INTRODUCTION

-Galactosidase (-D-galactoside galactohydrolase, EC) from Escherichia coli catalyzes hydrolytic and transgalactosylic reactions with -D-galactosides (Huber et al., 1976). The amino acid (Fowler and Zabin, 1978) and nucleotide (Kalnins et al., 1983) sequences have been determined. The enzyme is a tetramer, and each identical monomer (116,353 Da-1023 amino acid residues) functions independently (Cohn, 1957). Mg²⁺ or Mn²⁺ and Na⁺ or K⁺ are required for full catalytic efficiency (Tenu et al., 1972; Case et al., 1973; Huber et al., 1979). The three-dimensional structure of -galactosidase (Jacobson et al., 1994) shows that the active site is in a deep pocket within a distorted ``TIM'' barrel. His-540 is located in the wall of the active site cavity, and one of its nitrogens is at the edge of the active site cavity and appears not to be H-bonded to any other group in the free enzyme. A His equivalent to His-540 is conserved (Fig. 1) in every related -galactosidase that has been sequenced to date (Kalnins et al., 1983; Burchhardt and Bahl, 1991; Buvinger and Riley, 1985; David et al., 1992; Fanning et al., 1994; Hancock et al., 1991; Poch et al., 1992; Schmidt et al., 1989; Schroeder et al., 1991; Stokes et al., 1985). In addition, His-540 is only three residues removed from Glu-537, a residue that probably acts as a nucleophile in the catalysis scheme (Gebler et al., 1992). Recent unpublished results¹ indicate that His-540 is within H bonding distance of the C6 hydroxyl groups of substrate and transition state analog inhibitors.

-galactosidase.

The active site of -galactosidase has two subsites. The aglycone subsite lacks specificity, whereas the galactose subsite is very specific (Deschavanne et al., 1978; Huber et al., 1984). Losses in binding at the galactose subsite are very dramatic, and catalysis is completely eliminated if there are changes at positions C3 and/or C4. Changes at C6 also have large effects on binding, but sugars that have hydroxyl groups in the same orientation as D-galactose except for changes at the C6 group are still quite good substrates (Marshall et al., 1977). McCarter et al. (1992) found that the hydroxyl groups of galactose at C3, C4, and C6 each contributed at least 16.7 kJ/mol to binding and catalysis.

Compounds that resemble D-galactose but have a planar shape (e.g. -D-galactonolactone) or do not have the carbon equivalent to the C1 group that D-galactose has (e.g. L-ribose-furanose form) are good inhibitors of -galactosidase (Lalegerie et al., 1982; Huber and Brockbank, 1987). Their geometries probably resemble the putative planar oxo-carbonium ion transition state. The active sites of enzymes are usually structured to be more complementary to the transition state than to the substrate. Therefore, interactions between an enzyme and its transition state are very important (Fersht, 1985). In this paper we report the importance (for binding and catalysis) of the interactions of His-540 of -galactosidase from E. coli with the C6 hydroxyl group of both the ground and the transition states.

MATERIALS AND METHODS

Generation of -Galactosidases with Substitutions for His-540

All the mutagenesis procedures utilized the pBS SK⁺ variant (Stratagene). The primers used were 5-dC GAA TAC GCC (GA)A(AT) GCG ATG GGT A-3 (the altered codon is shown in bold and the two bases within parentheses indicate the degeneracy that was introduced). All cell cultures were propagated in LB medium at 37 °C (LB growth medium consisted of 1% (w/v) Tryptone, 1% (w/v) NaCl, and 0.5% (w/v) yeast extract (pH 7.5 at 25 °C)). All media were autoclaved (120 °C, 22 psi) for 20 min before using. If required, 5-bromo-4-chloro-3-indolyl -D-galactopyranoside (0.002%, w/v), isopropyl-thio--D-galactopyranoside (0.02%, w/v), ampicillin (50 µg/ml), and/or tetracycline (40 µg/ml) were added to the cooled autoclaved liquid agar before dispensing. The method used for site-directed mutagenesis was a modified version of Kunkel's dut- ung- method (Kunkel et al., 1987). A DNA fragment of the lacZ gene containing the codon to be mutated was excised from a plasmid containing the gene (pIP101) using SstI and ClaI and ligated into the pBS SK⁺ vector previously digested with the same endonucleases. The resultant subclone, pBS SK⁺ SC1.1, was transformed into competent E. coli RZ1032 (HfrKL16 PO/45, Zbd-279::Tn10, (lysA(61-62)), thi1, relA1, supE 44, dut1, ung1 (Kunkel et al., 1987)). Single-stranded DNA was isolated using a helper phage (VCS M13, Stratagene), and this was used as the template DNA for mutagenesis. The phosphorylated primers were annealed to the single-stranded DNA template. T4 DNA polymerase and T4 DNA ligase were added to the annealed primer-template mixture to synthesize and ligate the mismatched complementary DNA strand. The reaction mix was then transformed into a dut+ ung+ strain of E. coli (TG2 ( (lac-pro AB), supE, thi, hsd5, (srl-recA)306::Tn10(tet⁺) F(tra D36, proAB⁺, lacI^q, lacZ  M15) (Gibson, 1984)) or XL1-Blue (hsd5, recA1, endA1, gyrA46, thi, relA1, F(proAB⁺, lacI^q, lacZM15, Tn10(Tet^r)) (Bullock et al., 1987)). E. coli TG2 or XL1-Blue preferentially degrades the uracil template DNA strand containing the wild type codon and thereby enhances selection for the mutant codons. Colonies were screened for the desired mutations by sequencing. Confirmed mutants were back cloned from the pBS SK⁺ SC1.1 construct into the pIP101 expression vector using the restriction enzymes SstI and ClaI. The integrity of the mutation in the final pIP101 mutant plasmid was reconfirmed by DNA sequencing, and the integrity of the restriction sites were confirmed by restriction analysis. Unless otherwise described, all general molecular biology methods used were those of Sambrook et al. (1989).

The galactosidases (wild type and those with substitutions for His-540) were isolated as described in Cupples et al. (1990) except that the enzyme was passed through an FPLC Superose 6 column as a last step. Purity was determined by SDS-polyacrylamide gel electrophoresis (Fig. 2). All the enzymes were greater than 98% pure as judged by densitometer scans.

SDS-polyacrylamide gel electrophoresis gels of the purified -galactosidases.

The assay buffer (pH 7.0) consisted of TES² (30 mM) with 145 mM NaCl, and substrate (ONPG, PNPG, PNPA, or PNPF). The Mg²⁺ concentration in the assays was usually 1 mM. When the Mg²⁺ was to be absent, Mg²⁺ was left out and 20 mM EDTA was added. The reactions were started by adding enzyme. For most assays the reactions were followed at 420 nm as a function of time in a Shimadzu UV-2101 PC Spectrophotometer (25 °C). Extinction coefficients used were 2.65 mM¹ cm¹ for oNP and 6.50 mM¹ cm¹ for p-nitrophenol (at pH 7.0). A nonlinear regression computer program (American Chemical Society; Enzyme Kinetics) was used to determine the K_m and k_cat values. During purification, 2 mM ONPG was used routinely for the assays.

The probable mechanism of -galactosidase with inhibitor/acceptor present is shown on Scheme 1. Molecules that act as competitive inhibitors usually also act as acceptors (and thus are designated as A (Deschavanne et al., 1978; Huber et al., 1984)). K_i values can be obtained, but the acceptor action must be considered when calculating the K_i values. An equation

(Eq. 1)

to obtain K_i values was described by Deschavanne et al. (1978). It accounts for the acceptor reaction. Experiments at different concentrations of inhibitor (acceptor) were carried out and the K_i values were averaged.

The postulated reactions of -galactosidase in the presence of an inhibitor/acceptor.

One can also obtain information about rate-determining steps using inhibitors/acceptors. Alcohol and sugar acceptors can react (k₄) in place of water to form galactosyl-acceptor adducts (GA-A). If k₄ > k₃ and if k₃ is partially or fully rate-limiting, the apparent k_cat of the reaction in the presence of the acceptor will increase as a function of the acceptor concentration to a maximum (k₂k₄/(k₂ + k₄)) that depends on the values of k₂ and k₄ (Deschavanne et al., 1978). If the rate increases more than 10-fold it indicates that k₂ is at least 10 × as large as k₃ and, therefore, that k_cat (without the acceptor) is essentially equal to k₃. If k₄ > k₃ and if k₂ is rate-limiting, there will be no change in rate. If k₄ is smaller than k₃, the rate of the reaction will slow down regardless whether k₂ or k₃ is slower.

The energy needed to attain the transition state can be estimated (Fersht, 1974). The k_cat/K_m value (equal to k₂/K_s with -galactosidase) is a second order rate constant for the formation of the enzyme-transition state complex (starting from free enzyme and substrate) and thus RTln(k_cat/K_m) is the energy needed to attain the transition state. The equation below (Fersht and Leatherbarrow, 1987), therefore, gives the difference of the energetic contribution of a substituted side chain ``R'' to the formation of the transition state when compared to the wild type enzyme. The equation assumes that the activation energy required for bond breaking/making does not change significantly in the substituted enzyme as compared with the wild type enzyme.

(Eq. 2)

Thus, the differences in the k_cat/K_m values and the subsequent differences in the G_S values are very important indications of the contribution of a residue to catalysis. This theory applies for the formation of the first transition state. If there is an additional transition state and if more energy is required to attain that transition state, the net effect will be to slow down the rate. If the additional transition state needs less energy to form, the overall rate will not slow down (but it will of course also not be any faster).

Enzyme solutions were passed through a Superose 6 column pre-equilibrated with buffer (pH 7.0) made up in Milli-Q^TM water and containing 30 mM TES, 145 mM NaCl, and 10 µM MgSO₄. The enzymes were concentrated to 2 mg/ml using a Microsep^TM centrifugal concentrator (cut-off 30 kDa), and the solution was placed into Spectra/Por cellulose dialysis tubing (cut-off 12-14 kDa). The enzymes were then dialyzed extensively against the buffer containing 10 µM Mg²⁺. After the final buffer change, the enzyme was dialyzed for a 24-h period to ensure that an equilibrium was reached. The final concentration of the protein was again checked to account for changes that occurred during dialysis. The enzyme and final buffer dialysates were analyzed by atomic absorption to determine the amount of Mg²⁺.

The k_cat, K_m, and k_cat/K_m values of the substituted -galactosidases were determined at various pH values. Fixed time assays were used for all the pH profile analyses. The reactions were stopped with 2 volumes of 1 M Na₂CO₃. This increased the pH to 11 and stopped the reaction. Using this method, all the rates could be determined using a single extinction coefficient since the final measurements were done at the same pH. The extinction coefficients used for these assays were determined with oNP standards.

The enzymes (substituted and wild type) were placed (0.35 mg/ml) into a 50 mM sodium phosphate buffer (pH 7.0, 1 mM MgSO₄) and incubated at 52 °C. At various times, 60-µl samples were removed and diluted into an equal volume of ice-cold TES buffer (30 mM TES, 145 mM NaCl, 1 mM MgSO₄, pH 7.0). These enzymes were further diluted, if necessary, immediately before assaying.

RESULTS

Table I gives the k_cat, K_m, k_cat/K_m and G_S values for the substituted enzymes and the wild type enzyme with ONPG, PNPG, PNPA, and PNPF. Note that the G_S values with ONPG and PNPG are significantly larger than those with PNPA and PNPF.

Table I. Kinetic and energetic values for the action of the wild type and the substituted enzymes on galactosyl substrates and on substrates that are similar to galactose but do not have C6 hydroxyl groups H540F H540N H540E WILD ONPG kcat (s1) 0.1 0.7 0.66 620 Km (mM) 0.11 0.088 0.141 0.12 kcat/Km (s1 mM1) 0.91 8.0 4.7 5200  GS (kJ/mol)  21.4  16.1  17.4 PNPG kcat (s1) 0.1 0.3 1.2 90 Km (mM) 0.45 0.069 0.50 0.041 kcat/Km (s1 mM1) 0.22 4.4 2.4 2200  GS (kJ/mol)  22.8  15.4  16.9 PNPA kcat (s1) 11.9 2.0 8.2 211 Km (mM) 0.57 0.73 6.0 11.6 kcat/Km (s1 mM1) 20.9 2.7 1.3 18.2  GS (kJ/mol) +0.34  4.7  6.5 PNPF kcat (s1) 8.5 4.2 14.1 123.4 Km (mM) 2.9 1.2 13.4 16.9 kcat/Km (s1 mM1) 2.9 3.5 1.0 7.3  GS (kJ/mol)  2.3  1.8  4.9

Transition state analog inhibitors with C6 hydroxyl groups inhibited the substituted enzymes much more poorly than they inhibited wild type enzyme (Table II) (except for H540N--galactosidase interactions with L-ribose). On the other hand, there was not much difference between the inhibition of the wild type and the substituted enzymes by L-arabinolactone, which is similar to D-galactonolactone but does not have a C6 hydroxyl methyl group.

Table II. The Ki values (mM) with some transition state analog inhibitors H540F H540N H540E Wild type Transition state analogs with C6 hydroxyls L-Ribose 33 0.32 65 0.24 D-Galactonolactone 25 1.6 >500 0.13 Transition state analog without a C6 hydroxyl L-Arabonolactone 51 11 8 16

Table III shows the K_i values for competitive inhibition by various sugars and sugar derivatives. In general, the values of inhibition by the compounds were affected much more by the substitutions if a C6 hydroxyl was present than if it was not present. The differences were usually greater when the His was replaced by Phe than when replaced by Asn or Glu. Some of the inhibitors also acted as acceptors and increased the rates of reaction (Table IV). To ensure that these rate increases were due to transferolysis (acceptor reaction), TLC was done (data not shown) after reaction in the presence of 1 M methanol. In the case of every substituted -galactosidase, a compound that migrated as did authentic methylgalactopyranoside was produced in the presence of the methanol.

Table III. The Ki values (mM) with some substrate analog inhibitors H540F H540N H540E Wild type Substrate analogs with C6 hydroxyls IPTG 4.2 0.17 60 0.085 PETG 0.25 0.005 0.54 0.0015 Lactose 53 1.3 90 1.21 D-Talose 94 12 298 8 Glycerol 660 165 600 93 D-Glucose 970 325 2700 409 D-Galactose 1370 67 910 25 D-2-Deoxy-galactose 798 212 338 62 Substrate analogs without C6 hydroxyls D-Fucose 124 311 102 203 L-Arabinose 169 110 520 217

Table IV. Apparent kcat values (s1) of the substituted enzymes in the presence of various acceptors Acceptor H540F H540N H540E None 0.1 0.7 0.7 Methanol (1 M) 0.2 1.4 3.4 Glycerol (125 mM) 0.21 1.89 3.1 L-Arabinose (250 mM) 0.09 0.70 1.1 D-Galactose (250 mM) 0.16 0.75 2.2 D-Fucose (250 mM) 0.12 0.58 1.4 D-Talose (100 mM) 0.04 0.56 0.7 2-Deoxy-D-Galactose (150 mM) 0.16 0.56 1.4 D-Glucose (300 mM) 0.44 3.71 6.0 D-Glucose (infinite concentration) 15.0

Fig. 3 shows that except for magnitudes the dependence of k_cat/K_m on the pH for two of the substituted enzymes was not much different from wild type. There were, however, definite differences in the K_m and k_cat profiles.

The kinetic results are shown on Table V. The removal of Mg²⁺ had similar effects on the magnitude of the k_cat and K_m values in the substituted enzymes as it did in the wild type enzyme. In addition, H540F--galactosidase bound 0.49 Mg²⁺ per monomer, whereas wild type -galactosidase bound 0.53 Mg²⁺ per monomer when dialyzed in 10 µM Mg²⁺. The values would probably have been closer to 1 Mg²⁺ per monomer if higher Mg²⁺ concentrations had been used, but at higher concentrations of Mg²⁺, the differences of Mg²⁺ concentrations inside and outside the dialysis tubing would have been small, and it would have been hard to measure the differences.

Table V. The effects of Mg2+ on the Km (mM) and kcat (s1) values of the substituted -galactosidases and on wild type -galactosidase H540F H540N H540E Wild type kcat Km kcat Km kcat Km kcat Km 10 mM EDTA 0.02 1.91 0.03 0.6 0.04 0.67 39 0.57 1 mM Mg2+ 0.1 0.11 0.7 0.08 0.66 0.14 620 0.12

Fig. 4 shows that all three of the substituted -galactosidases were less stable than wild type -galactosidase. H540N--galactosidase was, however, a little more stable to heat than were the other two substituted enzymes.

DISCUSSION

The lacZ genes from E. coli and from some other related -galactosidases have been sequenced (Kalnins et al., 1983; Burchhardt and Bahl, 1991; Buvinger and Riley, 1985; David et al., 1992; Fanning et al., 1994; Hancock et al., 1991; Poch et al., 1992; Schmidt et al., 1989; Schroeder et al., 1991; Stokes et al., 1985), and they share significant homology. These -galactosidases also share extensive homology with prokaryotic (E. coli) and eukaryotic (rat, mouse, and human) -glucuronidases (Gallagher et al., 1988; Jefferson et al., 1986; Oshima et al., 1987; Nishimura et al., 1986; see Poch et al., 1992 for alignment with -galactosidases). Several His are conserved among the various sequences, and one of the conserved His is at position 540 (Fig. 1). Histidines are often important at the active sites of enzymes since they can readily form H bonds, they have the capacity to act as acid/base catalysts at pH values that are optimal for enzyme action, they are good nucleophiles, and they can function as ligands for metal binding. Their H bonding capacity means that they could be important for substrate binding and transition state stabilization. The His-540 of -galactosidase (E. coli) was substituted by Phe, Asn, and Glu in these studies. The Phe was introduced to eliminate the possibility of forming H bonds and because Phe and His are approximately of the same size and His does have some aromatic properties. The Asn was introduced because an amido group can form H bonds somewhat similar to those of His and because the distance between the -carbon of His and the ¹N of His is similar to the distance between the -carbon of Asn and the amido nitrogen of Asn (Fig. 5). Glu was introduced because we thought that it could interact somewhat like His since it can form H bonds, it can act as a general acid/base catalyst, it can act as a nucleophile, and it can function as a ligand for metal binding. It does, however, introduce a negative charge, and it has a different geometry than does His. The three-dimensional structure (Jacobson et al., 1994) clearly shows that His-540 is at the active site and that one of its nitrogens is available for hydrogen bonding.

N of His and the amido N of Asn are approximately equi-distant from their respective -carbons.

Substitutions for His-540 caused the k_cat values of the reactions with galactosyl substrates to decrease by about 1000-fold or more (Table I). For substrates that were identical to galactose but did not have C6 hydroxyl groups, the losses of the k_cat values were much smaller. In most cases the K_m values of the substituted enzymes were relatively small. Since K_m = (k₃·K_S)/(k₂ + k₃) in the case of -galactosidase, small K_m values either mean that binding of substrate is good or that k₃ is small relative to k₂. The k_cat/K_m values were decreased in a similar fashion to the way that the k_cat values were, but the differences with the substrates that did not have C6 hydroxyl groups were very small (15-fold at the most). The data shown on Table I indicate that the G_S values that resulted from substitutions for His-540 were about 10-20 kJ/mol larger when a C6 hydroxyl was present than when the hydroxyl was absent. ONPG and PNPG have intact C6 hydroxyl methyl groups, whereas PNPA is similar to D-galactose but does not have a C6 hydroxyl methyl group. PNPF is similar to D-galactose but it does not have a C6 hydroxyl group. The differences in G_S values with the two classes of substrate strongly suggest that His-540 is involved in binding to the C6 hydroxyl of the transition state. The salient point is that the deleterious effects of the substitutions for His-540 on the k_cat/K_m and G_S values were much smaller when the C6 hydroxyl group was absent. The differences in G_S values between substrates with and without C6 hydroxyl groups were greatest with H540F--galactosidase. This is expected since Phe can't form hydrogen bonds whereas Asn and Glu may be able to participate in H bonds.

In every case (except for the inhibition of H540N--galactosidase by L-ribose) substitutions for His-540 led to large losses of the binding ability in the cases of D-galactonolactone and L-ribose but not in the case of L-arabinolactone (Table II). L-Arabinolactone does not have a C6 hydroxyl methyl group while the other two compounds do. These compounds are considered to be transition state analog inhibitors (Huber and Brockbank, 1987). This is further evidence that His-540 stabilizes the transition state via interactions with the C6 hydroxyl of galactose. The relatively good inhibition of H540N--galactosidase by L-ribose may be due to favorable H bond interactions of some type that are allowed with the amido nitrogen of Asn.

There was also a differential effect on inhibition by substrate analog inhibitors (Table III) depending upon whether a C6 hydroxyl was present (except again for the inhibition of H540N--galactosidase by some of the compounds). These data show that His-540 also interacts with the C6 hydroxyl of the substrate in the ground state.

In every case, when Mg²⁺ was removed, losses in activity were similar to the effects on wild type -galactosidase (Table IV). In addition, equilibrium dialysis showed that H540F--galactosidase bound as much Mg²⁺ as did wild type -galactosidase in the presence of 10 µM Mg²⁺. Thus His-540 is not involved in Mg²⁺ binding.

The substituted enzymes had different pH profiles for K_m and k_cat whereas the k_cat/K_m profiles for the substituted enzymes were very similar to those of wild type (except for magnitude). The K_m value ((k₃·K_S)/(k₂ + k₃)) and the k_cat value (k₂k₃/(k₂ + k₃)) contain k₃, whereas the k₃ value in k_cat/K_m is canceled out (k_cat/K_m = k₂/K_S). Thus, these data suggest that the ``degalactosylation'' (k₃) reaction is affected differently by pH in the substituted enzymes than in wild type enzyme whereas the way that k₂ and K_S interact as the pH is changed is not affected by the substitutions.

Some of the ``inhibitors/acceptors'' significantly speeded up the rates of reaction with ONPG (Table IV; see also Scheme I). In particular glycerol and glucose were good activators. Since the rate increases, k<sub>3</sub> must be at least partially rate-limiting. An analysis was done on H540E--galactosidase to find the rate extrapolated to an infinite concentration of glucose. The extrapolated rate is equal to k<sub>2</sub>k<sub>4</sub>/(k<sub>2</sub> + k<sub>4</sub>) (Deschavanne et al., 1978), and the value of this for H540E--galactosidase was 15 s<sup>1</sup>. This means that k<sub>2</sub> is equal to or greater than 15 s<sup>1</sup> for this enzyme. Since the k<sub>cat</sub> without glucose for H540E--galactosidase was 0.66 s<sup>1</sup> and this value is at least 10-fold lower than 15 s<sup>1</sup>, the k<sub>3</sub> step (degalactosylation) must be rate-determining. The data (Table IV) indicate that k<sub>3</sub> may also be rate-determining for H540N--galactosidase. In the case of H540F--galactosidase, the value of k<sub>2</sub> is definitely larger than k<sub>3</sub> (its value is at least 0.44 s<sup>1</sup> which compares with the k<sub>cat</sub> value of 0.1 s<sup>1</sup>). The values of k<sub>2</sub> and k<sub>3</sub> of the wild type enzyme are roughly equal with ONPG as the substrate. These data, showing that k<sub>3</sub> is rate determining, show that the transition state of the k<sub>3</sub> (degalactosylation) step is destabilized more by the substitutions than is the transition state of the k<sub>2</sub> (``galactosylation'') step.

Enzymes that have site-specific mutations often have very low activities. Therefore, one must always be concerned that the results that one obtains for an enzyme solution of substituted enzyme could be from a small amount of wild type contamination during growth or purification rather than from the substituted enzymes themselves. In this case, the substituted enzymes had properties that were totally different from the properties of wild type -galactosidase, and this precludes the possibility that the activities noted were from wild type contamination.

There were no significant differences of the fluorescence and circular dichroism spectra between the substituted enzymes and wild type -galactosidase (data not shown). -Galactosidase is, however, a very large molecule, and it is difficult to detect small changes in the secondary and tertiary structure. Some perturbations of structure undoubtedly occur when substitutions for residues are made. The heat stability studies indicated that this was the case with the substitutions introduced here. It can also be argued that if His-540 interacts with the C6 group (as the evidence presented here strongly suggests), the G_S values for PNPA and PNPF should be zero since there should be no such interactions with those two substrates. The non-zero G_S values may result from perturbations of structure that affect other interactions with these substrates. Alternatively, the substituted residues themselves may interact in some other way with PNPA and PNPF to give the G_S values observed (e.g. there could be interactions between the Phe of H540F--galactosidase with the C6 methyl group of PNPF).

The studies of these substituted enzymes clearly showed that His-540 interacts (presumably by forming H bonds) with the C6 hydroxyl of the substrate and the transition state of -galactosidase. The data show that this is important both for binding the substrates and for stabilizing the transition state of the reaction. The magnitude of the effects observed shows that these interactions are obviously a significant component of the -galactosidase reaction. The data also showed that the stabilizing effect of His-540 on the C6 group of the transition state may be more significant for the degalactosylation step than for the galactosylation step. Recent structural work¹ shows that the His-540 side chain is indeed within H bonding distance of the C6 hydroxyl of galactosyl inhibitors.