Catabolism of Asialo-GM2 in Man and Mouse

Tay-Sachs disease is an inborn lysosomal disease characterized by excessive cerebral accumulation of GM2. The catabolism of GM2 to GM3 in man requires β-hexosaminidase A (HexA) and a protein cofactor, the GM2 activator. Thus, Tay-Sachs disease can be caused by the deficiency of either HexA or the GM2 activator. The same cofactor found in mouse shares 74.1% amino acid identity (67% nucleotide identity) with the human counterpart. Between the two activators, the mouse GM2 activator can effectively stimulate the hydrolysis of both GM2 and asialo-GM2 (GA2) by HexA and, to a lesser extent, also stimulate HexB to hydrolyze GA2, whereas the human activator is ineffective in stimulating the hydrolysis of GA2 (Yuziuk, J. A., Bertoni, C., Beccari, T., Orlacchio, A., Wu, Y.-Y., Li, S.-C., and Li, Y.-T. (1998) J. Biol. Chem. 273, 66–72). To understand the role of these two activators in stimulating the hydrolyses of GM2 and GA2, we have constructed human/mouse chimeric GM2 activators and studied their specificities. We have identified a narrow region (Asn106–Tyr114) in the mouse cDNA sequence that might be responsible for stimulating the hydrolysis of GA2. Replacement of the corresponding site in the human sequence with the specific mouse sequence converted the ineffective human activator into an effective chimeric protein for stimulating the hydrolysis of GA2. This chimeric activator protein, like the mouse protein, is also able to stimulate the hydrolysis of GA2 by HexB. The mouse model of human type B Tay-Sachs disease recently engineered by the targeted disruption of the Hexa gene showed less severe clinical manifestation than found in human patients. This has been considered to be the result of the catabolism of GM2 via converting it to GA2 and further hydrolysis of GA2 to lactosylceramide by HexB with the assistance of mouse GM2 activator protein. The chimeric activator protein that bears the characteristics of the mouse GM2 activator may therefore be able to induce an alternative catabolic pathway for GM2 in human type B Tay-Sachs patients.

In man, the degradation of the GM2 1 ganglioside requires lysosomal ␤-hexosaminidase A (HexA) and a protein cofactor, the GM2 activator. The physiological importance of the GM2 activator is demonstrated by the severe clinical manifestations and neural accumulation of GM2 in type AB Tay-Sachs disease caused by the deficiency of this protein cofactor (1,2). The recent studies of the mouse model of type B Tay-Sachs disease (Hexa Ϫ/Ϫ ), generated via homologous recombination in embryonic stem cells, did not show the severe neurological symptoms characteristic in the same disease found in man (3)(4)(5)(6)(7). In these studies, the mild manifestations were initially attributed to the GM2-degrading activity of mouse HexB as reported by Burg et al. in 1983 (8). In contrast, we have shown that the highly purified mouse HexB was not able to convert GM2 to GM3, but was able to slowly catalyze the conversion of GA2 to LacCer in the presence of the mouse GM2 activator (mM2act) (9). In the same report, we also showed that mM2act was able to effectively stimulate the hydrolysis of GA2 catalyzed by either human or mouse HexA and that the human GM2 activator (hM2act) was not effective in stimulating the hydrolysis of GA2.
To better understand the role of hM2act and mM2act in the degradation of GM2 and GA2, we have constructed a series of human/mouse chimeric GM2 activators and studied their ability to stimulate the hydrolysis of GM2 and GA2 by human HexA. Since hM2act is not effective in stimulating the hydrolysis of GA2 by HexA, the specific human/mouse chimeras that elicit this activity should reveal the amino acids that are responsible for stimulating the hydrolysis of GA2.

Construction of Human/Mouse Chimeric cDNAs by Exon Swapping
The numbering system for the deduced amino acids in the following constructs was based on the alignment of the nucleotide sequences of * This work was supported by National Institutes of Health Grant NS 09626. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ To whom correspondence should be addressed: Dept. of Biochemistry SL 43, Tulane University School of Medicine, 1430 Tulane Ave., New Orleans, LA 70112. Tel.: 504-584-2459; Fax: 504-584-2739; E-mail: yli@tmcpop.tmc.tulane.edu. 1 The abbreviations used are: GM2, GalNAc␤134(NeuAc␣233) Gal␤134Glc␤131Ј-ceramide; GM3, NeuAc␣233Gal␤134Glc␤131Јceramide; GA2, GalNAc␤134Gal␤134Glc␤131Ј-ceramide; LacCer, lactosylceramide; HexA and HexB, ␤-hexosaminidases A and B, respec-hM2act and mM2act using the PC/GENE computer software program (Fig. 1). Three gaps in the human sequence and seven gaps in the mouse sequence were inserted to obtain the best alignment. Therefore, the numerical assignment for the amino acids in the human sequence differed from that in the mouse sequence by a factor of ϩ4 amino acids. All the gaps inserted were contained within the first 27 amino acids of the human sequence at the propeptide region, which is removed by proteolysis during the maturation of GM2 activator protein (13). The insertion did not at all affect the construction of the chimeras since all of the cDNA constructs started from Ser 32 , the N-terminal amino acid of the mature human protein. The first set of cDNA constructs for the activator chimeras were generated from exon swapping ( Fig. 2A). For the names of all constructs, the prefix "p" denotes plasmid that contains the designated cDNA. ph 2 h 3 m 4 -This construct contains human exons 2 and 3 plus mouse exon 4. A T7-7 plasmid vector containing the cDNA (p513) encoding only the 162 amino acids of mature hM2act (14) was used as template to generate, by polymerase chain reaction (PCR), a 445-bp cDNA fragment. This human cDNA fragment encodes 9 amino acids of the pT7-7 expression vector (Met-Ala-Arg-Ile-Arg-Ala-Arg-Gly-Ser) plus human exons 2 and 3. The upstream primer was 5Ј-TAA-TAC-GAC-TCA-CTA-TAG-GGA-GA-3Ј (T7 primer) within the pT7 vector region, and the downstream primer (noncoding) was 5Ј-GAG-TAG-GTA-CCT-TCT-TTG-3Ј with a built-in KpnI restriction site (underlined). This cDNA segment was digested with BamHI and KpnI. The remaining 164-bp cDNA fragment encoding mouse exon 4 was obtained by restriction digestion of pMact with KpnI and HindIII. The two cDNA segments were subcloned into the pT7-7 expression vector at BamHI and HindIII sites. ph 2 m 3 m 4 -This construct contains human exon 2 and mouse exons 3 and 4. A 177-bp DNA fragment encoding the 9 amino acids of the pT7-7 expression vector plus the first 50 amino acids (Ser 32 -Lys 81 ) of the mature hM2act sequence (13) encoded by exon 2 of the human sequence (also see Fig. 1) was generated by PCR amplification using p513 as template. The upstream primer was the T7 primer as described above. The downstream primer (noncoding) was 5Ј-ACG-GTG-AGC-TCC-ACC-TTC-3Ј, which contained a SacI restriction site (underlined). The remaining 112 amino acids encoded by mouse exons 3 and 4 were obtained by restriction digestion of the mM2act cDNA (pMact) (9) with SacI and HindIII to yield a 344-bp cDNA fragment encoding Val 78 -Arg 189 of mM2act. This mouse cDNA segment and the above PCR fragment of human exon 2, which had been digested with BamHI and SacI, were subcloned into a pT7-7 expression vector at BamHI and HindIII sites. ph 2 m 3 h 4 -This construct contains human exon 2, mouse exon 3, and human exon 4. The cDNAs encoding human exon 2 and mouse exon 3 were obtained by restriction digestion of the construct ph 2 m 3 m 4 using KpnI and HindIII to generate the ph 2 m 3 fragment. The remaining cDNA of the 164-bp fragment encoding human exon 4 was generated by PCR using p513 as template. The upstream primer was 5Ј-GAA-GGT-ACC-TAC-TCA-3Ј, in which the KpnI site is underlined. The downstream noncoding primer was the T7-7 primer 5Ј-GAT-GAT-AAG-CTT-GGG-CTG-3Ј, corresponding to the 3Ј-untranslated region of the pT7-7 vector. The amplified human fragment (207 bp) was digested with KpnI and HindIII and ligated into ph 2 m 3 at KpnI and HindIII sites. pm 2 m 3 h 4 -This construct contains mouse exons 2 and 3 plus human exon 4. The construct ph 2 m 3 h 4 was digested with PvuII and HindIII to yield the 324-bp cDNA fragment encoding Ser 103 -Glu 138 of the mouse sequence followed by the h 4 segment. This fragment was subcloned into mouse pMact that had been digested by PvuII and HindIII.

Construction of Chimeras from ph 2 m 3 h 4 with a Modified m 3 Segment
From the initial experiments, the m 3 segment appeared to be important for eliciting the stimulatory activity for the enzymatic hydrolysis of GA2. Therefore, we subsequently modified the m 3 segment in ph 2 m 3 h 4 by including more human sequence and generated the following constructs (also see Fig. 2B). ph 2 m 3 h 4 -a-This construct contains an extended human sequence (25 amino acids from Val 82 to Cys 106 ) at the N terminus of m 3 . The 226-bp fragment encoding Ser 32 -Cys 106 of hM2act was excised from p513 by restriction digestion with BamHI and PvuII. The segment of mouse exon 3 encoding Ser 103 -Glu 138 plus human exon 4 encoding Gly 143 -Ile 193 was obtained by digestion of ph 2 m 3 h 4 with PvuII and HindIII. The two fragments were ligated into the pT7-7 vector at its BamHI and HindIII sites. The detailed strategies used to generate these constructs are described under "Experimental Procedures." Shaded boxes represent the mouse sequence, and white boxes represent the human sequence. In each box, the amino acids of hM2act at the junctions are shown in the upper corners and those of mM2act in the lower corners. The numbering system for the human and mouse sequences follows that shown in Fig.  1. For convenience, the name of each construct without "p" was used to designate the expressed protein. ϩ, Ͼ80% hydrolysis; Ϫ, Ͻ30% or no hydrolysis; ϩ/Ϫ, between 30 and 80% hydrolysis under the assay conditions specified in the legends to Figs. 4, 5, and 7. ph 2 m 3 h 4 -b-This construct contains an extended human sequence (36 amino acids from Thr 107 to Glu 142 ) at the C terminus of m 3 . The 323-bp cDNA fragment encoding Thr 107 -Ile 193 of the human sequence was generated by PCR using p513 as template. The upstream primer was 5Ј-GGC-AGC-TGT-ACC-TTT-GAA-3Ј, containing a PvuII restriction site (underlined); and the downstream primer was the T7-7 primer as described above. The PCR product was digested with PvuII and HindIII and subcloned into ph 2 m 3 h 4 that had been cut by PvuII and HindIII. ph 2 m 3 h 4 -a-SH-This construct contains an extended human sequence from both ends of m 3 and keeps only Ser 103 -Pro 117 in the mouse sequence. The p513 clone was used as template to generate, by PCR, a cDNA fragment covering Thr 121 -Ile 193 of hM2act. The upstream primer was 5Ј-CCT-CCC-GGG-GAG-CCC-TGC-3Ј, containing a point mutation in the Thr 121 codon to give Pro 117 of mM2act and also to generate a SmaI restriction site (underlined). The T7-7 primer as described above was used as the downstream noncoding primer. This fragment was digested using SmaI and HindIII. Another 374-bp cDNA fragment encoding Ser 32 -Cys 106 of the human sequence followed by Ser 103 -Pro 116 of the mouse sequence was generated from the ph 2 m 3 h 4 -a clone by PCR. The upstream primer was the T7 primer as described above, and the downstream noncoding primer was 5Ј-GCT-CTC-CCC-GGG-AGG-AAT-3Ј, in which the SmaI restriction site is underlined. This fragment was digested with BamHI and SmaI. Then, both cDNA fragments were inserted at the BamHI and HindIII sites of the pT7-7 vector. ph 2 m 3 h 4 -a-NI-This construct contains an additional extension of 7 amino acids (Thr 107 -Asp 113 ) of the human sequence in ph 2 m 3 h 4 -a-SH, leaving only Leu 110 -Pro 117 in the mouse sequence. The p513 clone was used as template to generate a cDNA fragment by PCR to span from Thr 107 to Ile 193 of the human sequence. The upstream primer was 5Ј-GGC-AGC-TGT-ACC-TTT-GAA-CAC-TTC-TGT-GAC-CTG-ATA-G-AC-GAA-TAC-ATT-3Ј with a built-in PvuII site (underlined), and the T7-7 primer described above was used as the downstream noncoding primer. Nine point mutations (indicated in boldface) were introduced into the upstream primer to generate the fragment encoding Leu 110 , Ile 111 , Glu 113 , Tyr 114 , and Pro 117 of the mouse sequence. The amplified cDNA fragment was digested with PvuII and HindIII and subcloned into p513 at the corresponding restriction site. ph 2 m 3 m 4 -a-NI-This construct contains the human sequence from Ser 32 to Asp 113 followed by the mouse sequence from Leu 110 to Arg 189 . The portion from Ser 32 to Asp 113 of the human sequence plus Leu 110 -Pro 117 of the mouse sequence was generated by digestion of ph 2 m 3 h 4a-NI with BamHI and SmaI. This cDNA fragment was subcloned into mouse pMact that was digested with BamHI and SmaI.

Construction of Chimeras for Identifying 5 Crucial Amino Acids in the Mouse Sequence
Since ph 2 m 3 h 4 -a-SH contains only 5 amino acids (Asn 106 , Ile 107 , Glu 113 , Tyr 114 , and Pro 117 ) that are different from the human sequence, the following constructs were prepared to evaluate the importance of these amino acids (see Fig. 6).
p513ML-This construct is basically the human sequence except with the changes of Met 117 and Leu 118 to Glu 113 and Tyr 114 , respectively, to match the mouse sequence. The upstream primer 5Ј-GGC-A-GC-TGT-ACC-TTT-GAA-CAC-TTC-TGT-GAT-GTG-CTT-GAC-GAA-T-AC-ATT-3Ј with a built-in PvuII site (underlined) was used along with the T7-7 primer described above to generate, by PCR, the cDNA encoding the amino acid sequence from Thr 107 to Ile 193 of the human sequence. The p513 clone served as template. The five point mutations in the upstream primer (indicated in boldface) were introduced to create the replacement of human Met 117 and Leu 118 with Glu 113 and Tyr 114 , respectively, at the corresponding positions of the mouse sequence. The amplified fragment was digested with PvuII and HindIII and subcloned into the p513 clone at its equivalent restriction site.
p513MLT-This construct is essentially the same as p513ML with an additional change of Thr 121 of the human sequence to Pro 117 at the corresponding position of the mouse sequence. The primers used to generate p513ML were also used to construct p513MLT. However, the template for PCR was ph 2 m 3 h 4 -a-NI instead of p513.
p513HFMLT-This construct contains the human sequence except with the replacement of 5 amino acids with the corresponding Asn 106 , Ile 107 , Glu 113 , Tyr 114 , and Pro 117 from the mouse sequence. p513MLT was used as template to generate a cDNA fragment by PCR. The upstream primer was 5Ј-GGC-AGC-TGT-ACC-TTT-GAA-AAC-ATC-TGT-3Ј with a built-in PvuII site (underlined). Two point mutations (indicated in boldface) were introduced to provide Asn 106 and Ile 107 of the mouse sequence. The noncoding downstream primer was the T7-7 primer as described above. The PCR product was digested with PvuII and HindIII and subcloned into p513 at its corresponding restriction sites.

Expression and Purification of Chimeric Human/Mouse GM2 Activators
Each construct was verified by sequencing the cDNA prior to transforming the competent E. coli BL21(DE3) cells. The E. coli transformants were inoculated into 150 ml of LB medium containing 1 mg/ml ampicillin and incubated overnight at 37°C. The overnight culture was diluted at a ratio of 1:33 with fresh LB/ampicillin medium (30 ml/1 liter) and grown for ϳ4 h at 37°C. Expression of GM2 activator protein was then induced by addition of isopropyl-1-thio-␤-D-galactopyranoside at a final concentration of 1 mM, and the culture was grown for an additional 6 h. The cells were harvested by centrifugation at 6000 rpm for 15 min using a GS3 rotor in a Sorvall RC5C centrifuge. Expression, refolding, and purification of the human/mouse GM2 activator chimeras were carried out as described previously for the refolding of hM2act (14).

Enzyme Assay
Human HexA activity was determined using fluorogenic substrates (4-methylumbelliferyl N-acetylglucosaminide and 4-methylumbelliferyl N-acetylglucosaminide 6-sulfate) according to Potier et al. (15). An appropriate amount of HexA was incubated with 1.5 mM 4-methylumbelliferyl N-acetylglucosaminide or 4-methylumbelliferyl N-acetylglucosaminide 6-sulfate in 50 mM sodium citrate buffer (pH 5.0) in a total volume of 50 l at 37°C. After a preset time, 1.5 ml of 0.2 M sodium borate buffer (pH 9.8) was added to stop the reaction. The released 4-methylumbelliferone was determined using a Sequoia-Turner Model 450 fluorometer. One unit of enzyme activity is defined as the amount that liberates 1 mol of 4-methylumbelliferone/min at 37°C. This fluorogenic assay was used only to standardize the amount of HexA for each experiment.

Enzymatic Hydrolysis of GM2 and GA2
For the hydrolysis of GM2 and GA2, the reaction mixture contained 3 nmol of substrate in 40 l of 10 mM sodium acetate buffer (pH 5.0). The reactions were initiated by adding 20 milliunits of human HexA and terminated by adding 40 l of ethanol. The mixtures were dried under vacuum using a SpeedVac, redissolved in 20 l of chloroform/ methanol (2:1, v/v), and applied to a TLC plate. The plate was developed with chloroform/methanol/water (60:35:8, v/v/v), sprayed with diphenylamine reagent (16), and heated at 115°C for 15-20 min to visualize the glycolipids. The quantitative analysis of the glycolipid bands on the TLC plate was carried out using a Scan Jet 2C/ADF scanner (Hewlett-Packard Co.) and the NIH Image 1.41 program.

Western Blot Analysis
The recombinant human/mouse chimeric GM2 activators were analyzed by SDS-polyacrylamide gel electrophoresis using 15% gel (17). Proteins were electrophoretically transferred onto a nitrocellulose membrane in 20 mM Tris and 150 mM glycine buffer (pH 8.0) containing 20% methanol at 70 V for 1 h using a Bio-Rad transfer apparatus. The nitrocellulose membrane was first soaked with 1% milk powder and then overlaid with rabbit anti-hM2act antibodies (1:1500) (18) or rabbit anti-mM2act antibodies (produced by Cocalico Biological, Inc.) as the primary antibody followed by horseradish peroxidase-conjugated goat anti-rabbit IgG (1:2000) as the secondary antibody. The membrane was then developed in 3.4 mM 4-chloro-1-naphthol containing 0.01% hydrogen peroxide to produce purple bands.

Alignment of the Deduced Amino Acid Sequences of hM2act
and mM2act-The nucleotide sequences of hM2act and mM2act downstream from the initiation codon were analyzed using the PC/GENE program (Fig. 1). With 10 gap insertions in exon 1 (three in the human sequence and seven in the mouse sequence), the two sequences obtained the best alignment and showed 67% identity in their nucleotides. These gaps are upstream of the sequence encoding the mature form of hM2act (13). A higher degree of similarity (74.1% identical and 9.9% similar) was found in the deduced amino acid sequences when they were compared as mature proteins from Ser 32 to the C terminus of the human sequence (Fig. 1). The comparison based on the mature protein is appropriate in this study since all the chimeras were expressed only as the mature proteins.
Stimulatory Activities of the Parent hM2act and mM2act for the Hydrolysis of GM2 and GA2-Although hM2act and mM2act share a very high degree of homology and both are active in stimulating the enzymatic hydrolysis of GM2, their stimulatory activities for the hydrolysis of GA2 are distinctly different. Fig. 3A shows the time course of GM2 hydrolysis by human HexA in the presence of either hM2act or mM2act. Under the same condition, 1 g each of hM2act and mM2act showed comparable stimulatory activity for the hydrolysis of GM2 throughout the 10-, 20-, and 30-min incubations. This indicates that these two activator proteins have similar potency in stimulating the hydrolysis of GM2. However, when the hydrolysis of GA2 (Fig. 3B) was examined in the presence of the same amount of each activator as that used in Fig. 3A, mM2act showed a much more pronounced stimulatory activity than that exerted by hM2act. This result corroborates our previous report (9) that mM2act was able to effectively stimulate the hydrolysis of GA2 and that hM2act was ineffective in stimulating this reaction. Image scanning of the TLC plate showed that the stimulatory activity of hM2act for GA2 hydrolysis was only ϳ15-18% of that of mM2act. Based on this difference, we reasoned that any human/mouse chimeric activator that can elicit stimulatory activity similar to that of mM2act for the hydrolysis of GA2 should be attributed to the specific mouse sequence in the chimeric protein. All chimeras were therefore examined for their stimulatory activities for the hydrolyses of GM2 and GA2 using 1 g of the protein.
Characterization of Chimeric GM2 Activators from ph 2 h 3 m 4 , ph 2 m 3 m 4 , ph 2 m 3 h 4 , and pm 2 m 3 h 4 -Our strategy was to exchange selected parts of the human sequence with the corresponding mouse sequence and to examine the chimeras for their increased ability to stimulate the hydrolysis of GA2. This strategy was proven to be effective. Alternatively, one can start from the mouse sequence and substitute it with certain parts of the human sequence. This strategy would result in chimeras with diminished stimulatory activity for GA2 hydrolysis. Because the stimulatory activity of the activator protein can also be attenuated by factors such as poor protein refolding, the second approach might not provide clear results.
Taking advantage of the fact that the cDNAs of hM2act and mM2act share a high degree of homology and identical intron/ exon junctions (19,20), we first constructed four chimeric constructs (ph 2 h 3 m 4 , ph 2 m 3 m 4 , ph 2 m 3 h 4 , and pm 2 m 3 h 4 ) according to the four possible exchanges of exons (Fig. 2). All four proteins expressed by these chimeric constructs were fully active in stimulating the hydrolysis of GM2 (Fig. 4A, lanes 4 -7). These results indicate that the exchange of the corresponding exon segments between the human and mouse sequences did not alter the stimulatory activity of the chimeric proteins for the hydrolysis of GM2. These results were not unexpected since both hM2act and mM2act could effectively stimulate this reaction. We were intrigued, however, by the differences in the ability of these chimeras to stimulate the hydrolysis of GA2 (Fig. 4B). Under the same incubation conditions, those proteins encoded by ph 2 m 3 m 4 , ph 2 m 3 h 4 , and pm 2 m 3 h 4 , but not that by ph 2 h 3 m 4 , showed significant activity to stimulate the hydrolysis of GA2 (Fig. 4B, lanes 4 -7; also summarized in Fig. 2A). The protein encoded by ph 2 h 3 m 4 and the parent hM2act showed only the basal stimulatory activity for GA2 hydrolysis (Fig. 4B,  lanes 3 and 4). These results strongly suggest the possibility that the peptide segment encoded by mouse exon 3 (m 3 ) may be responsible for exerting the stimulatory activity on GA2 hydrolysis.

Characterization of Chimeric Activator Proteins Expressed by ph 2 m 3 h 4 -a and ph 2 m 3 h 4 -b-
Since the m 3 segment of mM2act appeared to be important for the hydrolysis of GA2, we carried out further analysis of this region. The presence of a PvuII site in both the human and mouse sequences at the equivalent position enabled us to replace the h 3 sequence with a section of the m 3 region and to generate two additional chimeras, ph 2 m 3 h 4 -a and ph 2 m 3 h 4 -b (Fig. 2B). In addition to having human exons 2 and 4 in both chimeras, the middle segment of ph 2 m 3 h 4 -a contained human Val 82 -Cys 106 followed by mouse Ser 103 -Glu 138 , whereas ph 2 m 3 h 4 -b contained mouse Val 78 -Cys 102 followed by human Thr 107 -Glu 142 (Fig. 2B). Again, both proteins expressed by ph 2 m 3 h 4 -a and ph 2 m 3 h 4 -b were fully active in stimulating the hydrolysis of GM2 (Fig. 5A, lanes 4  and 5). However, only ph 2 m 3 h 4 -a showed the characteristic of mM2act to stimulate the hydrolysis of GA2 (Fig. 5B, lane 4). Judging from the substrate (GA2) remaining and the product (LacCer) formed, hM2act had only a basal activity and ph 2 m 3 h 4 -b had practically no stimulatory activity for the GA2 hydrolysis (Fig. 5B, lanes 3 and 5). These results indicate that Ser 103 -Glu 138 of the mouse sequence (see Fig. 1) may be responsible for eliciting the stimulatory activity for the hydrolysis of GA2. In this region, only 7 amino acids are different between the mouse and human sequences (Asn-Ile to His-Phe, Glu-Tyr to Met-Leu, Pro to Thr, Ser to Pro, and His to Arg, respectively), whereas 4 other amino acids (Ser-Tyr to Thr-Phe and Leu-Ile to Val-Leu) are similar (Fig. 1).
Characterization of the Proteins Expressed by ph 2 m 3 h 4 -a-SH, ph 2 m 3 h 4 -a-NI, and ph 2 m 3 m 4 -a-NI-Based on the above results, we proceeded to further narrow down the region between Ser 103 and Glu 138 of the mouse sequence. Using the specific primers and PCR, we generated the construct ph 2 m 3 h 4 -a-SH, which contained human exon 2 followed by a segment of human exon 3 up to Cys 106 and then the mouse sequence from Ser 103 to Pro 117 and, again, back to the human sequence from Gly 122 until the C terminus of the activator protein (Fig. 2B). This construct was similar to ph 2 m 3 h 4 -a except that Ser 120 and His 126 of the mouse sequence were changed to Pro 124 and Arg 130 , respectively, to match the human sequence. The protein obtained from this construct was fully active for the stimulation of GM2 hydrolysis (Fig. 5A, lane 6) and also had ϳ70% activity for the hydrolysis of GA2 as compared with that of mM2act (Fig. 5B, lanes 6 and 11). It was therefore necessary to further clarify the involvement of Ser 120 and His 126 of the mouse sequence in the stimulation of GA2 hydrolysis. We subsequently constructed two chimeras, ph 2 m 3 h 4 -a-NI and ph 2 m 3 m 4 -a-NI (Fig. 2B). The construct ph 2 m 3 h 4 -a-NI was basically identical to ph 2 m 3 h 4 -a-SH except for the substitution of Asn 106 and Ile 107 (NI denotes the one-letter code of the 2 amino acids) in the mouse sequence with the corresponding human His 110 and Phe 111 residues. Another chimera, ph 2 m 3 m 4 -a-NI, contained the human sequence from the N terminus up to Asp 113 followed by the mouse sequence from Leu 110 until the C terminus (Fig. 2B). As expected, the proteins obtained from these two constructs had the stimulatory activity for the hydrolysis of GM2 (Fig. 5A, lanes 7 and 8). However, both of them were very weak in stimulating the hydrolysis of GA2 (Fig. 5B,  lanes 7 and 8). Only using an increased amount of protein could we detect some hydrolysis of GA2 (Fig. 5B, lanes 9 and 10). Comparison of the stimulatory activities of ph 2 m 3 h 4 -a, ph 2 m 3 h 4 -a-NI, and ph 2 m 3 m 4 -a-NI (Fig. 5B, lanes 4, 7, and 8) clearly suggested that within the m 3 region, at least 2 amino acids (Asn 106 and Ile 107 , NI) in the mouse sequence are very important for eliciting the stimulatory activity for the hydrolysis of GA2. Both proteins obtained from ph 2 m 3 m 4 -a-NI (with the mouse Ser 120 and His 126 residues) and ph 2 m 3 h 4 -a-NI (without the mouse Ser 120 and His 126 residues) showed very poor activity in stimulating the hydrolysis of GA2. This indicates no or little involvement of Ser 120 and His 126 in exerting the stimulatory activity for the hydrolysis of GA2.
Characterization of the Chimeric Proteins Expressed by p513ML, p513MLT, and p513HFMLT-To further clarify the involvement of Asn 106 , Ile 107 , Ser 120 , and His 126 in mM2act for expressing the stimulatory effect on GA2 hydrolysis, it was necessary to generate three additional chimeras: p513ML, p513MLT, and p513HFMLT. The region of amino acid sequences between Lys 96 and His 137 of these chimeras together with those of hM2act (p513) and mM2act are listed in Fig. 6. The construct p513ML has 98.76% of the human sequence except for Met 117 and Leu 118 (ML), which were substituted with the corresponding Glu 113 and Tyr 114 residues of the mouse sequence. The chimera p513MLT contains the same changes as in p513ML plus the substitution of Thr 121 of the human sequence with Pro 117 of the mouse sequence (Fig. 6). The third construct, p513HFMLT contains the crucial Asn 106 and Ile 107 residues of the mouse sequence in addition to the changes in p513MLT. This construct has 96.91% of the human sequence except for the 5 amino acids (His 110 , Phe 111 , Met 117 , Leu 118 , and Thr 121 ) being replaced by the corresponding Asn 106 , Ile 107 , Glu 113 , Tyr 114 , and Pro 117 residues of the mouse sequence. All proteins obtained from these three chimeras were able to stimulate the hydrolysis of GM2 by HexA (Fig. 7A, lanes 5-7). However, only the protein expressed by p513HFMLT showed a full activity like mM2act in stimulating the hydrolysis of GA2 (Fig. 7B, lane 5). The protein expressed by ph 2 m 3 m 4 -a was included here for comparison because this protein showed the best activity among those examined in the previous experiments (Figs. 5B, lane 4; 7B, lane 4). These results underscore the importance of Asn 106 and Ile 107 in eliciting the stimulatory activity for the hydrolysis of GA2 by HexA. Although both the proteins obtained from the constructs p513MLT and p513ML were weak in stimulating the hydrolysis of GA2 (Fig. 7B, lanes  6 and 7), the former had better activity than the latter. Only with a prolonged incubation of 16 h (Fig. 7B, lanes 9 and 10) did these two proteins show detectable activity for this reaction.
Characterization of the Protein Expressed by p513ML-Ala 132 -During the attempt to produce the chimera p513ML by PCR, we accidentally obtained a construct that has the same sequence as p513ML, but that also contains a 3-bp insertion for an extra Ala at position 132. This construct is therefore called p513ML-Ala 132 . Interestingly, this mutant was able to stimulate neither the hydrolysis of GM2 (Fig. 7A, lane 8) nor the hydrolysis of GA2 (Fig. 7B, lane 8). Even with a prolonged incubation, the protein expressed by p513ML-Ala 132 was completely inactive in stimulating the hydrolysis of GM2 (Fig. 7A, lane 9) and GA2 (Fig. 7B, lane 11). This protein was used as a negative control for all incubations. DISCUSSION We have previously shown that although human HexA is capable of hydrolyzing GM2 in the presence of either hM2act or mM2act, it can only effectively degrade GA2 in the presence of mM2act (9). These results indicate that human HexA is capable of hydrolyzing both GM2 and GA2 and that mM2act plays a specific role in assisting the enzyme to hydrolyze GA2. Since hM2act and mM2act share a very high degree of homology, it is reasonable to search for the amino acids that are responsible for the differences in their ability to stimulate the hydrolysis of GA2. If such amino acids can be identified, it should be possible to construct a human/mouse chimera by replacing the specific amino acids in the human sequence with the mouse sequence, thereby converting the ineffective hM2act into an effective protein capable of stimulating the enzymatic hydrolysis of GA2. Through this engineering, hM2act can acquire an extra stimulatory activity for the hydrolysis of GA2, in addition to its native role for stimulating the degradation of GM2.
The alignment of the amino acid sequences of hM2act and mM2act from amino acid 32 to the C terminus showed 74.1% identity (Fig. 1). In addition, the characterization of the genomic structures of hM2act (19) and mM2act (20) revealed that the splice junctions of the two genes are completely conserved. This suggests that the introns in the chromosomes of the two species may contribute similarly to the processes of making the final mature proteins. Therefore, it is logical to use exon swapping of the human and mouse sequences to minimize the alteration of the structural integrity of the protein. By SDS-polyacrylamide gel electrophoresis under nonreducing conditions, all chimeric proteins showed one major band with Coomassie Brilliant Blue staining. All proteins cross-reacted with both anti-hM2act (18) and anti-mM2act (data not shown) antibodies in Western blot analysis. Although not quantitative, these results indicate that the chimeric proteins do preserve the necessary structural features to be recognized by the antibodies.
The initial results from the chimeras obtained by exon swapping (Fig. 3) supported our rationale concerning the possibility of locating the active domain in mM2act responsible for the stimulatory activity for GA2 hydrolysis. Our first set of results clearly indicated that the active domain was between Ser 103 and Glu 138 of the mouse sequence. The proteins expressed by ph 2 m 3 h 4 -a-NI and ph 2 m 3 m 4 -a-NI, both devoid of mouse Asn 106 and Ile 107 , showed a diminished activity to stimulate the hydrolysis of GA2 as compared with that expressed by ph 2 m 3 h 4 -a and mM2act (Fig. 5B, lanes 7 and 8). The importance of Asn 106 and Ile 107 in the mouse sequence for the hydrolysis of GA2 was further supported by the very weak stimulatory activity observed in the proteins expressed by the constructs p513ML and p513MLT, which did not contain these 2 amino acids. The protein expressed by p513MLT was more active than that expressed by p513ML (Fig. 7B, lanes 6 and 7) in stimulating the hydrolysis of GA2. These results indicate that Pro 117 in the mouse sequence is another important amino acid in eliciting the stimulatory activity for the hydrolysis of GA2. Substitution of Thr 121 in the human sequence with Pro to match the Pro 117 residue in the mouse sequence did enhance the chimeric pro- tein to stimulate the hydrolysis of GA2. The fact that the protein expressed by p513HFMLT was as active as mM2act in stimulating the hydrolysis of GA2 (Fig. 7B, lanes 5 and 12) indicated that the region from Ser 120 to the C terminus of the mouse sequence was not crucial for the stimulatory activity of GA2 hydrolysis. From the diminished activity of the protein expressed by ph 2 m 3 h 4 -a-SH for the hydrolysis of GA2, we further concluded that the presence of Pro 124 next to Cys 125 in the human sequence might form a conformation unfavorable for GA2 hydrolysis. The protein expressed by p513HFMLT has only 5 out of 162 amino acids (3.2%) different compared with hM2act. However, this protein has the full activity compared with that of mM2act in stimulating the hydrolysis of GA2. These results clearly indicate that the region between Asn 106 and Pro 117 of the mouse sequence is extremely important for the stimulatory activity for GA2 hydrolysis.
A recent study (21) reported the existence of four pairs of putative disulfide bonds in hM2act. They were Cys 39 -Cys 183 , Cys 99 -Cys 106 , Cys 112 -Cys 138 , and Cys 125 -Cys 136 . The latter two pairs of disulfide bonds were suggested to form a clamp to stabilize the region in hM2act that is equivalent to the mouse region between Asn 106 and Pro 117 . Our results indicate that this region in the mouse sequence is important for the stimulation of GA2 hydrolysis. The m 3 segment is rich in Cys residues that are completely conserved between hM2act and mM2act. A Cys 138 mutation in hM2act has been characterized in a human type AB Tay-Sachs patient by Schröder et al. (22) and Xie et al. (23). In this patient, the mutation of thymidine to cytidine at position 412 of the full-length hM2act cDNA (19) caused the substitution of Cys 138 with Arg and abolished the activator's stimulatory activity for GM2 hydrolysis. In a recent study, Xie et al. (24) further characterized the C138R substitution in hM2act. They suggested that the mutation specifically affected the domain in the activator protein that was responsible for the recognition of the activator-ganglioside complex by HexA. This suggestion agrees with our observation that the same region in the mouse sequence seems to be crucial for HexA to hydrolyze GA2. Xie et al. also suggested that the mutation might cause localized changes in the mutant protein without major changes in the secondary or tertiary structure of the protein. The fact that all of our chimeras showed good activities in stimulating the hydrolysis of GM2 indicates that the conformation of the chimeric proteins must remain very close to that of the native protein. Moreover, all of the chimeras in this study did not involve any Cys substitution; thus, no direct disturbance was introduced into their disulfide bond formation. The specific action of the mouse sequence between Asn 106 and Pro 117 for GA2 hydrolysis cannot be fully understood until the revelation of the crystal structures of hM2act, mM2act, and the chimeric protein expressed by p513HFMLT.
It is remarkable that the substitution of a very small region (5 amino acids) in the human activator sequence with the specific mouse sequence can create a mutant protein (p513HFMLT, 96.8% human and 3.2% mouse) effective in stimulating the degradation of both GM2 and GA2 carried out by human HexA. We have further examined the stimulatory activities of hM2act, mM2act, and the chimeric protein expressed by p513HFMLT in the hydrolysis of GM2 and GA2 carried out by human HexB. It is known that human HexB is not able to catalyze the hydrolysis of GM2. We also observed that human HexB was not able to hydrolyze GM2 in the presence of any one of the three activators. However, human HexB could hydrolyze GA2 in the presence of either mM2act or the protein produced by p513HFMLT, but not in the presence of hM2act. Although the hydrolysis of GA2 by HexB is relatively slow, at a rate ϳ5% of that by human HexA, this activity could be important in the classical Tay-Sachs patients who are devoid of HexA and have only HexB. This importance has been illustrated in the milder clinical conditions in the murine model for type B Tay-Sachs disease (targeted disruption of the Hexa gene) (3-7) whose activator is effective in stimulating HexB to hydrolyze GA2. In human Tay-Sachs patients, their hM2act activators are not able to stimulate HexB to hydrolyze GA2. Therefore, based on our results, the chimeric protein expressed by p513HFMLT should be theoretically able to induce an alternative catabolic pathway for GM2 in human through the degradation of GA2. Thus, the chimeric activator protein may be ultimately useful for type B Tay-Sachs patients who can be benefited from the induction of the alternative catabolic pathway for GM2.