Substitution of both histidines in the active site of yeast alcohol dehydrogenase 1 exposes underlying pH dependencies

Abstract

Histidine residues 44 and 48 in yeast alcohol dehydrogenase (ADH) bind to the coenzymes NAD(H) and contribute to catalysis. The individual H44R and H48Q substitutions alter the kinetics and pH dependencies, and now the roles of other ionizable groups in the enzyme were studied in the doubly substituted H44R/H48Q ADH. The substitutions make the enzyme more resistant to inactivation by diethyl pyrocarbonate, modestly improve affinity for coenzymes, and substantially decrease catalytic efficiencies for ethanol oxidation and acetaldehyde reduction. The pH dependencies for several kinetic parameters are shifted from pK values for wild-type ADH of 7.3–8.1 to values for H44R/H48Q ADH of 8.0–9.6, and are assigned to the water or alcohol bound to the catalytic zinc. It appears that the rate of binding of NAD⁺ is electrostatically favored with zinc-hydroxide whereas binding of NADH is faster with neutral zinc-water. The pH dependencies of catalytic efficiencies ($V/E_{\text{t}}{K}_{\text{m}}$) for ethanol oxidation and acetaldehyde reduction are similarly controlled by deprotonation and protonation, respectively. The substitutions make an enzyme that resembles the homologous horse liver H51Q ADH, which has Arg-47 and Gln-51 and exhibits similar pK values. In the wild-type ADHs, it appears that His-48 (or His-51) in the proton relay systems linked to the catalytic zinc ligands modulate catalytic efficiencies.

1. Introduction

The structures and functions of alcohol dehydrogenases (ADHs) have been intensively studied because of their roles in the metabolism of alcohols. ADH catalyzes transfer of a proton from an alcohol to solvent and of a hydride from an alcohol to NAD⁺ to form an aldehyde and NADH [1]. Biochemical studies of the mechanisms and substrate specificities of ADHs provide fundamental knowledge about catalysis and useful information for commercial applications, such as diagnostic reagents and production of alcohols. The present work investigates the enzymatic mechanism of Saccharomyces cerevisiae ADH1, with comparisons to horse liver ADH1E, when the two histidines that can participate directly in catalysis are substituted.

Three-dimensional structures and site-directed mutagenesis have facilitated studies of catalysis by yeast ADH1. X-ray crystallography and cryo-electron microscopy established structures for several forms of the yeast ADH1 that represent various steps in the general mechanism of alcohol dehydrogenases [2–4]. The structures show that histidine residues 44 and 48 are in positions of catalytic importance (Fig. 1). His-44 forms a hydrogen bond with O2A of the pyrophosphate of the NAD(H) [5]. His-44 was changed to an arginine residue in yeast ADH, which is present in many wild-type ADHs, such as horse liver ADH [6]. His-48 hydrogen bonds with the 2’-hydroxyl group of the nicotinamide ribose, and potentially acts in acid/base catalysis through the hydrogen bonded system connected to the oxygen of the substrate (see Fig. 1 for yeast ADH and Scheme 1 for horse liver ADH in Ref. [7]). In some ADHs, this residue is a tyrosine, threonine, or glutamate, and it was replaced with serine, glutamate, or glutamine residues in previous studies [5, 7–11]. In the present study both histidines were replaced to make yeast H44R/H48Q ADH, which should form a productive structure, as shown in Fig. 1. The double substitution might be expected to make a pH independent enzyme, or to expose the contributions of other enzymatic groups to the pH dependences, such as the water or alcohol ligated to the catalytic zinc.

Fig. 1.

Structure of yeast alcohol dehydrogenase with modeling of the H44R and H48Q substitutions. The wild-type enzyme complexed with NAD⁺ and 2,2,2-trifluoroethanol (TFE) has histidines 44 and 48 in green stick representation, and the modeling for Arg-44 and Gln-48 in ball and stick with blue and gray coloring (4w6z.pdb subunit A, Ref [2]). The catalytic zinc is just below the label H44R and linked to the O of TFE. The model is similar to the structure observed for the homologous horse liver ADH with the H51Q substitution (8g2l.pdb).

The roles of histidine residues in alcohol dehydrogenases and many other enzymes have been studied by chemical modification with diethyl pyrocarbonate, a toxic sterilizing reagent used to inactivate ribonucleases [12]. Although the reagent is relatively specific for histidine residues, its use to identify “essential” histidines needs to be evaluated and validated. Comparison of the reactions of the wild-type yeast ADH and variants with histidine residues substituted with residues that cannot react with diethyl pyrocarbonate provide a lesson in the interpretation of the modification studies.

A role for histidine residues in catalysis by yeast ADH was suggested because enzymatic reactions are pH dependent with pK values of 7–8, and the rate of inactivation by diethyl pyrocarbonate increased with pH with a pK value of 7.1 for free enzyme and a pK of 8.7 for the enzyme-NADH complex [13–16]. It was suggested that one histidine residue was involved in binding the coenzyme or in acid/base catalysis, although the identity of the residue was not established, and a three dimensional structure was not known at the time. His-44 and His-48 can be substituted conservatively (H44R, H48E, H48Q, and H48S) without completely knocking out activity (but with modest changes in the kinetics of catalysis), and it remains to be determined if these two residues are the ones reacting with diethyl pyrocarbonate and if there are other “essential” histidine residues [6, 7, 9]. Of the remaining 8 histidine residues, 6 are exposed to solvent (sequence numbers 15, 51, 113, 138, 171, and 240). His-15 is near the active site, but it does not appear to be involved in activity because the H15R substitution has only small effects on activity [17]. The nitrogens of the imidazole group of His-66 are ligated to the catalytic zinc and hydrogen bonded to Asp-46, and His-121 is buried with close interactions with protein residues, so that both of these histidines are probably not readily accessible to chemical modification [2, 4].

2. Experimental procedures

2.1. Materials

LiNAD⁺, Na₂NADH and the Klenow fragment of DNA polymerase I were purchased from Boehringer Mannheim Biochemicals; T4-DNA ligase was from International Biotechnologies Inc.; exonuclease III, T4-polynucleotide kinase and restriction enzymes were from Bethesda Research Laboratories or New England Biolabs; and TPCK-trypsin was from Cooper Labs. Deoxyribonucleotides, DEAE-Sepharose CL-6B, and Octyl-Sepharose CL-4B came from Pharmacia P-L Biochemicals; Amersham supplied radioactive nucleotides; 1-butanol-d₉ (98%) was obtained from MSD Isotopes; and Fluka Chemical Corp provided ethanol-d₅. Acetaldehyde was redistilled on the day of use. (4R)[4-²H]Nicotinamide adenine dinucleotide was freshly prepared and purified [18].

2.2. Mutagenesis

The gene for the fermentative yeast S. cerevisiae ADH1 (adc1, YOL086c, UniProtKB entry P00330) was obtained in the 16.5 kbp yeast shuttle vector YEp13, which contains the ampicillin resistance gene for selection in E. coli, parts of the 2μ circle for autonomous replication in yeast, and the LEU2 gene for selection in yeast [19–21]. The mutagenic oligonucleotides GTGTCTGTCGCACTGACTTG (for H44R) and TGACTTGCAAGCTTGGC (for H48Q), with sites of mutation underlined and the codon in italics, were made on a Beckman DNA synthesizer. The 1.6 kbp Sph I fragment containing the ADH1 gene was inserted into M13mp18RF. The double substitution was created sequentially, in two steps using the two-primer method [22]. The mutations were confirmed by dideoxy DNA sequencing [23]. Double stranded phage DNA was isolated from infected E. coli JM101 cells [24], and the Sph I fragment containing the mutagenized ADH1 gene was subcloned into the YEp13 vector. The plasmid was produced in E. coli HB101 and used to transform a strain of S. cerevisiae that did not produce active alcohol dehydrogenase (MATa leu2 his4 trp2 adc1–11 adr2–40 adm adr1–1) by the lithium acetate method [25]. Transformants were selected on minimal medium for acquisition of the LEU2 gene and on antimycin A (1 mg/ml) plates for expression of active fermentative ADH1.

2.3. Enzyme preparation and protein characterization

The H44R/H48Q ADH was produced aerobically by growing yeast in 2% Bacto-peptone, 1% Bacto-yeast extract and 2% glucose and purified as described previously [6]. The enzyme appeared to be nearly homogeneous on native polyacrylamide gel electrophoresis at pH 9.5 and on a denaturing gel with sodium dodecyl sulfate. Notably, on non-denaturing gel electrophoresis at pH 9.5, the enzyme migrated about 90 % as fast as wild-type enzyme toward the anode, as expected when a histidine residue is replaced by an arginine residue (see for example Ref. [17]). In contrast, the single H48Q substitution does not change the rate of migration. The physical behavior and thermostability at pH 8.0 and 60 °C of H44R/H48Q ADH are similar to that of the wild-type enzyme and suggest that the overall structure of the enzyme is not affected by the mutations.

The concentration of active enzyme sites (normality, N) was determined by titration with NAD⁺ in the presence of 10 mM pyrazole [26]. The difference spectrum of the enzyme-NAD⁺-pyrazole complex against enzyme and NAD⁺ showed the doublet of peaks at 284 and 292 nm and an extinction coefficient of 12 mM⁻¹cm⁻¹, typical for wild-type enzyme. The A₂₈₀/A₂₆₀ ratio was 1.8, indicating that the protein was free of contaminating nucleotides. Based on the active site titration and the protein concentration determined with 1.26 A₂₈₀ cm⁻¹ (mg/ml) ⁻¹, 67 % of the active sites in the protein could bind NAD⁺ and pyrazole.

The protein modification was confirmed by isolating and sequencing the peptide containing residues 45–59. The pyridylethylated [27] protein was digested with TPCK-trypsin, and the chromatogram of the peptides separated on a Synchropak RP-P reverse phase column was almost the same as for wild-type and H48Q ADHs [9], except that a peptide with residues 39–59 was missing, and a new peptide with residues 45–59 eluted earlier, as was found for the H44R ADH [6]. The sequencing confirmed the presence of Gln-48, and by inference the cleavage after Arg-44 by trypsin.

2.4. Enzyme kinetics

Enzyme activity was routinely determined with a standard assay in 85 mM Na₄P₂O₇, 18 mM glycine, and 6.5 mM semicarbazide hydrochloride buffer at pH 9 and 30 °C with 1.75 mM NAD⁺ and 550 mM ethanol [28]. Kinetic experiments used a physiological buffer of 83 mM potassium phosphate with 40 mM KCl at pH 7.3 and 30 °C. ADH activity was determined by measuring the change in absorbance at 340 nm for NADH on a Cary 118C spectrophotometer interfaced to an IBM PC/XT computer equipped with a Data Translation 2805 A/D board. A FORTRAN program was used to estimate initial velocities by a linear or parabolic fit of the time course. Kinetic constants were estimated with the programs HYPER, SEQUEN, COMP, NONCOMP and UNCOMP, and the errors on the kinetic constants for the appropriate fit were less than 25 %, indicating that the values were well-determined [29]. In initial velocity studies, the concentrations of both substrates were varied, and product inhibition studies held the concentration of one substrate constant while varying the other substrate and the inhibitor concentrations. The ranges of concentrations used are listed in the tables.

3. Results

3.1. Modification of histidine residues with diethyl pyrocarbonate

Diethyl pyrocarbonate selectively reacts with histidine residues, and the resulting carbethoxyimidazole group absorbs light at 240 nm and provides a method to estimate the number of modified histidines [12, 30]. Treatment of wild-type and the three substituted yeast ADHs led to loss of activity with different rates and extents of inactivation (Table 1). Each of the H44R and H48Q substitutions partly protect against inactivation, and the H44R/H48Q double substitution protects even better as compared to the other enzymes. The loss of activity reflects the contribution to the catalytic activity of the residues that are not substituted, so that the decrease in activity of H44R ADH might indicate that His-48 contributes about 70% of the activity of the wild-type enzyme. However, it appears that the effects of the single substitutions are not quite additive for the double H44R/H48Q substitution. The activity of the partly inactivated H44R ADH was restored after treatment with 22 mM hydroxylamine for 21 h at room temperature, indicating that histidine residues were probably involved in the inactivation. In contrast, lysine residues, such as Lys-206, which interacts with the adenosine ribose of NAD (Fig. 1), probably were not modified. Inactivation by modification of cysteine or tyrosine residues is also not likely with the low concentration of diethyl pyrocarbonate that was used.

Table 1.

Inactivation of yeast ADHs by diethyl pyrocarbonate^a.

Parameter	Wild-type	H44R	H48Q	H44R/H48Q
Inactivation constant (M−1s−1)	180	31	25	11
Initial enzyme activity (s−1)	400	60	98	77
Final enzyme activity (s−1)	12	18	54	49
Residual activity (% of initial)	3.1	30	55	64

^aEnzymes with known concentrations of protein (5–10 μN subunits) were treated with 90 μM diethyl pyrocarbonate (except 120 μM for H44R/H48Q ADH) in 33 mM sodium phosphate buffer, pH 8.0, at 30 °C, conditions that favor specific reaction with histidine residues [30]. The modification of histidine residues was monitored spectrophotometrically at 240 nm, and samples were withdrawn from a duplicate reaction at various times and diluted into a buffer with 1 mg/ml bovine serum albumin (to stop the reaction) for assay of the enzyme activity. The reactions were essentially complete by 20 min. The rate constants for inactivation were computed according to the equation: $\text{Δ}{A}_{\text{t}}=\text{Δ}{A}_0\text{exp}\left(\left(-k/k^{\prime }\right)C_0\left[1-\text{exp}\left(-k^{\prime }\text{t}\right)\right]\right)$, where $\text{Δ}{A}_{\text{t}}$ is the difference in enzyme activity at time t and the Final enzyme activity $\left(A_{\text{t}}-A_{\text{f}}\right),\text{Δ}{A}_{\text{o}}$ is the total difference between the Initial and Final enzyme activities $\left(A_0-A_{\text{f}}\right),C_{\text{o}}$ is the initial concentration of reagent, k’ is the rate constant for hydrolysis of reagent (7.7 × 10⁻⁴ s⁻¹ [31]), and k is the bimolecular rate constant for the inactivation of the enzyme [32]. Final and initial activities are turnover numbers in the standard assay [28]. The concentration of carbethoxyhistidine residues was calculated with an extinction coefficient of 3200 M⁻¹cm⁻¹ at 240 nm and used to prepare Fig. 2. Time courses for wild-type, H44R, and H48Q ADHs are illustrated in Ref. [33].

The number of histidine residues modified during the inactivation is related to the fractional activity for wild-type, H44R, H48Q, and H44R/H48Q ADHs (Fig. 2). The early phase of each reaction is relatively fast compared to the time required for the enzyme assays, and some data points are interpolated. The slopes of the lines and the extrapolated net incorporation at the end of the reaction are most informative. Modification of a particular histidine apparently does not totally inactivate the wild-type enzyme, and it appears that there are three phases of reaction, with about one histidine reacting with little loss of activity, modification of one more being associated with loss of about one-half of the activity, and modification of an additional histidine being associated with almost complete inactivation. Modification of either His-44 or His-48 apparently results in loss of some activity. H44R ADH seems to have one histidine that can be modified without loss of activity, and modification of one more histidine, probably His-48, decreases activity to 70 % of the initial value. Modification of two residues in H48Q ADH decreases the residual activity, probably because of modification of His-44 and reaction of a second “nonessential” residue at the same time. H44R/H48Q ADH is more resistant to inactivation. The studies show that the H44R, H48Q, and H44R/H48Q substitutions partly protect against inactivation by diethyl pyrocarbonate, but extensive reaction of other histidines can further decrease activity.

Fig, 2.

Correlation of loss of activity with the number of histidine residues modified in yeast ADHs. The fractional activity (see Table 1) at any time is (A_t – A_f)/(A_o –A_f), as for a typical first order reaction. The ADHs are identified by the symbols: ●, green, wild-type; ▲, black, H44R; ■, red, H48Q; *, gold, H44R/H48Q.

3.2. Kinetic characterization

Initial velocity and product inhibition studies for H44R/H48Q ADH yielded kinetic constants for both forward and reverse reactions. Intersecting initial velocity patterns indicate a sequential bi bi mechanism where both substrates bind to the enzyme before products are formed. Product inhibition studies showed that the coenzymes were mutually competitive. Inhibition by ethanol against varied concentrations of acetaldehyde was best described as competitive, and inhibition by acetaldehyde against varied ethanol concentrations was also best described as competitive. The amino acid substitutions caused modest (~2-fold) to large (up to 50-fold) changes in kinetic constants as compared to wild-type enzyme (Table 2). The agreement between the independently determined equilibrium constant for the overall redox reaction and the equilibrium constants calculated with the Haldane relationship show that the kinetic constants are self-consistent. The coenzymes bind more tightly (decreased $K_{\text{ia}}$ and $K_{\text{iq}}$), apparently because the ionic interaction of Arg-44 with the pyrophosphate and the hydrogen bonding of Gln-48 with the nicotinamide ribose are better. The effects of the H44R substitution on the free energy for binding of NADH are almost additive. The turnover number in the forward reaction ($V_1/{\mathrm{E}}_{\mathrm{t}}$) is decreased by the substitutions, perhaps because release of NADH is slower. In contrast, trifluoroethanol binds less tightly and seems to be directly related to decreased catalytic efficiency with ethanol ($V_1/{\mathrm{E}}_{\mathrm{t}}{K}_{\mathrm{b}}$). The Michaelis constant for acetaldehyde is increased and is reflected in decreased catalytic efficiency for acetaldehyde reduction ($V_2/{\mathrm{E}}_{\mathrm{t}}{K}_{\mathrm{p}}$). Michaelis constants are useful to characterize enzymes, but they are complex kinetic constants, whereas catalytic efficiencies are more informative mechanistically [34][35][36].

Table 2.

Kinetic constants for yeast alcohol dehydrogenase with substitutions of histidine residues^a.

Constant	Wild-typeb	H44Rb	H48Qc	H44R/H48Qd
$K_{\text{a}}$ (μM), NAD+	160	150	100	320
$K_{\text{b}}$ (mM), ethanol	21	66	18	31
$K_{\text{p}}$ (mM), acetaldehyde	0.74	4.7	12	22
$K_{\text{q}}$ (pM), NADH	94	10	110	42
$K_{\text{ia}}$ (μM), NAD+	950	260	410	230
$K_{\text{iq}}$ (μM), NADH	31	16	13	3.5
KbKia/Ka (mM)e	96	180	27	36
KpKiq/Kq (mM)e	0.21	1.8	0.55	1.1
$K_{\text{ib}}$ (mM)e	120	1600	100	e
Kip (mM)e	1.5	e	e	e
$V_1/E_{\text{t}}$ (s−1)	360	60	30	24
$V_2/{\mathrm{E}}_{\mathrm{t}}$ (s−1)	1800	460	3000	1040
$V_1/E_{\text{t}}{K}_{\text{b}}$ (mM−1 s−1)	17	0.91	1.7	0.77
$V_2/{\mathrm{E}}_{\mathrm{t}}{k}_{\text{p}}$ (mM−1 s−1)	2400	98	250	47
$V_1/{\mathrm{E}}_{\mathrm{t}}{K}_{\mathrm{b}}{K}_{\text{ia}}$ (mM−2 s−1)	18	3.5	4.1	3.4
$V_2/{\mathrm{E}}_{\mathrm{t}}{k}_{\text{p}}{K}_{\text{iq}}$ (mM−2 s−1)	77000	6100	19000	6800
Keq (pM)	12	29	11	12
Activity (V/Et, s−1)	400	60	98	77
Ki, CF3CH2OH (mM)	2.5	19	33	55 f

^a$K_{\text{a}}$, $K_{\text{b}}$, $K_{\text{p}}$, and $K_{\text{q}}$ are the Michaelis constants for NAD⁺, ethanol, acetaldehyde, and NADH, respectively. K_i values are the inhibition constants (dissociation constants for NAD⁺ and NADH). $V_1/E_{\text{t}}$ is the turnover number for ethanol oxidation and $V_2/{\mathrm{E}}_{\mathrm{t}}$ is the turnover number for acetaldehyde reduction. The kinetic constants are described for the ordered bi bi mechanism [36, 38, 39]. The buffer was 83 mM potassium phosphate with 40 mM KCl and 0.25 mM EDTA at pH 7.3 and 30 °C. Standard errors of fits were 10–20% of the listed values. The equilibrium constant is calculated from the Haldane equation, $K_{\text{eq}}=V_1{K}_{\text{p}}{K}_{\text{iq}}\left[{\text{H}}^{+}\right]/V_2{K}_{\text{b}}{K}_{\text{ia}}$, where [H⁺] = 5 × 10⁻⁸ M at pH 7.3. The experimental value is 9.8 pM [40]. Activity is the turnover number in a standard assay [28] at 30 °C, based on active site titration and is used to calculate the concentration of enzyme subunits in every kinetic study.

^bFrom Ref. [6].

^cFrom Ref. [9]. Combined initial velocity and product inhibition studies.

^dCombined initial velocity and product inhibition studies (Supplementary Data Table 1S).

^eDetermined by product inhibition studies. Ethanol was a noncompetitive inhibitor against varied concentrations of acetaldehyde for H44R and H48Q ADHs, but acetaldehyde appeared to be competitive against varied concentrations of ethanol, which means that the slope inhibition constants, $K_{\text{ia}}{K}_{\text{b}}/K_{\text{a}}$ and $K_{\text{iq}}{K}_{\text{p}}/K_{\text{q}}$, were determined, and the intercept inhibition constant, $K_{\text{ib}}$, was determined, but not $K_{\text{ip}}$. For H44R/H48Q ADH, inhibition by ethanol or acetaldehyde against each other was best described as competitive because the intercept inhibition constants were 7–10 times larger than the slope inhibition constants, and the errors were 40 % of the values.

^fDetermined from the uncompetitive (intercept) inhibition constant (K_ii = 140 mM) against varied concentrations of NAD⁺ and 50 mM ethanol (B) and corrected with $K_i=K_{\text{ii}}/\left(1+\left[\text{B}\right]/K_{\text{b}}\right)$ to yield the dissociation constant for binding of trifluoroethanol to the enzyme-NAD⁺ complex.

It is interesting that the turnover number for acetaldehyde reduction ($V_2/{\mathrm{E}}_{\mathrm{t}}$) by the singly substituted H48Q enzyme is increased somewhat, but turnover is decreased in the H44R or H44R/H48Q ADHs. An increased turnover number might lead to faster fermentation. The common yeast strain used for commercial fermentations, S. carlsbergensis, expresses an ADH1 with differences in four residues, which are not in the active site, and has a 2-fold faster $V_2/{\mathrm{E}}_{\mathrm{t}}$. as compared to wild-type S. cerevisiae ADH1 [37]. There is no obvious explanation for the slightly higher activity of these enzymes.

The overall kinetic mechanism seems to be the same as for the wild-type, H44R and H48Q ADHs, but the mutually competitive inhibition by ethanol and acetaldehyde against each other would be consistent with the “Theorell-Chance” mechanism originally proposed for horse liver alcohol dehydrogenase where concentrations of central ternary complexes with a coenzyme and a substrate were not kinetically significant [41]. Previous studies on product inhibition of horse liver and wild-type yeast ADHs provided evidence for noncompetitive inhibition by ethanol and concluded that the mechanism can be described as preferentially ordered with coenzymes binding before the second substrate binds [39, 42, 43]. For the H44R and H48Q ADHs, some small intercept effects for inhibition by ethanol ($K_{\text{ib}}$) were noted, fitting with noncompetitive inhibition. Mutually competitive inhibition by the substrates can be explained by weak binding of substrates to the enzyme-coenzyme complexes [44]. Although kinetic complexity (e.g., isomerizations of enzyme forms) precludes exact calculations [34], the dissociation constant for ethanol from the enzyme-NAD⁺-ethanol complex can be approximated by K_d,B = K_ibK_aV₂/K_iaV₁, and for wild-type, H44R and H44Q ADHs, the values are 0.10 M, 7.1 M, and 2.4 M, respectively [9]. Noncompetitive effects are not well determined if the intercept inhibition constants are much larger than the slope inhibition constants. For H44R/H48Q ADH, if $K_{\text{ib}}$ is 10 times larger than K_bK_ia/K_a, K_d,B would be calculated to be about 22 M, which is not a realistic estimate.

The mechanism for H44R/H48Q ADH does not fit rapid equilibrium random because the $K_{\text{iq}}$ values determined with three different concentrations of ethanol (2, 25, and 250 mM) were about the same. Furthermore, low concentrations of the dead-end inhibitor trifluoroethanol were uncompetitive against varied concentrations of NAD⁺, suggesting that NAD⁺ binds before the alcohol binds, but higher concentrations of the alcohol can bind to the free enzyme to produce a noncompetitive effect. With wild-type yeast ADH, high concentrations of ethanol can lead to random binding of NAD⁺ and ethanol [14, 42].The kinetic studies suggest that the mechanism of H44R/H48Q ADH approximates ordered bi bi.

3.3. Substrate isotope effects

Studying the hydrogen transfer of substrates with D substituting for H gives isotope effects that are informative about changes in rate-limiting steps in the mechanism. Previous studies showed that the H48Q substitution increased the isotope effects as compared to wild-type enzyme, indicating that hydride transfer from ethanol became somewhat more rate-limiting in catalytic turnover (Table 3). In contrast, the H44R substitution decreased the isotope effects as compared to wild-type enzyme. A preliminary study (with single substrate concentrations of 1 and 550 mM ethanol) suggests that the H44R/H48Q ADH resembles wild-type enzyme where hydride transfer is only partly rate-limiting for catalysis. Reduction of 1 mM acetaldehyde with 0.15 mM NADH gave an isotope effect of ~2 (for ${{}^{\mathrm{D}}V}_2/K_{\text{p}}$) over the pH range, similar to the wild-type and H48Q ADHs [9]. The isotope effects are not consistent with a simple rapid equilibrium random mechanism where hydride transfer is rate-limiting and H/D isotope effects are expected to be >4, but are consistent with a preferentially ordered reaction with some rate limitation by release of products or isomerizations of enzyme complexes.

Table 3.

Isotope effects for oxidation of ethanol by yeast ADHs^a.

	Wild-typeb		H48Qb		H44Rb		H44R/H48Qc
PH	${{}^{\mathrm{D}}V}_1$	${}^{\mathrm{D}}V/K_{\text{b}}$	${{}^{\mathrm{D}}V}_1$	${{}^{\mathrm{D}}V}_1/K_{\text{b}}$	${{}^{\mathrm{D}}V}_1$	${{}^{\mathrm{D}}V}_1/K_{\text{b}}$	${{}^{\mathrm{D}}V}_1$	${{}^{\mathrm{D}}V}_1/K_{\text{b}}$
6	2.0	3.6	3.9	4.8	3.5	3.5	2.8	ND
7	1.8	3.2	4.2	5.2	1.7	2.4	4.2	3.2
8	1.8	3.0	3.3	4.2	1.2	1.5	2.9	1.6
9	1.4	2.6	2.9	3.2	1.1	1.1	2.3	2.1
10	1.3	2.0	2.3	3.4	1.1	1.0	1.8	3.1

^aSuperscript D indicates the ratio of rates with H and D substrates. ${{}^{\mathrm{D}}V}_1$ and ${{}^{\mathrm{D}}V}_1/K_{\text{b}}$ include secondary effects since fully deuterated alcohol was used. Errors were typically 10 % except for H44R/H48Q ADH where the concentration of substrate was not varied and assays were only done in duplicate.

^bInitial velocities were studied in sodium pyrophosphate/phosphate buffers at 30 °C. For determinations of ${{}^{\mathrm{D}}V}_1$ and ${{}^{\mathrm{D}}V}_1/K_{\text{b}}$, ethanol or ethanol-d₅ concentrations were varied from 15–200 mM while the NAD⁺ concentration was fixed at 10 mM (from Refs. [6] and [9]).

^cEthanol concentrations were 1 mM for $V_1/K_{\text{b}}$ and 550 mM for V₁ with 2 mM NAD⁺, using 0.1 M sodium salts of Good’s buffers with 0.25 mM EDTA: MES (2-(N-morpholino)ethanesulfonate), pH 6; MOPS (3-(N-morpholino)propanesulfonate), pH 7; TAPS ((N-tris(hydroxymethyl)methyl-3-aminopropanesulfonate), pH 8 – 9; CAPSO (N-cyclohexyl-2-hydroxyl-3-aminopropanesulfonate), pH 10 [47].

The isotope effects on $V_1/K_{\text{b}}$ for butanol oxidation are about 4 for the wild-type and H48Q ADHs, suggesting rate-limiting hydride transfer for this alcohol [9, 45]. For the H44R/H48Q enzyme, with 5 mM 1-butanol or 1-butanol-d₉ (K_m ~ 30 mM) and 2 mM NAD⁺, the average isotope effect was 4.2 over the pH range from 6.5 to 10.5. Catalytic efficiency with butanol is about 20-fold lower than with ethanol for wild-type ADH [46] and H44R/H48Q ADH, consistent with more rate limiting hydride transfer for both enzymes.

3.4. pH dependence of kinetic constants

Substituting both histidine residues in the active site is expected to alter the pH dependencies of the kinetic constants, as observed previously for the individual H44R and H48Q ADHs [6, 9]. After both histidine residues are substituted, new pH dependencies might be exposed. Resulting pK values might be interpreted with the knowledge of ionizable groups remaining in the enzyme. Fig. 3 provides pH dependencies for the forward and reverse reactions determined from initial velocity studies with the pyrophosphate/phosphate or carbonate buffers (Supplementary Data, Tables 2S and 3S). Some data were collected also with the Good’s buffers [47] for evaluation of potential effects of buffers. The two different buffer systems gave similar results for the parameters that were determined, and the fits in Table 4 used all of the available data.

Fig. 3.

pH dependencies for H44R/H48Q ADH. (A) The data for the forward reaction are from Supplementary Data Table 2S. (B) Data for the reverse reaction are given in Supplementary Data Table 3S. Data points are logarithms of the units. Symbols and units (A) $V_1/E_{\text{t}}{K}_{\text{b}}$ (Δ, pyrophosphate buffers; ∇, Good’s buffers), mM⁻¹s⁻¹; 1/K_ia (□), mM⁻¹; $V_1/{\mathrm{E}}_{\mathrm{t}}$ (◊, pyrophosphate buffers; +, Good’s buffers), s⁻¹; $V_1/E_{\text{t}}{K}_{\text{a}}$ (○), mM⁻¹s⁻¹; (B) $V_2/{\mathrm{E}}_{\mathrm{t}}{K}_{\text{p}}$ (Δ, pyrophosphate buffers; ∇, Good’s buffers), mM⁻¹s⁻¹; 1/K_iq (□), mM^–1; $V_2/{\mathrm{E}}_{\mathrm{t}}$ (◊), s⁻¹; $V_2/{\mathrm{E}}_{\mathrm{t}}{K}_{\text{q}}$ (○), mM⁻¹s⁻¹. The lines are drawn from the computer fits given in Table 4. The association constants (1/K_ia and 1/K_iq) were fitted so that the pK values represent free enzyme.

Table 4.

pH Dependence of kinetic constants for forward and reverse reactions of H44R/H48Q and wild-type yeast ADHs^a.

	H44R/H48Q		Wild-typeb
Kinetic constant	pK	Limiting values or (at pH 7)	pK or slope	Limiting values or (at pH 7)
NAD+ and ethanol
$V_1/E_{\text{t}}$ (s−1)	pK1 7.0 ± 0.2 pK2 9.2 ± 0.3	$k_{\text{a}}$ 32 ± 9 $k_{\text{b}}$ 230 ± 50	7.3 ± 0.1	$k_{\text{a}}$ 190 ± 14 $k_{\text{b}}$ 530 ± 16
$V_1/E_{\text{t}}{K}_{\text{b}}$ (mM−1s−1)	8.5 ± 0.1	$k_{\text{b}}$ 12 ± 2	7.8 ± 0.2	$k_{\text{a}}$ 3.5 ± 0.9 $k_{\text{b}}$ 41 ± 5
$K_{\text{ia}}$ (mM)	~pH independent	0.16 (at pH 7)	8.1 ± 0.2	$k_{\text{a}}$ 0.54 ± 0.06 $k_{\text{b}}$ 4.9 ± 0.9
V1/EtKa (mM−1s−1)	8.03 ± 0.04	$k_{\text{b}}$ 1200 ± 80	−0.11 ± 0.02	1570 (at pH 7)
NADH and acetaldehyde
$V_2/{\mathrm{E}}_{\mathrm{t}}$ (s−1)	9.6 ± 0.1	$k_{\text{a}}$ 850 ± 60	7.8 ± 0.2	$k_{\text{a}}$ 3600 ± 300 $k_{\text{b}}$ 1200 ± 200
V1/EtKp (mM−1s−1)	9.5 ± 0.1	$k_{\text{a}}$ 32 ± 3	7.7 ± 0.1	$k_{\text{a}}$ 3200 ± 200
$K_{\text{iq}}$ (μM)	8.9 ± 0.2	$k_{\text{a}}$ 7.9 ± 1.4	7.4 ± 0.2	$k_{\text{a}}$ 15 ± 2 $k_{\text{b}}$ 410 ± 80
$V_2/{\mathrm{E}}_{\mathrm{t}}{k}_{\text{q}}$ (μM−1s−1)	8.8 ± 0.1	$k_{\text{a}}$ 16 ± 2	7.8 ± 0.2	32 ± 3

^aThe data points for H44R/H48Q ADH are shown in Fig. 3 with the lines calculated from the fits of the data to the logarithmic forms of the equations derived for a simple pH dependence where two forms of enzyme, E and EH, are related by the pK value, and $k_{\text{a}}$ represents the rate or dissociation ($K_{\text{ia}}$ or $K_{\text{iq}}$) constants for reaction of EH, and $k_{\text{b}}$ represents the constants for reaction of E. If only one form of enzyme is reacting, the equations are $k_{\text{obs}}=k_{\text{a}}/\left(1+K/\left[{\text{H}}^{+}\right]\right)$ or $k_{\text{obs}}=k_{\text{b}}/\left(1+\left[{\text{H}}^{+}\right]/K\right)$, and if both forms react, the “WAVE” equation applies: $k_{\text{abs}}=\left(k_{\text{a}}+k_{\text{b}}K/\left[{\text{H}}^{+}\right]\right)/(1\left.+K/\left[{\text{H}}^{+}\right]\right)$. $V_1/E_{\text{t}}$ data for H44R/H48Q ADH were fitted to $k_{\text{obs}}=\left(k_{\text{a}}+k_{\text{b}}{\text{K}}_1/\left[{\text{H}}^{+}\right]\right)/(1+{\text{K}}_1/{\text{H}}^{+}]+\left[{\text{H}}^{+}\right]/{\text{K}}_2)$, where $k_{\text{a}}$ and $k_{\text{b}}$ are the activities of monoprotonated and unprotonated forms of enzyme, and pK₁ describes the deprotonation of H₂E, and pK₂ is for deprotonation of HE. See also Refs. [6, 7]. A nonlinear least squares program was used for fitting of the logarithmic forms of the equations. Fits to more complicated mechanisms with four or five parameters were also tested, but the simplest mechanism with low standard errors (e.g., pK ± < 0.2) was considered to be acceptable and informative.

^bpH dependence graphs and data for wild-type enzyme were reported previously [6, 7, 9, 14]. Data for H44R ADH are given in [6] and data for H48Q ADH in [9].

The H44R/H48Q substitutions altered the pH dependencies in the forward and reverse reactions as compared to wild-type ADH. The wild-type enzyme appears to show a general dependence with pK values in the range of 7.3 to 8.1 for most kinetic parameters in both the forward and reverse reactions. For H44R/H48Q ADH, the pH dependence for $V_1/{\mathrm{E}}_{\mathrm{t}}$ is more complicated than for wild-type or H48Q ADHs [9], but turnover is controlled by multiple unimolecular steps in the forward reaction for the simplest ordered bi bi mechanism [34, 36]. The other pH dependencies are quite simple, as compared to the “WAVL” functions for wild-type ADH, which has limiting values at low and high pH. The binding of NAD⁺ (1/K_ia) is pH independent, in contrast to wild-type ADH with tighter binding at low pH than at high pH. Because the general mechanisms of yeast and horse liver ADHs include isomerizations of enzyme complexes, rate constants for individual steps in the mechanism cannot be calculated, but the kinetic parameters are relevant for analyzing the reactions of the protein complexes in the mechanism [4, 39, 48, 49].

For the reverse reaction, the pK values of 7.4 to 7.8 for all of the parameters for the wild-type enzyme have been shifted and simplified with pK values of 8.8 to 9.6 for the H44R/H48Q ADH. It appears that only one ionizable group in H44R/H48Q ADH is controlling steps in each direction of the reaction.

4.0. Discussion

4.1. “Essential” histidine residues

Previous studies with diethyl pyrocarbonate modification suggested that one histidine per subunit was essential for enzyme activity of wild-type yeast ADH based on an extrapolation such as shown in Fig. 2, but the present results indicate that modification of two histidines leads to more complete inactivation [15, 16]. The previous studies showed that reaction of the apoenzyme was faster as pH was increased, reaching a maximum rate above a pK value of 7.1, consistent with reaction of unprotonated imidazole groups. Binding of NADH partially protected against inactivation and shifted the pK value to 8.7, which was explained by proposing that the “essential” histidine is involved in binding of NADH [16]. This interpretation is consistent with the observation that His-44 interacts with the pyrophosphate O2A of NADH and the proposal that increasing the pK of the imidazole group decreases the rate of reaction with diethyl pyrocarbonate. The results are also consistent with modification of His-48, which interacts with the 2’-hydroxyl group of the nicotinamide ribose and participates in the hydrogen bonded network to the catalytic zinc-water, and the idea that binding of the nicotinamide group of NADH close to the zinc shifts the pK for the system [2–4].

The structures and active sites of horse liver ADH1E and yeast ADH1 are homologous, so that H44R yeast ADH1 would be essentially similar to the horse ADH with Arg-47, and both ADHs would have only one accessible histidine in the active site [2, 5]. An early study found that diethyl pyrocarbonate modified four (of the seven) histidine residues per subunit in horse ADH and decreased enzymatic activity, but it was concluded that none of these histidines was in the active site. However, formation of complexes with NADH with or without the inhibitor isobutyramide protected against inactivation [50]. A complication in this study was that reaction of diethyl pyrocarbonate produced an initial slight increase in activity and that treatment with hydroxylamine did not restore activity, suggesting that lysine residues may have reacted. The carbethoxy modification (CH₃CH₂OCO−) would neutralize the lysine epsilon amino group. It was later shown that modification of the epsilon amino group of Lys-228 with cyanate to produce a neutral carbamoyl substitution (NH₂CO−) can increase the activity of the enzyme [51]. In a subsequent study, this complication was avoided by first acetimidylating (CH₃-C(=NH₂⁺)−) all accessible lysine residues so that reaction of histidine residues could be studied [32]. This study also showed that four histidines per subunit could be modified, and that one or two histidines were involved in catalytic activity. The ternary complex with NADH and isobutyramide was inactivated more slowly than apoenzyme and modification of one histidine was associated with inactivation. Modification of an additional one led to more complete inactivation. The rate of modification of histidines in the ADH increased as pH increased, described by a pK of 6.8, but inactivation was approximately described by a pK of 9.6, which was attributed to ionization of the catalytic zinc-water, which interacts through the proton relay system with His-51.

The studies of the reaction of diethyl pyrocarbonate with these ADHs show that histidines are involved in activity, but establishing the stoichiometry and the contributions to catalysis are problematic. Chemical modifications are useful for identifying amino acid residues that are involved in activity, and they can facilitate structural interpretations and mechanistic studies [52]. Nevertheless, the concept of an “essential” amino acid residue is outmoded because it is based on chemical modifications that decrease enzyme activity and might just block the active site. Site-directed mutagenesis provides more information about the contribution of a residue to catalysis, but extensive kinetic and structural analyses are still required for each modified enzyme [53]. The site-directed mutagenesis studies on yeast alcohol dehydrogenase show that histidine residues 44 and 48 contribute to catalysis, but are not absolutely “essential” for activity.

4.2. Overall enzyme structure and mechanism

X-ray crystallography and cryo-electron microscopy have established structures for several forms of the yeast ADH in the general mechanism of alcohol dehydrogenases as shown in Scheme 1 [2–4]. The yeast enzyme is a tetramer of chemically identical subunits that are arranged as “back-to-back” dimers of subunits with the active sites exposed to solvent, but each subunit can have different conformations for the protein and coordination geometries of the catalytic zinc. In the apoenzyme conformation (E forms), the active site cleft between the catalytic and coenzyme binding domain is open, and the catalytic zinc has an “alternative” coordination with Cys-43, His-66, Cys-153, and Glu-67. The closed forms (F) have the protein domains locked around the coenzyme, and the zinc has the “classical” coordination with Cys-43, His-66, Cys-153, and water or alcohol displaces Glu-67 (see Scheme 4 in Ref. [7]). The apoenzyme (E, 7kcq.pdb) binds NAD⁺ to make E-NAD⁺ (5env.pdb, subunit B), which isomerizes to the closed conformation (F-NAD⁺). Alcohol binds to make F-NAD⁺-Alc and deprotonates to the presumed intermediate alkoxide (F-NAD⁺-Alk, 7kcb.pdb, 4w6z.pdb subunits A and C), which transfers hydrogen and yields F-NADH-aldehyde. After aldehyde is released, F-NADH (7kc2.pdb) isomerizes to produce the open conformation of E-NADH (7kjy.pdb), which releases NADH. This is a minimal mechanism that does not specify states of protonation of the various forms, other intermediates in each step and possible alternative pathways.

Scheme 1.

General Mechanism for Alcohol Dehydrogenases

Horse liver ADH has a similar mechanism, except that the classical coordination of the catalytic zinc was observed by X-ray crystallography in the apoenzyme, and only the structure of the enzyme-NADH complex (4xd2.pdb) showed evidence for alternative positions for the catalytic zinc [54, 55]. Structures of several complexes with alcohol and aldehyde analogs have been determined [55–58].

Structural information was not obtained for the yeast ADHs with the substituted histidines, but the behavior of the proteins are similar (purification, stability, electrophoresis) to that of the wild-type enzyme. The substitutions of the residues on the surface of the protein should not generate steric conflicts. The global, overall structure is probably not altered by the substitutions. Horse liver ADH1E with the H51Q or K228R/H51Q substitutions crystallize in the same space group as wild-type enzyme, and the protein structures are closely superimposable (8g2l.pdb, and 1qv6 and 1qv7 [11]). Local interactions are expected to change with H44R/H48Q ADH, but the kinetics of the enzymatic reaction are very similar to those of the wild-type enzyme, and the general mechanism above should apply. Wild-type ADHs have some alternative pathways that appear with very high concentrations of ethanol, but the inhibition patterns and isotope effects suggest that there are no major changes in the mechanisms of the substituted ADHs. Nevertheless, modification of residues involved in binding of substrates and proton transfers produces significant changes in the kinetic constants and pH dependences for the mechanism.

4.3. Interpretation of the pH dependencies

4.3.1. Similarity of horse liver H51Q and yeast H44R/H48Q ADHs

Assignment of the pK values to particular amino acid residues in wild-type enzyme is difficult because there are many ionizable groups, but it might be simpler when both histidine residues in the active site have been substituted with residues that do not ionize in the pH range from 5.5 to 10.0. Furthermore, yeast H44R/H48Q ADH is structurally homologous to horse H51Q ADH, which has also Arg-47 instead of His-44 as in the wild-type yeast ADH. Ternary complexes of both the yeast and horse ADHs have the catalytic zinc bound to water or alcohol and a lysine residue, Lys-206 in yeast (Fig. 1) or Lys-228 in horse [58], interacting with the 3’-hydroxyl group of the adenosine ribose of the coenzyme. Both of these enzymes have similar pH dependencies (Supplementary Data Table 4S), and considering all of the information should provide a more comprehensive understanding of the pH dependencies. Yeast H44R ADH and horse ADH1E also have the homologous arginine and histidine residues in the active sites, but some pH dependencies for the kinetics are different (Supplementary Data Table 5S). The following discussion considers the steps in the mechanism in Scheme 1, starting with apoenzyme in each direction.

4.3.2. Interpretation of the pH dependencies for alcohol oxidation

The estimated rate constant for binding of ${\mathrm{N}\mathrm{A}\mathrm{D}}^{+}\left(k_{\mathrm{o}\mathrm{n}}=V_1/{\mathrm{E}}_{\mathrm{t}}{K}_{\mathrm{a}}\right)$ to $\mathrm{H}44\mathrm{R}/\mathrm{H}48\mathrm{Q}\mathrm{A}\mathrm{D}\mathrm{H}$ increases with $\mathrm{p}\mathrm{H}$ and reaches a maximum above a $\mathrm{p}\mathrm{K}$ of 8.0, which could reasonably be assigned to ionization of a water bound to the catalytic zinc because the zinc-hydroxide would favorably interact with the positively-charged nicotinamide ring. Binding of NAD⁺ to $\mathrm{H}48\mathrm{Q}\mathrm{A}\mathrm{D}\mathrm{H}$ also increases as $\mathrm{p}\mathrm{H}$ is increased with a $\mathrm{p}\mathrm{K}$ of 7.3 [9]. In contrast, the binding to the wild-type ADH is essentially $\mathrm{p}\mathrm{H}$ independent (Table 4). Transient kinetics show that binding of ${\mathrm{N}\mathrm{A}\mathrm{D}}^{+}$ to horse liver H51Q ADH has a WAVL dependence with a faster rate at higher $\mathrm{p}\mathrm{H}$ and a $\mathrm{p}\mathrm{K}$ value of 8.0 [10]. However, the binding of NAD⁺ is not a simple bimolecular reaction for the horse enzymes because the initial enzyme-NAD⁺ complex isomerizes (changes conformation [4, 59]) as shown by transient kinetics studies [10, 49, 60].

Steady-state kinetics also suggest that the horse and yeast ADH-NAD⁺ complexes isomerize. For a simple ordered bi bi mechanism, the rate constant for binding of NAD⁺ to apoenzyme is $k_1=V_1/{\mathrm{E}}_{\mathrm{t}}{K}_{\mathrm{a}}$, and the rate constant for dissociation of the E-NAD⁺ complex is $k_{-1}=V_1{K}_{\mathrm{i}\mathrm{a}}/{\mathrm{E}}_{\mathrm{t}}{K}_{\mathrm{a}}$ [36, 38]. However, the calculated value of $k_{-1}$ is substantially less than the turnover number in the reverse reaction $\left(V_2/{\mathrm{E}}_{\mathrm{t}}\right.$) over the whole $\mathrm{p}\mathrm{H}$ range for H44R/H48Q ADH (Supplementary Data, Tables 2S and 3S). This inconsistency was noted for wild-type horse liver and yeast ADHs and attributed to an isomerization of the enzyme-NAD⁺ complex [39, 48]. For the mechanism in Scheme 1, the forward and reverse rate constants can be defined as $k_1$ and $k_{-1}$ for the binding of NAD⁺ and $k_2$ and $k_{-2}$ for the forward and reverse isomerization of E-NAD⁺ to F-NAD⁺. (See “Minimal kinetic mechanism” in Supplementary Data.) Then the overall rate constant for binding NAD⁺ calculated from $V_1/{\mathrm{E}}_{\mathrm{t}}{K}_{\mathrm{a}}$ is $k_{\text{on}}=k_1{k}_2/\left(k_{-1}+k_2\right)$, and the overall rate constant for dissociation of NAD⁺ calculated from $V_1{K}_{\mathrm{i}\mathrm{a}}/{\mathrm{E}}_{\mathrm{t}}{K}_{\mathrm{a}}$ is $k_{\mathrm{o}\mathrm{f}\mathrm{f}}=k_{-1}{k}_2{k}_{-2}/\left(k_{-1}+k_2\right)\left(k_2+k_{-2}\right)$, with $K_{\mathrm{i}\mathrm{a}}=k_{-1}{k}_{-2}/k_1\left(k_2+k_{-2}\right)$. Another indicator of the isomerization is that the apparent bimolecular rate constant for binding of NAD⁺ calculated from $V_1/{\mathrm{E}}_{\text{t}}{K}_{\mathrm{a}}$ at pH8 is less than diffusion controlled for wild-type yeast and horse ADHs. (For comparison, binding of NADH is about 10-times faster than binding of NAD⁺ to these ADHs.) Furthermore, the H51Q or H48Q substitutions decrease this overall rate constant by 10–30 fold ([10], Table 2).

The isomerization of E-NAD⁺ with horse liver ADH is coupled to the release of a proton from the initial E-NAD⁺ complex (with a $\mathrm{p}\mathrm{K}$ of 7.3), presumably from the zinc-water and involving His-51 [49, 59–61]. The rate constant for binding of NAD⁺ is modestly $\mathrm{p}\mathrm{H}$ dependent. The H51Q substitution in horse liver ADH alters the $\mathrm{p}\mathrm{H}$ dependence for the rate of binding of NAD⁺ to expose a pK of 8.0 and decreases the rate constant $k_2$ for the forward isomerization that is estimated by transient kinetics [10]. Because $k_2$ is small relative to the other first order rate constants, the expressions for the rate constants become simplified: $k_{\text{on}}\approx k_1{k}_2/k_{-1},K_{\mathrm{i}\mathrm{a}}\approx k_{-1}/k_1$, and $k_{\text{off}}\approx k_2$. When $K_{\text{ia}}$ is $\mathrm{p}\mathrm{H}$ independent, the $\mathrm{p}\mathrm{H}$ dependence for $k_{\text{on}}$, with a $\mathrm{p}\mathrm{K}$ value of 8, can be assigned to the ionization of the zinc-water associated with the isomerization $\left(k_2\right)$ of the E-NAD⁺ complex.

For wild-type yeast ADH, the conformational change appears to be coupled to the change in coordination of the catalytic zinc (see Scheme 4 in Ref. [7]). The mechanism for the change of zinc coordination might involve direct attack of water to form a bipyramidal-pentacoordinate intermediate and displacement of Glu-67 (see Fig. 4, in Ref. [62]), which is facilitated by interaction with Thr-45 and His-48 in the proton relay system [2, 62–64]. Binding of water to the initial E-NAD⁺ complex and ionization of the water might control the isomerization to F-NAD⁺ by stabilizing the interaction with the positively-charged nicotinamide ring in the closed conformation [3, 4]. Nevertheless, the apparent rate constant for binding of NAD⁺ to wild-type yeast ADH is almost pH independent, or weakly bell-shaped (Table 4 [6][7]), which might be because a protonation of His-44 and His-48 favors binding of the coenzyme, and hydroxide bound to zinc favors binding of the positively-charged nicotinamide ring. Raising or lowering the pH would disrupt the favorable zwitterionic interactions. The H44R/H48Q substitutions seem to expose the underlying pH dependence of the water bound to the catalytic zinc.

The equilibrium binding constant of NAD⁺ to yeast H44R/H48QADH($1/K_{\text{ia}}$) is $\mathrm{p}\mathrm{H}$ independent, and it follows that the rate of dissociation ($k_{\text{off}}=k_{\text{on}}{K}_{\text{ia}}$) must also be more rapid as the $\mathrm{p}\mathrm{H}$ is increased. If the binding of NAD⁺ is controlled by a slow isomerization, $k_{\text{off}}\approx k_2$, and the $\mathrm{p}\mathrm{H}$ dependence would be the same as for $k_{\text{on.}}$. As described above, the $\mathrm{p}\mathrm{H}$ dependence could be related to the change of coordination of the zinc and the ionization of the zinc-water. For wild-type yeast $\mathrm{A}\mathrm{D}\mathrm{H},K_{\mathrm{i}\mathrm{a}}$ increases modestly above a $\mathrm{p}\mathrm{K}$ of 8.1, which could be related to deprotonation of the histidine residues because the H44R and H48Q substitutions modestly increase affinities for NAD⁺ above pH 9 [6, 9]. The $\mathrm{p}\mathrm{H}$ dependence for $K_{\text{ia}}$ was not determined for horse liver H51Q ADH.

After the yeast H44R/H48Q and horse H51Q ADHs bind NAD⁺, the $\mathrm{p}\mathrm{H}$ dependence for oxidation of ethanol ($V_1/{\mathrm{E}}_{\mathrm{t}}{K}_{\mathrm{b}}$) is described by a $\mathrm{p}\mathrm{K}$ value of 8.5 (or 8.4). This kinetic parameter includes the binding of alcohol to form alkoxide, hydrogen transfer, release of the aldehyde (F-NAD⁺ to F-NADH), and isomerizations of enzyme complexes. The pH dependence can be explained by ionization of the zinc-water in F-NAD⁺ or of the zinc-alcohol in F-NAD⁺-Alc to form the alkoxide (F-NAD⁺-Alk), which is energetically favored to transfer hydride to NAD⁺. With the available data, the pK value cannot be assigned to a particular enzyme form because the pH dependence for the isomerization of the enzyme-NAD⁺ complex is included in the kinetic expression for $V_1/E_{\text{t}}{K}_{\text{b}}$. Transient kinetic studies with wild-type yeast ADH show that binding of NAD⁺ and 2,2,2-trifluroethanol releases 0.5 proton per active site at pH 7.6, and likewise horse liver ADH releases 1.0 proton at pH 8, suggesting that alkoxide is formed [43, 60]. Transient oxidation of alcohols with horse ADH is associated with release of a proton [61]. The binding of alcohol to the horse liver ADH-NAD⁺ complex might occur by the double displacement exchange mechanism, as suggested by structural studies with the wild-type yeast and horse ADHs [3, 4, 55], human ADH3 [64, 65][66][67], and computational studies [63].

The conventional pH dependence observed for $V_1/E_{\text{t}}{K}_{\text{b}}$ (slope of 1 below the pK, Fig. 3A) is somewhat unexpected because previous studies with yeast enzymes with the H48S or H48Q substitutions provided linear pH dependencies with slopes of 0.4–0.5, which were interpreted as being due to specific hydroxide catalysis when His-48 is not available to act as a base [7][9]. Inspection of a model of the yeast enzyme with the H48S or H48Q substitutions suggests that His-44 might allow a solvated hydroxide to attack the 2’-hydroxyl group of the nicotinamide residue, whereas Arg-44 might alter access of hydroxide. Horse H51Q ADH shows a typical pH dependence with a pK of 8.4 [11].

The turnover numbers for the forward reaction ($V_1/E_{\text{t}}$) catalyzed by the wild-type yeast and horse ADHs have relatively flat pH dependencies, with modulation by a group with a pK of 6.5 or 7.0, and mostly controlled by the rate constant for dissociation of the enzyme-NADH complex [6, 43, 48, 68]. In contrast, yeast H44R/H48Q and horse H51Q ADHs show a larger effect of pH, and the data can be fit to a function with two poorly determined pK values of ~7.1 and ~9.2, or less-well to straight lines. Previous studies with the yeast H48S and H48Q ADHs showed that pH dependencies over a range of ~ 4 pH units for log $V_1/E_{\text{t}}$ best fit straight lines with slopes of 0.32 and 0.36, respectively, and for these enzymes the isotope effects for oxidation of ethanol suggest that hydride transfer is partly rate-limiting for turnover [7, 9]. For H44R/H48Q yeast and H51Q horse ADHs, the pH dependencies for $V_1/E_{\text{t}}$ can result from kinetic complexities because multiple unimolecular steps control the reaction and shift pK values for ionizations of various forms of enzyme.

4.3.3. Interpretation of the pH dependencies for acetaldehyde reduction

The pH dependencies for the several kinetic constants for the reaction with NADH and acetaldehyde catalyzed by H44R/H48Q ADH are similar to one another (Fig. 3B and Table 4) and to the pH dependencies for the reactions of horse H51Q enzyme (Supplementary Data Table 4S). The pK values are relatively high, and it is possible that Lys-206 in yeast ADH or Lys-228 in horse ADH, as well as the ligand, water or alcohol, to the catalytic zinc are involved. Binding of NADH to wild-type horse liver ADH is slower at pH values above a pK of 9.2, which was assigned to the zinc-water, although pH dependencies in the pH range from 10–12 could involve Lys-228 [68, 69]. However, with the horse wild-type, H51Q, K228R, and H51Q/K228R ADHs, the rate constants for binding decrease above a pK of ~10, suggesting that Lys-228 is not responsible and that the zinc-water can be involved [10].

The rate constant for binding of NADH to H44R/H48Q yeast ADH, approximated by $V_2/{\mathrm{E}}_{\mathrm{t}}{k}_{\text{q}}$, is fastest at low pH, decreasing above a pK value of 8.8, which could be because ionization of the zinc-water during the isomerization of the enzyme-NADH complexes, i.e., the change in protein conformation and zinc coordination to form F-NADH, is not favorable. The enzyme-NADH complex should be more stable with a neutral water bound to the catalytic zinc near the neutral dihydronicotinamide ring. In contrast, the binding of NAD⁺ could be stabilized electrostatically by zinc-hydroxide, with a pK value of 8.0, at high pH. The pK values for coenzyme binding can differ because the local environments in the initial and isomerized complexes with the coenzyme are different. Binding of NADH to the wild-type horse ADH is most rapid at pH below a pK of 9.2, which is assigned to the zinc-water in the apoenzyme, and the pK shifts to 11.2 in the enzyme-NADH complex, apparently because of electrostatic effects [68, 70–72]. Binding of NADH to horse H51Q ADH has a similar pH dependence, with a pK 9.5, and thus it appears that the state of ionization of His-51 does not control the reaction [10]. Although binding of NADH is accompanied by the conformational change, the isomerization has been too fast to measure [55, 61].

The equilibrium binding constant for NADH (1/K_iq) for H44R/H48Q ADH shows a modest pH dependence that is not well-defined because of the relatively high errors on some of the determinations. The rate constant for dissociation, calculated from V₂K_iq/E_tK_q, is essentially pH independent, with a value of 120 s⁻¹, which is larger than the turnover number in the forward reaction. Structural studies with wild-type yeast ADH show that binding of NADH is associated with a conformational change, but kinetic evidence for an isomerization is not available [4].

The reaction of acetaldehyde with the enzyme-NADH complex, $V_2/{\mathrm{E}}_{\mathrm{t}}{K}_{\text{p}}$, is faster at low pH below a pK of 9.5, which might be related to the protonation of the alkoxide formed by hydride transfer before the alcohol dissociates or to the pH dependence for binding of NADH with isomerization to F-NADH. The maximum magnitude of $V_2/{\mathrm{E}}_{\mathrm{t}}{K}_{\text{p}}$ is 100-fold smaller and the pK value is shifted up by 2 units for H44R/H48Q ADH as compared to wild-type ADH, suggesting that His-48 is very important for this reaction. The forward and reverse reactions have opposite pH dependencies for $V/E_{\text{t}}{K}_{\text{m}}$, but the pK values are different because the reacting forms of enzyme and the commitments to catalysis (“stickiness”) are different [35, 73].

The turnover number in the reverse direction, $V_2/{\mathrm{E}}_{\mathrm{t}}$ also has a modest pH dependence that is controlled by some group with a pK of 9.6. Because this kinetic parameter is controlled by unimolecular steps after formation of the ternary complex with NADH and acetaldehyde, as well as by isomerizations, it is problematic to assign the pK to a particular step.

5. Conclusions

The substitutions of the histidine residues in the active sites of the yeast and liver ADHs expose simpler pH dependencies for the enzymatic reaction. The ionization of water or alcohol bound to the catalytic zinc appears to determine the pH dependencies for several steps in the forward and reverse reactions for both yeast H44R/H48Q and liver H51Q ADHs. The previous studies with the H44R, H48S, H48E, and H48Q substitutions in yeast ADH and H51Q in horse liver ADH showed the pH dependencies were altered relative to the wild-type ADHs and that the histidines are involved in catalytic reactions [6, 7, 9–11]. However, assignment of pK values to particular groups and evaluations of their contributions to catalysis are difficult. The studies suggest that histidine residues modulate activity, but the substituted amino acid residues themselves can make different contributions to catalysis, and further studies are required to quantify them. For instance, the amido NH₂ group of Gln-48 in yeast ADH probably interacts with the 2’ and 3’ hydroxyl groups of the nicotinamide ribose and potentially hinders proton transfer by the hydrogen bond system that includes the hydroxyl group of Thr-45 and the hydroxyl group of water or alcohol ligated to the catalytic zinc.

Further study is also required to understand the mechanism for substrates (alcohol or aldehyde) to bind to the catalytic zinc because the wild-type yeast apoenzyme has an alternative coordination of the catalytic zinc (with Glu-67), and the isomerization (conformational change) of enzyme-coenzyme complexes is linked to the change in coordination. In contrast, horse liver ADH apoenzyme has the classical coordination of the catalytic zinc, and the exchange of ligands is still required to form the ternary complexes with substrates. It is reasonable to suggest that the carboxylate group of Glu-67 in yeast ADH or Glu-68 in horse ADH participates in the exchange of water and substrates accompanied by changes in the zinc coordination [4, 63, 65].

Substitutions of protein residues close to the catalytic zinc affect catalytic efficiencies and pH dependencies. The E67Q substitution in yeast ADH decreased catalytic efficiencies by ~100-fold and produced a pH dependence with a pK of 6.9 and some evidence for participation by hydroxide at higher pH for ethanol oxidation [18]. Substitution of Asp-46, which is buried and interacts with His-66, which is ligated to the zinc, with asparagine decreased catalytic efficiencies by ~1000-fold and showed a pK of 6.3 for ethanol oxidation. Even though replacing a negatively-charged oxygen in a carboxylate with –NH₂ might not affect the overall structure, local ionic and structural changes could alter the electrostatic environment and affect the isomerization of enzyme-substrate complexes. The substitution of Cys-153, a ligand to the catalytic zinc, with aspartic acid (found in some ADHs) decreased catalytic efficiencies by ~10-fold and somewhat altered the pH dependencies [74]. In this case, model building suggested that the C153D substitution could be accommodated in the apoenzyme and holoenzyme forms and modestly affect the electrostatics of the zinc.

The proton relay system apparently modulates the pH dependencies due to the zinc-ligand. His-48 (or 51) is linked to the zinc water (or alcohol) so that the microscopic pKs of the histidine or the zinc-ligand cannot be established easily. (Two macroscopic pKs are expected, and ionization of one group will affect the microscopic pK values of the other group [75].) Nevertheless, the shifts in the pH dependencies by the substitutions with glutamine or serine demonstrate the connection in the proton relay system [7, 9, 11]. Disrupting the proton relay system with the T48A, T48C or T48G substitutions decreases catalytic activity by more than 5000-fold, and the pH dependencies of T48G ADH are significantly altered [62]. Thr-45 cannot contribute directly to acid/base catalysis, but it connects the zinc-ligand to the 2’-hydroxyl group of the coenzyme and participates in efficient proton transfer.

Although the pH dependencies for the H44R/H48Q and H51Q ADHs can be explained with the relatively simple mechanism involving the catalytic zinc, some distal amino acids might participate. Because diethyl pyrocarbonate can inactivate the H44R/H48Q ADH, other histidines might be involved in catalysis. Further research is required to determine the mechanisms for the changes in metal coordination and protein conformation and to estimate rate constants for individual steps in the general mechanism. It is not clear why apoenzyme forms of many tetrameric ADHs have the inverted zinc coordination whereas the holoenzymes have the classical coordination [2, 4]. The contribution of the zinc to catalysis, as compared to other metals, has been studied in the horse liver enzyme, and further studies are needed [76–80].

H44R/H48Q substitutions partially protect against inactivation by diethyl pyrocarbonate.

H44R/H48Q substitutions modestly increase affinity for coenzymes at pH 7.3.

H44R/H48Q substitutions decrease catalytic efficiencies 20 to 50-fold.

H44R/H48Q substitutions expose simpler pH dependencies for catalytic reactions.

Ionization of the catalytic zinc-water or alcohol controls steps in the mechanism.

Abbreviations:

ADH

alcohol dehydrogenase

H44R

substitution of His-44 with arginine

H48Q

substitution of His-48 with glutamine

H44R/H48Q

substitutions of both histidines in yeast ADH

H51Q

substitution of His-51 with glutamine in horse liver ADH1E