Identification of domains in human beta-hexosaminidase that determine substrate specificity.

The lysosomal β-hexosaminidases are dimers composed of α and β subunits. β-Hexosaminidase A (αβ) is a heterodimer, whereas hexosaminidase B (ββ) and S (αα) are homodimers. Although containing a high degree of amino acid identity, each subunit expresses a unique active site that can be distinguished by a differential ability to hydrolyze charged substrates. The site on the β-subunit primarily degrades neutral substrates, whereas the α-subunit site is, in addition, active against sulfated substrates. Isozyme specificity is also exhibited with glycolipid substrates. Among human isozymes, only β-hexosaminidase A together with the GM2 activator protein can degrade the natural substrate, GM2 ganglioside, at physiologically significant rates. To identify the domains of the human β-hexosaminidase subunits that determine substrate specificity, we have generated chimeric subunits containing both α- and β-subunit sequences. The chimeric constructs were expressed in HeLa cells to screen for activity and then selected constructs were produced in the baculovirus expression system to assess their ability to degrade GM2 ganglioside in the presence of GM2 activator protein. Generation of activity against the sulfated substrate required the substitution of two noncontinuous α-subunit sequences (amino acids 1-191 and 403-529) into analogous positions of the β-subunit. Chimeric constructs containing only one of these regions linked to the β-subunit sequence showed either neutral substrate activity only (amino acids 1-191) or lacked enzyme activity entirely (amino acids 403-529). Neither the chimeras nor the wild-type subunits displayed activator-dependent GM2-hydrolyzing activity when expressed alone. However, one chimeric subunit containing α amino acids 1-191 fused with β amino acids 225 to 556, when co-expressed with the wild-type α-subunit, showed activity comparable with that of recombinant β-hexosaminidase A formed by the co-expression of the α- and β-subunits. This result indicates that the β-subunit amino acids 225-556 contribute an essential function in the GM2-hydrolyzing activity of β-hexosaminidase A.

The human ␤-hexosaminidases (EC 3.2.1.52) are dimeric lysosomal enzymes composed of two subunits, ␣and ␤, that share about 60% of their amino acid sequence (1,2). The subunits are synthesized as precursors in the endoplasmic reticulum where amino-terminal signal peptides are removed, Nlinked glycosylation occurs, disulfide bonds are formed, and the subunits are folded and assembled. The subunits dimerize and form three isozymes; ␤-hexosaminidase A (␣␤), ␤-hexosaminidase B (␤␤), and ␤-hexosaminidase S (␣␣). Dimerization of the subunits is required for acquisition of enzymatic activity. When properly folded and assembled the enzymes are transferred to the Golgi apparatus for synthesis of the mannose 6-phosphate recognition marker. Mannose 6-phosphate receptors then target the enzymes to lysosomes where the precursor subunits are proteolytically processed to their mature forms (for reviews see Refs. 3 and 4).
The ␤-hexosaminidases participate in the degradation of glycoproteins, glycolipids, and proteoglycans through the removal of terminal ␤-glycosidically linked N-acetylglucosamine or Nacetylgalactosamine residues. The capacity of this enzyme system to degrade this range of substrates is due, in part, to a unique active site with distinct specificities carried by each subunit (5). The ␤-subunit active site predominantly hydrolyzes neutral substrates, whereas the ␣-subunit active site can also react with negatively charged substrates. In addition, certain glycolipid substrates, such as G M2 ganglioside, 1 show a strict isozyme dependence for their degradation. Of the three isozymes, only the heterodimer ␤-hexosaminidase A, together with the G M2 activator protein, is able to degrade G M2 ganglioside significantly (6). The G M2 activator protein functions by binding the ganglioside substrate and interacting with ␤-hexosaminidase A effecting release of the terminal N-acetylgalactosamine residue from the ganglioside (5,7,8).
The importance of this enzyme system is demonstrated by the consequences of mutations in the genes that encode the ␣-subunit, the ␤-subunit, or the G M2 activator. Mutations in the HEXA gene cause Tay-Sachs disease as a result of a deficiency of the ␣-subunit and a resulting absence of ␤-hexosaminidase A and S. Mutations in the HEXB gene encoding the ␤-subunit cause Sandhoff's disease and result in the absence of ␤-hexosaminidase A and B. Defects in the GM2A gene result in G M2 activator deficiency. In each of these disorders there is a massive accumulation of G M2 ganglioside in neuronal lysosomes leading to severe neurodegeneration.
The goal of this work was to identify the domains that confer distinctive substrate specificity to the hexosaminidase isozymes. Our approach was to create chimeric hexosaminidase subunits by interchanging analogous regions of the ␣and ␤-subunits. We expected that some chimeric subunits would be enzymatically active due to the high degree of structural similarity shared between the two subunits. Using these chimeras, we have defined two noncontiguous sequences on the ␣-subunit that, substituted into the ␤-subunit, confer the ability to degrade a charged substrate. Further, we have localized the region of the ␤-subunit required for activator-dependent G M2 ganglioside degradation by ␤-hexosaminidase A.

EXPERIMENTAL PROCEDURES
Construction of Chimeric Enzymes-The human ␤-hexosaminidase ␣and ␤ subunit cDNA were subcloned into pBluescript II KS (pBS) (Stratagene) as described (9) to produce pBS␣ and pBS␤, respectively. The open reading frame of the ␤-subunit coding sequence (1) in construct pBS␤ begins at the second methionine in the sequence shown in Fig. 1. To avoid confusion, the numbering system used for the ␤-subunit begins with the first methionine shown in Fig. 1. Translation initiation from any of the first 3 methionines has been shown to produce an active enzyme (10).
To construct chimeras, regions of pBS␤ were removed by digestion at unique restriction enzyme sites and were replaced with the analogous ␣-subunit cDNA sequence. To make the ␣␤1 construct, pBS␤ was digested with NcoI (cDNA site) and XbaI (polylinker site). The corresponding ␣-subunit cDNA sequence, nucleotides 1-573, was amplified with primers containing a NcoI site (restriction sites underlined), 5Ј-CCCCCATGGCATCCAGAGTGTCCAGGATGC-3Ј, and a XbaI site, 5Ј-GGGTCTAGACAGCGGGCCATGACAAGCTCCAGGCTTTGG-3Ј. The resulting fragment was cut with NcoI and XbaI and subcloned into the NcoI/XbaI-digested pBS␤. To make the ␤␣1 enzyme construct, nucleotides 1206 -1590 of the ␣-subunit cDNA sequence were amplified by primers containing a Bsu36I site (5Ј-CCCCCTGAGGAGCTGGAACTG-3Ј) and an XhoI site (5Ј-GGGGGCTCGAGGGCTCAGGTCTGT-TCAAACTCCTGCTCACAG-3Ј). The resulting fragment was digested with Bsu36I and XhoI, purified by low melting point agarose gel electrophoresis and subcloned into Bsu36I-(cDNA site) and XhoI-digested (polylinker site) pBS␤. The ␣␤␣1 construct was made by subcloning the polymerase chain reaction amplified sequence of the ␣-subunit cDNA (nucleotides 1-573) into NcoI-(cDNA site) and XbaI-digested (polylinker site) ␤␣1. To make the ␣␤␣2 construct, nucleotides 1149 -1590 of the ␣-sequence were amplified with a primer containing a HindIII site (5Ј-AGCAAAGCTTCAGCCAGACACAATCATACA-3Ј) and the XhoI site containing primer used to make ␤␣1. The resulting ␣-subunit cDNA fragment was then digested with HindIII and XhoI and subcloned into HindIII-(cDNA site) and XhoI-digested (polylinker site) ␣␤1. To make the ␣␤␣3 and ␤␣2 constructs, nucleotides 1519 -1590 of the cDNA sequence were amplified with a primer containing a PstI site (5Ј-ACCGCTGCAGGTTGCTGAGGCGAGGTGTCCAG-3Ј) and the XhoI containing primer used to make the ␣␤␣2 and ␤␣1 constructs. The polymerase chain reaction fragment was then digested with PstI and XhoI, gel purified, and subcloned into PstI (cDNA site) and XhoI (polylinker site) digested pBS␤ or ␣␤1. The sequences of the polymerase chain reaction generated regions of cDNA constructs were checked by dideoxy-sequencing with Sequenase (U. S. Biochemical Corp.) for potential errors generated during the procedure. Standard molecular biology techniques were performed according to Sambrook et al. (11).
Cell Culture-HeLa cells were grown in Dulbecco's modified Eagle's medium (DMEM) with 10% fetal bovine serum (FBS). Sf9 cells were grown in spinner flasks at 26°C in Grace's media containing 10% FBS. High-Five cells were grown in monolayer in Grace's medium with 10% serum and were passaged by pipetteing medium over the monolayer of cells and replating at a 1:10 dilution.
Expression of Chimeric Enzyme Constructs-Chimeric enzyme constructs were expressed in HeLa cells using the vaccinia/T7 expression system (12) as has been described (9). HeLa cells were plated 1 day prior to the transfection in 6-well plates at a density of 2 ϫ 10 5 cells/well in DMEM with 10% FBS. Before infection, 0.5 ml of recombinant vaccinia virus (5.2 ϫ 10 9 plaque-forming units/ml) was digested for 0.5 h with 0.5 ml of crystalline trypsin phosphate-buffered saline, pH 7.4 at 37°C. The vaccinia/trypsin mixture was then diluted with 0.5 ml of 1% FBS/ DMEM for each well used in the experiment. The cells were then washed in DMEM without serum and incubated for 0.5 h in 0.5 ml of the diluted vaccinia/trypsin mixture. The transfection mixture was prepared by mixing, for each well, 17 l of Lipofectin (Life Technologies, Inc./BRL) and 5 g of DNA in 1.5 ml of serum-free medium and incu-bating for 0.5 h at room temperature before adding to the cells. After a 5 h of incubation with the cells, 2 ml of 10% FBS containing DMEM was added, and the following day cell lysates were assayed for ␤-hexosaminidase enzyme activity with 4-methylumbelliferyl-2-acetamido-2deoxy-␤-D-glucopyranoside (MU-GlcNAc) and 4-methylumbelliferyl-6sulfo-2-acetamido-2-deoxy-␤-D-glucopyranoside (MU-GlcNAc-6-SO 4 ) as described (10).

Generation of Recombinant Baculovirus and Protein Expression-
The baculovirus containing the human ␤-subunit cDNA has been described (13). A baculovirus containing the human ␣-subunit was generated by the same procedure (13). To produce baculovirus containing ␣␤1 and ␣␤␣1, Sf9 cells were first co-transfected with linearized Autographa californica nuclear polyhedrosis virus DNA and pBlueBacII (Invitrogen) containing the appropriate construct using cationic liposomes. Resulting recombinant virus was purified from the transfection supernatant by two rounds of plaque purification. Because the pBlueBac II vector co-expresses ␤-galactosidase, recombinant plaques were distinguished by the blue color in the presence of 5-bromo-4-chloro-3-indoyl ␤-D-galactoside and the absence of occlusion bodies. Recombinant viral plaques were further identified by screening for MU-GlcNAc and MU-GlcNAc-6-SO 4 hydrolyzing activity. Viral stocks were generated by large scale infection of 2 ϫ 10 8 Sf9 cells in suspension culture and titered in High-Five cells (Invitrogen) using the Quick-titer kit from Kemp Biotechnologies Inc.
Large scale expression of ␤-hexosaminidase and chimeric enzyme constructs was accomplished in High-Five cells (Invitrogen) by Kemp Biotechnologies Inc. For co-infection studies, the optimal multiplicity of infection for co-expression of chimeric and wild-type cDNA constructs was determined by isoelectric focusing of the High-Five cell conditioned medium (described below). The cells were incubated at the appropriate multiplicity of infections in 5 ml of serum-free Grace's medium for 4 h on a rocking platform. After the incubation, the cells were washed and incubated with 5 ml of serum-free medium for 3 days.
Purification of ␤-Hexosaminidase and Chimeric Enzymes-The enzymes were obtained from the serum-free culture medium of infected cells. The enzymes were partially purified on a column of concanavalin A-Sepharose (Pharmacia Biotech Inc.) (13). Column fractions containing MU-GlcNAc hydrolyzing activity were pooled and concentrated with an Amicon PM-10 membrane. The resulting concentrate was run on an 8% precast Tris-glycine gel (Novex) and stained with Coomassie Blue to evaluate purity and to estimate enzyme protein concentration.
Western Blotting-Cell lysates or media were analyzed by either 8 or 12% SDS-Tris/glycine polyacrylamide gels or pH 3-10 isoelectric focusing pre-cast gels from Novex. Isoelectric focusing gels were electrophoretically transferred to a nitrocellulose membrane in 0.7% acetic acid at 10 V for 1 h using an Novex transfer apparatus. SDS-Tris/ glycine gels were transferred in 39 mM glycine, 48 mM Tris base, 0.037% SDS, 20% methanol at 250 milliamps for 0.5 h at 4°C. Membranes were probed with goat anti-human ␤-hexosaminidase A (14) as 1°C antibody followed by horseradish peroxidase-conjugated rabbit anti-goat IgG as 2°C antibody according to the protocol in the ECL Western detection kit from Amersham Corp.
Determination of K m and V max -The K m and V max values for recombinant ␤-hexosaminidase S and B and the chimeric enzymes ␣␤␣1 and ␣␤1 were determined by the graphic method of Lineweaver and Burk (15). Enzyme activities were measured as described (5) using various concentrations of the synthetic substrates MU-GlcNAc (0 -2.5 mM in 100 mM citrate buffer, pH 4.5) and MU-GlcNAc-6-SO 4 (0 -1.0 mM in 50 mM citrate buffer, pH 4.0), respectively. Enzyme kinetics of all recombinant proteins followed Michaelis-Menten theory. The apparent V max values were calculated for the pure enzymes based on their protein percentage of the concanavalin A fraction as determined by SDS-polyacrylamide gel electrophoresis analysis. The Coomassie Blue-stained protein bands of concanavalin A purified recombinant hexosaminidases (ϳ10 g/lane) from the insect cell expression system were compared with various amounts of bovine serum albumin (0.5-10 g) by densitometric scanning. The purities determined this way were 74% for ␤-hexosaminidase S, 43% for ␤-hexosaminidase B, 43% for ␣␤␣1, and 80% for ␣␤1.
Assay for the Degradation of G M2 Ganglioside-Aliquots of serumfree High-Five cell expression medium were dialyzed against water containing 0.1% NaN 3 . Assays were incubated for 4.5 h at 37°C in 80 mM citrate buffer, pH 4.0, 0.01% NaN 3 containing 10 g of bovine serum albumin, 5 nmol 3 H-G M2 ganglioside (3.6 Ci/mol) tritiated in the Gal-NAc moiety, 10 l of recombinant G M2 activator protein purified from Escherichia coli (16) (1.6 g) with 25 l of dialyzed expression medium. After the incubation 500 l of 1 mM aqueous GalNAc was added, and the solution was loaded onto rp-18 columns that were washed with 2 ϫ 1 ml, 1 mM GalNAc-solution. The combined effluents containing the liberated radioactive N-acetyl-D-galactosamine were collected in scintillation vials, and their radioactivity was measured.

RESULTS
The ␣and ␤-subunits of ␤-hexosaminidase share about 60% of their amino acid sequence (1, 2) (Fig. 1). Chimeric hexosaminidase subunits were constructed by interchanging cDNA sequences encoding analogous segments of the ␣and ␤-subunits ( Figs. 1 and 2). The constructs were positioned adjacent to the T7 polymerase promoter in the pBluescript vector so that expression of the constructs could be driven by T7 polymerase after their transfection into HeLa cells. In the representative experiment shown in Fig. 3, the extracts from the transfected HeLa cells were first assessed by Western blot analysis to ensure protein expression (Fig. 3A) and then for enzyme activity using the neutral (MU-GlcNAc) and the charged (MU-GlcNAc-6-SO 4 ) synthetic substrates. Polypeptides, reactive with anti-␤-hexosaminidase A, were expressed from each chimeric construct that were similar (60 -70 kDa) but not identical in electrophoretic mobility to the wild-type subunit precursors. The wild-type enzymes demonstrated the expected specificity: ␤-hexosaminidase S, produced by expressing the wild-type ␣-subunit, cleaved both the MU-GlcNAc and MU-GlcNAc-6-SO 4 substrates. By contrast, ␤-hexosaminidase B, produced from expressing the ␤-subunit, showed a marked preference for the neutral MU-GlcNAc substrate and, correspondingly, a very low ratio of sulfated to neutral substrate cleaved. Chimeric construct, ␣␤1, containing the amino-terminal residues 1-191 of the ␣-subunit fused to the carboxylterminal amino acids 225-556 of the ␤-subunit, was active against the MU-GlcNAc with relatively little activity against the MU-GlcNAc-6-SO 4 substrate in a manner similar to wildtype ␤-subunit. A number of other ␣-␤ constructs were prepared containing both larger and smaller segments of the ␣-subunit at the amino terminus, but they were all catalytically inactive (not shown). Also inactive was construct ␤␣1, which consisted of amino-terminal ␤-subunit sequence up to amino acid 431 linked to ␣-subunit sequence from amino acid 403 to the carboxyl terminus ( Fig. 1 and 2).
In an attempt to prepare an active chimera with MU-Glc-NAc-6-SO 4 activity, we modified construct ␣␤1 by replacing the ␤-subunit carboxyl-terminal portion with the analogous ␣-subunit sequence 403-529 so that this construct contained the same amino-terminal segment as ␣␤1 (active only against MU-GlcNAc) and the same carboxyl-terminal segment as ␤␣1 (in- active). This construct, ␣␤␣1, with both the amino and carboxyl ends comprised of ␣-subunit sequence, now exhibited activity against the sulfated substrate and showed a ratio of sulfated to neutral substrate cleaved similar to the wild-type ␣-subunit. The ability to degrade the sulfated substrate was maintained when the carboxyl-terminal segment was increased to 146 amino acids of ␣-sequence in ␣␤␣2 but was lost when the ␣-segment was reduced to 23 amino acids in ␣␤␣3. Chimera ␣␤␣3 was, however, active against the neutral MU-GlcNAc substrate. Construct ␤␣2, the same as ␣␤␣3 but without the amino-terminal ␣-segment, was also active against MU-Glc-NAc only. Construct ␤␣2 was the only catalytically active construct out of several prepared with ␤-sequence at the amino terminus fused to ␣ sequence at the carboxyl terminus (not shown). These data (summarized in Fig. 2) indicate that the minimum regions required for MU-GlcNAc-6-SO 4 activity include ␣-subunit sequences 1-191 and 403-529; other constructs in which these regions are replaced with ␤ sequence are either active only against MU-GlcNAc or inactive.
We next investigated the activator-dependent G M2 ganglioside hydrolyzing activity of chimeric hexosaminidase subunits. However, expression of the enzymes using the vaccinia-T7 polymerase system in HeLa cells or in COS-1 cells after insertion of the constructs in SV-40-based expression vectors proved unsatisfactory because of a high background G M2 ganglioside degrading activity inherent in the human and monkey cells. We, therefore, chose the baculovirus expression system because of the potential for very high levels of protein production and because of very low background activity. We analyzed two representative chimeric enzymes, ␣␤␣1 and ␣␤1, in this manner. Apparent K m and V max values for MU-GlcNAc and MU-GlcNAc-SO 4 were first determined for the chimeric enzymes and for recombinant ␤-hexosaminidase S and B produced by expression of the wild-type ␣and ␤-subunit, respectively (Table I). Like ␤-hexosaminidase B, chimeric construct ␣␤1 displayed a lower K m with MU-GlcNAc compared with MU-Glc-NAc-SO 4 . By contrast, ␤-hexosaminidase S and the chimeric construct, ␣␤␣1, both of which display activity against the sulfated substrate, showed a lower K m with MU-GlcNAc-6-SO 4 relative to MU-GlcNAc. All enzymes developed the greatest maximal velocity with the MU-GlcNAc substrate. However, the ratio of V max with MU-GlcNAc to V max with MU-GlcNAc-6-SO 4 was considerably larger for recombinant ␤-hexosaminidase B (ϳ150) and ␣␤1 (ϳ60) than for recombinant ␤-hexosaminidase S (ϳ2) and ␣␤␣1 (ϳ3).
None of the constructs (␣-subunit, ␤-subunit, ␣␤␣1, or ␣␤1) produced significant activator-dependent G M2 ganglioside degrading activity when expressed alone (Fig. 4). In contrast, co-expression of the wild-type ␣and ␤-subunits, to produce ␤-hexosaminidase A yielded significant activator-dependent G M2 ganglioside hydrolyzing activity. We next co-expressed the chimeric subunits with each wild-type subunit in order to assemble heterologous subunits. The assembly of the various subunit combinations was verified by isoelectric focusing (not shown). However, only the combination of ␣␤1 and the wildtype ␣-subunit resulted in activity comparable with the coexpression of wild-type ␣and ␤-subunits. All other combinations were not effective in the activator-dependent G M2 ganglioside degradation assay. This result indicates that the ␤-subunit amino acids 225-556 contribute an essential function in combination with the wild-type ␣-subunit for effective activator-dependent degradation of G M2 ganglioside.

DISCUSSION
The ␣and ␤-subunits of ␤-hexosaminidase are the product of a gene duplication event (1), and as a result, they show a high degree of structural and functional similarity. It is also likely that the structure of the active site in each subunit is very similar, although differences obviously exist that are responsible for their characteristic specificity. In addition to their active sites, the subunits express other distinctive functions related to the binding of the G M2 ganglioside activator complex (6). The structural similarity of the subunits has enabled the creation of a series of enzymatically active chimeric polypeptides and has allowed us to attribute the functional differences of the subunits to discrete polypeptide segments.

FIG. 3. Expression of ␤-hexosaminidase chimeric subunits in HeLa cells.
A, HeLa cells were infected with vaccinia virus expressing T7 polymerase and subsequently transfected with the indicated constructs or with pBluescript. After 18 h, a portion of the cell extracts (10 l) were electrophoresed on a SDS-polyacrylamide gel. The polypeptides were electrophoretically transferred to nitrocellulose and then visualized with antibody to ␤-hexosaminidase A as described under "Experimental Procedures." The molecular mass standards (in kilodaltons) are indicated to the left of the photograph. The same HeLa cells extracts were assayed with MU-GlcNAc-6-SO 4 (C) and MU-GlcNAc (B). One unit of enzyme activity is defined as the activity that releases 1 nmol of 4-methylumbelliferone/mg of cell protein/h.  dase A and S, can effectively degrade sulfated substrates exemplified by MU-GlcNAc-6-SO 4 . Although exhibiting a similar specificity (5), it should be pointed out that there may be some differences between the ␣-subunit active site in the homodimer, ␤-hexosaminidase S, and the ␣-subunit active site in the context of the heterodimer, ␤-hexosaminidase A. The chimeric subunit, ␣␤1, with the 191 amino-terminal amino acids of the ␣-subunit linked to the carboxyl-terminal 332 amino acids of the ␤-subunit showed activity against the MU-GlcNAc but not the MU-GlcNAc-6-SO 4 substrate. Other bipartite chimeras were either totally inactive or only active with the MU-GlcNAc substrate. However, when the 127 carboxyl-terminal amino acids as well as the 191 amino-terminal amino acids of the ␣-subunit were included to produce the tripartite construct, ␣␤␣1, significant activity against the MU-GlcNAc-6-SO 4 substrate was observed. Both ␣-subunit termini (amino acids 1-191 and 403-529) were required for activity against the sulfated substrate because ␤␣1, which contained the same carboxyl-terminal ␣-sequence as ␣␤␣1 but without the aminoterminal ␣-sequence, was inactive. The requirement for two discontinuous stretches of ␣-sequence for MU-GlcNAc-6-SO 4 activity is reminiscent of the cathepsin D/pepsinogen chimeras described by Kornfeld and co-workers (17) that required two different regions of cathepsin D for effective phosphorylation of its mannose residues. In this case, these two noncontinuous amino acid sequences of cathepsin D were contiguous in the three-dimensional structure forming a compact protein recognition domain. Therefore, it is possible that in the three dimensional structure of the ␣-subunit, the amino and carboxyl termini form a portion of the active site for MU-GlcNAc-6-SO 4 such that amino acid residues from one or both of these regions are involved in the binding and/or hydrolysis of the substrate. Alternatively, residues from these regions may not be directly involved with the substrate but may contribute to the structure of the active site indirectly by influencing the folding and overall conformation of the subunit. In light of the difference in specificity of the subunits, it is interesting the amino and carboxyl termini contain the most differences in amino acid sequence between subunits (Fig. 1).
Clearly, a three-dimensional structure of ␤-hexosaminidase, when available, will greatly facilitate the interpretation of these results.
Although the active sites on ␤-hexosaminidase subunits have been well characterized kinetically (5), there is only minimal information concerning their structure. Recently, however, Glu-355 in the ␤-subunit was photoaffinity labeled with an active site-directed inhibitor implicating Glu-355 and the amino acids in the immediate vicinity as binding or active site residues (Fig. 1) (18). This evolutionarily conserved area (Fig.  1) may be involved in a catalytic mechanism shared between the ␣and ␤-subunits and other hexosaminidases (18), so it is not surprising that the region is excluded from the segments that confer MU-GlcNAc-6-SO 4 specificity, a unique property of the ␣-subunit. A second amino acid implicated in the catalytic activity of ␤-hexosaminidase is Arg-178 of the ␣-subunit, the residue mutated in the B1-variant of Tay-Sachs disease (19,20). Mutagenesis of the corresponding amino acid in the ␤-subunit impairs catalytic activity again, suggesting a shared function of the this residue between subunits. Clearly Arg-178 and surrounding residues alone are not sufficient for MU-GlcNAc-6-SO 4 activity because chimera ␣␤1, which contains the aminoterminal 191 amino acids of the ␣-subunit, is only active against MU-GlcNAc.
The degradation of G M2 ganglioside by ␤-hexosaminidase A requires the G M2 activator protein (reviewed in Ref. 6). The activator binds the ganglioside in a 1:1 complex and the complex interacts directly with the enzyme allowing the ␣-subunit active site to remove the terminal GalNAc moiety from the tetrasaccharide moiety of G M2 ganglioside. The ␣-subunit possesses a binding site for the activator-ganglioside complex. However, the ␤-subunit must also contribute to this interaction, because only ␤-hexosaminidase A of the three isozymes can carry out the reaction at physiological rates. We found that neither chimeric subunits ␣␤1 nor ␣␤␣1, when expressed alone, demonstrated activator-dependent ganglioside degradation activity. In this regard, they were no more effective than the wild-type subunits expressed singly. However, when these chimeric subunits were expressed in combination with each wildtype subunit, we found that ␣␤1 together with the ␣-subunit caused the activator-dependent degradation of the ganglioside about as well as the co-expression of the wild-type ␣and ␤-subunits. This result shows that the specific function of the ␤-subunit in activator-dependent ganglioside degradation resides in the carboxyl-terminal two-thirds of the precursor polypeptide (amino acids 225-556). The result also indicates that the amino-terminal one-third (amino acids 1-224) does not provide a ␤-subunit-specific function because it can be replaced by the corresponding segment from the ␣-subunit. In addition, the chimeric enzymes also shed some light on the important regions of the ␣-subunit in activator-dependent ganglioside degradation. The inability of ␣␤␣1 to substitute for the ␣-subunit in a heterodimer with the ␤-subunit indicates that amino acids 192-402 of the ␣-subunit, which are missing from ␣␤␣1, impart an essential function for ganglioside hydrolysis.
We have established a structure-function relationship of discrete regions of the ␤-hexosaminidase subunits with the ability to degrade particular substrates. By substituting smaller segments into these chimeras and through site-directed mutagenesis within these regions, it should be possible to define individual residues that participate in determining the specificity of the hexosaminidase isozymes. Ultimately, a complete understanding of the interaction of various substrates with ␤-hexosaminidase will require a combination of approaches including chimera analysis and affinity labeling together with a three-dimensional structure of the enzyme.