Conformational adaptability of the active site of beta-galactosidase. Interaction of the enzyme with some substrate analogous effectors.

The action of different effectors, glycosides, and alcohols on the reactions catalyzed by beta-galactosidase is analyzed in this paper. Effectors as large as tri- and tetrasaccharides have no effect on the enzyme activity, suggesting that the binding site has rather small size. Most of the beta-galactosides produce a competitive inhibition. The other compounds assayed behave either as noncompetitive inhibitors, and they are deadened inhibitors, or as uncompetitive inhibitors which exhibit a better affinity for the chemical intermediate than for free enzyme; nearly all of them give transfer products. The analysis of the data indicates that the active center of beta-galactosidase is made up of two subsites: a galactose and a glucose subsite. This latter site is in a more favorable conformation in the galactosylenzyme than in free enzyme; possibly it might even by generated by the galactose binding. Conformational rearrangements of the active center deduced from the inhibition data have been directly observed by differential spectroscopy. The conformational adaptability of the enzyme and its consequence for the functional properties of beta-galactosidase are discussed.

The action of different effecters, glycosides, and alcohols on the reactions catalyzed by P-galactosidase is analyzed in this paper. Effecters as large as tri-and tetrasaccharides have no effect on the enzyme activity, suggesting that the binding site has rather small size. Most of the /3-galactosides produce a competitive inhibition. The other compounds assayed behave either as noncompetitive inhibitors, and they are deadened inhibitors, or as uncompetitive inhibitors which exhibit a better affinity for the chemical intermediate than for free enzyme; nearly all of them give transfer products.
The analysis of the data indicates that the active center of /3-galactosidase is made up of two subsites: a galactose and a glucose subsite. This latter site is in a more favorable conformation in the galactosylenzyme than in free enzyme; possibly it might even be generated by the galactose binding. Conformational rearrangements of the active center deduced from the inhibition data have been directly observed by differential spectroscopy. The conformational adaptability of the enzyme and its consequence for the functional properties of /3-galactosidase are discussed.
Although a great amount of data has been accumulated on /3-galactosidase during these last years (l-16), little is known about the active site. In order to identify the essential residues, active site-directed reagents have been used (10)(11)(12). Until now, this approach has not been successful. The only amino acid blocked which has led to a loss of activity of the enzyme was shown to be a methionine, but Naider et al. demonstrated that it is not an essential group (12).
The active site of an enzyme is not limited to the few amino acid residues involved as catalytic groups. It extends over a larger part of the molecule, including many amino acids able to interact with substrate in the binding process.  (17)(18)(19). Such an approach implicitly supposes a rigid structure for an enzyme. However, even if located at the active site level, the flexibility of proteins could lead to more dynamic situations, the structure of the site being different in free enzyme and in the complexes. This induced fit, as proposed for the first time by Koshland (201, could be very important for the enzymatic reaction. In such a situation the interaction between different compounds and the enzyme could not allow a suitable interpretation for the mapping of the active site of the enzyme, but it could reveal the conformational states of the enzyme in the different complexes.
In the present paper, the effect of different alcohols and glycosides on P-galactosidase is reported. This work was initiated as an approach for mapping the rigid active site structure of the enzyme, but the results reported must be discussed in terms of conformational adaptability of the enzyme.

THEORY
It was previously shown that /3-galactosidase-catalyzed reactions proceed through a chemical intermediate (ES') which occurs simultaneously with the liberation of the first product of the reaction (5, 21). An analogous substrate could bind to the free enzyme or to the -ES' complex, but not to the Michaelis complex in which all the binding subsites are occupied (see "Discussion" below). Furthermore, the substrate itself, when it binds to the ES' complex, could be either a dead-end inhibitor or an acceptor initiating transfer product. Scheme I represents the minimum pathway: eat, 0 k',K", Thus, when the plot of l/k,,,, , versus Z is linear, one can conclude that k, is negligible.
The different parameters k4 and K", can be calculated from Equations 4 and 6, using the values of k2 and k13 previously determined (5, 6, 21).

MATERIALS AND METHODS
Enzyme -p-Galactosidase was prepared from Escherichia coli strain EOl as previously described (2).

RESULTS
On the basis of the kinetic patterns, the assayed compounds can be divided into four categories of compounds: the compounds which have no action on the enzyme reaction; neither K, nor V, are modified by their presence; the competitive effecters which modify K, but not V,; the noncompetitive effecters which affect V, and practically not K,; they are able to bind to both free enzyme and ES' complex with comparable binding constants; and the uncompetitive effecters which bind only to the ES' intermediate; the K,IV, ratio remains constant. In this work, the concept was extended to the compounds exhibiting a significantly better affinity for ES' than for E, and not only for compounds giving inhibition which affect K, and V, to the same extent and give parallel linear Lineweaver-Burk plot as classically referred. Compounds without Effect on PGalactosidase -In the presence of trehalose, stachyose, raffinose, and D-(+)-melezitose, neither V, nor K, are modified. For stachyose and rafinose, which are oligosaccharides, and cr-galactosides the size and the structure of the molecule could be sufficient to explain their failure to bind to the enzyme. The same explanation could account for the behavior of n-(+)-melezitose, which is an cY-glucoside. Trehalose, which is an a-glucoside, but only a disaccharide, could not bind to the enzyme however (Table I).
Competitive Inhibitors of PGalactosidase -Most of the /3-gala&sides behave as competitive inhibitors of P-galactosidase. However, an exception has to be noted. The hydroxyethyl-P-n-thiogalactoside is able to bind on the ES' intermediate but with a weaker affinity than on free enzyme. Monosaccharides such as gala&se and deoxygalactose, and linear alcohols such as mesoerythritol and dulcitol, inhibit the enzyme competitively. Fig. 1 gives a typical plot obtained for lactose.
Noncompetitive Effecters of PGalactosidase -The following '1 hxt)r compounds are very poor but noncompetitive inhibitors: maltose, n-(-)-lyxose, cellobiose, L-( -)-arabinose, and L-( -)-arabitol. K, and K", are similar and equal to 0.1 M or more (Table I). In Fig. 2, as an example, the effect of maltose on o-Np-Gal hydrolysis is presented. Both plots K,, ,/keat, , versus Z and lkat. I versus Z give straight lines, which allow the All the other compounds exhibit the same behavior; they are dead-end inhibitors without formation of transfer products. D-( + )-Arabitol also has similar affinity for both free enzyme and chemical intermediates; however, when it is present, the rate of ES' decomposition increases. This compound activates the reaction (Fig. 3) as a nucleophilic reagent does (5, 21).
Uncompetitive Effecters of pGalactosidase -The other compounds (melibiose, saccharose, fucose, glucose, mesoinositoll exhibit significantly better affinity for the chemical intermediate than for free enzyme. Among them, only melibiose is unable to give transfer products. For the other effecters, the plot lilz,,, I versus Z is not linear; therefore kq is not negligible. The results corresponding to the effect of glucose on the hydrolysis of o-Np-Gal by the enzyme are shown in Fig. 4.
It seems important to point out that in all the cases where transfer products are observed, the effector or acceptor exhibits better affinity for ES' than for free enzyme.

DISCUSSION
Two Binding Subsites of Enzyme-The major part of the glycosides can bind to the free enzyme, as is shown by the increase of K,,,, ,IK,,,, , . Some of them bind to the enzyme. substrate complexes, since they give a decrease of the k cat,, value. Such binding might be nonspecific. However the liberation of the first product of the reaction restores an available subsite in the ES' complex, and the inhibitor effect could be ascribed to the specific binding of inhibitors to this subsite. This hypothesis is confirmed by a comparison of glucose effects on o-Np-Gal and p-Np-Gal hydrolysis. If the inhibitory effect of glucose is related to a decrease of the ES' hydrolysis step, this effect must be more important on o-Np-Gal hydrolysis, where KS and kf3 are of the same order of magnitude, than on p-Np-Gal hydrolysis, where the k, step is the limiting one in the absence of effector (5). Such is the case (Figs. 4 and 5). The difference of behavior of glucose on the kp and kf3 steps reveals the appearance, in the ES' complex, of a 5 I I new subsite which in the ES complex is occupied by the aglycon part of the substrate. Since lactose is the physiological substrate of p-galactosidase, the subsite could be designated as the "glucose subsite," which exists in addition to the "galactose subsite." Binding of Effecters to Free Enzyme-The data seem to indicate that there are very definite requirements in the binding of the effecters at the active center of P-galactosidase. The binding site is probably rather small since effecters as large as tri-and tetrasaccharides are not able to interact with enzyme. Variations as large as lo4 are observed in the K, values. P-Gala&sides exhibit the best affinity for free enzyme, better than galactose itself, indicating that the aglycon part in the glucose subsite participates significantly in binding. Among the monosaccharides, galactose is the best inhibitor. In D-(+)-fucose and in L-(-)-arabinose, the replacement of the CH,OH group in position 5 by smaller groups (CH, and H, respectively) increases the K, value, underlining the importance of CH,OH group in this position. For the binding of disaccharides, the enzyme does not seem to distinguish between a-and P-glucosides on the one hand, and a-galactosides on the other hand. The K, values are of the same order of magnitude for maltose, cellobiose, and melibiose. Among the linear alcohols, the behavior of mesoerythritol is noteworthy. It exhibits a better affinity than other alcohols in C, and Cg, or than glycerol as studied by Van der Groen et al. (9).
Binding of Effecters to Galactosylenzyme -Among the various effecters tested, the best inhibitors of free enzyme, /3-ngalactosides, are unable to bind significantly to the ES' intermediate, compared to their binding of free enzyme, except for hydroxyethyl-/3+thiogalactoside; but, in this case, the affinity is about 10 times weaker for ES' than for E. As /3-n-galactosides, galactose binds only to free enzyme with significant affinity. Other monosaccharides such as glucose and fucose, as well as disaccharides such as saccharose and melibiose, have a higher affinity for the galactosylenzyme than for free enzyme. The K", values for the different compounds differ maximally by two orders of magnitude. No definite specificity can be deduced from the various values.
In some cases, binding of the inhibitors leads to the formation of transfer product. The compounds which exhibit a higher affinity for the ES' complex than for free enzyme induce such reaction (melibiose is however an exception). Formation of transfer products has already been demonstrated (1,22). In these last cases as in our work, this formation occurs through intermolecular reaction mechanisms. But it is not the only way to obtain transfer products; with lactose for instance, Huber et al. (13) described a direct intramolecular transfer for the formation of allolactose.
Induced Conformation -The results clearly indicate the existence of two binding subsites. The K", values can be easily associated with a binding to the glucose subsite. However, the K, values are more ambiguous, since a binding on one or the other subsite could provide competitive behavior. Although some effecters are able to bind both to free enzyme and to the intermediate, it is not possible to know if they bind to the same subsite in the two species of the enzyme. The K, values of galactose, fucose, arabinose, and glucose exactly reflect the K, values determined for the corresponding substrates, o-Np-Gal, o-nitrophenylarabinoside, nitrophenylfucoside, and o-nitrophenylglucoside, which are, respectively, 0.3 mM, 3 mM, 4.3 mM (21, and 6 mM. This result suggests that the same subsite is implicated in the binding of the glycosidic part of the substrates and of the monosaccharides. More favorable values of K", compared to the K, of the same compounds suggest the occurrence of conformational changes from free enzyme to the complex. The glucose subsite could be generated even by such conformational rearrangements induced by gala&se binding. Such an argument could be provided by considering the energy of interaction of lactose to the enzyme. Binding of lactose involves a free energy variation of -4 kcal/mol. Interaction of galactose with free enzyme (AG = -1.9 kcal/mol) and interaction of glucose on galactosylenzyme (AG = -2 kcal/mol) could account for this value. The additive binding energies of gala&se plus glucose (AG = -0.3 kcal/mol) to the free enzyme do not account for free energy of lactose binding. An alternative explanation could also account for free energy of lactose binding; it involves synergistic interactions between ligands without any conformational change of the protein.
Conformational changes strongly suggested by the action of effecters on enzyme kinetics were directly investigated.
As shown in Fig. 6o, it was possible to detect a differential spectrum between free enzyme and the complex obtained by saturating the enzyme with isopropyl-P-n-thiogalactoside.
The spectral change is characterized by a very small amplitude, which prevents quantitative analysis. A differential spectrum was also observed between free enzyme and the intermediate obtained in the presence of n-galactal (Fig. I%). These results are strongly in favor of conformational changes of the enzyme upon ligand binding, and allow to discard any explanation requiring a rigid enzyme.
probably each step of the reaction catalyzed by p-galactosidase involves very discrete conformational changes localized at the active center. These changes are very important, since their occurrence limits the rate of the overall reaction (5, 7) and further effector binding. The interaction of the gala&se moiety at its specific subsite induces the formation of a second subsite capable of binding sugars and alcohols of appropriate 0 0 z ; 0 1' -: -Y Q -5 FIG. 6. Difference spectra induced by ligand binding in P-galactosidase (a) with 1.58 mM isopropyl-/3-n-thiogalactoside or (b) with 9 mM n-galadal. Experimental conditions. soectra were determined with doible compartment cells, each comPartment with a 0.438-cm path length. Enzyme concentration was 1.85 mg/ml, in 10 mM Tes, pH 7, 0.145 M NaCl, and I mM MgSO, at 25".

configuration.
The results presented here have shown that the structure of the active site of the enzyme has a rather dynamic conformation, which changes with the various steps of the reaction pathway. In the galactosylenzyme, the glucose subsite either appears, or adopts a more favorable conformation for binding the corresponding effecters. Does this conformational adaptability provide a suitable situation for transgalactosylation?
A question which can be asked is the following: are those conformational effects interpretable in terms of induced fit (20) or in terms of a priori conformational states of the protein (each stabilized by ligand binding)? On the basis of these experiments it is not possible to conclude. It is even quite difficult to imagine some conclusive experiments allowing to choose one theory over the other. Nevertheless, the concept of "dynamic specificity" developed elsewhere (24) could also apply to such a dynamic occurrence of a second binding site following the first steps of the reaction which it catalyzes.