Specificity and Mechanism of Metal Ion Activation in UDP-galactose:β-Galactoside-α-1,3-galactosyltransferase*

UDP-galactose:β-galactosyl-α1,3-galactosyltransferase (α3GT) catalyzes the synthesis of galactosyl-α-1,3-β-galactosyl structures in mammalian glycoconjugates. In humans the gene for α3GT is inactivated, and its product, the α-Gal epitope, is the target of a large fraction of natural antibodies. α3GT is a member of a family of metal-dependent-retaining glycosyltransferases that includes the histo blood group A and B enzymes. Mn2+activates the catalytic domain of α3GT (α3GTcd), but the affinity reported for this ion is very low relative to physiological levels. Enzyme activity over a wide range of metal ion concentrations indicates a dependence on Mn2+ binding to two sites. At physiological metal ion concentrations, Zn2+ gives higher levels of activity and may be the natural cofactor. To determine the role of the cation, metal activation was perturbed by substituting Co2+and Zn2+ for Mn2+ and by mutagenesis of a conserved D149VD151 sequence motif that is considered to act in cation binding in many glycosyltransferases. The aspartates of this motif were found to be essential for activity, and the kinetic properties of a Val150 to Ala mutant with reduced activity were determined. The results indicate that the cofactor is involved in binding UDP-galactose and has a crucial influence on catalytic efficiency for galactose transfer and for the low endogenous UDP-galactose hydrolase activity. It may therefore interact with one or more phosphates of UDP-galactose in the Michaelis complex and in the transition state for cleavage of the UDP to galactose bond. The DXD motif conserved in many glycosyltransferases appears to have a key role in metal-mediated donor substrate binding and phosphate-sugar bond cleavage.

UDP-Gal:␤-galactosyl-␣-1,3-galactosyltransferase (␣3GT), 1 a Golgi membrane-bound enzyme, catalyzes the synthesis of galactosyl ␣-1-3-galactosyl ␤-OR structures in glycoconjugates (Refs. 1 and 2; see Fig. 1). ␣3GTcd and the products of its action are found in most mammals but not in humans and their closest relatives, the old world monkeys and apes (3,4). In these species the gene for ␣3GT is mutationally inactivated (4,5), and the absence of active enzyme allows the production of antibodies against the product of ␣3GT action, the ␣-Gal epitope (Fig. 1). Primates lacking this enzyme have natural antibodies (1-3% of circulating IgG, designated anti-Gal) that bind ␣-Gal (5) and facilitate immune defenses against pathogens but also present a barrier to the xenotransplantation of organs from mammalian species that produce active ␣3GT (6).
Like other glycoprotein glycosyltransferases, ␣3GTs are type-2 membrane proteins with a short N-terminal cytosolic domain, a transmembrane helix, a stem, and a C-terminal catalytic domain (7)(8)(9)(10). Only a few subgroups of glycosyltransferases that function in the processing of glycoproteins and glycolipids show global homology in primary structure. At present, a 2.4-Å structure of the catalytic domain of ␤-1,4-galactosyltransferase I is the only reported three-dimensional structure of an enzyme of this type (11). ␣1,3-GT is homologous to histo blood group glycosyltransferases A and B and to Forssman glycolipid synthase but is not significantly similar in overall sequence to ␤-1,4-galactosyltransferase I (12)(13). Nevertheless, these and other divalent cation-dependent glycosyltransferases share a DXD sequence motif that is thought to represent at least part of a cation binding site (14 -16), suggesting that they may share a metal binding domain or substructure.
Recombinant forms of bovine and other ␣3GTs have been previously expressed (18,19); the cytosolic and transmembrane domains together with a 67-residue stem can be deleted to produce a fully active soluble enzyme (18). We have described the use of a bacterially expressed soluble form of the catalytic domain of bovine ␣3GT (residues 80 -367; ␣3GTcd) for the enzymatic production of substantial amounts of ␣-Gal-containing oligosaccharides (20). Here we have investigated the role of the metal cofactor in the catalytic mechanism using the recombinant enzyme. Like many other glycoprotein glycosyltransferases, ␣3GTcd requires a divalent cation for activity that has been thought to be Mn 2ϩ (1,2). However, the reported affinity of the enzyme for this ion (K d of 6 mM) exceeds the physiological concentration of Mn 2ϩ by about 3 orders of magnitude, raising questions about the cation dependence in vivo. Studies of metal activation over a wide concentration range reveal a high affinity binding site for Mn 2ϩ and other metals, including Zn 2ϩ , which is present at higher levels in biological systems.
Glycosyltransferases that, like ␣3GT, catalyze "retaining" reactions are expected to utilize double displacement mechanisms in which the UDP to galactose bond is cleaved, with formation of an intermediate before transfer of galactose to an acceptor. The mechanism of the reaction catalyzed by ␣3GTcd is sequential, UDP not being released before the completion of catalysis. However, a low level of UDP-galactose hydrolytic activity indicates that the UDP to galactose bond can be cleaved in the absence of a carbohydrate acceptor, possibly via formation of an oxycarbenium intermediate. The properties of ␣3GT activated by different metal ions and a mutant with a substitution in a metal binding sequence motif indicate a role for the metal ion in UDP-galactose binding and catalysis.

EXPERIMENTAL PROCEDURES
Construction of pET15b E80r␣GT-The expression plasmid was generated from a cDNA of bovine ␣1,3-GT in pSV-SPORT vector provided by Dr. L. Inverardi, Diabetics Research, Cell Transplant Center, University of Miami School of Medicine. This clone had a deletion corresponding to Tyr-64 to Phe-95 that included a 15-residue sequence previously reported to be essential for enzyme activity. The region required for activity was restored by polymerase chain reaction using the primers designated "extension" and "BGT-C." The product was gel-purified and used as a template in a second amplification using primers BE80GT-N and BGT-C as follows. Extension (coding), GAAAGCAAGCTTAAGCTATCGGACTGGTTCAACCCATTTAA-ACGC; BE80GT-N (coding), CGAATATCATATGGAAAGCAAGCTTAAGC-TATCG; BGT-C (complementary), CGCGGATCCCAAAGTCAGACATTA-TTTCTAACCAC.
The product from the second polymerase chain reaction was directly used for TA cloning. Positive colonies were isolated and screened by restriction mapping with NdeI and BamHI. A clone with the appropriate insert was selected and digested with the same restriction enzymes. The insert was purified by agarose gel electrophoresis and cloned into a preparation of pET15b vector that had been previously cleaved with NdeI and BamHI. The product was transformed into Escherichia coli DH5␣ competent cells, and plasmids were prepared and characterized by restriction mapping and DNA sequencing.
Bacterial Expression and Purification-Cultures of E. coli BL21(DE3) transformed with pET15b r␣GT were grown in LB medium containing 100 g/ml ampicillin with rapid shaking (250rpm) at 37°C. When the A 600 nm of the culture reached 0.8 -1.0, isopropyl-1-thio-␤-S-galactopyranoside was added to a concentration of 0.4 mM to induce expression of T7 RNA polymerase and recombinant ␣3GTcd. The cultures were then incubated at 27°C for 20 h with slow shaking (200 rpm) and harvested by centrifugation at 3,000 rpm for 20 min. Bacteria were washed with washing buffer (20% sucrose, 20 mM Tris-HCl, pH 8.0), suspended in 50 mM Tris-HCl, pH 8.0, containing 1 mM EDTA, 0.1 M NaCl, and lysed using a French pressure cell. The soluble fraction was collected by centrifugation at 12,000 rpm for 30 min.
The supernatant was applied to a to Ni 2ϩ column at 4°C, and the column was subsequently washed with 10 volumes of 20 mM Tris HCl buffer, pH 7.9, containing 5 mM imidazole and 0.5 M NaCl followed by 6 volumes of 20 mM Tris-HCl buffer, pH 7.9, containing 60 mM imidazole and 0.5 M NaCl. The enzyme was finally eluted with 6 volumes of 20 mM Tris-HCl buffer, pH 7.9, containing 500 mM imidazole and 0.5 M NaCl. Purified recombinant ␣3GTcd was precipitated between 10 and 80% saturation with ammonium sulfate (can be stored in 20 mM MES buffer, pH 6.0, containing 50% glycerol at Ϫ20°C for up to 6 months).
Mutagenesis-Mutagenesis was carried out by the polymerase chain reaction megaprimer method (21) with modifications described previously (22,23). Mutant coding sequences were generated by amplifications using a T7 promoter and mutagenic and T7 terminator primers. The final product of amplification was cleaved using NdeI and BamHI and cloned into a pET15b vector that had been previously treated with the same enzymes. Mutants were characterized by automated DNA sequence analysis of the entire coding sequence in the expression vec-tor, and the proteins were expressed as described for the wild-type enzyme.
Activity Measurements-␣3GT assays typically included 2-4 g/ml enzyme, 50 mM MES buffer, pH 6.0, 10 mM lactose, 0.3 mM UDP-[ 3 H]galactose (specific activity, 500 cpm/nmol), 0.1% bovine serum albumin, and metal cation in a total volume of 100 l and were incubated at 37°C for 5-15 min. In steady-state kinetic studies, the concentrations of metal ion, lactose, and UDP-galactose were varied with other substrates and cofactors at fixed concentrations. Blanks were reactions from which the acceptor substrate is omitted. Studies at high enzyme concentrations indicated that the enzyme has a low level of UDPgalactose hydrolase activity, which contributes to the backgrounds measured in this way; however, because hydrolase activity is only 0.25% of the transferase activity, the effects of the hydrolase activity on the calculated transferase activities is insignificant. Assays were terminated by adding ice-cold 0.1 M EDTA (100 l). The reaction mixture was then applied to a 2-ml AG1-X8 (Bio-Rad) column, and the radioactive product was eluted with 0.5 ml followed by 1 ml of water. The eluate was collected in a plastic vial, mixed with 10 ml of EcoLume (ICN Biomedicals, Costa Mesa, CA), and counted in a liquid scintillation counter (LKB). One unit of enzyme activity is defined as the amount of enzyme that catalyzes the transfer of 1 mol of galactose from UDPgalactose to lactose/min at 37°C. Conditions were chosen so that Ͻ25% of the radioactivity from the UDP-[ 3 H]galactose was incorporated into product in most experiments. Kinetic data were analyzed by fitting to appropriate rate equations using the curvefitter program of Sig-maPlot™. For metal activation studies, data were fitted to equations describing activation through binding to a single metal binding site (Equation 1) or activation to V 1 by binding to site 1 with affinity K 1 and to V 2 by additional binding to site 2 with a K d of Data from studies in which the two substrates are varied were fitted to equations for a symmetrical sequential initial velocity pattern associated (3) and an asymmetrical initial velocity pattern (4).
CD Spectroscopy-Near and far UV CD spectra of recombinant proteins were determined with a JASCO J-710/720 spectropolarimeter. Twenty spectra were scanned for each sample at a speed of 100 nm/min, averaged, and smoothed. Near UV CD spectra (250 -320 nm) were determined using a cell with a path length of 1 cm, and far UV spectra (200 -250 nm) were determined using a cell with a path length of 0.1 cm. Proteins were dissolved in 10 mM MES, pH 6.0, containing 50% glycerol at concentrations between 0.15 and 0.5 mg/ml. Far UV CD data were analyzed to estimate secondary structure composition using the k2d neural network program (24).

RESULTS
Preparation and Properties of Bovine Recombinant ␣3GTcd-␣3GTcd (residues 80 -367 of bovine ␣1,3GT) was expressed cytoplasmically in E. coli as a soluble active enzyme and was purified, reproducibly, in yields of more than 7 mg/liter bacterial culture. Active enzyme is also produced by expression at 37°C, but the yield is lower by about 20% than at 27°C. The enzyme migrates as a single component on SDS gel electrophoresis with an apparent molecular weight of 36,000 (data not shown). After precipitation between 10 and 80% saturation with ammonium sulfate, ␣3GTcd is soluble at concentrations of up to 7.5 mg/ml in 20 mM MES, pH 6.0, containing 50% (v/v) glycerol. A form of ␣3GTcd with an additional N-terminal truncation of six residues was also expressed. In this case, when the expression was at 27°C, a similar yield was obtained, but at 37°C, essentially no active enzyme was produced. Steady state kinetic studies with the purified smaller enzyme, in which the concentrations of UDP-galactose or lactose were varied at a fixed concentration of the second substrate at 10 mM Mn 2ϩ , showed that the apparent K m and V m values were closely similar to those of the larger enzyme (data not shown). Fig. 2 shows the near and far UV CD spectra of ␣3GTcd. CD spectra have not been previously reported for either natural or recombinant forms of this enzyme. The near UV spectrum has a similar magnitude to those of other proteins and can be expected to be that of the correctly folded catalytic domain. The far UV CD spectrum was analyzed using the k2d neural network program to give estimates of 28% ␣ helix and 38% ␤ sheet; the predicted spectrum from the analysis is a reasonable fit to the experimental data (Fig. 2). The secondary structure predicted from the sequence from a consensus of the PREDATOR (25,26), PHD (27), and Quadratic Logistic (28) methods is 27% ␣ helix and 18% ␤ sheet. Differences in the ␤ sheet content obtained from these analyses may reflect the inaccuracy of ␤-structure predictions from far UV CD spectra resulting from the relatively weak ellipticity of ␤ sheets compared with ␣ helices as well as the inherent limitations of sequence-based secondary structure predictions. The predictions are similar to the secondary structure composition of the ␤-1,4-GT catalytic domain: 25% helix, 20% ␤ sheet (11). The form of ␣GTcd with the additional six-residue truncation had CD spectra closely similar to those of the larger enzyme.
␣3GTcd Has High and Low Affinity Binding Sites for Mn 2ϩ and Other Metals-As previously reported for natural ␣3GT, the bacterially expressed enzyme is inactive in the absence of metal ions and is strongly activated by Mn 2ϩ . Activity measurements in the presence of other metal ions at concentrations of 10 mM and in the absence of Mn 2ϩ indicated that Co 2ϩ and Fe 2ϩ also activate the enzyme but Zn 2ϩ , Ca 2ϩ , and Mg 2 do not.
Detailed studies with Fe 2ϩ were not performed because of its tendency to oxidize in the assay system, but the kinetic properties of the Co 2ϩ -activated enzyme were characterized in detail (see below and Table I). Enzyme activities at Mn 2ϩ concentrations from 10 M to 15 mM at fixed concentrations of lactose (10 mM) and UDP-galactose (0.3 mM) did not fit well to a rate equation describing the dependence of activity on metal binding to a single site, giving residuals that vary systematically with log[Mn 2ϩ ]. However, the rate equation for a two-site model of metal activation gave a better fit (Fig. 3B). The results of this analysis indicate that ␣3GTcd is activated by binding of Mn 2ϩ to a high affinity site with an apparent K d value of 80 Ϯ 20 M; additional binding to a second site with an apparent K d of 2.3 Ϯ 0.5 mM gives a 6-fold higher maximum activity at the substrate concentrations used in these assays.
This result led us to re-investigate the effects of other metal ions over a broader range of concentrations (10 -10,000 M). Although no activity was observed with Ca 2ϩ or Mg 2ϩ at any concentration, Zn 2ϩ was found to activate up to a concentration of 500 M and to progressively inhibit at higher concentrations. In the lower concentration range, Zn 2ϩ produces a similar level of activity to Mn 2ϩ and is more effective than Co 2ϩ (Fig. 3). The data obtained with Zn 2ϩ do not fit well to an equation describing activation by binding to a high affinity site and inhibition resulting from binding to a second lower affinity site, and the line through these data in Fig. 3A was generated by a nonlinear spline method. Inhibition by Zn 2ϩ may result from multisite binding and/or denaturation at higher concentrations. The activity profile with Co 2ϩ fits well to the equation for single-site activation with a K d of 4.8 Ϯ 0.5 mM and a lower maximum activity than for Mn 2ϩ . It is possible that this cation also binds to two sites, but that the enzyme form with a single metal ion bound to the high affinity site is inactive or has extremely low activity. The apparent binding constants for activation by different cations derived from these studies with fixed substrate concentrations are summarized in Table II. To determine whether ␣3GT, when partially activated by Mn 2ϩ binding to the high affinity site, can be further activated by a different metal binding to a second lower affinity second site, different metal ions (Ca 2ϩ , Mg 2ϩ , Zn 2ϩ , Ni 2ϩ , Cu 2ϩ , Te 3ϩ , and Eu 3ϩ ) were added at concentrations of 10 mM in the presence of 80 M Mn 2ϩ . None enhanced the activity. In the enzyme activated by 10 mM Mn 2ϩ , Cu 2ϩ or Eu 3ϩ were found to strongly inhibit activity at micromolar concentrations, indicating that they bind to a high affinity cation binding site(s). Because Eu 3ϩ is a very potent inhibitor, it was further characterized and found to be a mixed inhibitor with respect to Mn 2ϩ (0.5-6 mM) with a small intercept effect and large slope effect in a double reciprocal plot (Fig. 4). In contrast, Eu 3ϩ is a competitive in-FIG. 2. Near and far UV CD spectra of recombinant ␣3GTcd. Spectra were determined as described at a protein concentration of 0.5 mg/ml at 25°C. The spectrum is shown as a continuous line; the fitted far UV CD spectrum obtained from the analysis with k2d is shown as a discontinuous line.
hibitor against Co 2ϩ (2.2 to 15 mM; data not shown). These results are consistent with the presence of two metal binding sites for Mn 2ϩ and a single site for Eu 3ϩ ; it appears also that a binding of Co 2ϩ to a single site is required for catalysis. An apparent K i of 0.5 M was calculated from the inhibitory activity against Co 2ϩ and K ii and K is values for inhibition against Mn 2ϩ were 11.3 Ϯ 1.  Table III. In each case, the data fitted best to Equation 3, and in double-reciprocal plots for velocity and concentration of either substrate at a series of fixed concentrations of the second substrate, gave a family of intersecting lines (Fig. 5). In a double-displacement mechanism, K ia is zero (Equation 4), and double-reciprocal plots produce families of parallel lines. These results indicate that the catalytic mechanism is sequential so that both substrates bind to the enzyme before any product is released (29). In a sequential mechanism, substrate binding can be random or ordered, and if ordered, acceptor binding can precede UDP-galactose binding or vice versa.    ␣3GTcd Catalyzes a Low Rate of UDP-galactose Hydrolysis, Indicating That It Does Not Bind Acceptor before Donor Substrate in an Obligatory Order-UDP-galactose hydrolase activity was initially noticed in the form of high backgrounds ( 3 H release from UDP-galactose in the absence of acceptor into a product that does not bind to the anion exchange resin) at higher enzyme concentrations. The activity increases linearly with enzyme concentration and time, up to 15 min. Hydrolase activity was characterized using a 10 -20-fold higher concentration of enzyme than in standard assays; it is Mn 2ϩ -dependent (data not shown) and also displays saturation kinetics with varying concentrations of UDP-galactose (Fig. 6). The k cat for hydrolysis is 0.25% of the corresponding transferase activity with lactose as acceptor.
The hydrolytic activity of ␣3GTcd indicates that the enzyme binds the donor substrate and catalyzes cleavage of the bond between UDP and galactose in the absence of a carbohydrate acceptor. Although the enzyme complexes for the hydrolysis reaction may differ structurally from those for the transfer reaction, the hydrolase activity is inconsistent with a sequential mechanism with obligatory ordered binding of acceptor before donor substrate. The remaining alternative mechanisms are random sequential or ordered sequential binding of donor and acceptor.
The kinetic parameters listed in Table III were calculated by fitting to Equation 3 designating UDP-galactose as substrate A and lactose as substrate B. The same rate equation applies to an ordered sequential mechanism and random equilibrium mechanism; the values for K ib ϭ K ia ϫ K b /K a given in Tables I  and III represent the K d of the enzyme-acceptor complex in the case of a random equilibrium mechanism.
Metal Cofactor Substitution Affects UDP-galactose Binding and, More Strongly, Catalytic Efficiency-Because Mn 2ϩ , Co 2ϩ , and Zn 2ϩ are active as cofactors for ␣3GTcd, the role of the metal ion can be investigated by determining the effects of substituting different metals on different kinetic parameters. The three metal ions were used at concentrations of 10 mM, 10 mM, and 300 M, respectively, levels that gave optimal activities in assays conducted at fixed substrate concentrations. Table I shows that the metal substitution has a relatively small but significant effect on k cat and on the K m for acceptor (lactose), but K ia , the K d of UDP-galactose from the enzyme-UDPgalactose complex in an ordered or random equilibrium sequential mechanism, was increased 10-fold, and k cat / K ia ϫ K b , a parameter reflecting catalytic efficiency in a bisubstrate reaction (30), was reduced 11-38-fold.
These three metal ions are also effective cofactors for the low hydrolase activity of ␣3GTcd. Hydrolase activity was measured at a range of UDP-galactose concentrations with the same fixed concentrations of these metals (Fig. 6). As also shown in Table  I, metal substitution has similar effects on this activity, perturbing k cat and the K m for UDP-galactose, which results in a larger reduction in catalytic efficiency (k cat /K m ), as observed with the transferase activity.    Table I. specific conserved features in the adjacent sequence, is found in many metal-dependent glycosyltransferases (14 -16). The aspartates have been shown to be crucial for activity in enzymes that are highly divergent in function and sequence: yeast MNN1 ␣-1,3-mannosyltransferase and large clostridial cytotoxins (15,16). A similar sequence, DXH, is essential in UDP-GalNAc:polypeptide GalNAc-transferase (17). This region, which appears to represent part of the binding site for a divalent cation cofactor (15), is present in ␣3GTcd as D 149 VD 151 . The role of this region in ␣3GTcd was investigated by expressing mutants with substitutions for each of the aspartates: D 149 N and D 151 N. Both enzymes were expressed in soluble form and isolated in pure form in reasonably good yields (about 4 mg/liter bacterial culture), but no catalytic activity was detected in either protein. Near and far UV CD spectra of the D 149 N mutant are closely similar to those of the wild-type enzyme (Fig. 7), indicating that the loss of activity is a direct effect of the mutation, but the D 151 N mutant shows large changes around 265 nm in the near UV range and 230 nm in the far UV range. Difference spectra for this mutant are characteristic of an effect on tertiary, rather than secondary, structure that affects the environment of a buried tryptophan (see "Discussion"). These results indicate that Asp 149 has an essential role in catalysis, whereas Asp 151 may be required to form the correct structure in this region for activity. Val 150 is less conserved and has a side chain that is unlikely to have a direct role in catalysis. To perturb the structure associated with the essential aspartates but retain catalytic activity so that quantitative changes in kinetic parameters can be measured, a mutant of ␣3GTcd was constructed and expressed containing a Val 150 to Ala substitution. The V 150 A mutant was also expressed in a soluble form in good yield and purified. It is less active than the wild-type enzyme under standard assay conditions, yet its CD spectrum was closely similar to that of the wild-type protein (Fig. 7). Tables I and II summarize its metal activation and kinetic properties. The activities with varying [Mn 2ϩ ] at fixed standard concentrations of the two substrates were found to fit best to the equation for metal binding to a single site. The apparent affinities of the enzyme for Mn 2ϩ , Zn 2ϩ , and Co 2ϩ were reduced 2-4-fold, and the extrapolated activity at saturating [Co 2ϩ ] was very low, in keeping with an effect on cation activation. The values of kinetic parameters determined at 10 mM Mn 2ϩ indicate that the mutation produces a 2-fold reduction in k cat , a 6-fold increase in the K m and K i for UDP-galactose, and a more than 20-fold decrease in catalytic efficiency (k cat /K ia ϫ K b ). DISCUSSION Expression of ␣3GTcd at 27°C produces a stable enzyme in high yield that is sufficiently soluble (up to 7.5 mg/ml) and stable to be a suitable subject for structural analysis. The effects of temperature on the expression of the fully active variant with six additional residues deleted from the N terminus indicates that residues 80 -85 enhance stability but not structure or activity.
The activity of ␣3GTcd is modulated by high and low affinity binding sites for metal cations (K a of Ͼ10 3 and 10 5 -10 6 M Ϫ1 ) as previously observed with ␤-1,4GT (31). Inhibition by Eu 3ϩ and Cu 2ϩ is associated with K I values in the low micromolar range. The involvement of two metal ions in catalysis has been observed in several metalloenzymes that catalyze phosphoryl transfer reactions including 3Ј-5Ј DNA exonuclease (32), alkaline and acid phosphatases (33,34), purple acid phosphatase (35,36), and phosphoprotein Ser/Thr phosphatase, PP-1 (37). In these enzymes, the metal ions form a bimetal center that in purple acid phosphatase and PP-1 has been found to act as a ligand for the phosphate, facilitating its stabilization and correct orientation in the active site and also in generating a hydroxide nucleophile involved in catalysis (35)(36)(37). An analogous role for the metal as a ligand for phosphate(s) in the donor substrate in ␣3GT is supported by the present results.
In vivo only the high affinity site may be relevant since the cellular concentrations of the ions that can activate ␣3GTcd are in the micromolar range. In this concentration range, Zn 2ϩ is as effective as Mn 2ϩ (Fig. 3A). The concentrations of different metals in the trans-Golgi lumen, the cellular site of action of ␣3GT, are unknown. However, body fluids that originate in part from this compartment such as milk have Zn 2ϩ concentrations (50 -200 M) that exceed those of Mn 2ϩ by about 2 orders of magnitude (38). Co 2ϩ is irrelevant as an activator in vivo because of its low abundance relative to the level required for activation. It is therefore possible that, as with methionine aminopeptidase I (39), the biologically active cofactor has been misidentified, and Zn 2ϩ is the relevant cofactor for ␣3GT. However, previous studies with ␤-1,4GT-1 show that the enzyme in Golgi vesicles is optimally activated by lower concentrations of Mn 2ϩ and that an endogenous high molecular weight molecule participates in the activation of this glycosyltransferase in vivo (40). Although the present studies indicate that Zn 2ϩ is the likely natural cofactor(s) of ␣3GT, the possibility of the involvement of an endogenous macromolecule cannot be discounted.
There is limited information presently available on the mechanisms of glycoprotein glycosyltransferases that catalyze retaining reactions. The steady state kinetic properties of ␣3GTcd indicate, as for ␤-1,4GT (41), a sequential mechanism in which metal cofactor, donor, and acceptor bind enzyme before catalysis, with no release of UDP before the transfer of galactose to an acceptor. Thus, a double displacement mechanism in which UDP is released with formation of an interme- diate covalently bound ␣-galactosyl derivative can be eliminated. Present data do not distinguish whether substrate binding is random or ordered. Although the low level of UDPgalactose hydrolase activity makes obligatory binding of acceptor before donor substrate seem unlikely, it cannot be eliminated since we have not established that hydrolase activity is a catalytically competent step in the overall reaction. Previous studies with bovine ␣3GTcd also show that the enzyme binds more strongly to UDP-Sepharose than ␤1,4GT-1 but does not bind to lactose-or N-acetyllactosamine-Sepharose (2); however, this does not prove that there is obligatory binding of donor before acceptor substrate and may simply reflect the much weaker affinity of the enzyme for acceptor as compared with donor substrate. Ordered sequential binding of donor and acceptor substrates has been reported for ␤1,4GT-1 and ␣-3 fucosyltransferase (41,42), but further studies are required to determine whether substrate binding to ␣3GT is ordered or random.
The effects of different metal cofactors on the steady state kinetic parameters show that the metal strongly affects the affinity for UDP-galactose (K ia ) but has a lesser effect on affinity for acceptor substrate (K b and K ib ); this is borne out by the kinetic parameters for UDP-galactose hydrolysis with different cofactors (Table I). Although there are large standard errors in the values for the catalytic efficiency for galactose transfer, k cat /K ia ϫ K b , since these are compounded from the errors in three experimentally determined parameters, it is clear that metal substitution has its largest effect on this parameter. For UDP-galactose hydrolysis, the substitution of Co 2ϩ or Zn 2ϩ produces increases in the K m similar to those for K ia for the transferase reaction and greater reductions in catalytic efficiency (k cat /K m ); the K m for UDP-galactose will closely approximate the K d of the E⅐S complex for hydrolysis because of the low k cat . As discussed above, the transition state for the ␣3GTcd-catalyzed reaction may be similar to those in other retaining glycosyltransferase such as glycogen phosphorylase and its close relative, maltodextrin phosphorylase. In these enzymes, where the reaction is analogous to the reverse of the ␣3GTcd-catalyzed reaction, the monosaccharide in the enzymebound oligosaccharide substrate that is transferred is rotated relative to its preferred structure in solution, and C1-O bond cleavage results in formation of an oxycarbenium intermediate (43). A similar process may occur with the UDP-galactose substrate of ␣3GT; the formation of such an intermediate independent of acceptor substrate binding is consistent with the low level of hydrolytic activity. Data supporting cleavage of the nucleoside diphosphate-monosaccharide bond before transfer have been previously reported for two inverting transferases, ␣-1,3-fucosyltransferase V (43, 44) and ␤-1,4GT-1 (45,46), suggesting that this may be a common intermediate in retaining and inverting transferases. A plausible role for the metal would involve an interaction with the phosphates of the UDP-sugar and of the UDP formed by cleavage of the UDP to galactose bond.
The aspartates of the D 149 VD 151 motif of ␣3GTcd are essential for catalysis as found for analogous residues in yeast MNN1 ␣-1,3-mannosyltransferase (15) and clostridial cytotoxins (16) and for the aspartate and histidine of an analogous DXH sequence in polypeptide GalNAc transferase (17). The Asp 151 to Asn mutation changes the near and far UV CD spectra, indicating that this substitution perturbs the local structure. Difference spectra for the mutant show a peak in the far UV CD spectrum around 230 nm and a broad peak in the near UV CD spectrum centered around 265 nm (Fig. 7); both of these peaks are consistent with an effect of the mutation on the environment of a buried tryptophan side chain (47). Since residues of this sequence motif are essential for function in highly divergent cation-dependent glycoprotein glycosyltransferases, which are otherwise not significantly similar in sequence, it appears that these enzymes may be distantly related or share a homologous domain (14). Sequence comparisons suggest that the catalytic domains of glycoprotein-processing glycosyltransferases have modular structures and are composed of domains of different origins (17). The elucidation of high resolution crystallographic structures for multiple representatives is needed to clarify this issue. At present, the only known structure is that of the truncated catalytic domain of ␤-1,4GT-1 (11); the structure of ␣3GTcd is presently unknown, but a model for a putative sugar nucleotide binding domain of ␣3GTcd has been described based on the use of a T4 phage DNA-modifying glucosyltransferase (48) as a template. However, the phage enzyme appears to be structurally unrelated to ␤-1,4 GT (11), whereas secondary structure topology predicted for ␣3GTcd from multiple alignment methods and the secondary structure content indicated by the far UV CD spectrum are consistent with an ␣/␤ fold similar to that found in ␤1,4GT-1. Also, the D 252 VD 254 sequence in ␤1,4GT-1 has a role in binding the UDP-galactose substrate (11), in agreement with the role for D 149 VD 151 in ␣3GTcd indicated by the present study which shows that the Val 150 to Ala substitution reduces the affinity for UDP-galactose and more strongly lowers the catalytic efficiency for transferase and hydrolase activities. Although there is insignificant global similarity between the sequences of the catalytic domains of ␣3GTcd and ␤1,4GT family, this apparent similarity between the two enzymes in structure-function relationships suggests that part of their structures may have a common origin. Unfortunately, the level of resolution of the structure of ␤1,4GT-1 does not provide any information about bound metal ions. The properties of the Val 150 to Ala mutant of ␣3GTcd suggest that the mutation disrupts properties of the enzyme that are associated with cation activation, supporting the view that this region has a role in binding the metal ion and UDP-galactose.