Identification of candidate active site residues in lysosomal β-hexosaminidase A

The β-hexosaminidases (Hex) catalyze the cleavage of terminal amino sugars on a broad spectrum of glycoconjugates. The major Hex isozymes in humans, Hex A, a heterodimer of α and β subunits (αβ), and Hex B, a homodimer of β subunits (ββ), have different substrate specificities. The β subunit (HEXB gene product), hydrolyzes neutral substrates. The α subunit (HEXA gene product), hydrolyzes both neutral and charged substrates. Only Hex A is able to hydrolyze the most important natural substrate, the acidic glycolipid GM2 ganglioside. Mutations in the HEXA gene cause Tay-Sachs disease (TSD), a GM2 ganglioside storage disorder. We investigated the role of putative active site residues Asp-α258, Glu-α307, Glu-α323, and Glu-α462 in the α subunit of Hex A. A mutation at codon 258 which we described was associated with the TSD B1 phenotype, characterized by the presence of normal amounts of mature but catalytically inactive enzyme. TSD-B1 mutations are believed to involve substitutions of residues at the enzyme active site. Glu-α307, Glu-α323, and Glu-α462 were predicted to be active site residues by homology studies and hydrophobic cluster analysis. We used site-directed mutagenesis and expression in a novel transformed human fetal TSD neuroglial (TSD-NG) cell line (with very low levels of endogenous Hex A activity), to study the effects of mutation at candidate active site residues. Mutant HEXA cDNAs carrying conservative or isofunctional substitutions at these positions were expressed in TSD-NG cells. αE323D, αE462D, and αD258N cDNAs produced normally processed peptide chains with drastically reduced activity toward the α subunit-specific substrate 4MUGS. The αE307D cDNA produced a precursor peptide with significant catalytic activity. Kinetic analysis of enzymes carrying mutations at Glu-α323 and Asp-α258 (reported earlier by Bayleran, J., Hechtman, P., Kolodny, E., and Kaback, M. (1987) Am. J. Hum. Genet. 41, 532-548) indicated no significant change in substrate binding properties. Our data, viewed in the context of homology studies and modeling, and studies with suicide substrates, suggest that Glu-α323 and Asp-α258 are active site residues and that Glu-α323 is involved in catalysis.

The ␤-hexosaminidases (Hex 1 , EC 3.2.1.52) are lysosomal hydrolases that catalyze the cleavage of terminal ␤-N-acetylglucosamine or ␤-N-acetylgalactosamine residues on a broad spectrum of glycoconjugates. The major Hex isozymes in humans are: Hex A, a heterodimer composed of one ␣ and one ␤ subunit and Hex B, a homodimer of two ␤ subunits. A third isozyme, Hex S, is composed of two ␣ subunits, which are unstable and not normally found in most tissues. The ␣ and ␤ subunits are structurally related, sharing 60% amino acid identity in the mature form (1,2). Both subunits are catalytically active with different but overlapping substrate specificities (3). The ␤ subunit, in Hex A and Hex B, hydrolyzes neutral substrates, whereas the ␣ subunit, in Hex A and Hex S, hydrolyzes neutral substrates as well as substrates bearing a negative charge either on the terminal sugar (e.g. GlcNac-SO 4 ) or on a distinct residue (e.g. sialic acid) (4). The latter includes the most important natural substrate, the sialic-acid containing glycosphingiolipid, GM 2 ganglioside, found mainly in neuronal tissue. Only Hex A catalyzes cleavage of the terminal ␤-Nacetylgalactosamine on GM 2 ganglioside in the presence of the substrate-specific protein cofactor, the GM 2 -activator protein (5).
Mutations in the HEXA, HEXB, and GM2A genes, encoding the ␣ and ␤ subunits of Hex A and the GM 2 activator, respectively, lead to a group of inherited neurodegenerative diseases, collectively known as the GM 2 gangliosidoses, that are characterized by lysosomal accumulation of GM 2 ganglioside mainly in neuronal tissue. These disorders range in severity from Tay-Sachs disease (TSD), a progressive and fatal neurodegenerative disorder of infancy, to clinically milder or later onset forms of GM 2 gangliosidosis occurring in patients with some residual enzyme activity (reviewed in Gravel et al. (6)).
Although more than 70 mutations at the human HEXA locus, and 14 at the HEXB locus have been described (6), few have revealed information about the location or properties of the ␣ and ␤ subunit active sites. A subset of HEXA mutations, known as the B1 mutations, lead to production of normal amounts of mature Hex A, which is deficient in ␣ subunit catalytic activity without affecting ␤ subunit activity (7). The B1 mutations are thus compatible with normal maturation of Hex subunits and delivery of structurally intact Hex A to the lysosome. The presence of a mature Hex A, unable to hydrolyze charged substrates, suggested that the B1 biochemical pheno-type might be associated with mutations at or near the ␣ subunit active site (4,7,8). The first B1 mutations described were ␣R178H and ␣R178C (7)(8)(9). In vitro generated mutation at the homologous site to Arg-␣178 in the ␤ subunit, Arg-␤211, resulted in production of mature Hex B devoid of catalytic activity (10). More recently, we identified a third B1 mutation, ␣D258H (11). Both Arg-␣178 and Asp-␣258 have been proposed as candidates for participation at or near the active site of the ␣ subunit.
The mechanism of cleavage of glycosidic bonds by Hex remains unclear. Sinnott (12) has proposed that glycosyl hydrolases employ an acid-catalysis mechanism involving the participation of two acidic residues (a proton donor and a nucleophile) at the active site, as identified in glycoside cleavage by ␤-glucosidases (13)(14)(15)(16), ␣-glucosidases (17), ␤-glucanases (18 -20), ␤-galactosidases (21)(22)(23), and chitinases (24,25) and the ␤-Nacetylglucosaminidase (chitobiase) (26). In contrast, based on inhibition studies using nitrogen-containing substrate inhibitors, Legler and Boothagen (27) and Legler et al. (28) proposed that the mechanism of Hex catalysis involves both an acidic and a basic residue and depends on a transition state in which the glycosidic O of the substrate is joined with the acetyl group of the pyranose ring, a mechanism fundamentally different from that of other glycosidases.
In this study, we investigated candidate active site residues in the ␣ subunit of Hex A through analysis of mutations that affect catalytic activity without disrupting maturation of the enzyme. We introduced conservative mutations at residue Asp-␣258, as well as at three glutamic acid residues (Glu-␣307, Glu-␣323, and Glu-␣462) which are evolutionarily invariant in family 20 of glycosyl hydrolases using the classification system of Henrissat (29). We then studied the impact of these substitutions on enzyme activity and maturation after transfection of mutant cDNAs using an SV40-transformed neuroglial (NG) cell line established from a fetus with Tay-Sachs disease which is devoid of endogenously expressed ␣ subunits but produces functional ␤ subunits.

EXPERIMENTAL PROCEDURES
Cell Culture-NG141 (NG) and NG125 (TSD-NG), SV40-transformed fetal TSD and normal neuroglial cell lines, respectively, were provided by L. Hoffman and S. Brooks (Kingsbrook Jewish Medical Center, Brooklyn, NY). Cells were cultured in ␣-minimum essential medium with 15% fetal calf serum and antibiotics.
Recombinant Plasmids-The Escherichia coli ␤-galactosidase gene, in pSVL␤gal (Clontech), was used as a reporter gene to control for transfection efficiency. The cloned human HEXA cDNA (1) was inserted into pRCCMV and pREP4 (Invitrogen) to produce pRCCMV␣ and pREP4␣. Preparation of PRCCMV␣ involved: 1) subcloning the HEXA cDNA from PSVL␣ into pCRII (Invitrogen) as an XhoI/BamHI fragment, 2) subcloning of an NsiI HEXA cDNA-containing fragment PCRII subclone into the PstI site of pBluescript (Stratagene), 3) subcloning an XhoI/XbaI fragment from pBluescript into a PCRII intermediate, and 4) cloning a NotI/XbaI HEXA cDNA-containing fragment from the PCRII subclone into PRCCMV. Preparation of pREP4␣ involved cloning the XhoI/BamHI HEXA cDNA-containing fragment from PSVL␣ into pREP4. Construction of pCMV␣ required creating a HEXA cDNA flanked by NotI sites. The HEXA cDNA insert in pRCCMV␣ has a NotI site at the 5Ј end of the gene. This insert was subcloned into the BamHI site of pBluescript (KSϪ, Stratagene) introducing a second NotI site downstream from the 3Ј end of the gene. This step permitted the cloning of the insert into the unique NotI site of pCMV␣. All plasmids were purified on Qiagen columns prior to transfection.
Site-directed Mutagenesis-Mutations were introduced into the HEXA cDNA using a modified protocol of the Clontech Transformer Mutagenesis Kit. The second screening step for mutant clones was omitted. Mutant clones were identified after the first screen by PCR amplification of samples of isolated disrupted bacterial colonies followed by restriction enzyme digestion or allele-specific hybridization of amplified product to identify mutant genotypes. Mutations were introduced into pSVL␣, a cassette containing the altered sequence was subcloned into pBS(KS)HEXA, and the full-length mutant cDNA was subsequently subcloned into pCMV. Mutant pCMV␣ inserts were sequenced (Pharmacia T7 sequencing kit) and the plasmids purified on Qiagen columns prior to transfection.
Transfection-For transient expression in cell line NG 125, subconfluent T175s (175 cm 2 , Sarstedt) were harvested by trypsinization and washed twice with 1 ϫ phosphate-buffered saline. The cell pellet was resuspended in Optimem medium (Life Technologies, Inc., containing 5% fetal bovine serum) to obtain a final concentration of 6 ϫ 10 6 cells/ml. Cell suspension (800 l), 20 g of pCMV␣, and 2 g of pSVL␤gal were placed in a 0.4-cm cuvette, mixed, placed on ice for 5 min, and pulsed (500 microfarads, 400 V) using a Bio-Rad electroporation apparatus. The time constant was between 16 and 18 ms. After 15 min on ice, 800 l of ␣-minimum essential medium (without antibiotics) was added to the suspension. For the ␤-galactosidase qualitative assay, 250 l of transfected cell suspension were plated on 12 multiwell plates. The remainder of the transfected cell suspension was grown for 48 h in a T75 (75 cm 2 ).
Enzyme and Protein Assays-A qualitative ␤-galactosidase assay was performed to determine the percentage of surviving cells which expressed bacterial enzyme. Multiwell-plated cells were incubated as described by Lake (30). After 24-h incubations the number of blue cells was estimated microscopically. Harvested cells were lysed by freezethawing in 0.25 M Tris-HCl (pH 7.4) and protein determined by the Bradford method (Bio-Rad). A fluorescent assay (31) was adapted for quantitation of ␤-galactosidase activity in transfected cell lysates. The reaction mixture contained 3 l of 100 ϫ magnesium solution (4.5 M 2-mercaptoethanol, 0.1 M MgCl 2 ), 100 l of 0.5 mM 4 methyl-umbelliferyl ␤-D-galactoside, and approximately 2-5 g of lysate protein in 0.1 M sodium phosphate buffer (pH 7.5) in a volume of 334 ml. After incubation at 37°C for 15 min, fluorescence, due to release of 4MU, was determined using a Perkin-Elmer spectrofluorimeter (excitation wavelength, 360 nm; emission, 447 nm). Hexosaminidase activity was also determined fluorometrically using either 4MUGS (␣ subunit substrate) (32) or the 4MUG (␤ subunit) substrate.
Western Blot Analysis-The enhanced chemiluminescence (ECL) Western blotting kit from Amersham Corp. was used to detect the presence of the Hex A ␣ subunit with a polyclonal rabbit anti-human Hex A antibody. Both the primary and secondary (rabbit Ig, horseradish peroxidase-labeled antibody) antibodies were used at a 1:5000 dilution.
Chromatofocusing of Transfected Cell Extracts-The hexosaminidase isoenzyme profile in cell lysates was determined by chromatofocusing using the Pharmacia Polybuffer Exchanger (PEB) system according to a modified protocol of O'Dowd et al. (33). Transfected cell extracts were freeze-thawed (three times) in 0.025 M imidazole buffer (pH 7.4), and the protein concentration was determined by the Bradford method. All steps were carried out at 7°C. Approximately 1 ml of PEB74 slurry was used to prepare a column in a 1-ml syringe. The column was washed with the equivalent of 1.5-2 ϫ bed volume with 0.025 M imidazole. The protein extract was added after passing 2 ϫ 100 l of Polybuffer (pH 4.0) through the column. Four hundred-microliter fractions were collected. Sodium citrate buffer (0.13 M, pH 3.46) was used to elute Hex S after a pH of 4.0 was reached using a pH gradient.
Kinetic Analysis-The K m and K i for wild type and mutant Hex A proteins were determined using 4MUGS substrate at concentrations from 1.0 to 7.5 mM. For inhibition studies, the competitive inhibitor N-aetylglucosamine-6-phosphate (Sigma) was added to a final concentration of 10 mM.

RESULTS
Expression of the HEXA Gene in TSD-NG Cells-The endogenous activity of NG cells toward the ␣-specific substrate 4MUGS was 715 Ϯ 13 nmol/mg/h. In untransfected TSD-NG cells, the rate of 4MUGS hydrolysis is Ͻ1% of that in NG cells. This trace activity is probably due to the residual action of Hex B on 4MUGS (32).
In order to maximize ␣ subunit expression through transfection, several vectors carrying the HEXA gene cDNA were assessed for their ability to drive expression of enzymatic activity when transfected into TSD-NG cells. Transfection efficiency (percent surviving cells catalyzing 5-bromo-4-chloro-3-indoyl ␤-D-galactoside hydrolysis) was determined to be 10 -20%. Hexosaminidase activity in cells transfected with 20 g of plasmids pSVL␣, pRCCMV␣, pREP4␣, or pCMV␣ was at least 10-fold higher than activity in mock-transfected cells ( Table I) (Table I) showed that activity continued to increase throughout the 72-h period. The pCMV␣ plasmid expressed over a 48-h incubation period post-transfection was selected for all subsequent experiments.
Western blot analysis of transfected (pCMV␣ and pSVL␤gal) and mock-transfected (pSVL␤gal) cell extracts confirmed that the increase in 4MUGS activity in TSD-NG cells is associated with ␣ subunit expression (Fig. 1, lane N). Both precursor and mature ␣ subunit are absent in mock-transfected TSD-NG cell extracts (Fig. 1, lane M) and present in cells transfected with pCMV␣.
In order to determine whether the ␣ subunits encoded by the HEXA cDNA were expressed as the heterodimeric enzyme Hex A (␣␤) or the homodimeric species Hex S (␣␣), the Hex isozyme forms in transfected TSD-NG cell lysates were resolved by chromatofocusing. Fig. 2a illustrates the chromatofocusing profile of TSD-NG cells transfected with pCMV␣ and assayed with 4MUG (all Hex) and 4MUGS (Hex A and Hex S) to detect all Hex isozyme species. All three hexosaminidase isoenzymes were present, and eluted at their expected pIs. Hex S was the isoenzyme present in greatest abundance followed by Hex A and Hex B respectively. This high expression of the nonphysiological Hex S in TSD-NG cells is the most likely due to the massive overexpression of ␣ subunits in the face of limiting, endogenously produced ␤ subunits. Total Hex activity in pCMV␣ transfected TSD-NG cells (harvested at 48 h posttransfection) is 5-6-fold greater than endogenous Hex activity in untransfected normal NG cells (Table I). Given an efficiency of transfection of 10 -20%, the transfected cells express up to 60-fold greater Hex activity than normal untransfected NG cells. Despite overwhelming ␣ subunit synthesis, a significant proportion (22%) of the transfected gene product is expressed as the Hex A heterodimer as shown by the chromatofocusing profile (Fig. 2).
Expression of HEXA Mutations-In order to evaluate the expression of mutant ␣ subunits in TSD-NG cells, we initially examined mutations known to cause the infantile acute (␣R170W) (11) subacute (␣G250D) (35) and chronic (␣G269S) (37) forms of GM 2 gangliosidosis. The mutant cDNAs were expressed in TSD-NG cells (Table II) and cell lysates assayed for Hex A specific activity using 4MUGS. Hex A activity in lysates of ␣G269Sand ␣G250D-transfected TSD-NG cells was Ͻ4% and 1.5%, respectively, of the activity measured after transfection with wild type pCMV␣. These Hex A activities were 9-and 4-fold greater than mock-transfected activity measured with the 4MUGS substrate. In contrast, expression of ␣R170W exhibited no significant 4MUGS activity above that of mock transfected cells. Western blot analysis showed that the ␣ subunit synthesized by all three mutant cDNAs appeared in the precursor form (Fig. 1, lanes A, J, and I). Only the ␣G269S mutation appeared to be associated with expression of a small amount of mature ␣ subunit (Fig. 1, lane A). These results are compatible with previous findings on ␣G269S (34) and ␣G250D (35) in COS cell expression studies and demonstrate complete inactivation of Hex A by the ␣R170W mutation consistent with the infantile TSD phenotype. In the latter case, the R170W substitution is associated with expression of an ␣ subunit precursor (Fig. 1, lane I) but not with its maturation and targeting to the lysosome.
Analysis of Putative Active Site Residues-We next examined the expression in TSD-NG cells of mutations at amino acid residues which are candidates for participation in the Hex A ␣ subunit catalytic site. To assess the role of mutation at residue Asp-␣258 (␣D258H), first described in association with the B1 biochemical phenotype of TSD, HEXA cDNA constructs carrying the conservative or isosteric substitutions ␣D258H, ␣D258E, or ␣D258N were expressed in TSD-NG cells and Hex A activity measured in transfected cell lysates. All three mutations resulted in negligible or trace amounts of Hex A activity, with ␣D258N exhibiting 2-fold background and the others the same activity as measured in mock-transfected cells (Table  III, Experiment 1). Western blot analysis of transfected cell extracts (Fig. 3), revealed that maturation of pro-␣ chains carrying the ␣D258H mutation was reduced. In contrast, expression ␣D258N and ␣D258E mutant cDNAs resulted in production of mature ␣ subunit. Chromatofocusing of ␣D258N cell extracts revealed that this mutation significantly reduced 4MUGS hydrolysis by Hex A and Hex S (data not shown). The 4MUG hydrolysis of Hex A was reduced to a lesser extent. This result was expected since 4MUG hydrolysis is catalyzed by both the ␣ and ␤ subunits of Hex A.
We then studied the effects of mutation at conserved residues Glu-␣307, Glu-␣323, and Glu-␣462, predicted as candidate active site residues by homology studies and hydrophobic cluster analysis of family 20 glycosyl hydrolases. Fig. 4, a and  b, illustrate multiple alignment of the two regions of family 20 enzymes which produce significant alignment. Asterisks indicate the three evolutionarily invariant glutamate residues identified as candidate active site residues using the classification system of Henrissat. Other invariant amino acids are also indicated.
HEXA cDNA constructs carrying the conservative substitutions ␣E307D, ␣E323D, or ␣E462D were expressed in TSD-NG cells and Hex activity measured in transfected cell lysates. All three substitutions had dramatic effects on catalytic activity (Table III, Experiment 2). ␣E323D did not exhibit any 4MUGS activity above background whereas ␣E462D and ␣E307D exhibited 14-and 60-fold 4MUGS activity above the background activity obtained in mock transfected cells respectively. West-  ern blot analysis of transfected cell extracts showed that only the ␣E323D and ␣E462D mutations were compatible with synthesis of significant amounts of precursor and mature ␣ subunit (Fig. 5). Chromatofocusing of ␣E323D and ␣E462D cell extracts confirmed that the Hex A isoenzyme is processed normally but has a catalytically defective ␣ subunit that is unable to hydrolyze 4MUG or 4MUGS (Fig. 2, b and c). The ␣E307D mutation, however, dramatically reduces ␣ subunit maturation.
Kinetic experiments were performed to evaluate the impact of the ␣E323D mutation on the binding affinity of Hex A for the substrates 4MUGS and for the competitive inhibitor N-acetylglucosamine-6-PO 4 (Fig. 6). The Lineweaver-Burke plots for the wild type enzyme are monophasic, representing the contribution of a single active site to the hydrolysis of 4MUGS. In contrast, the kinetics of hydrolysis of this substrate by the Hex A ␣E323D enzyme are biphasic revealing a significant contribution of a second active site (the ␤ subunit) at high substrate concentrations. When the kinetic parameters of the high affinity site are evaluated the K m for 4MUGS is 2.5 mM for Hex A ␣E323D compared to 1.7 mM for the wild type enzyme. Similiarly K i values for the competitive inhibitor are 15.6 mM for the ␣E323D-substituted Hex A and 11.0 mM for wild type Hex A. We showed, previously, that the K m for Hex A in fibroblasts from a patient carrying the D258H allele (and a null allele) was identical to that of wild type enzyme (7). DISCUSSION ␤-Hexosaminidase catalyzes the hydrolysis of its glycoside substrates by a "retaining" mechanism in which the anomeric configuration of the substrate remains unchanged (12,36). Most retaining glycosidases employ an acid catalysis mechanism in which two amino acids function respectively as proton donor and nucleophile (12). Alternatively, glycosaminidases may employ a second mechanism, termed substrate-assisted catalysis, in which an acidic residue and a basic residue par-   ticipate in substrate cleavage. In this study, we attempted to identify acidic residues in the Hex A ␣ subunit which may participate in the active site of Hex A.
Since our studies involved the use of a novel TSD-NG expression system, we first evaluated expression in this system of wild type and mutant HEXA cDNAs. Transient expression of mutant HEXA cDNAs in COS1 cells (reviewed in Brown and Mahuran (34)) has several disadvantages which are overcome in the TSD-NG expression system. COS1 cells exhibit a high level of endogenous Hex A and Hex B activity which compromises the analysis of low levels of Hex activity expressed by mutant HEXA cDNAs. Furthermore, because the cells are of heterologous origin, human HEXA gene mutations are expressed in COS1 cells as the non-physiological isozyme Hex S (7,37,38), an ␣␣ homodimer which differs from Hex A both in stability and substrate specificity. As a result, the activity of mutant ␣ subunits produced following transfection of mutant HEXA cDNAs, does not correlate well with clinical severity of disease or biochemical phenotypes observed in patients (34,35).
In contrast, TSD-NG cells, produce no endogenous ␣ subunits. Endogenous expression of human ␤ subunits in these cells, however, permits dimerization with transfected human ␣ subunits, to form active Hex A. When wild type ␣ subunits are overexpressed in TSD-NG cells, ␤ subunit concentration limits the formation of Hex A. As the ␤ subunits become depleted, excess ␣ subunits dimerize to form Hex S, distinguishable from Hex A by chromatofocusing. The presence of both the precursor (67 kDa) and mature (54 kDa) forms of ␣ subunit following transfection confirms that newly synthesized ␣ subunit can be normally processed and targeted to the lysosome, but also indicates that excess ␣ subunits are poorly processed in the absence of ␤ subunits.
We evaluated the expression in TSD-NG cells of ␣ subunits carrying mutations associated with acute (␣R170W), subacute (␣G250D), and chronic (␣G269S) forms of GM 2 gangliosidosis, respectively, and demonstrated that residual enzyme activity was inversely correlated with disease severity. Results for the ␣R170W mutation confirmed our previous report (11), that this mutation causes the classical infantile form of TSD. We proceeded to use this expression system to evaluate the role of candidate active site residues.
The four acidic amino acids we examined were chosen on the basis of (a) occurrence in the B1 variant of Tay-Sachs disease (Asp-␣258) or (b) conservation in family 20, a sequence related family of glycosyl hydrolases (␣307, ␣323, and ␣462). The TSD-B1 phenotype is believed to be caused by mutations which lead to substitution of residues with catalytic function (4,7,10). These include the basic residue Arg-␣178 (4, 8 -10) (which is invariant in family 20 enzymes) and the acidic Asp-␣258 (7,10). Conservative substitutions at these positions have little affect on K m but dramatically reduce the V max of Hex A (7). The residues Glu-␣307, Glu-␣323, and Glu-␣462 were selected on the basis of their evolutionary invariance in family 20 glycosyl hydrolases (39), a classification system (29) based on amino acid sequence similarity. Members of homology families generally share folding characteristics and catalytic residues within families are strictly conserved (40). The three glutamate residues are invariant at homologous positions in all members of family 20. The position corresponding to Asp-␣258 shows variation in a single enzyme which carries the functionally related acidic residue glutamate at this position.
Analysis of residual enzymatic activity and ␣ subunit maturation in cells transfected with HEXA cDNAs carrying conserv- ative or isosteric substitutions at residues Glu-␣307 and Glu-␣462 led us to conclude that these residues are unlikely to have catalytic functions. Residual Hex A activity associated with both mutations was too high to be compatible with the loss of either an acid catalyst or a proton donor. Furthermore, the ␣E307D substitution prevented ␣ subunit maturation (Fig. 5). In contrast, both the isosteric substitution ␣D258N and the conservative substitution ␣E323D were associated with production of mature Hex A with drastically reduced enzyme activity. Since maturation of the ␣ subunit is dependent on dimerization with the ␤ subunit, we were also able to conclude that the ␣D258H and ␣E323D mutations are compatible with production of normal amounts of ␣␤ heterodimer and that loss of activity is not the result of compromised dimerization. Furthermore, kinetic analysis of residual Hex A activity in ␣E323D transfected cells (Fig. 6) or ␣D258H in patient fibroblasts (7) demonstrates that loss of enzymatic activity is the consequence of decrease in k cat rather than a change in K m .
Our proposal that Asp-␣258 and Glu-␣323 are catalytic residues is in accord with results of other structure/function studies. The x-ray crystallographic structure for Serratia marcescens chitobiase, the first member of family 20 glycosyl hydrolases to be crystallized, was recently reported (26,40,42). Residues Glu-519 and Glu-740 of chitobiase are not located at or near the active site of the bacterial enzyme suggesting, by analogy, that the homologous residues, Glu-␣307 and Glu-␣462, in the ␣ subunit of Hex A would also not be present at the the ␣ subunit active site. Most significantly, the x-ray crystallographic data (26) identify Glu-540, the residue corresponding to Glu-␣323, as the acid catalyst in the S. marcescens enzyme.
Members of family 20 glycosyl hydrolases share a conserved central region which aligns in all members of the family (Fig.  4). According to the crystal structure of chitobiase, this region comprises the catalytic domain which has an ␣/␤ barrel fold (26). The modeling of the catalytic domain of Hex A (Fig. 7) (26), brings Arg-␣178, Asp-␣258, and Glu-␣323 in proximity within the substrate binding cleft (facing toward the center of the ␣/␤ barrel) and spatially arranged to facilitate catalysis. In contrast, Glu-␣307, Glu-␣462, and Asp-␣163 which corresponds to Asp-␤196, a residue recently proposed as an active site residue in the ␤ subunit (44), appear to be remote from the catalytic domain. While it is likely that the model, illustrated in Fig. 7, will contain some errors in its proposed structure of Hex A, the high degree of conservation of residues in and around the active site suggest that the model is much less prone to errors in the active site pocket of the enzyme.
A role for Glu-␣323 at the Hex A active site is also supported by the studies of Liessem et al. (43). Using a mechanism-based pyrrolidine substrate analog, this group identified residue Glu-␤355, which corresponds to Glu-␣323, as the only reactive residue at or near the ␤ subunit active site. The role of Arg-␣178 is less clear. Given the high pK a of arginine (12.0) it is unlikely that this residue functions as a nucleophile at the pH optimum of Hex (3.9 -4.2). The role of active-site arginine may be to maintain the acid catalyst in its protonated state at a pH which is significantly higher than the pK a for dicarboxylic amino acids.
The extensive sequence similarity of the ␣ and ␤ subunits underscores the structural and catalytic similarity of two subunits of Hex A. Chimeric Hex enzymes, generated by the fusion of different segments of the ␣ and ␤ subunits have recently been used to identify domains required for substrate specificity (41,45). The structural modelling of Hex should now permit the more precise localization of other residues involved in the active site as well as in other functions of the enzyme, such as dimerization, substrate binding and activator recognition.