Enhancement of the activity of horse liver alcohol dehydrogenase by modification of amino groups at the active sites.

Abstract Reaction of horse liver alcohol dehydrogenase with imidoesters or cyanate at pH 8 significantly increases the activity of the enzyme, as assayed in high concentrations of NAD+ and ethanol at pH 9.0. Methyl picolinimidate activates the enzyme 19-fold and modifies about 50 of its 60 amino groups, as determined by spectral and amino acid analyses. When the active sites are protected with NAD+ and pyrazole (or NADH and isobutyramide) methyl picolinimidate activates only 2-fold, although most of the amino groups still react; after removal of the reagents by gel filtration, the partially substituted enzyme could be activated 11-fold more by methyl picolinimidate or 2-fold more by 14C-cyanate with the modification of about three amino groups per active site. A similar experiment with ethyl acetimidate in the first step and methyl picolinimidate in the second step gave similar results. Product inhibition studies show that the reactions catalyzed by both the native and picolinimidylated enzymes at pH 9.0 conform to the same mechanism, ordered bi bi. The modified enzyme has 12- to 53-fold larger Michaelis and inhibition constants for NADH and NAD+ and 12- and 30-fold larger turnover numbers. The rate-limiting step in either the forward or the reverse reaction with the native enzyme is the breakdown of the enzyme-coenzyme complex; the picolinimidylated dehydrogenase probably gives higher maximum velocities because the complexes dissociate faster. Picolinimidylation of the enzyme does not greatly affect the binding of AMP, ADP, or adenosine 5'-diphosphoribose, but markedly decreases the binding of NAD+, NADH, and 3-acetylpyridine adenine dinucleotide. The reactivities of the essential —SH groups and the zinc ions at the active sites of the enzyme are not affected by picolinimidylation. These results indicate that the amino groups that can be modified are not required for the catalytic activity of the enzyme and that there is probably at least one amino group near the binding site for the nicotinamide ring of the coenzyme.

BRYCE V. PUPP From The Rockefeller University, New York, New York 10021 SUMMARY Reaction of horse liver alcohol dehydrogenase with imidoesters or cyanate at pH 8 significantly increases the activity of the enzyme, as assayed in high concentrations of NADf and ethanol at pH 9.0. Methyl picolinimidate activates the enzyme 19-fold and modifies about 50 of its 60 ammo groups, as determined by spectral and amino acid analyses. When the active sites are protected with NADf and pyrazole (or NADH and isobutyramide) methyl picolinimidate activates only Z-fold, although most of the amino groups still react; after removal of the reagents by gel filtration, the partially substituted enzyme could be activated 11-fold more by methyl picolinimidate or Z-fold more by 1Gcyanate with the modification of about three amino groups per active site. A similar experiment with ethyl acetimidate in the tirst step and methyl picolinimidate in the second step gave similar results.
Product inhibition studies show that the reactions catalyzed by both the native and picolinimidylated enzymes at pH 9.0 conform to the same mechanism, ordered bi bi. The modified enzyme has 1% to 53-fold larger Michaelis and inhibition constants for NADH and NAD+ and 12-and 30fold larger turnover numbers.
The rate-limiting step in either the forward or the reverse reaction with the native enzyme is the breakdown of the enzyme-coenzyme complex; the picolinimidylated dehydrogenase probably gives higher maximum velocities because the complexes dissociate faster.
Picolinimidylation of the enzyme does not greatly affect the binding of AMP, ADP, or adenosine 5'-diphosphoribose, but markedly decreases the binding of NAD+, NADH, and 3-acetylpyridine adenine dinucleotide. The reactivities of the essential -SH groups and the zinc ions at the active sites of the enzyme are not affected by picolinimidylation.
These results indicate that the amino groups that can be modified are not required for the catalytic activity of the enzyme and that there is probably at least one amino group near the binding site for the nicotinamide ring of the coenzyme. *  Studies on the amino acid residues at the active sites of horse liver alcohol dehydrogenase (alcohol : NADf oxidoreductase, EC 1.1.1. 1) have indicated that carboxymethylation of 1 cysteine residue per site inactivates the enzyme (2, 3) even though the modified enzyme can still interact with NADH and ethanol (4). Other amino acids have not been directly implicated, but Kosower has predicted that the c-amino group of a lysine residue participates in the binding of coenzyme and substrate (5). We have studied the effects on the enzyme of the imidoesters, methyl picolinimidate and ethyl acetimidate, and cyanate, which form stable derivatives with primary amino groups (6, 7) but not with other functional groups of proteins, such as the cysteinyl -SH groups. MPI' was introduced by Benisek and Richards (8) for attaching metal-chelating groups onto enzymes.

METHYL PICOLINIMIDYL-PICOLINIMIDATE AMINE
We thought that the zinc ion at the active site of the enzyme (9-12) might bind MPI and facilitate its reaction with a nearby amino group.
Unexpectedly, reaction of the enzyme with the reagents increased the activity of the enzyme. The chemical and kinetic studies reported here allow us to estimate the number of amino groups at the active sites, to explain the enhancement of the enzymic activity, and to implicate amino groups in the activity of alcohol dehydrogenase.
The reagent was stored at -10". A 0.1 M solution was prepared by the addition of 12.5 ~1 of reagent (density 1.1 g per ml) to each ml of reaction mixture.
Enzyme Assays-The enzymes were routinely assayed in 1 ml of 85 mu Na4P207, 6.5 mu semicarbazide hydrochloride, 18 mM glycine, 550 mu ethanol, and 1.75 mu NAD+, at pH 9.0 and 25' (Boehringer Mannheim, 1968 catalogue). A fresh solution of YNAD+ was prepared each day and added to a solution of the other compounds.
Enzyme solutions were diluted (if necessary) in 1 mg per ml of bovine serum albumin in buffer at pH 7.7 to 9.0, and measured volumes were introduced into the assay mixture on a plastic spoon.
The time required for a change of 0.1 A at 340 rnp was determined; 10 pg or less of enzyme were assayed.
Enzymes were also assayed according to Dalziel (14)  Kinetic Stud&---The buffer for the product inhibition studies was 10 mu Na4P20? and 20 mu glycine, pH 9.0. Acetaldehyde was redistilled on the day of use. Reagent grade 95% ethanol was also redistilled. NAD+ was purified (15) for some of the experiments.
Solutions of substrates were prepared daily. Solutions of NAD+ and adenine nucleotides were neutralized before use. Concentrations of NADf and NADH were determined from the absorbances at 260 rnp and 340 rnp, respectively (16). Enzyme activity was determined in a total volume of 2 ml in l-cm cuvettes, after the reaction had been initiated by the addition of 20 or 50 ~1 of enzyme with an adder-mixer (17). Initial velocities (of NADf reduction or NADH oxidation) were determined from the tangents to the curves recorded with a Zeiss spectrophotometer PM& II equipped with a TE-converter so that 0.2 A (at 340 rnp) could be recorded linearly; the recorder speed was varied from 1 to 4 inches per min.
Solutions and reaction mixtures were kept at 25" with circulating water. Enzyme was diluted in 1 mg per ml of bovine serum albumin in 10 IM Na4P20T and 20 mu glycine, pH 9.0, with or without 1 mg per ml of reduced glutathione, or in 0.05 M Tris-HCl, pH 8.0, and kept in ice. Dilute solutions of the picolinimidylated enzyme lost about 20% of their activity (in the routine assay) during the 2 to 3 hours required for the kinetic studies; therefore, each assay point was corrected for the actual activity at that time.
The normality of the enzyme was calculated on the basis of an equivalent weight of 40,000 (18  The column (0.6 X 11 cm) of Spinco AA-27 resin was eluted with 0.38 N sodium (citrate) buffer, pH 5.28, at 52" and 50 ml per hour.
Similar results were obtained with a column (0.9 X 6 cm) of Spinco PA-35 resin eluted with 0.38 N sodium (citrate) buffer, pH 5.26, at 55" and 75 ml per hour.
inhibitor, data were fitted to the equations u = VS/(K + S) by means of a least squares method and on the assumption of equal variance for the velocities.
Slopes (K/V) and intercepts (l/V) were plotted against inhibitor concentration; weighted least squares were fitted to a line, a parabola, and a "two-one" function as a means of determining which type of inhibition the data fitted best.
In every case but one a line gave the best fit. We distinguished between competitive and noncompetitive inhibition by applying t tests to the intercepts (21). If the probability was greater than 5% that the two intercepts differing most were equal, competitive inhibition was assumed. If the probability was less than 1% that two intercepts were equal, noncompetitive inhibition was assumed. Suitably diluted diquots were then assayed at 25' in 20 rnM Na4P207 and 40 mM glycine, pH 9.0, with 1. The activity is given relative to a control that contained no MPI.
analysis on the assumption that the enzyme has a molecular weight of 80,000 (18)  Picolinimidyllysine was eluted as a broad peak at about 2.2 times the elution volume of arginine ( Fig. 1) and was assumed to have about the same color value as arginine; the sum of t'he lysines and picolinimidyllysines (calculated on this assumption) for the PI-enzyme equaled the number of lysines in the native enzyme.
About 60% of the picolinimidyllysine present in the PI-enzyme is converted to lysine by hydrolysis in 6 M HCI at 110" in 22 hours. For accurate determination of the picolinimidyllysine content, therefore, the values found after hydrolysis at different times were extrapolated with first order kinetics to zero time.

Activation of Alcohol Dehydrogenase by ModiJication of Amino
Groups-As shown in Fig. 2 3. Determination of the incorporation of picolinimidy1 groups into picolinimidylated enzyme by difference spectrophotometry.
Enzyme, 10 mg, in 2 ml of 0.25 M N-ethylmorpholine-HCI buffer, pH 8, 5% ethanol, and 10 mM phosphate was activated 17-fold by reaction with 0.1 M MPI at 27" for 3 hours. The reagent and buffers were exchanged for 50 mM sodium phosphate and 0.5 mM EDTA, pH 7.6, on a column (0.9 X 31 cm) of Sephadex G-25, fine, at room temperature.
From the absorption spectrum (corrected for light scattering due to slight turbidity) and the protein concentration, the molar absorptivities were calculated.
The molar absorptivities of native enzyme (LADH) determined in the same buffer were subtracted from those of PI-enzyme (PI-LADH), and the difference is presented (O-O). The number of picolinimidyl groups incorporated was calculated from the molar absorptivity-of %-butylpicolinamidine hydrochloride at 262 11111. namelv 5790 M-I cm-l (88): the absorntion expected for 52 such groups fs given (a---•j." assay mixture had no effect on the activity observed. The modified enzyme was 6 times as active as the unmodified enzyme when assayed with NADH and acetaldehyde but only 1.2 times as active in the assay used by Dalziel (14) with NAD+ and 8 mM ethanol.
Reaction of the enzyme with MPI for more than 3 to 4 hours led to a progressive loss of activity, with a half-time of 5 hours. Reaction at pH 8 gave more activation (as determined in the routine assay) and a more stable product than reaction at pH 7, 9, or 10. If the PI-enzyme produced by a a-hour reaction at pH 8 was freed of reagents by gel filtration through a column of Sephadex G-25 equilibrated with 0.5 M N-ethylmorpholine-HCI, pH 8.0, it had a half-life of 40 hours at 25" and slowly precipitated.
The modified enzyme was similarly stable in 0.05 M sodium phosphate and 0.5 mM EDTA, pH 7.6, at 5". The isolated PI-enzyme reduced 50 pmoles of NADf per min per mg in the routine assay and 7.6 in the assay of Dalziel; the unmodified enzyme reduced 2.4 and 5.5 pmoles, respectively.
The PI-enzyme had markedly enhanced ultraviolet absorption with a maximum at 264 mp. From the increased absorbance of the PI-enzyme and the known absorptivity of the picolinimidyl group, the number of such groups incorporated was determined (Fig. 3). The PI-enzyme was found to have about 52 of its 60 lysine residues modified.
The difference spectrum of PI-enzyme against native enzyme agrees fairly well with the spectrum calculated for 52 picolinimidyl groups (Fig. 3). Amino acid analyses of 22-, 47-, and 74-hour acid hydrolysates indicated that the PI-enzyme had about 48 residues of picolinimidyllysine. The accuracy of the incorporation data may not be better than &lo'%, but these data show that most of the e-amino groups of liver alcohol dehydrogenase were modified by reaction with

MPI.
The or-amino groups of the enzyme are probably acetylated (29) and hence cannot react with MPI.
The activation is apparently due to a chemical modification of the enzyme.
Reaction of Amino Groups at Active Sites- Fig.   4 shows that the activation of the enzyme by MPI was less when the active sites of the enzyme were protected by the prior formation of binary or ternary complexes of the enzyme.
The coenzymes alone protected only partially even though the active sites should have been almost fully occupied by coenzyme, since the dissociation constants for NAD+ and NADH are 51 pM and 1 pM, respectively (in buffers of low ionic strength, at pH 8 (30)). The ternary complexes were activated only 2-fold; the concentrations of NADH and isobutyramide, or NAD+ and pyrazole, far exceeded the dissociation constants (30-32). After reaction of t,he enzyme and MPI in the presence of NAD+ and pyrazole (as in Fig. 4, but for 4 hours) and removal of the MPI, NAD+, and pyrazole by gel filtration, the enzyme was found to have been activated only 2-fold, but nevertheless to have 50 f 5 picolinimidyl groups per molecule. Further reaction of the coenzyme-free, partially substituted PI-enzyme with MPI gave 11-fold more activation, for an over-all activation of 22-fold (Fig. 5). It appears that the activation is due to the modification of a few amino groups at the active sites of the enzyme.
If the coenzyme-free, partially substituted PI-enzyme was treated with cyanate instead of with MPI (Fig. 5), a 2-fold activation was observed, for an over-all activation of 4-fold, as assayed in 0.55 M ethanol.
(However, as assayed in 17 mM ethanol with 1.7 mM NAD+ at pH 8.8 (4) and 25", cyanate inactivated this PI-enzyme with a half-time of 90 min. Native enzyme was activated slightly by cyanate and then inactivated, with precipitation, with a half-time of 56 min as shown in the routine assay.) Using radioactive cyanate in the experiment of  in the presence of NAD+ and pyrazole was freed of reagents by filtration through a column (0.9 x 40 cm) of Sephadex G-25 medium, at room temperature equilibrated with 0.5 M N-ethylmorpholine-HCl, pH 8.0. Then the solution was made 0.1 M in MPI, and the reaction proceeded for 4 hours at 25". The protein was freed of reagents and the incorporation of picolinimidyl groups was determined as in Fig. 3.
In a similar experiment (Fig. 6) In each lettered Jigure, the lower part is a Lineweaver-Burk plot of the primary data.
V has units of AAs4,, per min, and the reciprocal of the concentration of the varied substrate is indicated on the figure. The lines are least square fits of the data to a hyperbola. The upper part of each figure is a replot of secondary data from the lower part.
K/V is the slope of the line from the Lineweaver-Burk plot and has units of mM.min (AA~&~. l/V is the intercept from the primary plot.
The lines in the secondary plots are least squares fits to a line.
A (after correction for 60% loss of this derivative during hydrolysis). This determination is less accurate than the spectrophotometric one because it depends upon integration of a low, broad peak (compare Fig. 1); moreover, extrapolation of values from timed hydrolysates was not made. Kinetics of Reactions Catalyzed by Native and Picolinimidylated Enzyme-The kinetic basis for the enhanced activity of the PI-enzyme was investigated by means of product inhibition studies.
The native and modified dehydrogenases were compared at pH 9.0 since the activation was observed with the routine assay at pH 9.0. Fig. 7   respectively. b The apparent maximum velocities were corrected with the general equation V' = V(l + Km/S) where X represents the concentration of nonvaried substrate, and V' the corrected velocity.
( Fig. 7, A and B). Ethanol was a linear noncompetitive inhibitor (Fig. 7C). For inhibition by acetaldehyde (Fig. 70), the closest intercepts were not significantly different, but the farthest ones were. The data did not fit linear competitive inhibition as well as they fit linear noncompetitive inhibition, so we concluded that acetaldehyde was a linear noncompetitive inhibitor. Fig. 8 presents the product inhibition experiments with the PI-enzyme. NAD+ (Fig. 8A) and NADH (Fig. 8B) gave linear competitive inhibition. For inhibition by ethanol (Fig.  8C) the intercepts of Lines 1, I, S, and 4 were significantly different, and the replots of the slopes and intercepts fitted a line almost as well as a parabola.
The data fitted the over-all equation for linear noncompetitive inhibition better (lower variance and smaller standard errors) than they fitted slope-parabolic noncompetitive inhibition or linear competitive inhibition. In another experiment with ethanol concentration from 0 to 100 mM the replots fitted a line better than a parabola.
Thus we concluded that ethanol gave linear noncompetitive inhibition over the concentration range studied, although use of a wider range may reveal that the inhibition is more complex. Acetaldehyde gave linear noncompetitive inhibition (Fig. 80).
E$ect of Picolinimidylation on Binding of Parts of NAD+ Molecule-The weaker binding of the PI-enzyme to NAD+ or NADH could be due to the disruption of one or more of the interactions between the enzyme and the coenzyme. To locate the binding region affected, we studied the binding of portions of the NAD+ molecule and NAD+ analogues (Table II).
AMP, ADP, and adenosine 5'-diphosphoribose had essentially the same dissociation constants with the native enzyme as with PI-enzyme, whereas NAD+, NADH, and 3-acetylpyridine adenine dinucleotide bound much less tightly to PI-enzyme than to native enzyme.
It appears that a picolinimidyl group interferes with the binding of the nicotinamide portion of the NADf molecule to the PI-enzyme.
E$ects of Picolinimiclylation on Reactivity of Groups at Active Sites-The PI-enzyme could be activated because picolinimidyl groups have increased the reactivity of the -SH groups that may be involved in the catalytic mechanism. Imidazole, for instance, increases the rate of inactivation of the native enzyme by iodoacetate and iodoacetamide which react with the -SH groups (35). But the native and picolinimidylated enzymes were inactivated by iodoacetate at the same rate (2. patterns consistent with an ordered bi bi mechanism (see "Discussion"), and the data in Figs. 7 and 8 were used to calculate the kinetic constants in the steady state rate equation for this mechanism (33). The constants (Table I) obtained for the native enzyme in this study are quite different (2 to 8 times larger) from those obtained by Dalziel (34) from initial velocity studies at pH 9; the different buffers used could account for the discrepancies.
Comparison of the results from this study shows that all of the constants for the PI-enzyme are larger than those for the unmodified enzyme.
Of particular interest are the observations that picolinimidylation increases the dissociation constants for NADH and NAD+ (Ki, and Ki,) 12 and 53 times and also the turnover numbers (V/Et) 12 and 30 times.
The increased VI/E8 and Ki, of the PI-enzyme are apparently due to modification of amino groups at the active sites, for enzyme that was picolinimidylated in the presence of NAD+ and pyrazole (Fig. 4) -12), and picolinimidyl groups of PI-enzyme could chelate the zinc ions (8) and alter their properties.
If this were so, the picolinimidyl groups should compete with the metal ion chelator, 2,2'-bipyridine (II), for the two or three available coordination positions of the zinc ions bound to the enzyme (10,36), and the modified and unmodified enzymes should bind the chelator differently.
For this study, we used enzyme that was acetimidylated in the presence of NAD+ and pyrazole and then picolinimidylated (Fig. 6). This derivative contained about five picolinimidyl groups and, unlike PI-enzyme, did not form the slightly turbid solutions that interfere with sensitive spectral studies.
The dissociation constants of 2,2'-bipyridine and the enzymes and the extinction coefficients of the complexes were determined by the procedure used by Sigman (II), except that a 0.05 M sodium phosphate and 0.5 mM EDTA buffer, pH 7.5, was used. The native and picolinimidylated enzymes had the same dissociation constants, about 0.3 mM, and the same extinction coefficients at 308 mp, about lo4 M-' cm-l. Apparently, the picolinimidyl groups do not affect the accessibility of the zinc ions at the active sites of PI-enzyme.
Since acetimidylation and carbamylation also activate the native enzyme, chelation of the zinc ions is not responsible for the activation.

DISCUSSION
Amino Groups at Active Sites-The increase in the activity (Fig. 2) of the enzyme after modification of amino groups (Fig.  3) and the protection against this activation furnished by NAD+ and pyrazole or by NADH and isobutyramide (Fig. 4) are most simply explained by the hypothesis that there are amino groups at or very near the active sites of the enzyme.
From the differential labeling experiments (Figs. 5 and 6) we conclude that there are about six amino groups at the active sites of the dimeric enzyme (18), or about three per active site.
This interpretation should be qualified by some limitations in the data. Incorporation of reagents can be determined with an accuracy of perhaps ltl0 to 20%.
Incomplete substitution of amino groups not at the active sites while the active sites are blocked with NADf and pyrazole could leave a small fraction of the large number of amino groups unreacted; these would then react when NAD+ and pyrazole were removed to give some extrinsic incorporation. For example, if 50 amino groups outside the active sites reacted to the average extent of 95$&, 2.5 eq of amino groups would still be free to react in the next step.
Since one reagent was used to block the amino groups not at the active sites and another was used for those at the active sites (MPI and cyanate, or ethyl acetimidate and MPI), differential reactivity of amino groups could raise or lower the number apparently at the active sites.
This possibility is not very likely since the use of two different pairs of reagents gave the same number of amino groups.
Changes in conformation of the enzyme when it binds NAD+ and pyrazole could, of course, be invoked. X-ray diffraction (37) and optical rotatory dispersion studies (38,39) indicate that the enzyme may change conformation when it forms ternary complexes, but the interpretations are still tentative, and we do not know whether the exposure of amino groups would be affected if a conformational change does occur. Finally, we are assuming that the reagents reacted only with amino groups (6,7). None of the results can eliminate the possibility that the reagents reacted with one group per active site (other than an amino group) with special reactivity. This possibility seems unlikely, however, for several reasons. Both cyanate and the imidoesters gave similar results. The difference spectrum of enzyme that was acetimidylated in the presence of NAD+ and pyrazole and then picolinimidylated (Fig. 6) was typical of picolinamidines (e.g. Fig. 3). MPI activated the enzyme and modified amino groups at about the same rates; both reactions were essentially complete in 3 hours (Fig. 2). It would be coincidental if another functional group had the same properties as an e-amino group.
Kinetic Basis for Activation-The enhanced activity of the PI-enzyme could be due to a change in the mechanism of the enzymic reaction or simply to an increase in the rate of the rate-limiting step. The mechanism for the native enzyme is predominantly ordered bi bi at pH 7.15 (19,40), although it is probably partly random (41)(42)(43). The ordered bi bi mechanism for the forward reaction can be represented by the following scheme (33)  The rate constants on the left of the arrows are for the forward reaction and those on the right are for the reverse reaction. The rate-limiting step in either the forward or the reverse reaction is most probably the breakdown of the enzyme-coenzyme complex (34,42,44,45). In fact, the ternary complexes form, interconvert, and break down so fast over the pH range of 6 to 9 (34) that the reactions were described for many years by the Theorell-Chance mechanism (34,46,47). The product inhibition patterns presented in Figs. 7 and 8 show that the reactions catalyzed by both the native and picolinimidylated enzymes at pH 9.0 conform to the ordered bi bi mechanism.
The data are inconsistent with both the simple Theorell-Chance and rapid equilibrium random bi bi mechanisms since ethanol and acetaldehyde are linear noncompetitive inhibitors (19,33). Moreover, for the PI-enzyme the simple rapid equilibrium random bi bi mechanism is excluded by the observation that, in 8.  8A). If the PI-enzyme had this mechanism, the apparent K;, should have increased as the ethanol concentration increased.
The simple random bi bi mechanism cannot be excluded, but there is no evidence for it (such as nonlinear reciprocal plots and hyperbolic noncompetitive inhibition (19,33)). More complicated mechanisms, such as random with dead end complexes, also cannot be excluded. The simplest interpretation of the data is that the basic mechanism is unchanged by picolinimidylation and that the increase in activity is due to the faster breakdown of the enzyme-coenzyme complexes.  The agreement is good here, also (Table III). The rates of association of the enzyme and coenzymes, ks and kl, are essentially unchanged. However, the calculations in Table III expose an inconsistency in that the calculated kZ is less than V2/EI for both the native and picolinimidylated enzymes.
The over-all rate of the rea,ction cannot be faster than the slowest step.
The inconsistency probably is not caused by impurities in the coenzymes (34), for purified NAD+ gave the same inhibition and Michaelis constants as the best, grade of commercial NAD+ in our experiments; also, the impurities in commercial grades of NADf and NADH did not affect initial rate data at, pH 9.0 (34,48). On the other hand, the presence of inactive (dead end) E.NAD+ or isomeric E .NAD+ complexes in the mechanism could account for the discrepancy (33