Active Site Residues of Human β-Glucuronidase

Human β-glucuronidase (hGUSB) is a member of family 2 glycosylhydrolases that cleaves β-d-glucuronic acid residues from the nonreducing termini of glycosaminoglycans. Amino acid sequence and structural homology of hGUSB and Escherichia coliβ-galactosidase active sites led us to propose that residues Glu451, Glu540, and Tyr504 in hGUSB are involved in catalysis, Glu451 being the acid-base residue and Glu540 the nucleophile. To test this hypothesis, we introduced mutations in these residues and determined their effects on enzymes expressed in COS cells and GUSB-deficient fibroblasts. The extremely low activity in cells expressing Glu451, Glu540, and Tyr504 hGUSBs supported their roles in catalysis. For kinetic analysis, wild type and mutant enzymes were produced in baculovirus and purified to homogeneity by affinity chromatography. Thek cat/K m values (mm −1·s−1) of the E540A, E451A, and Y504A enzymes were 34,000-, 9100-, and 830-fold lower than that of wild type hGUSB, respectively. High concentrations of azide stimulated the activity of the E451A mutant enzyme, supporting the role of Glu451 as the acid-base catalyst. We conclude that, like their homologues in E. coli β-galactosidase, Glu540 is the nucleophilic residue, Glu451 the acid-base catalyst, and Tyr504 is also important for catalysis, although its role is unclear. All three residues are located in the active site cavity previously determined by structural analysis of hGUSB.

Lysosomal ␤-glucuronidase (EC 3.2.1.31) is an essential catabolic enzyme that is involved in the degradation of sulfated glycosaminoglycans. Deficiency of ␤-glucuronidase (GUSB) 1 in humans produces a mucopolysaccharide storage disease referred to as mucopolysaccharidosis type VII (Sly syndrome) (1,2). In the absence of GUSB, chondroitin sulfate, dermatan sulfate, and heparan sulfate are only partially degraded and accumulate in the lysosomes of many tissues. The GUSB enzyme, synthesized as an 80-kDa glycoprotein monomer precursor (653 amino acids), is processed to a 78-kDa monomer by proteolytic cleavage, removing 18 amino acids from the C terminus (3,4). Mature GUSB is normally a homotetramer, but there is evidence that the homodimer can also be enzymatically active (5).
Glycosidases function by using one of two general mechanisms leading either to retention or inversion of the anomeric configuration at the hydrolysis site (6,7). In either case, two acidic residues, usually two glutamic acids, participate directly in catalysis. One amino acid acts as a catalytic nucleophile and the other as an acid-base catalyst or the proton donor. A number of mutations in hGUSB result in complete to partial loss of in vitro activity and have been associated with different disease phenotypes (8 -12). However, none of these amino acids have been established as essential to the catalytic mechanism.
Based on the effects of salt, pH, and group-specific chemical reagents on the activity, Wang and Touster (19) proposed that a carboxylic acid and a carboxylate anion are the catalytic functional groups. Another approach to predict candidate catalytic residues is by sequence comparison with homologous enzymes whose active site residues have been identified. According to a recent classification based on amino acid sequence similarity, hGUSB was placed into family 2 together with Escherichia coli ␤-galactosidase (EGAL) (13,14). X-ray crystal structure (15), inhibitor studies (16), and site-directed mutagenesis (17) studies of EGAL unequivocally established that the important catalytic residues include Glu 537 as the nucleophile and Glu 461 as the acid-base catalyst. Tyr 503 was also found to be important for catalysis, but its role is not yet clear (18). From a sequence comparison of hGUSB with EGAL and a number of additional bacterial ␤-galactosidases, the candidate residues which correspond to the Glu 537 , Glu 461 , and Tyr 503 in EGAL were identified as Glu 540 , Glu 451 , and Tyr 504 , respectively, in hGUSB ( Fig. 1). Hydrophobic cluster assay, where homologous folds in glycosidases within the same family were compared, also predicted that Glu 540 might be the nucleophile, whereas Glu 451 might be the acid catalyst in hGUSB (20). On the other hand, by comparing the x-ray crystal structure of hGUSB with those of lysozyme and EGAL, Jain et al. (21) proposed that the Asp 207 -Glu 451 pair might form the nucleophile-acid-base catalyst pair in hGUSB, analogous to the Glu 35 -Asp 52 pair in lysozyme.
In this report, we modified all the candidate residues proposed to be important for catalysis in hGUSB using site-directed mutagenesis. From the enzymatic activity and kinetic analyses, we concluded that Glu 540 -Glu 451 , not Asp 207 -Glu 451 , forms the nucleophile-acid catalyst pair in hGUSB. Furthermore, Tyr 504 , the residue analogous to Tyr 503 in EGAL and also located in the active site cavity of hGUSB (20), is also important for catalysis.

EXPERIMENTAL PROCEDURES
Materials-M13mp18, DEAE dextran, and nucleotides for DNA sequencing were from Amersham Pharmacia Biotech. Enzymes for molecular biology were from Omega and Promega, except for Sequenase®, which was from U. S. Biochemical Corp. Chloroquine and 4-methylumbelliferyl-␤-glucuronide were from Sigma. LipofectAMINE was from Life Technologies, Inc. Tissue culture medium was from Life Technologies, Inc. Tran 35 S-label was from ICN, IgGsorb was from the Enzyme Center (Malden, MA), and EN 3 HANCE was from NEN Life Science Products. The bicinchoninic acid protein assay kit was from Pierce.
Construction of Mutant cDNAs-The mutations were generated with a single-strand mutagenesis system from Amersham Pharmacia Biotech in M13 vector and with a double-strand system using CLONTECH For all mutants except D207A, a 428-base pair fragment between SacII (1322) and SacI (1750) was excised from the respective mutant clone and swapped with the wild type fragment in human cDNA that had been previously cloned into the EcoRI site of expression vector pJC119RI (22). For the D207A mutant, a 262-base pair fragment generated by digestion with ApaI (478) and BglII (740) was exchanged with the normal fragment between ApaI-BglII in pJC119RI vector. All mutant fragments transferred into the wild type were verified to exclude undesired mutations by DNA sequencing using the dideoxy chain termination method (23). The full-length mutant cDNAs constructed in this way were subcloned into the EcoRI site of expression vector pCAGGS (24) or Backpack-8, a baculovirus transfer vector (25).
Transfection, Metabolic Labeling, and Immunoprecipitation-COS-7 cells (26) were transfected with cDNAs in pJC119RI using the DEAEdextran method. Mouse GUSB-deficient 3521 cells were transfected with the cDNAs in pCAGGS using 6 l of LipofectAMINE and 3 g of plasmid DNA in a total volume of 200 l in 35-mm Petri dishes. Media were collected, and cells were solubilized in 0.6 ml of 0.25% sodium deoxycholate 76 -78 h after the start of transfection. Metabolic labeling with Tran 35 S-label, chase with unlabeled media, and harvest of cells and media followed by immunoprecipitation with anti-hGUSB antibody were done as described previously (3).
Clonal recombinant baculoviruses containing wild type and E540A, E451A, and Y504A mutant cDNAs were produced according to the manufacturer's instructions (CLONTECH) by transfecting SF21 cells in p35 Petri dishes with SauI-digested viral genome and transfer vector, Backpack-8 containing the wild type, or mutant cDNA. The plaques obtained with the wild type hGUSB were screened by GUSB activity, whereas the plaques obtained for mutants were screened by enzymelinked immunosorbent assay using a polyclonal anti-hGUSB antibody. The recombinant baculoviruses containing the wild type or the mutant cDNAs were amplified in SF21 cells. A 500-ml SF21 culture was then infected with medium containing the amplified virus to produce the wild type or mutant GUSB enzymes, which were secreted into medium.
Affinity Chromatography-The wild type and the mutant enzymes produced in the infected SF21 culture medium were purified by affinity chromatography on monoclonal antibody columns of 1-ml bed volume as described (3,27). After elution, the enzymes were dialyzed against 20 mM Tris-HCl containing 50 mM NaCl, pH 7.5. Purity of the enzyme preparations was determined by SDS-polyacrylamide gel electrophoresis (28), followed by staining with Coomassie Blue R-250.
Gel Filtration-The purified enzymes were passed over a TSK G3000 SW column connected to an isocratic high pressure liquid chromatography system. Column equilibration and sample elution were carried out with 0.1 M phosphate buffer containing 0.1 M Na 2 SO 4 , pH 6.7.
Assay of Mutant Proteins-␤-Glucuronidase activity was determined using 4-methylumbelliferyl-␤-D-glucuronide. One unit is the amount of activity that releases 1 nmol of 4-methylumbelliferone/h (29). Protein concentrations were measured by bicinchoninic acid protein assay kit according to the manufacturer's instructions using bovine serum albumin as standard.
Kinetics-The pH profiles for wild type and mutant enzymes were determined by adding 10 l of the enzymes to 100 l of 12.5 mM 4-methylumbelliferyl ␤-D-glucuronide in the buffers of the respective pH levels (0.1 M sodium acetate, pH 3-5.5, 0.1 M Tris-HCl, pH 6.0 -8.0) followed by incubation at 37°C for 30 min. The kinetic parameters were determined by assaying ␤-glucuronidase activity at 37°C in 0.2 M acetate buffer at the respective pH optima with 0.5, 1, 2, and 4 mM 4-methylumbelliferyl ␤-D-glucuronide. The K m and V max were obtained from a double-reciprocal plot of initial substrate concentration versus rate of product formed.

Expression in COS and 3521
Cells-The mutants constructed by changing the wild type residues Glu 451 , Tyr 504 , and Glu 540 were E451A, E451Q, Y504A, Y504H, Y504F, E540A, E540Q, and E540D. For comparison, we also made changes in Tyr 508 , Glu 515 , and Asp 207 , which were alternate candidates for the nucleophile and acid-base catalyst, producing E515A, Y508A, and D207A. All of these mutants were transiently expressed in COS cells from the SV40 late promoter in vector pJC119. ␤-Glucuronidase activity was measured in cells and medium, and the sum of activity in cells and medium produced by each mutant during the 76 h following transfection was determined ( Table I). The mutants thought to be candidates for involvement in catalysis were also expressed in 3521 cells, a mouse cell line with no endogenous ␤-glucuronidase activity (3,10). Unlike with COS cells, where endogenous GUSB activity makes it difficult to demonstrate the low activity of mutant  enzymes, the 3521 cell transfections with wild type and mutant cDNAs expressed in pCAGGS vector allow characterization of low activity mutant enzymes (24). Table I shows that different mutants of Glu 540 , the residue homologous to nucleophilic residue Glu 537 in EGAL, all had greatly reduced residual activity when expressed in COS cells and 3521 cells. The D207A mutant, which Jain et al. (21) had suggested might be the nucleophile, had 1.5% (COS cells) and 4.9% (3521 cells) of wild type activity, more than would be expected if it were the nucleophilic residue.
The E451A mutant transfections produced only 0.6% of wild type activity in both cell types, whereas the E451Q mutant produced 1.5% (COS cells) and 5.9% (3521 cells) of wild type activity. These low activities are consistent with Glu 451 being the acid-base catalyst. The higher activity of E451Q could be explained by a low level of lysosomal deaminase activity converting the Q to E. These results contrast with the relatively high activity of the E515A mutant, which excludes it as an important residue in catalysis.
The three different mutants of Tyr 504 , which is homologous to the required 503 in EGAL, had activities ranging from 0.1-0.5% in COS cells and 0.6 -2.4% in 3521 cells. By contrast, Y508A had 14.7% of the wild type activity in COS cells. Taken together, these data suggest that Glu 540 , Glu 451 , and Tyr 504 in hGUSB have comparable roles in catalysis as their homologues in EGAL, i.e. that Glu 540 is the nucleophilic residue, Glu 451 is the acid-base catalyst, and Tyr 504 is also important for catalysis like Tyr 503 in EGAL, whose role in catalysis is not yet defined, although it clearly is located in or near the active site in EGAL (15) as is Tyr 504 in the active site of GUSB (21).
To compare synthesis, processing, and secretion of wild type and mutant enzymes, we carried out metabolic labeling of transfected COS cells. As shown in Fig. 2, when labeled for 1 h, the biosynthesis of all the mutant enzymes tested appeared comparable with that of the wild type. After a 24-h chase, all but one mutant (E540D) showed processing to the mature form (which is known to involve removal of the C-terminal 18 amino acid residues) and were secreted in amounts comparable to that of the wild type enzyme. E540D was exceptional. Even though some of the E540D enzyme was secreted, no processed enzyme was evident, and the intracellular enzyme appeared to be more rapidly degraded (Fig. 2, lanes 5 and 6). These observations could mean that some of the E540D mutant enzyme was retained in the endoplasmic reticulum and underwent endoplasmic reticulum-mediated degradation (31) or that it was rapidly degraded after delivery to lysosomes. The fact that the other mutants were processed normally to the mature form and secreted into the medium indicated that they were properly folded and were recognized by receptors in the secretory and lysosomal targeting pathways and by the processing enzyme(s) in endosomes and/or lysosomes.
Purification and Kinetic Parameters-To purify and characterize the mutant enzymes without contaminating wild type endogenous enzyme, the E540A, E451A, and Y504A mutants were transferred to the baculovirus genome by homologous recombination and produced in SF21 insect cells. The enzymes secreted into the growth medium were purified by affinity chromatography. As shown in Fig. 3, the purified mutant enzymes had the same migration patterns on SDS-PAGE as the wild type, except for Y504A, which had slightly faster migration. This faster mobility of Y504A could be due to a different level of glycosylation in SF21 cells. In fact, when the mutant enzyme was expressed in COS cells (Fig. 2, lanes 19 -21), it showed the same mobility as the wild type enzyme. Size exclusion chromatography on TSK gel revealed that all three mutant enzymes were tetrameric (not shown).
The kinetic parameters obtained from the assays of the purified wild type and mutant enzymes are presented in Table II. The k cat /K m values were decreased 33,000-fold in E540A, 9,100fold in E451, and 830-fold in Y504A mutant enzymes. However, the K m values were similar to the wild type hGUSB. These kinetic studies are consistent with our assignments of Glu 540 as the nucleophile, Glu 451 as the acid-base catalyst, and Tyr 504 as an important active site residue whose role is still not established.
pH Optima and Heat Stability-The enzyme activities of the wild type and mutant enzymes at different pH levels are shown in Fig. 4. Y504A had a broad pH activity profile between pH 3.0 and 8.0, similar to that of the wild type enzyme. The E540A mutant enzyme had an optimum at pH 5.0 instead of 4.5. The E451A enzyme showed a broader pH optimum (pH 4.0 -5.0), and its activity drop with increasing pH was more gradual than that of wild type GUSB. It retained 35% of its activity at pH 8. Fig. 5 compares the heat stability of wild type and mutant enzymes. Human GUSB is relatively stable to heat inactivation. Less than 40% of the activity was inactivated by heating to 68°C for 2 h at pH 7.5 in 75 mM NaCl and 5 mg/ml bovine serum albumin. Y504A activity was as stable as the wild type up to 2 h. E540A activity was inactivated at a faster rate than the wild type. The E451A enzyme was completely inactivated within 30 min at 68°C. The greater heat lability of the E540A and the E451A enzymes suggests that the carboxyl groups of Glu 540 and Glu 451 contribute to stability of the enzyme. Jain et al. (21) suggested from structural analysis that Glu 540 forms salt bridges to His 385 and Arg 382 , which probably contribute to stability of the wild type enzyme, and Glu 451 is surrounded by three Asn residues.
Effect of Sodium Azide-MacLeod et al. (32) proposed that stimulation of activity of mutant enzymes by azide can be used to identify the acid-base catalyst in retaining type hydrolases for substrates that do not require protonic assistance for initial bond cleavage. They inferred from stimulation of E127A in exoglucanase/xylanase that this residue was the acid-base catalyst. When we studied the effects of azide on wild type and mutant GUSB activities, wild type hGUSB was found to be inactivated with increasing sodium azide concentrations (Fig.  6). Such inhibition was not noted by MacLeod et al. (32), and its basis is unclear. Like the wild type enzyme, the E451A mutant enzyme also showed inhibition by azide and lost 70% of its original activity in 50 mM azide. However, concentrations of azide between 50 mM and 0.5 M stimulated activity of the E451A enzyme. Activity was 4-fold greater at 500 mM azide than that seen at 50 mM azide. Further increase in azide concentration to 1 M inhibited the E451A enzyme like the wild type enzyme. Azide had no effect on the extremely low activity of the E540A mutant enzyme. DISCUSSION Recent classification of glycosyl hydrolases based on comparison of amino acid sequences in the active sites placed hGUSB into family 2 together with EGAL (6,7). The active site of the latter has been studied in great detail, and two glutamate/ glutamic acid residues (Glu 537 and Glu 461 ) were identified to be involved in catalysis (15)(16)(17)(18). Recently, the three-dimensional structure of EGAL confirmed that Glu 537 and Glu 461 are in the active site cleft and positioned at a distance that would be consistent with them forming the nucleophile and acid-base catalyst pair and their participating in the retaining type catalysis. In addition, early mutational studies had identified Tyr 503 in EGAL as an important catalytic residue. Although structural studies show that Tyr 503 forms part of the active site cavity, its role in catalysis is still unknown (16). Amino acid sequence comparison of hGUSB with mouse, rat, or E. coli GUSB and EGAL revealed the three residues in hGUSB that correspond to Glu 537 , Glu 461 , and Tyr 503 in EGAL to be Glu 540 , Glu 451 , and Tyr 504 , respectively (Fig. 1).
The data presented here support the predictions based on homology to active site residues in EGAL and those based on hydrophobic cluster analysis (20) implicating Glu 540 and Glu 451 as the nucleophile/acid-base pair involved in catalysis of hGUSB. Furthermore, based on alignment of a variety of retaining-type glycosyl hydrolases (32) and on hydrophobic cluster assay of regions surrounding the catalytic amino acids  4. pH profiles of the wild type (Wt) and mutant enzymes. The enzymes were produced in SF21 cells using a baculovirus system and purified as in Fig. 3. Enzymes were assayed in 1-h incubations with 4-methylumbelliferyl-␤-D-glucuronide as substrate in buffer at the pH levels specified. The activities shown on the y axis are expressed as the maximum percentage of activity obtained for that enzyme. identified for a few retaining O-glycosyl hydrolases, similar motifs were present in over 150 glycosyl hydrolases. In all for which the nucleophilic residue has been identified, the putative proton donor (acid-base catalyst) is located upstream of the nucleophile and is preceded immediately by an invariant Asn residue and also preceded by a conserved Trp residue five residues upstream. In GUSB, Glu 451 is upstream of Glu 540 and is located in the sequence WSVANEP.
Another piece of evidence consistent with Glu 451 being the acid-base residue is the increase in activity of the E451A mutant enzyme in the presence of azide. MacLeod et al. (32) suggested from the increase in k cat with some substrates seen with E127A mutant exoglucanase/xylanase in the presence of 60 mM to 2 M azide that azide can occupy a vacant anionic site created by removal of the acid-base catalyst (E127A in their case and E451A in GUSB) and react rapidly with the glucosylenzyme intermediate, increasing the steady state rate and forming the glycosyl azide product. We interpret the stimulation of E451A hGUSB by 50 -500 mM azide to support its role as the acid-base catalyst, although inhibition of the wild type enzyme by azide, the mechanism of which is not yet clear, complicates the interpretation of this experiment. Furthermore, it has been noted in the retaining type glycosyl hydrolases for which crystal structures are available (20) that both active site residues are located in the C-terminal TIM barrel. The x-ray crystal structure of hGUSB placed residues Glu 451 and Glu 540 in the C-terminal TIM barrel formed by residues 343-642 and located both residues in the active site cleft (21).
Using another approach to characterize the mechanism of hydrolysis of GUSB and to identify its active site residues, Wong et al. (35) recently determined that hGUSB is a retaining acid hydrolase using NMR analysis of the product of hydrolysis. In addition, by analysis of the labeled peptides after hydrolysis of hGUSB-substrate analogue complex, they concluded that the nucleophile in hGUSB is Glu 540 , as suggested by the studies reported here. Despite remarkable conservation of the catalytic residues identified in the active site of hGUSB and in most of the other O-glycosyl hydrolases, there is evidence of extensive evolutionary diversification in the residues that confer substrate specificity (20).
It is interesting that no patient so far reported to have the Sly syndrome has been found to have a mutation involving Glu 540 , Glu 451 , or Tyr 504 . Possibly such active site mutations would produce severe enough clinical consequences to be incompatible with survival. However, three mutations have been reported in patients that affect two residues close to the active site cleft in which these three residues reside (11). These are R382H, R382C, and Y508C. Although none of the MPS VII mutations reported affect the three active site residues described here, most of the MPS VII mutations that have been characterized affect residues that are conserved in both mammalian and E. coli GUSBs. Presumably these mutations affect folding and/or stability of the mutant enzyme rather than catalysis.