/ml pepstatin A. The suspension was homogenized with a Dounce homogenizer for 1 min at 4°C and sonicated (6 times for 15 s) at 100 watts on ice using a Braun-sonic 1510 (B. Braun Melsungen AG). The lysate was centrifuged at 20000 ϫ g for 20 min, and the supernatant, considered to be the crude cytosolic fraction, was stored at Ϫ20°C.
For immunodetection, protein samples were separated on 7.5% SDS-PAGE gels and transferred onto a nitrocellulose membrane (30). The membrane was incubated for 1 h in TBST (10 mM Tris-Cl, pH 7.5, 100 mM NaCl, 0.1% Tween 20 (w/v)) containing 5% milk powder and for a further 1-2 h in the same buffer containing primary antibody (1:10,000 dilution). The membrane was washed with TBST and incubated with donkey anti-rabbit IgG-horseradish peroxidase conjugate from Amersham Pharmacia Biotech (1:10,000 dilution) for 1 h in TBST containing 5% milk powder. After washing with TBST, the antigen-antibody complex was detected using ECL from Amersham Pharmacia Biotech.

RESULTS AND DISCUSSION
DNA Sequence Analysis of SpHex-Originally, clone 36 was classified as a ␤-N-acetylhexosaminidase on the basis of its substrate specificity (1). In a search for enzymes related to human ␤-N-acetylhexosaminidase that might be useful for structure/function analysis, we determined the nucleotide sequence and deduced amino acid sequence of the enzyme (SpHex) encoded by clone 36. This sequence did not include an initiating Met codon, indicating that a portion of the 5Ј-end of the gene was missing. This was not surprising, given that SpHex had to be produced as a ␤-galactosidase fusion to allow its expression in E. coli (1). However, Western blot analysis demonstrated that the molecular masses of the recombinant and native SpHex were indistinguishable (Fig. 1, lanes 3 and  4), suggesting that only a small portion of the 5Ј-end of the SpHex gene was missing from the clone. An additional clone (ps5HEX-1.0) containing the 5Ј-end of the gene was obtained, and 284 bp of new sequence was determined. The complete coding sequence revealed a 1683-bp open reading frame, encoding 561 amino acids.
Comparison of the deduced amino acid sequence with the SWISS-PROT data base using MaxHom identified 12 ␤-Nacetylhexosaminidases and one chitobiase, all family 20 enzymes (Table I). SmChb, which has not been deposited in the SWISS-PROT data base, was retrieved from the Brookhaven Protein Data Bank and compared with the SpHex by pairwise alignment using MaxHom (Table I). Thus, not only did SpHex have the substrate specificity of a ␤-N-acetylhexosaminidase, its primary amino acid sequence was most similar to those glycosyl hydrolases that are members of family 20 (chitobiases and ␤-N-acetylhexosaminidases) (9, 10).
Multiple sequence alignment (ClustalW1.7) of the 15 family SpHex are predicted to be the substrate-binding residue and catalytic acid residue, respectively. The conservation of these two residues suggested that the catalytic mechanism may be conserved between SpHex and SmChb. To help substantiate this hypothesis, a three-dimensional model of SpHex was constructed by comparative molecular modeling. Comparative Molecular Modeling-A pairwise alignment between the SmChb and SpHex using MaxHom revealed a sequence identity of 30%. When the alignment was restricted to SmChb domain 3, an ␣/␤-barrel structure that contains the enzyme active site on the C-terminal end, the sequence identity was 45%. Domain 3 is the largest of four SmChb domains. Alignment of SpHex residues Pro 149 -Pro 498 with domain 3 encompassed 350 residues (63%) of the total amino acid sequence of SpHex.
Although the multiple alignment of all 15 family 20 enzymes, including SpHex, revealed high conservation for specific residues predicted to be involved in catalysis, SmChb, Vibrio harveyi chitobiase, Alteromonas sp. ␤-N-acetylhexosaminidase, and Vibrio vulnificus ␤-N-acetylhexosaminidase contained many large insertions not present in the other family 20 members. Using the x-ray structure of SmChb, the insertions in this four-member subfamily were found localized to loop regions that were absent in the other family members, including human HexA and SpHex. Alignment of the enzymes listed in Table I, which exclude this subfamily of sequences with the exception of the SmChb, greatly improved the multiple sequence alignment (Fig. 2). All deletions in the SpHex model localized primarily to surface loop structures of SmChb and did not disrupt secondary structural elements composing the ␣/␤-barrel.  Fig. 3, A and  B. PROCHECK analysis before and after energy minimization demonstrated that energy minimization dramatically improved the overall geometry and relaxed most of the bad contacts within the model. Of the 10 bad contacts remaining, none occur within the core of the model; only one contact, between the last two residues in the model, is less than 2.5 Å (Leu 497 C 3 Pro 498 N, 1.3 Å). A Ramachandran map indicated that 94.3% of the residues had , angles within core and allowed regions, 3.5% had , angles within generously allowed regions, and 2.1% (6 residues) had disallowed , angles. The residues with disallowed , angles (Val 214 , Leu 255 , Ala 369 , Leu 404 , Ala 433 , Thr 474 ) all reside in loop structures on the outer surfaces of the model, primarily at locations spliced together to compensate for deletions.

Analysis of the SpHex Model-The final energy-minimized model of SpHex is shown as a ribbon diagram in
The modeled SpHex is similar in structure to SmChb, forming an eight-stranded ␣/␤-barrel where the seventh ␣-helix of the barrel has been replaced by an extended loop structure (Fig. 3, A and B). ␣-helices surrounding the hydrophobic core of SpHex contain hydrophobic and hydrophilic residues that pack against the hydrophobic enzyme core and contact the solvent, respectively. Interestingly, the model predicts a significantly more compact ␣/␤-barrel structure for SpHex as opposed to SmChb. The compact ␣/␤-barrel is also seen for theoretical models constructed for human HexA and -B (9). Because SmChb contains many extended loop regions as well as large domains that are not encoded by human HEXA, HEXB, or the gene encoding SpHex, it appears that the tertiary structure of HexA and -B may be significantly more similar to SpHex than SmChb.
Because the chitobiase x-ray structure was solved in complex with chitobiose, we were able to build our model complexed with the same substrate (Fig. 3, A and B). Of the natural substrates known to be hydrolyzed by family 20 enzymes, chitobiose is probably the most analogous to 4-MUG. It is therefore likely that chitobiose and 4-MUG complex with the SpHex in a similar fashion; thus, our kinetic data, using 4-MUG as substrate (see below), should directly support the comparative modeling results.
As expected, the active site architectures of the SpHex model and the SmChb are highly conserved. The salient features, as identified in SmChb (9), include Glu 314 (SmChb, Glu 540 ), Arg 162 (SmChb, Arg 349 ) (Fig. 3C), and three Trp residues Trp 344 , Trp 442 , and Trp 408 (Fig. 3D). The SpHex model indicates that the protonated carboxyl group of Glu 314 is within hydrogen bonding distance (2.5 Å) of the oxygen in the glycosidic linkage of the substrate. Thus, Glu 314 is perfectly situated to donate its proton to the glycosidic oxygen, thereby aiding in the hydrolysis of the substrate. The ␦-guanido group of Arg 162 appears to directly bind the substrate by forming two hydrogen bonds, each at a distance of 2.6 Å, to OH-3 and OH-4 of the nonreducing N-acetylglucosamine sugar of the chitobiose. As described for the SmChb x-ray structure, the active site model of SpHex appears to lack an appropriate residue that could act as the nucleophilic base in an acid/base reaction for this enzyme. As described for the SmChb catalytic mechanism (9), the nucleophile involved in the SpHex catalytic mechanism may be provided by the substrate itself. This is evidenced by a particular substrate-enzyme hydrogen bond, which is predicted to occur in both SmChb and SpHex only when the substrate is twisted into an energetically unfavorable conformation as described below.
A notable hydrogen bond occurs between the acetamido-O-6 of the nonreducing sugar of the chitobiose substrate and the OH of the phenolic ring on Tyr 393 of SpHex (SmChb, Tyr 683 ) (Fig. 3C). This hydrogen bond would only occur when the chitobiose is in the energetically unfavorable 4-sofa conformation as described for the SmChb mechanism by Tews et al. (9). The twisting of chitobiose into a 4-sofa conformation is proposed to be essential for the substrate assisted catalytic mechanism for it allows the acetamido-O-7 atom of the nonreducing N-acetylglucosamine to move within 3.0 Å of the C-1 atom of the same sugar, an ideal position for the nucleophilic base (9). This twisting of the chitobiose may be partially stabilized by hydrogen bonding to Tyr 393 in SpHex, as it is by the equivalent residue (Tyr 683 ) in SmChb. To see a similar hydrogen bonding interaction occurring in the SpHex model provides additional support for the catalytic mechanism of SmChb and helps validate the quality of the model. The active site pocket of our model contains Trp residues that pack against the hydrophobic hexose rings of the chitobiose substrate (Fig. 3D). The indole ring of Trp 442 is parallel to the hexose ring of the nonreducible N-acetylhexosamine of the chitobiose substrate, exactly as is seen the in SmChb x-ray structure. Although Trp 408 may provide a hydrophobic environment for the reducible N-acetylglucosamine sugar of the substrate, the indole ring of this Trp is not parallel to the hexose ring of the sugar as is seen in the SmChb x-ray structure.
Enzyme Purification and Digestion with Factor Xa Protease-The pmHEX-1.8 constructs of wild type and mutant SpHex produced high level fusion protein expression in E. coli, yielding approximately 0.5 mg/100 ml culture. The final protein concentrations for all purifications were brought to 1-2 mg/ml by eluting the fusion protein with 0.5 ml of elution/storage buffer. Purified wild type and mutant SpHex-MBP fusion protein was stable for at least 48 h at room temperature and at least 2 weeks at 4°C.
The purified SpHex-MBP fusion protein was digested with factor Xa protease to assess the effects of the MBP on enzyme activity. There was no difference in the pH optimum or K m of undigested or digested enzyme (data not shown). Western blot analysis using the anti-SpHex-MBP antibody indicated that factor Xa cleaved the SpHex-MBP fusion protein into a 55-kDa protein predicted to be recombinant SpHex and the 43-kDa MBP (Fig. 1, lane 3). Anti-SpHex-MBP antibody also detected a 55-kDa protein in the cytosol of S. plicatus (Fig. 1, lane 4), indicating that the recombinant and native SpHex are of the same approximate molecular mass. Two higher molecular weight proteins were also detected in the S. plicatus cytosol; they are thought to be a result of nonspecific cross-reactivity of the antibody to other proteins in the fraction. Interestingly, SpHex was detected in the spent medium of S. plicatus cultures after approximately 5 days of growth, suggesting that S. plicatus may secrete this enzyme into the medium upon expression (Fig. 1, lane 5).
Kinetic Analysis of Wild-type and Mutant SpHex-Based upon multiple sequence alignments of family 20 enzymes, previous mutagenesis studies of human HexA (44 -47) and the SmChb structure (9), we created the following mutations in SpHex: Arg 162 3 His, Asp 246 3 Asn and Glu 314 3 Gln. Since the kinetics of the SpHex-MBP fusion and factor Xa-digested fusion protein were indistinguishable from each other when using 4-MUG as substrate, kinetic analysis was restricted to the SpHex-MBP fusion protein. The resulting kinetic analyses (Table II) are consistent with our predicted structure of SpHex (Fig. 3, C and D). The model of SpHex predicts that the Arg 162 3 His mutation would significantly affect the substrate binding capacity of the enzyme (Fig. 3C). Indeed, this mutation increased the K m 40-fold as compared with wild type SpHex. The observed 5-fold decrease in the V max observed for this mutation probably reflects the inability of the enzyme to efficiently dock the substrate. By virtue of conservation, not only does this mutation provide direct biochemical evidence to support the predicted role of Arg 349 in SmChb, it is the first example of this conserved Arg acting as a substrate-binding residue. This same mutation is associated with the B1 variant form of Tay-Sachs disease in humans, but previous studies of the mutation in the human enzyme showed that the equivalent Arg residue in human HexA is not involved in substrate bind-ing (46,47). The discrepancy may reflect the difficulties in obtaining accurate kinetic data from the human enzyme.
The proposed catalytic acid residue Glu 314 , when mutated (Glu 314 3 Gln), reduced V max 296-fold, and the K m of this mutant was reduced 7-fold, implying that Glu 314 is the proton donor in the catalytic reaction of SpHex. The 7-fold decrease in K m suggests this mutant may saturate with substrate more readily than the wild type SpHex. Our kinetic data and modeling of SpHex, the SmChb structure (9), theoretical models of human HexA and -B (9), and recent mutagenesis studies of human HexA and -B (10,11) are consistent with the Glu residue acting as the proton donor. Sequence and structural similarities support the idea that the catalytic mechanism within family 20 is conserved.
In SmChb, Asp 448 appears to indirectly hydrogen-bond with the chitobiose substrate through a water molecule and Asp 539 . The equivalent residue in SpHex is Asp 246 . Comparative modeling suggests that Asp 246 may fulfill a similar role in this enzyme. The mutation Asp 246 3 Asn results in a 2-fold decrease in V max and a 2-fold increase in K m . In SpHex, Asp 246 most likely forms an indirect hydrogen bond with the substrate through Glu 313 . Tews et al. (9) suggested that the analogous Asp residue in chitobiase (Asp 448 ) also forms a hydrogen bond with the substrate through a water molecule situated in a pocket between this residue and the substrate. Unfortunately, our model does not predict the positions of water molecules within the structure. However, a similar sized pocket exists within the SpHex model, where a 6.0-Å distance between O2A of the nonreducible N-acetylhexosamine of the chitobiose and O2D of Asp 246 exists. This 6.0-Å distance potentially allows for a water molecule to situate itself between Asp 246 and the substrate, creating a similar hydrogen bonding pattern as seen for Asp 448 in the SmChb. In light of the modeled structure of SpHex, it is not surprising to observe that the Asp 246 3 Asn mutation has a less substantial effect upon the kinetics than do the Arg 159 3 His and Glu 314 3 Gln mutations.
pH Profile Analysis-The pH profiles of wild type and mu- FIG. 4. Analysis of pH profiles for wild type and mutant SpHex. Catalytic activity for wild type and mutant SpHex was measured in 10 mM Na 2 HPO 4 , 6 mM citric acid, 0.3% bovine serum albumin ranging from pH 2.0 to 9.0 using 4-MUG as a substrate as described under "Experimental Procedures." After a 45-min incubation at 25°C, each reaction was quenched with 0.1 M glycine-carbonate buffer, pH 10, and the fluorescent product, 4-MU, was measured by fluorometry. Activity is indicated as an average percentage of maximal activity (n ϭ 3).

TABLE II
Kinetic studies of recombinant SpHex Recombinant wild type and mutant SpHex-MBP fusion protein activities were determined by incubating a reaction mix containing 1-75 ng of affinity-purified SpHex-MBP fusion protein with 12 concentrations of 4-MUG (0.033-5.0 mM) in citrate/phosphate buffer, 0.3% bovine serum albumin, pH 3.0, in 30 l at 25°C for 45 min. The K m and V max values were determined using the direct linear plot method (29 tant SpHex can be seen in Fig. 4. The pH optimum of wild type SpHex was found to be 3.0. Although the pH optimum of this enzyme is quite broad, an environment more acidic than pH 3.0 resulted in a steady loss of activity. Interestingly, the Glu 314 3 Gln mutation changed the pH profile from that of a broad bell-shaped curve to more a hyperbolic shaped pH curve with no distinctive peak in the pH profile. According to our model and Tews et al. (9), Glu 314 must be protonated in order for catalysis to occur. This is reflected in the acidic pH optimum seen for wild-type SpHex. Gln does not undergo ionization; hence, the loss of a distinct peak at pH 3.0 suggests that the ionization state of Glu 314 is responsible for the pH dependence seen for the wild-type enzyme. The Asp 246 3 Asn mutation shifted the pH optimum from 3.0 to 3.75. This shift contradicts results from human HexA studies that showed that the analogous mutation in HexA resulted in a more acidic pH optima (48). The reason for the pH optimum shift could not be explained. Finally, the Arg 162 3 His mutation had a dramatic affect on the pH optimum for the SpHex, shifting it from pH 3.0 to pH 6.5. No clear explanation could be identified for this change; the drastic nature of this pH shift may suggest that the mutation has allowed for an alternative catalytic mechanism.
Our studies conclude that the SpHex belongs to the family 20 glycosyl hydrolases and that this family may have a conserved catalytic mechanism. With our theoretical modeling and the ease with which this enzyme can be genetically manipulated and expressed, there exists the potential for complex structure/ function analysis to be carried out. With the apparent structural and kinetic conservation within family 20 enzymes, future studies of SpHex may provide an alternative and potentially more revealing method of understanding mechanisms involved in other family 20 enzymes including human HexA and -B.