Identification of an Active Site Arginine in Rat Choline Acetyltransferase by Alanine Scanning Mutagenesis*

Kinetic as well as chemical modification studies have implicated the presence of an active site arginine in choline acetyltransferase, whose function is to stabilize coenzyme binding by interacting with the 3 (cid:42) -phosphate of the coenzyme A substrate. In order to identify this residue seven conserved arginines in rat choline acetyltransferase were converted to alanine by site-directed mutagenesis, and the properties of these mutants were compared with the wild type enzyme. Substitution of arginine 452 with alanine resulted in a 7–12-fold increase in the K m for both CoA and acetylcholine as well as k cat , with little change in the K m for dephospho-CoA. Product inhibition studies showed choline to be a competitive inhibitor with respect to acetylcholine, indicat- ing R452A follows the same Theorell-Chance kinetic mechanism as the wild type enzyme. Similar results were obtained with R452Q and R452E, with the latter showing the largest changes in kinetic parameters. These findings are consistent with Arg-452 mutations increasing the rate constant, k 5 , for dissociation of the coenzyme from the enzyme. Direct evidence that arginine 452 is involved in coenzyme A binding was obtained by showing a 5–10-fold decrease in affinity of the R452A mutant for coenzyme A as determined by the ability to protect against phenylglyoxal inactivation as well as thermal inactivation. The enzyme choline acetyltransferase (ChAT, 1 catalyzes the transfer of the acyl group from acetyl-CoA to choline resulting in the formation of the neurotransmitter acetylcholine. Mechanistic studies on the enzyme suggest a con-certed reaction in contrast to similar enzymes that transfer 30 concentrator (Amicon). The purified enzyme was either used immediately or stored at (cid:50) 80 °C in 20 m M sodium phosphate buffer, pH 7.6, containing 40% of glycerol. Assay of ChAT Activity— Enzymatic activity was determined by the fluorometric assay of Hersh et al. (19). This assay involves coupling the reverse ChAT reaction (CoA (cid:49) AcCh (cid:51) AcCoA (cid:49) Ch) to citrate synthase and malate dehydrogenase. Assays conducted under high salt condi-* 10 m M potassium phosphate buffer (pH 7.4), m M sodium chloride, 0.125 m M NAD, 0.5 m M L -malate, 0.1 m M DL -dithio-threitol, 1.5 units of pig heart citrate synthase, 4 units of pig heart malate dehydrogenase, and variable levels of acetylcholine chloride and CoA. routine assays 25 m M acetylcholine chloride and 0.1 m M CoA were used. Reactions were initiated by the acetylcholine chloride, and NADH formation was monitored continuously an Optical Technologies fluorometer. low salt conditions identical stained standard. Western blot analysis was used quantitate type enzyme

The enzyme choline acetyltransferase (ChAT, 1 EC 2.3.1.6) catalyzes the transfer of the acyl group from acetyl-CoA to choline resulting in the formation of the neurotransmitter acetylcholine. Mechanistic studies on the enzyme suggest a concerted reaction in contrast to similar enzymes that transfer acyl groups via an acyl enzyme intermediate. The kinetic mechanism for the enzyme approximates a Theorell-Chance mechanism (1), although by using isotope exchange at equilibrium a random component in the reaction has been detected (2). Chemical modification studies have implicated histidine, cysteine, and arginine as active site residues. Thus inactivation studies with dithiobis-4-nitro-2-carboxylate led to the proposal that the enzyme contains an active site cysteine (3)(4)(5)(6)(7); however, modification of this residue by methylation showed it is not essential for catalysis (8). Similarly, an active site histidine was implicated by inactivation studies with diethylpyrocarbonate while an active site arginine was implicated by inactivation studies with phenylglyoxal (7,9). It has been suggested that the active site histidine serves as a general acid/base catalyst (10), while the active site arginine is postulated to be involved in binding interactions with the 3Ј-phosphate of the substrate CoA (9).
With the availability of cDNA clones for ChAT from Drosophila melanogaster (11), porcine spinal cord (12), rat brain (13,14), mouse brain (14), and Caenorhabditis elegans (15) it has now become possible to use site-directed mutagenesis to identify these active site residues and to study their function. We have previously used site-directed mutagenesis to analyze the functionality of three conserved histidines in Drosophila ChAT and have shown that one of these residues, His-426, is essential for catalysis (16). We have now used a similar approach to search for the active site arginine thought to be involved in coenzyme A binding. Seven conserved arginines were individually changed to alanine and the resultant mutants characterized. The properties of mutant enzymes containing substitutions at arginine 452 are consistent with this arginine serving as an active site residue.

EXPERIMENTAL PROCEDURES
Materials-Citrate synthase and malate dehydrogenase, both from pig heart, were obtained from Sigma. Ni-nitriloacetate-agarose was obtained from Qiagen, while Immobilon-P nylon membranes were obtained from Millipore Corp. A rabbit anti-human ChAT antiserum prepared as described previously was employed for Western blot analyses. Alkaline phosphatase conjugated to goat anti-rabbit IgG was obtained from Bio-Rad.
Site-directed Mutagenesis-Oligonucleotide-directed mismatch mutagenesis was performed by the "dual primer" method of Zoller and Smith (17). The conserved arginine residues and the changes that were made in them are listed in Table I. The mutated cDNA fragments were inserted into the expression vector pQE9 as described (18), and final confirmation of the nucleotide changes in the clones was achieved by dideoxy chain-terminating DNA sequencing using Sequenase version 2.0 from U. S. Biochemical Corp.
Expression and Purification of Recombinant ChAT-Escherichia coli SG12036 expressing recombinant rat ChAT was grown at 22°C for 4 days in LB media. Cells were harvested by centrifugation and frozen at Ϫ80°C until further use. Enzyme purification was achieved as described previously (18). This procedure involves application of an E. coli extract, prepared in 20 mM sodium phosphate buffer, pH 7.6, containing the recombinant enzyme on a column of Ni-nitriloacetate-agarose equilibrated with the extraction buffer. The column is washed batchwise with buffer containing increasing concentrations of NaCl from 0 to 2 M and lastly with buffer containing 5 mM imidazole, pH 7.4. The enzyme, eluted with an imidazole gradient from 5 to 150 mM, was applied to a column of blue-agarose (Amicon), previously equilibrated with 20 mM sodium phosphate buffer, pH 7.6. After washing the column with equilibration buffer, the enzyme was eluted with a linear salt gradient from 0 to 1 M and concentrated/dialyzed in a Centricon 30 concentrator (Amicon). The purified enzyme was either used immediately or stored at Ϫ80°C in 20 mM sodium phosphate buffer, pH 7.6, containing 40% of glycerol.
Assay of ChAT Activity-Enzymatic activity was determined by the fluorometric assay of Hersh et al. (19). This assay involves coupling the reverse ChAT reaction (CoA ϩ AcCh 3 AcCoA ϩ Ch) to citrate synthase and malate dehydrogenase. Assays conducted under high salt condi-tions contained 10 mM potassium phosphate buffer (pH 7.4), 250 mM sodium chloride, 0.125 mM NAD, 0.5 mM L-malate, 0.1 mM DL-dithiothreitol, 1.5 units of pig heart citrate synthase, 4 units of pig heart malate dehydrogenase, and variable levels of acetylcholine chloride and CoA. For routine assays 25 mM acetylcholine chloride and 0.1 mM CoA were used. Reactions were initiated by the addition of acetylcholine chloride, and NADH formation was monitored continuously using an Optical Technologies fluorometer. Assays conducted under low salt conditions were identical except that NaCl was omitted from the reaction.
Characterization of Purified ChAT-Enzyme purity was monitored by SDS-PAGE (20), with the protein stained by the alkaline silver staining procedure (21). The concentration of purified ChAT was determined by the Bradford method (22) using bovine serum albumin as standard. For R250A, Western blot analysis was used to quantitate the amount of enzyme using known amounts of purified wild type enzyme as a standard. Briefly, proteins were electrophoretically transferred to an Immobilon-P nylon membrane as described by Towbin et al. (23). After transfer, the membranes were incubated for 1 h in blocking buffer (10 mM Tris-HCl buffer, pH 7.4, 150 mM sodium chloride, 5% nonfat dry milk, and 0.2% Nonidet P-40) and then overnight with affinity-purified anti-ChAT antibodies, diluted into fresh blocking buffer. The following morning the membrane was washed three times with 10 mM Tris-HCl buffer (pH 7.4), 150 mM sodium chloride, 0.25% sodium 7-deoxycholate, 0.1% SDS, followed by another three washes with 10 mM Tris-HCl (pH 7.4) buffer containing 150 mM NaCl. To visualize the immunoreactive protein a goat anti-rabbit antiserum coupled with alkaline phosphatase was employed. 5-Bromo-4-chloro-3-indolyl phosphate coupled with pnitroblue tetrazolium chloride were used as the color reagents.
Determination of Kinetic Constants-Preliminary estimates of the K m values for CoA, dephospho-CoA, and acetylcholine were made with each mutant at substrate concentrations saturating (10 ϫ K m ) for the wild type enzyme. Where the K m appeared higher than the wild type enzyme, the kinetic measurements were redetermined at a higher fixed substrate concentration to ensure saturation. This protocol could not be used for determining the K m of CoA and dephospho-CoA under low salt conditions with the R452A mutant since the concentration of acetylcholine, used as its chloride salt, needed to achieve a saturating level would have raised the anion concentration into the range used at high salt. Thus to determine the K m for CoA and dephospho-CoA with this mutant under low salt conditions, the concentration of acetylcholine was varied at fixed variable levels of CoA or dephospho-CoA, maintaining the concentration of acetylcholine below 10 mM. The data were then replotted as 1/V max (equivalent to an infinite acetylcholine concentration) versus 1/CoA (or dephospho-CoA). Kinetic constants were determined by fitting the data to the weighted least squares kinetic programs of Cleland. The same V max was obtained with either substrate indicating saturation was indeed achieved.
Chemical Modification of ChAT by Phenylglyoxal-Enzyme was incubated at 37°C with 5 mM phenylglyoxal in 20 mM sodium phosphate buffer, pH 7.6, containing 0.25 M NaCl and 1 mg/ml bovine serum albumin. The reaction was initiated by the addition of phenylglyoxal, and aliquots of 6 l were withdrawn at various times and added to an assay mixture containing 10 mM sodium phosphate buffer, pH 7.6, 10 mM choline iodide, and 10 mM radiolabeled acetyl-CoA in a total volume of 50 l. Acetylcholine formation was measured by the method of Fonnum (24).
Heat Inactivation of ChAT and Substrate Protection-Enzyme was incubated at the desired temperature in 20 mM sodium phosphate buffer, pH 7.6, in the presence or absence of CoA at the indicated concentration. Aliquots were withdrawn at various times, diluted with an equal volume of ice-cold 20 mM sodium phosphate buffer, pH 7.6, containing 2 mg/ml bovine serum albumin. The samples were maintained on ice until assayed by the standard fluorometric assay.

RESULTS
Previous studies have implicated an active site arginine residue to be involved in the binding of coenzyme A to the enzyme ChAT. Comparison of the amino acid sequences of ChAT from rat, Drosophila, and C. elegans, deduced from their respective cDNAs, revealed seven potential conserved arginines (Table I). In order to determine which, if any, are involved in coenzyme binding we utilized a rat ChAT cDNA and alanine scanning in which each of these putative active site arginines was separately changed to an alanine by site-directed mutagenesis. Each cDNA was constructed in an expression vector containing an N-terminal hexahistidine fused to the ChAT protein. Previous studies (18) have established that this modification has no effect on the kinetic properties of the enzyme. Expression of the recombinant enzymes in E. coli, in all but one case, led to their facile purification by a two-step procedure involving metal affinity chromatography followed by dye binding chromatography as described under "Experimental Procedures." Fig. 1 shows representative preparations of the purified enzymes in which it can be seen that essentially homogeneous enzyme is obtained. In one case, the conversion of arginine 99 to alanine resulted in low levels of expression of this mutant, and difficulties were subsequently encountered in purifying this recombinant protein. Therefore, the properties of this mutant were analyzed in E. coli extracts rather than with purified enzyme.
The kinetic properties of the mutant enzymes were initially compared with the wild type enzyme in the presence of 0.25 M NaCl, conditions previously shown to give maximal activity (25). The results of this kinetic analysis are summarized in Table II. Comparing the various alanine substitutions, the most significant effect observed is with the mutant R452A in which the K m for CoA increases more than 12-fold, the K m for acetylcholine increases ϳ8-fold, and k cat increases ϳ7-fold. Smaller increases in the K m for CoA and acetylcholine were observed with the mutants R453A, R458A, and R463A. These studies were further extended by replacing arginine 452 with glutamine, which is a near isosteric replacement, and also with glutamate, which produces a reversal in charge at this position. The kinetic properties of the R452Q mutant were essentially the same as the R452A mutant, while the R452E mutant showed greater increases in the K m for CoA (54-fold), the K m for acetylcholine (60-fold), and k cat (ϳ15-fold) (Table II). We also tested the effect of charge reversal on the adjacent arginine, arginine 453, and the effect of replacing both arginines 452 and 453 with glutamine. Charge reversal at position 453 produced an enzyme form similar to R452A or R452Q, while replacing both arginines produced an enzyme form in which the K m for CoA increased more than 170-fold, the K m for acetylcholine increased ϳ70-fold, and k cat increased ϳ17-fold.
Since it has been proposed that an active site arginine functions to interact with the 3Ј-phosphate of coenzyme A (9), the effect of each mutation with dephospho-CoA as substrate was examined, since no such interaction should occur with this substrate. With the wild type enzyme dephospho-CoA exhibits a K m that is ϳ10-fold higher than coenzyme A (Table II). In each of the mutants the K m for dephospho-CoA was similar to that obtained with the wild type enzyme, this being particularly notable with R452E and R452Q/R453Q in which the K m for dephospho-CoA increased less than 4-fold as compared with changes in the K m for CoA of more than 50-fold for these mutants. Thus at this initial level of analysis Arg-452 appears to be a likely candidate as an active site residue, with arginines 453, 458, and 463 also being possible candidates.
It has been previously observed that anions affect the kinetic parameters of the ChAT reaction (26). That is, anions act to increase V max but at the same time increase the K m for both substrates (25). It has been suggested that this effect can be attributed, at least in part, to anions interfering with the binding of the 3Ј-phosphate of coenzyme A to an active site arginine (9). Thus a similar kinetic analysis was conducted with the wild type enzyme and the four candidate active site arginine mutants under conditions of low ionic strength. The results of this analysis are shown in Table III. In agreement with previous studies (25,26), it can be seen that with the wild type enzyme decreasing the anion concentration decreases the K m for CoA ϳ12-fold but decreases the K m for choline less than 3-fold and lowers k cat by approximately 2-fold. In contrast decreasing salt has little effect on the K m for dephospho-CoA, a finding consistent with the proposal that anions compete for the interaction at the 3Ј-phosphate of coenzyme A with an arginine.
The effects of decreased ionic strength are different with the R452A mutant. The K m for CoA is reduced only ϳ3-fold by lowering the ionic strength of the assay, while there is little change in the K m for acetylcholine and k cat is essentially unchanged. As with the wild type enzyme the K m for dephospho-CoA is barely affected by changes in anion concentration. The other putative active site arginine mutants all exhibit changes in their kinetic properties that are similar to those observed with the wild type enzyme except that k cat remained unchanged. Again these results are consistent with arginine 452 as the active site arginine interacting with CoA.
In order to test for a change in kinetic mechanism between the wild type enzyme and the R452A mutant we determined the inhibition pattern with choline as a product inhibitor. The ChAT reaction follows primarily a Theorell-Chance kinetic mechanism, characterized by competitive inhibition between the inner substrate pair choline and acetylcholine (25). As shown in Fig. 2, choline acts as a competitive inhibitor with respect to acetylcholine for both the wild type enzyme and for the R452A mutant, a finding suggesting that both enzymes exhibit the same Theorell-Chance kinetic mechanism. However, the K i for choline was greater for the mutant (1.1 mM) as compared with the wild type enzyme (0.1 mM). Although not shown, competitive inhibition between choline and acetylcholine was also observed with the most severely affected mutant R452Q/R453Q with the K i for choline increased to 12 mM.
We next analyzed the ability of each of the alanine mutants to react with the arginine-specific reagent phenylglyoxal. As shown in Fig. 3 treatment of the wild type enzyme with 5 mM phenylglyoxal results in biphasic inactivation. There is a rapid initial decline of activity, which is too fast to measure but which results in a loss of ϳ35% enzyme activity. This is followed by a slower loss in enzyme activity, which plateaus at ϳ90% inactivation. A replot of the data as log (% activity remaining)   versus time from 0.1 to 45 min is linear assuming an end point of 90% inactivation. From this analysis a half-time of ϳ11 min was obtained. As shown in Fig. 3 the slow phase of inactivation was prevented by inclusion of acetyl-CoA in the inactivation reaction at a concentration in the range of the K m for CoA. Acetylcholine was without effect. Inclusion of acetyl-CoA at concentrations greater than 50 times the K m for CoA had no effect on the fast phase (data not shown). Conversion of arginines 99, 312, and 453 to alanine had no significant effect on phenylglyoxal inactivation. In each of these cases the same extent of enzyme inactivation was observed in the rapid phase, and the half-time for the slow phase varied only slightly from that observed with the wild type enzyme (10, 12, and 10 min for alanine substitutions at arginines 99, 312, and 453, respectively). Changing arginine 452 to alanine resulted in a biphasic inactivation curve similar to the wild type enzyme except that the secondary phase was faster (t 0.5 ϳ7 min) and inactivation went to completion Fig. 3. Acetyl-CoA at the same concentration used to protect the wild type enzyme from inactivation was ineffective with this mutant; however, increasing the acetyl-CoA concentration to 13.6 M was able to afford protection (Fig.  3). Although not shown, acetylcholine at 15 mM had no effect on phenylglyoxal inactivation.
Changing arginine 463 to Ala resulted in a considerably more rapid rate of inactivation by phenylglyoxal (Fig. 3). In this case the initial rapid and secondary phases of the reaction could not be distinguished. However, using 2.5 M acetyl-CoA to protect against inactivation, the biphasic nature of the inactivation process became apparent (Fig. 3). Changing Arg-250 to Ala totally eliminated the secondary phase of inactivation by phenylglyoxal (Fig. 3).
In order to provide additional evidence that Arg-452 is involved in CoA binding, we measured the ability of coenzyme A to protect the enzyme against thermal inactivation. The wild type enzyme is inactivated at 48°C under low ionic strength conditions with a t 0.5 of 60 s. As shown in Fig. 4, CoA afforded partial protection against thermal inactivation exhibiting a K d value of ϳ0.4 M, a value similar to the kinetic K m of 0.25 M listed in Table II. The R452A mutant was more thermolabile being inactivated at 48°C with a t 0.5 of ϳ20 s. At 44°C the t 0.5 was 35 s, and CoA also provided partial protection against thermal inactivation; however, in this case the estimated binding constant was shifted to ϳ6 M.

DISCUSSION
Previous studies have shown that the enzyme choline acetyltransferase is inactivated by arginine-specific reagents in a reaction protected by the substrate acetyl-CoA (9). This observation, in conjunction with the results of kinetic studies which showed that dephospho-coenzyme A was poorly bound by the enzyme (6), led to the proposal of an active site arginine that interacts with the 3Ј-phosphate of coenzyme A. Since the results of chemical modification experiments are often equivocal or ambiguous, we utilized alanine scanning to systematically search for this active site arginine by replacing seven conserved arginines with alanine. The results of the kinetic analysis of these mutant enzymes are consistent with arginine 452 being the likely candidate. The predicted properties of such a mutant would include a decreased affinity for CoA with little change in the affinity of the mutant for dephospho-CoA. In accordance with this predicted behavior conversion of arginine 452 to alanine, glutamine, or even glutamate caused a 12-50-fold increase in the K m for CoA but less than a 3-fold change in the K m for dephospho-CoA. Thus in contrast to the wild type enzyme, where the K m for CoA is ϳ10-fold lower than the K m for dephospho-CoA, the K m values for CoA and dephospho-CoA are nearly the same in R452A and R452Q. With R452E dephospho-CoA becomes a better substrate than CoA. The observation that the properties of R452A and R452Q are quite similar rules out any complications that might have been introduced as a result of a change in the side chain volume when substituting the non-isosteric alanine for arginine. Thus this data is consistent with arginine 452 interacting with the 3Ј-phosphate of coenzyme A. The effects of changing arginine 452 by mutagenesis may be blunted by the presence of an adjacent arginine in position 453. Arginine 453 may also be directly involving in coenzyme binding or alternatively may realign in the Arg-452 mutants so that it can participate in coenzyme binding. This might explain why the most dramatic effects are seen with the R452Q/R453Q double mutant in which both arginines have been substituted.
At first glance it might seem surprising that the K m for both CoA and choline as well as k cat would be increased by mutations at position 452. In the Theorell-Chance kinetic mechanism, which the Arg-452 mutant appears to obey, the ratelimiting step in the reaction is the release of product from the enzyme, k 5 in Scheme 1. The K m values for coenzyme A and acetylcholine reflect their respective affinity in the steady state and are defined as the ratio of k 5 /k 1 and k 5 /k 3 , respectively. The data are consistent with mutations at Arg-452 decreasing the affinity of the enzyme for CoA (or acetyl-CoA) by loss of the interaction with the 3Ј-phosphate. Kinetically this is manifested as an increase in the rate constant k 5 (k cat ), which represents dissociation of the coenzyme from the enzyme. An increase in k 5 would also be expected to result in an increase in K m(acetylcholine) since this kinetic constant is directly proportional to k 5 . With R452A and R452Q there is a constant 7-12-fold increase in k cat , K m(acetylcholine) , and K m(CoA) , a finding consistent with this proposal. The greater effects on K m(acetylcholine) and K m(CoA) than on k cat as seen in R452E and R452Q/R453Q may reflect other structural changes in the enzyme or may be due to the presence of inactive enzyme since the activity of these mutants was found to be rather labile. The increase in kinetic constants seen in R453E could likely result from ionic interactions between the introduced glutamate and the arginine at position 452. This would result in a decreased interaction of arginine 452 with the coenzyme, making this mutant resemble an Arg-452 mutant.
In low salt the K m(CoA) decreases ϳ12-fold, while the K m for dephospho-CoA changes less than 2-fold. Conversion of arginine 452 to alanine results in a relatively small decrease in the K m for CoA and eliminates the increase in V max . If as postulated, k 5 is increased by anions as a result of disrupting the interaction of an active site arginine with the 3Ј-phosphate of CoA, this effect would be expected to be eliminated when arginine is converted to alanine. Thus, as predicted, the R452A mutant, in contrast to the wild type enzyme, is refractory to salt effects.
Further evidence that arginine 452 is involved in coenzyme A binding is the demonstration that acetyl-CoA protection against phenylglyoxal inactivation requires coenzyme concentrations approximating the K m for CoA for the R452A mutant and is ineffective at the lower acetyl-CoA concentration that protects the wild type enzyme. Similarly protection against heat inactivation requires CoA concentrations in the range of its K m for the R452A mutant, while a considerably lower concentration protects the wild type enzyme. It is interesting to note that the residue which reacts with phenylglyoxal appears to be arginine 250. Changing this residue to alanine has no effect on the kinetic properties of the enzyme, indicating that its reaction with phenylglyoxal results in either a conformational change or steric hindrance. Modification of this residue does not result in a loss in activity, supporting the conclusion that this is not a critical residue for catalysis. Changing arginine 452 or 463 to alanine increased the rate as well as the extent of inactivation by phenylglyoxal. This indicates that these mutations permitted either an additional residue to react with phenylglyoxal or resulted in a structural change such that reaction with arginine 250 totally blocked activity.
Interestingly, conversion of three other arginines, Arg-453, -458, and -463, to alanine also affects the kinetics of the reaction, albeit to a considerably lesser extent. These arginines are clustered in the primary sequence. Thus it is possible that these residues are situated in the vicinity of or actually constitute part of the active site in three-dimensional space, participating in coenzyme A binding. It is worth noting that x-ray crystallographic studies have shown that there are multiple interactions such as electrostatic interactions, hydrophobic interactions, and hydrogen bonds between coenzyme A and the enzyme citrate synthase (27). At least three arginines form salt bridges with the three charged phosphate moieties; Arg-46 and Arg-324 (through a water molecule) interact with the 5Јdiphosphate while Arg-164 interacts with the 3Ј-phosphate in addition to possible hydrogen bonding interactions.
In summary alanine scanning of conserved arginines in the choline acetyltransferase reaction has provided evidence for arginine 452 being an active site residue that functions by interacting with the 3Ј-phosphate of the substrate CoA/acetyl-CoA. This arginine lies within a highly conserved region of the enzyme, whose sequence varies little among such divergent species as mammals, Drosophila, and the nematode, C. elegans. Direct evidence for this proposal must await the determination of the three-dimensional structure of the enzyme.