A Pro504 → Ser Substitution in the β-Subunit of β-Hexosaminidase A Inhibits α-Subunit Hydrolysis of GM2Ganglioside, Resulting in Chronic Sandhoff Disease*

The GM2 gangliosidoses are caused by mutations in the genes encoding the α- (Tay-Sachs) or β- (Sandhoff) subunits of heterodimeric β-hexosaminidase A (Hex A), or the GM2 activator protein (AB variant), a substrate-specific co-factor for Hex A. Although the active site associated with the hydrolysis of GM2 ganglioside, as well as part of the binding site for the ganglioside-activator complex, is associated with the α-subunit, elements of the β-subunit are also involved. Missense mutations in these genes normally result in the mutant protein being retained in the endoplasmic reticulum and degraded. The mutations associated with the B1-variant of Tay-Sachs are rare exceptions that directly affect residues in the α-active site. We have previously reported two sisters with chronic Sandhoff disease who were heterozygous for the common HEXB deletion allele. Cells from these patients had higher than expected levels of mature β-protein and residual Hex A activity, ∼20%. We now identify these patients’ second mutant allele as a C1510T transition encoding a β-Pro504 → Ser substitution. Biochemical characterization of Hex A from both patient cells and cotransfected CHO cells demonstrated that this substitution (a) decreases the level of heterodimer transport out of the endoplasmic reticulum by ∼45%, (b) lowers its heat stability, (c) does not affect its K m for neutral or charged artificial substrates, and (d) lowers the ratio of units of ganglioside/units of artificial substrate hydrolyzed by a factor of 3. We concluded that the β-Pro504 → Ser mutation directly affects the ability of Hex A to hydrolyze its natural substrate but not its artificial substrates. The effect of the mutation on ganglioside hydrolysis, combined with its effect on intracellular transport, produces chronic Sandhoff disease.

The hydrolysis of G M2 ganglioside (G M2 ) 1 requires the proper synthesis, intracellular transport, and protein-protein interactions of three different gene products. Two of these, encoded by the evolutionarily related HEXA (15q23-q24 (1)) and HEXB (5q13 (2)) genes, are the ␣and ␤-subunits of heterodimeric ␤-hexosaminidase A (Hex A), respectively. The third gene product is a small heat-stable protein, the G M2 activator protein (activator), encoded by the GM2A gene (5q31. 3-33.1 (3)). Mutations in any one of these genes can result in the storage of G M2 and one of the family of human diseases known as the G M2 gangliosidoses. HEXA mutations are associated with Tay-Sachs disease, HEXB with Sandhoff disease, and GM2A with the AB variant form (reviewed in Ref. 4).
The G M2 gangliosidoses show extreme variability in clinical expression. Typically, the earlier the age of onset of clinical symptoms the more severe the disease. A nomenclature based on the different clinical phenotypes and recognizing the dominance of the encephalopathy (rather than only the age of onset) has been suggested (4): acute (the classical infantile form), subacute (late infantile and juvenile forms), and chronic (adult and chronic forms). The most common, acute form is a severe neurological disorder that usually results in death within 4 years. Mutations associated with this classical phenotype prevent the formation of any functional Hex A. It is now generally believed that the broad range of less severe phenotypes result from small variations in the levels of residual Hex A activity, on the order of 0 -5% (5). Healthy individuals with ϳ10% residual Hex A activity have been described (reviewed in Ref. 4).
While patients with the acute form of G M2 gangliosidosis are deficient in Hex A activity, their total Hex activity, as measured by neutral artificial substrates, is significant. Tay-Sachs patients often have nearly normal levels of activity due to the presence of the homodimeric Hex B isozyme (␤␤), and Sandhoff patients have 1-5% of normal levels from homodimeric Hex S (␣␣) (reviewed in Ref. 4). Since any dimeric combination of ␣and/or ␤-subunits produces an active isozyme, each subunit must contain a potential active site. The characteristics of the two sites have been examined in the two homodimers (6) and in a novel form of Hex A with an inactive ␤-subunit due to an Arg 211 3 Lys substitution (7). These data indicate that the presence of the ␤-subunit affects the K m and V max of the ␣ active site toward neutral substrates (7). Whereas both subunits of Hex A are equally capable of hydrolyzing neutral ␤-GlcNAcor ␤-GalNAc-containing substrates, e.g. 4-methylumbelliferyl-␤-N-acetylglucosamine (MUG), only isozymes containing an ␣-subunit can efficiently hydrolyze negatively charged ␤-GlcNAc-6-sulfate-containing substrates, e.g. meth-ylumbelliferyl-␤-N-acetylglucosamine-6-sulfate (MUGS), i.e. the MUG/MUGS hydrolysis ratio for Hex B is ϳ300, for Hex A ϳ4, and for Hex S ϳ1 (7). The specificity of the Hex isozymes for G M2 ganglioside indicates that it should also be considered as a negatively charged substrate, presumably due to the sialic acid residue attached to the penultimate Gal residue. However, whereas Hex S as well as Hex A, but not Hex B, can hydrolyze G M2 in the presence of detergents, only Hex A is functional in vivo with the G M2 activator-G M2 ganglioside complex (reviewed in Ref. 4). Thus, some component(s) of the ␤-subunit are necessary for the hydrolysis of G M2 in vivo.
The exact role of the activator remains controversial (8 -10). However, it is generally agreed that it binds both the lipid and oligosaccharide portions of G M2 , extracting or at least lifting the ganglioside out of the membrane, and then the complex interacts with Hex A for hydrolysis (11). Hex B can hydrolyze the neutral, asialo derivative of G M2 , G A2 , in the presence of detergent, but it has little activity in the presence of activator. Furthermore, it has been reported that the activator, even in the absence of G M2 , can slightly inhibit the hydrolysis of MUGS by both Hex A and Hex S (reviewed in Refs. 12 and 13). These data indicate that at least a portion of the binding site for the complex is also located in the ␣-subunit. The required elements of the ␤-subunit may function by increasing the affinity of Hex A for the complex and/or correctly orient the complex, allowing the efficient hydrolysis of the terminal sugar from the ganglioside. Furthermore, these ␤-elements may act directly by interacting with the complex or indirectly by affecting the conformation of the ␣-subunit. Other functions associated with the ␤-subunit include greatly increasing the stability of the resulting dimer and facilitating the transport of the ␣-subunit out of the endoplasmic reticulum (ER) (reviewed in Refs. 4 and 14).
To date, all missense mutations except those at two codons, in either HEX gene, result in normal levels of mutant mRNA but paradoxically with a dramatic reduction in both mature ␤and/or ␣-protein and Hex B and/or Hex A activity in patient cells. This is believed to be the result of a strict "quality control system" in the ER that prevents the transport and increases the degradation rate of missfolded proteins or unassembled subunits (unlike ␤-subunits, ␣-subunits have an apparently low affinity for each other) (reviewed in Refs. 4 and 14). In several cases, it has been demonstrated that subunits with missense mutations, even those associated with the most severe clinical phenotype, are not totally incapable of forming a partially functional Hex A but may be prevented from doing so by their retention and degradation in the ER (15,16). Thus, the major detrimental effect caused by most HEX missense mutations is at the level of intracellular transport rather than structural changes specifically affecting some aspect of enzyme function. The exceptions to this conclusion are the missense mutations at ␣-Arg 178 (17,18) and ␣-Asp 258 (19), which produce the B1 biochemical phenotype. Patients with the most common Arg 178 3 His substitution were originally thought to have an activator defect, because they express normal levels of both Hex A and Hex B activities, as assayed with neutral (common) substrates, e.g. MUG. However, unlike the normal Hex A found in the true AB variants, Kytzia et al. (20) found that B1 variant-Hex A was inactive toward an ␣-specific GlcNAc 6-sulfatecontaining substrate (as well as G M2 ganglioside even in the presence of added activator protein), and they suggested the presence of a mutation at or near the active site of the ␣-subunit. This hypothesis has been demonstrated to be correct for substitutions at either residue based on mutational and expression studies of the aligned ␤-residues, i.e. ␤-analogs, ␤-Arg 211 (21,22) and ␤-Asp 290 (23), and on molecular modeling of human Hex using the structure of bacterial chitobiase (24).
We described 10 years ago two sisters of French Canadian ancestry with a chronic Sandhoff phenotype (25). We have also previously reported that these patients are heterozygous for the common 16-kb 5Ј HEXB deletion allele, which does not transcribe ␤-mRNA (26). In this report, we characterize the second mutant allele in these patients, a missense mutation in exon 13 of the HEXB gene that results in a Pro 504 3 Ser substitution. This mutation produces a novel biochemical phenotype that impacts directly on the ability of Hex A to hydrolyze G M2 . This is the first report of a mutation in the ␤-subunit that affects the ability of Hex A to hydrolyze its natural but not its artificial substrates and localizes essential elements of the ␤-chain for natural substrate hydrolysis to its C terminus.

MATERIALS AND METHODS
Preparation of Genomic DNA-Cultured fibroblasts were lysed by directly adding 1.0 ml of DNAZOL Reagent (Life Technologies, Inc.) to the 10-cm 2 culture dish. The lysate was then transferred into an Eppendorf tube, and insoluble cell debris was removed by brief centrifugation. The genomic DNA in the supernatant was precipitated with ethanol and resuspended in 10 mM Tris-HCl buffer containing 1 mM EDTA, pH 7.4 (27).
RNA Isolation and Reverse Transcription-Total RNA was isolated by using TRIzoI Reagent (Life Technologies, Inc.), as described by Hou et al. (28). Two g of total RNA were used to synthesize the single strand cDNA according to the SUPERSCRIPT TM II procedure (Life Technologies). Briefly, RNA was first denatured at 70°C for 10 min and then incubated at 42°C for 50 min with 200 units of SUPERSCRIPT II and 0.2 g of random primers in 20 l of 50 mM Tris-HCl (pH 8.3), 75 mM KCl, 3 mM MgCl 2 , 20 mM dithiothreitol, and 0.5 mM each of four dNTPs. Two l of this mixture were directly used for PCR to synthesize and amplify double strand cDNA.
DNA Amplification and Direct Sequencing-Amplification of exonic and intron/exon junctions from genomic DNA and cDNA fragments was performed by PCR as described previously (27). The reactions were carried out in a 100-l volume of 0.1-0.5-g genomic DNA or 2 l of cDNA (by reverse transcription), 10 mM Tris-HCl (pH 8.3), 50 mM KCl, 1.5 mM MgCl 2 , 0.01% gelatin, 0.2 mM each of four dNTPs, 0.5 g of each primer, and 2.5 units of AmpliTaq TM Taq polymerase. Amplification was achieved by incubation in a DNA Thermal Cycler (Perkin-Elmer) for 30 cycles, each consisting of 30 s of denaturation at 94°C, 30 s of annealing at 55-60°C, and 1-3 min of extension at 72°C. The region around exon 13, found to be heterozygous for Pro 504 3 Ser mutation in genomic DNA and homozygous in cDNA was amplified by PCR using oligonucleotides 129 (exon 10, sense; GGTTTTGGATATTATTGCAAC-CATAAA) and 14A (3Ј-untranslated region, antisense; TCAATCAATA-AAAATATTTTATTC). The resulting PCR products from genomic DNA and cDNA were 799 and 716 bp, respectively. PCR products were purified by utilizing the Geneclean Kit (Bio 101, Inc., Vista, CA), and direct sequencing was performed with [␣-35 S]dATP using a modification of the Sequenase TM protocol (U.S. Biochemical Corp.), as described by McInnes et al. (27).
Generation of Mutant Constructs-The wide type constructs, pREP4-␣ and pEFNEO-␤, have been reported (7). The mammalian expression vectors pREP4 (Invitrogen) and pEFNEO (kindly supplied by Dr. Anson) (29), have hygromycin B and neomycin (G418) resistance markers, respectively. To construct the mutant cDNA into pCD vector, a 636-bp product by reverse transcription-PCR from patient fibroblast (as described above) containing ␤-Pro 504 3 Ser was digested with PflMI at a site 5Ј to the mutation and BanI at a site 3Ј to the mutation. The middle fragment of 387 bp was purified and subcloned into pCD␤43 (21) treated with PfoMI/partial BanI. To generate the mutant pEFNEO-␤-Pro 504 3 Ser, a 2.0-kb fragment, partially digested by BamHI from pCD-␤-Pro 504 3 Ser, was isolated and subcloned into the BamHI site of the pEFNEO-␤ vector. The mutation was verified by DNA sequencing. A construct encoding an Asp 208 3 Asn substitution in the ␤-cDNA insert of pEFNEO has previously been reported (23). In permanently transfected CHO cells, this construct produces only soluble, monomeric, precursor ␤-subunits (23). We now used this transfected clonal CHO cell line as a control for the ER-retention on mutant ␤-protein.
Cell Culture and DNA Transfection-CHO cells were grown in minimal essential medium with 10% FCS and antibiotics at 37°C in 5% CO 2 . Transfections were performed using Lipofection from Life Technologies, as described previously (7). Transfected cells were also grown in serum-free medium containing 10 mM NH 4 Cl, which diverts proteins targeted to the lysosome to the secretory pathway, for 1 and 2 days, and the Hex activity was measured using MUG.
Hex Activity Assay-Cells were lysed in a buffer of 10 mM Tris-HCl, pH 7.5, and 5% glycerol through five sets of freeze-thaw cycles. Protein from cell lysate was quantitated by the Lowry method (30). Hex activity from cell lysates was determined using a ␣-chain-specific substrate MUGS and the common substrate MUG (15).
Western Blotting-The protein (amounts loaded are indicated in each figure) from cell lysates or DEAE fractions were resolved by SDS-PAGE using a Bio-Rad minigel system (31). Proteins were transferred to nitrocellulose overnight at 4°C. The filter was blocked in 5% skim milk and then incubated overnight with primary antibody, rabbit anti-human Hex A (7,28). Nitrocellulose was washed four times with 1% skim milk and incubated with a secondary antibody, horseradish peroxidaseconjugated goat anti-rabbit IgG for 1 h. The filter was developed and exposed to Hyperfilm using the ECL system (Amersham Pharmacia Biotech).
Separation of Hex Isozymes-Proteins (3 mg) from lysates of patient or normal fibroblasts or from control or transfected CHO cells were applied to a 1.0-ml column of DEAE CL-6B (Amersham Pharmacia Biotech). The unbound Hex B fraction was collected with 10 mM sodium phosphate, pH 6.0. Hex A was eluted by applying 0.15 M NaCl in 10 mM sodium phosphate, pH 6.0 (7). Three-ml fractions were collected and assayed for Hex activity.
Kinetic Analysis-The K m value was determined by varying the concentration of the substrates from 0.125 to 4.0 mM for MUG and from 0.05 to 2.5 mM for MUGS. Also, 10 experimental points were used for each K m determination. The normal and mutant Hex A from transfected CHO or patient cells were purified away from the other Hex isozymes by DEAE ion exchange chromatography (see above). Kinetic constants were calculated using a computerized nonlinear least squares curve fitting program for the Macintosh, KaleidaGraph TM 3.0 (7).
G M2 Hydrolysis Assay-[ 3 H]G M2 ganglioside (20 nmol), labeled in the C-6-position of its N-acetylgalactosamine moiety (32), was incubated in the presence of 2.0 g of recombinant activator protein from bacteria (33,34) at 37°C for 18 h in 10 mM citrate buffer (pH 4.1), 0.5% human serum albumin, and 10 mM GlcNAc (carrier), with 0, 50, 100, and 200 units of Hex A (nmol of MUGS hydrolyzed/h), from normal or patient fibroblasts or produced from human cDNAs (normal ␣ with normal or mutant ␤) in transfected CHO cells (final volume of 100 l). The hydrolyzed product from G M2 , i.e. [ 3 H]GalNAc, was separated from the unreacted G M2 substrate by passage through a positively charged ion exchange minicolumn of 0.6 ml of AG3X4 (acetate form) resin. The unbound fraction containing [ 3 H]GalNAc was determined by liquid scintillation counting, as described previously (7).
Thermal Stability Study-The wild-type and mutant Hex A or Hex B isozymes, which had been separated by DEAE chromatography, were added to 700 l of preheated citrate phosphate buffer (pH 4.1) with 0.3% human serum albumin. The heat denaturation was performed at 45°C, and aliquots (100 l) were removed at intervals of 0, 15, 30, 45, 60, 75, and 90 min for Hex A and 0, 30, 60, 90, 120, 150, and 180 min for Hex B, placed on ice, and assayed for enzyme activity. The wild-type and mutant Hex A from transfected CHO cells were also tested for their residual MUGS activity after incubation at 37°C for 18 h, under conditions that mimicked the natural substrate assay above.
Intracellular Localization of ␤-Proteins Quantified Using Indirect Immunofluorescence-Nontransfected CHO cells or CHO cells transfected with constructs encoding (a) the wild-type ␤-cDNA (lysosomal localization control), (b) the Pro 504 3 Ser substitution, or (c) an Asp 208 3 Asn substitution (ER localization control) were grown at 37°C in 5% CO 2 on glass slide covers in a 10-cm 2 culture dish. After 24 h of incubation, the cells were fixed and gently permeabilized with 100% cold methanol at Ϫ20°C for 30 min. The fixed cells were then washed in phosphate-buffered saline, blocked with 1% bovine serum albumin, and incubated with the primary polyclonal anti-Hex B antibody (35), diluted 1:200 for 1 h. The secondary antibody, a green fluoresceinlabeled goat-anti-rabbit IgG F(abЈ)2), diluted 1:100, was then added for 1 h, either alone or in combination with a 1:10,000 dilution of propidium iodide, which in addition to nuclear DNA also stains the cytoplasmic RNA and marks the position of the ER with the red fluorescence. The cells were then washed three times with phosphate-buffered saline and mounted with elvanol. In control cultures, the preimmune rabbit IgG substituted for the primary antibody. The slides were analyzed, and the proportion of ␤-protein present in the ER or endosome/lysosome was determined using a fluorescent microscope (Olympus Vanox-AH-3, magnification ϫ 800) and two narrow band filters to detect the green and red fluorescence separately. An additional broad spectrum filter was also used for the simultaneous detection of the fluorescein-tagged green Hex B and the nucleic acids labeled with red propidium iodide fluorescence. In this setting, the overlapping of the red and green labels in the cytoplasm is marked by yellow fluorescence and indicates the colocalization of Hex B and ER (36). Multiple images of the same cell obtained with all of the above mentioned filters were captured with the CCD camera (Optronix), stored in a Macintosh 9500 computer, and quantitatively analyzed using the Image Pro Plus program (Media Cybernetics, Silver Spring, MD) according to the manufacturer's instructions. In each of the three experimental groups (wild type ␤, ␤-Pro 504 3 Ser, and ␤-Asp 208 3 Asn), images of 50 cells were analyzed, and results were statistically evaluated to give quantitative measurements of the percentage of each ␤-protein that resides in the ER and/or lysosome.

RESULTS
Direct sequencing of the exons and exon/intron junctions of the HEXB gene revealed that the patients were heterozygous for a C1510T transition in exon 13 (ϩ2 bp from intron 12) at the codon for Pro 504 , which results in its conversion to a Ser codon (Fig. 1A). We have previously reported that the patients were also heterozygous for the common 16-kb 5Ј partial HEXB deletion allele, ⌬16kb (26). To confirm that this missense mutation was not part of the deletion allele, we also sequenced the ␤-cDNA. In this case, the patients appear to be homozygous for the missense mutation (Fig. 1B). Since a HaeIII site was predicted to be lost in the presence of the C 3 T transition, the direct sequencing results from both genomic DNA ( Fig. 2A) and cDNA (Fig. 2B) were confirmed by HaeIII digestions of a strategic PCR fragment from both patients and normal individuals. In addition to the genomic PCR fragments from the five normal individuals shown in Fig. 2A, samples from at least 45 other normal individuals were analyzed and found not to contain this mutation (data not shown). Thus, the C 3 T transition in the Pro 504 codon is not present in either the 16-kb deletion or any of the 100 normal HEXB alleles we analyzed.
Interestingly, the residual Hex A activity present in the patients' fibroblasts, ϳ20%, is only about half that found in cells from an obligate carrier of Sandhoff disease (acute form), 5-10-fold higher than the average levels of five cell lines from subacute patients (Table I), and even slightly higher than those reported for asymptomatic individuals with low Hex A activity (10 -15%) (5,37,38). We also investigated the levels of ␣and ␤-CRM in the patient's cells and compared them to levels found in cell lines from a normal individual, a subacute patient (2.5% residual Hex A activity), and an acute patient (0% residual Hex A activity) (Fig. 3). The apparent levels of mature ␤-CRM in these samples were consistent with the decreased Hex A and B activities reported in Table I, indicating that the specific activity of the mutant Hex isozymes for artificial substrates had not changed. However, it was also apparent that there was a great increase in the ratio of precursor/mature forms of the ␣and/or ␤-polypeptides, suggesting that the ␤-Pro 504 3 Ser mutation results in the retention of a significant amount of newly synthesized pro-␤-chains in the ER (39,40) and probably a more rapid turnover rate (41). To confirm that the mutant mature polypeptides were not being degraded in the lysosome, normal and patient cells were grown in media containing leupeptin, which has been shown to inhibit the turnover of mutant ␤-chains in the lysosome (21,22). No dramatic increase in either Hex activity or mature ␤-CRM was observed (Fig. 4).
To fully characterize the biochemical effects of the Pro 504 3 Ser mutation, CHO cells were permanently co-transfected with two cDNAs encoding the normal ␣and the mutant ␤-polypeptides. A high producing clone was isolated and grown (Fig. 3). The Hex isozymes from the lysate of these cells were separated by ion exchange chromatography (Fig. 3). Since several mutations linked to the chronic form of G M2 gangliosidosis have been shown to produce a less heat-stable isozyme, as well as an increased retention of the mutant subunit in the ER (15, 16, 42,  The chromatofocusing profile of lysate from one of these cell lines has previously been reported (27). d The range of numbers presented are the range of activities measured in the two affected sisters' cell lines.
e This MUG/MUGS ratio, 1.5:1 (7) plus the slightly lower pH of elution (47) suggests that this activity is from a small amount of pro-Hex S; mature Hex S is well separated from Hex A by this procedure (27,47) (the pro-␣-chain loses basic amino acids during maturation (50)). 43), the T1 ⁄2 values of both Hex isozymes carrying the mutant ␤-subunit were determined at 45°C (Table II). Consistent with these previous observations, the ␤-Pro 504 3 Ser substitution decreases the heat stability of both the A and B isozymes (Table  II).

FIG. 3. Western blot analyses using an anti-human Hex A antibody of the ␣and ␤-polypeptides in the total cell lysates (the amount of protein loaded is given directly below the sample lane) from co-transfected CHO cells (with wild type
The ability of the ␤-Pro 504 3 Ser substitution to inhibit ER to Golgi transport was directly confirmed by immunofluorescence microscopy (Fig. 5). CHO cells permanently transfected with either the wild-type (Fig. 5B) or mutant ␤-cDNAs encoding the Pro 504 3 Ser (Fig. 5D) or Asp 208 3 Asn (Fig. 5C) substitutions were examined. The latter substitution results in only monomeric, precursor ␤-chains in transfected cells (23) and thus serves as a control for ER retention (39,44). Quantitative measurements of the green versus yellow (overlapping of the red and green labels in the cytoplasm, i.e. ER) fluorescence in 50 cells from each group indicated that virtually all of the Asp 208 3 Asn ␤-protein (97 Ϯ 2%) is present in the ER of transfected cells, compared with 60 Ϯ 10% of the ␤-Pro 504 3 Ser protein and 15 Ϯ 5% of the wild-type ␤-chain (Fig. 6). Furthermore, CHO cells co-transfected with the wild type ␣ and ␤ and those transfected with ␣and ␤-Pro 504 3 Ser were grown in 10 mM NH 4 Cl, and the MUG activity was measured in the media. Wild-type transfected cells secreted 1.1 ϫ 10 4 nmol/ h/ml on day 1 and 2.4 ϫ 10 4 units on day 2. Cells expressing the mutant ␤-cDNA secreted 0.18 ϫ 10 4 units on day 1 and 0.28 ϫ 10 4 units on day 2. Thus, diverting the mutant Hex from the lysosomes to the secretory pathway did not result in a large increase in activity, confirming that the loss of activity and ␤-CRM occurs at an early point in protein transport, i.e. the ER.
Due to the relatively high levels of residual, mutant Hex A activity, i.e. ␣␤-Pro 504 3 Ser (␣␤*, Hex A*), that we found in both our patients' cells and in co-transfected CHO cells, it remained difficult to explain why the patients should present with any disease phenotype. One possibility would be that the ␤-mutation had some direct effect on the function of the Hex A* isozyme. To address this question, we first examined the kinetic behavior of Hex A* with the common and ␣-specific substrates, MUG and MUGS, respectively. Kinetic analysis confirmed that Hex A* has the same apparent K m values as the wild type isozyme for these artificial substrates (Table II). We next tested the ability of the Hex A* to hydrolyze its natural substrate, the G M2 activator-G M2 ganglioside complex. Using samples of Hex A and Hex A* that contained the same number of MUGS units, we found that the mutant isozyme is 3-fold less active toward the natural substrate than is the wild type Hex A (Fig. 7, Table II). Furthermore, we confirmed that the residual Hex A in the patient's fibroblasts also had a decreased activity toward G M2 as compared with MUGS (Table II, G M2 /  MUGS ratio). Finally, we tested the stability of the wild type and mutant Hex A over the 18-h, 37°C incubation period used in the G M2 hydrolysis assay. Both forms of Hex A lost some activity toward MUGS over this time period; however, at the end of 18 h. the residual activity of the mutant form was only 15% less than the wild type (Table II). DISCUSSION We have previously demonstrated that two French Canadian patients with chronic Sandhoff disease are heterozygous for the common ⌬16kb HEXB allele (26,27). Since this allele produces no ␤-mRNA, the uncharacterized, second allele must be responsible for the 15-25% residual Hex A activity (using MUG) we reported to be present in the patients' fibroblasts and for their mild chronic phenotype. Western blotting with anti-␤ antiserum has also indicated a similar reduction in the amount of mature ␤-protein (27). In this report, we identify the second allele as a C1510T transition encoding a Pro 504 3 Ser substitution. We demonstrate that this substitution is not found in   (46). In the same report, they characterized another mutation associated with subacute Sandhoff disease, which was also characterized by the Hex Aϩ/Hex BϪ biochemistry. Both of these patients were also heterozygous for the ⌬16kb allele; thus, their biochemical phenotype was due to their second allele (26). In both cases, the second allele produced a partial splicing defect in the HEXB gene encoding an elongated ␤-polypeptide, i.e. a duplication of bp Ϫ16 to ϩ2 of IVS-13 3 exon 14 (asymptomatic) and g-26a IVS-12 (subacute) (38). It was also shown that the residual activity present in these patients' samples was from a small amount of properly spliced ␤-mRNA encoding the wild type protein. Activity measurements indicated that the asymptomatic individuals had twice as much residual Hex A activity as the subacute patients, 10 and 5% respectively (38). In this and other reports (26,27,47), we have included the cell line from the above subacute patient (g-26a IVS-12; ⌬16kb) in our analyses. In our hands, the residual Hex A activity in this line is 2-3% of normal, using artificial substrates (Table I) (27,47). This would suggest that Hex A levels of 4 -6% of normal should prevent G M2 storage and disease. This estimate is close to that set as the "critical threshold" by Sandhoff and colleagues (5,37). Given this critical threshold and our previous Hex A activity data, it has been difficult to explain why our two patients present with chronic G M2 gangliosidosis. Two possibilities were considered: first, that the ␤ mutation is somehow affecting the ␣ active site, lowering its activity toward MUGS and G M2 , e.g. a new type of B1-variant; second, the mutant ␤-subunit is affecting the ability of Hex A to bind the G M2 activator-G M2 ganglioside complex.
We now report the reexamination of residual Hex A* activities using the ␣-specific MUGS substrate (Table I) and the evaluation of both ␣and ␤-CRM levels in patients' cells using an anti-Hex A antiserum ( Fig. 3 and 4). These analyses confirmed our previous data, particularly that the levels of MUGS activity from Hex A* are 9 -23% of normal, and the ␤-CRM present in the cell line from the aforementioned subacute patient is much less than half that present in cells from our patient (Fig. 3). Thus, this substitution does not appear to  Table II. Note that for this assay the human activator is "species-specific"; i.e. endogenous CHO cell Hex A is virtually nonfunctional. specifically affect the ␣ active site, e.g. through some induced conformational change. However, to fully eliminate this possibility, we determined the K m of Hex A* for both the MUG and MUGS substrates. These were found to be normal (Table II).
Finally, we assessed the ability of the Hex A* produced in transfected CHO cells and semipurified from one of our patient's fibroblasts to hydrolyze its natural substrate, the G M2 activator-G M2 ganglioside complex (Fig. 7, Table II). These data demonstrate that the ␤-Pro 504 3 Ser mutation reduces the ability of the Hex A* to hydrolyze ganglioside in the presence of human activator by 3-fold (Table II). If this 3-fold reduction in the specific activity of Hex A* toward G M2 ganglioside, but not MUG or MUGS, is factored into our residual Hex A activity measurement (Table I), the patients' Hex A* activity is reduced to 3-9% of normal. This is very close to the critical threshold values we discussed above and is consistent with the chronic phenotype observed in the patients.
Recently, we (48) and others (49) have reported the characterization of ␣-␤ fusion proteins. Although some of our conclusions differed, both studies concluded that the C terminus of the ␤-polypeptide is important for the correct binding of the activator-ganglioside complex. The characterization of this novel, naturally occurring mutation strengthens these conclusions and identifies the region surrounding Pro 504 as the area in the C terminus most likely to be responsible for this function.