Specific base catalysis by yeast alcohol dehydrogenase I with substitutions of histidine-48 by glutamate or serine residues in the proton relay system

Abstract

His-48 in yeast alcohol dehydrogenase I (His 51 in horse liver alcohol dehydrogenase) is a highly conserved residue in the active sites of many alcohol dehydrogenases. The imidazole group of His-48 may participate in base catalysis of proton transfer as it is linked by hydrogen bonds through the 2'-hydroxyl group of the nicotinamide ribose and the hydroxyl group of Thr-45 to the hydroxyl group of the alcohol bound to the catalytic zinc. In this study, His-48 was substituted with a glutamic acid residue to determine if a carboxylate could replace imidazole or to a serine residue to determine if the exposure of the 2'-hydroxyl group of the ribose to solvent would allow proton transfer to water without base catalysis. At pH 7.3, the H48E substitution increases affinity for NAD⁺ and NADH 17- or 2.6-fold, but decreases catalytic efficiency ($V∕K_{\mathrm{m}}$) on ethanol by 70-fold and on acetaldehyde by 6-fold relative to wild-type enzyme. The H48S substitution increases affinity for coenzymes by 2-fold and decreases ($V∕K_{\mathrm{m}}$) on ethanol and acetaldehyde only by ~3-fold. The substituted enzymes show substrate deuterium isotope (H/D) effects of 3 to 4 for turnover number ($V_1$) and catalytic efficiency ($V_1∕K_{\mathrm{b}}$) for ethanol oxidation, indicating that hydrogen transfer is partially rate-limiting and suggesting a somewhat more random mechanism for binding of ethanol and NAD⁺. For reduction of acetaldehyde, the deuterium isotope effects are small, and the kinetic mechanism appears to be ordered for binding of NADH first and acetaldehyde next. The pH dependencies for H48E and H48S ADHs can be described by a mechanism with pK values of about 6-7 and 9. However, the pH dependencies for oxidation of ethanol and butanol by the H48S enzyme are also simply described by a straight line, with slopes of log $V_1∕K_{\mathrm{b}}$ against pH of 0.37 or 0.43, respectively. The linear dependence apparently represents catalysis by hydroxide that has a low activity coefficient due to the protein environment, or to a kinetically complex proton transfer. The effects of the substitutions of His-48 show that this residue contributes to catalysis, although many dehydrogenases also have other residues.

1. Introduction

Alcohol dehydrogenases (EC 1.1.1.1) oxidize alcohols by catalyzing proton transfer of the proton of the hydroxyl group to solvent to form an alkoxide and hydride transfer from a carbon of an alcohol to the nicotinamide ring of NAD⁺ [1]. The X-ray structure of horse liver ADH complexed with NAD⁺ and an alcohol showed a hydrogen-bonded system connecting the oxygen of the alcohol bound to the zinc, the hydroxyl group of Ser-48, the 2’-hydroxyl group of nicotinamide ribose, and the imidazole group of His-51, as shown in Scheme 1 [2]. Studies of the pH dependence of catalysis and chemical modification suggested that the histidine residue participates in catalysis [3-7]. Substitution of His-51 with a glutamine residue in human ADH1B and horse liver ADH1E modestly decreases catalytic efficiency ($V_1$/$K_{\mathrm{m}}$) for oxidation of ethanol and alters pH dependencies [8-10].

Scheme 1.

Proton Relay System in Horse Liver ADH

Structures for yeast ADH1 (the constitutive enzyme involved in fermentation of glucose) also show the proton relay system in ternary complexes, where Thr-45 and His-48 are the homologous residues as shown in Fig. 1 [11-13].

Fig. 1.

Stereo view of the complex with NAD⁺ and trifluoroethanol (TFE) determined by X-ray crystallography at 2.4 Å (4w6z.pdb, subunit A).

Oxidation of alcohols by yeast alcohol dehydrogenase is maximal above a pK of about 7.6 [14-19]. The pK could be due to the ionization of the alcohol bound to the catalytic zinc [20, 21] or to His-48 (His-51 in horse liver ADH). Diethyl pyrocarbonate inactivates both yeast and liver alcohol dehydrogenases, supporting a role for histidine in catalysis [6, 15, 22].

Although sequence alignments show that the histidine residue is common in many ADHs, tyrosine is found at the homologous position in plant ADHs, ADH3 (glutathione-dependent formaldehyde dehydrogenases) and cod liver ADH [23-25]. Threonine is also found in human ADH2 [26]. Some medium chain glucose dehydrogenases have a glutamate [27-29]. Therefore, histidine is not absolutely essential for activity, but contributions to catalysis of other amino acids at this position should be evaluated.

Preliminary conference proceedings reported that the H48Q substitution in yeast ADH decreases catalytic efficiency with ethanol by a factor of about 10-fold at pH 7.3 and changes the pH dependence [30]. The H48E substitution decreases catalytic efficiency by about 65-fold and also alters the pH dependence [31]. Now that structures of apo and holoenzyme complexes of yeast ADH1 have been determined, there is a better basis for explaining the structural effects of substitutions of His-48.

For the present study, we used molecular modeling to show that the glutamate residue could replace His-48 and form a good hydrogen bonded system. Glu-48 could act as proton acceptor and either maintain acid/base catalysis or electrostatically block transfer and have differential effects on alcohol oxidation or aldehyde reduction. We also studied the H48S substitution, which should open up the proton relay system, possibly allowing a water to participate, but knocking out acid/base catalysis by His-48. Fig. 2 shows the models. We used site-directed mutagenesis to change the coding DNA for the enzyme and detailed steady state kinetics to study the catalytic mechanism.

Fig. 2.

Structural models of the proton relay system in yeast ADH with substitutions of His-48. The models are based on the structure of the complex with NAD⁺ and trifluoroethanol (TFE) determined by X-ray crystallography at 2.4 Å (4w6z.pdb, subunit A). The top model shows Glu-48 interacting with the nicotinamide ribose, where Asp-53 is rotated slightly to avoid close contacts with Glu-48. The bottom model shows Ser-48, which does not interact with the coenzyme, but a water molecule could fit between the hydroxyl groups.

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.; T4-polynucleotide kinase and all restriction enzymes were from New England Biolabs. Deoxynucleotides, DEAE-Sepharose CL-6B, and octyl-Sepharose CL-4B came from Pharmacia P-L Biochemicals; Amersham supplied radioactive nucleotides; Fluka Chemical Corp provided ethanol-d₆.The oligonucleotide primers were synthesized locally with a Beckman DNA synthesizer. (4R)[4-²H]Nicotinamide adenine dinucleotide was prepared as described previously [32]. Ethanol was redistilled, and acetaldehyde was redistilled on the day of use.

2.2. Site-directed mutagenesis

The single stranded template in M13mp18RF phage including the gene for ADH1 (adc1, YOL086c, UniProtKB entry P00330) in the yeast shuttle vector YEp13 [33, 34] was mutated with the oligonucleotide primer CTGACTTGGAAGCTTGGCAC for the H48E substitution or with the primer CTGACTTGTCTGCTTGGCAC for the H48S substitution (underline marks the site of mutation) by standard methods [32, 35, 36]. The mutations were confirmed by sequencing in the singe-stranded M13 phage [37]. The mutation for H48E created a new Hind III restriction site. The fragment generated by Sph I restriction was subcloned into the yeast shuttle vector YEp13 and transfected into E. coli XL1-Blue cells for production of the plasmid, which was then transfected into an ADH-negative strain of S. cerevisiae (adh1-11 adr2 leu2 trp1) with selection for growth without leucine and production of the substituted ADH [38-40].

2.3. Enzyme purification

Yeast were grown to full density (OD₆₀₀ = 12) at 30 °C in 6 1.5-L flasks of YPD medium (1% yeast extract, 2% peptone, 2% glucose), with vigorous shaking so that the yeast did not require fermentation to grow. The enzyme was purified by precipitation with protamine sulfate and PEG4000 and chromatography on DEAE-Sepharose CL-6B and Octyl-Sepharose CL-4B [19, 40]. The final enzyme preparations appeared to be free of other proteins by denaturing SDS and high pH nondenaturing polyacrylamide gel electrophoresis [41]. The concentration of purified protein was estimated using A₂₈₀ = 1.26 cm⁻¹ (mg/ml)⁻¹. The concentration of active sites was determined by spectrophotometric titration with NAD⁺ in the presence of 10 mM pyrazole [42]. The turnover numbers ($V∕E_{\mathrm{t}}$, s⁻¹) were calculated from the enzyme subunit concentrations and the activity in a standard enzyme assay [43].

2.4. Steady-state kinetics

A buffer designed to mimic physiological conditions, of 83 mM potassium phosphate, 40 mM KCl and 0.25 mM EDTA, pH 7.3, was used at 30 °C for most kinetics experiments [44]. Enzyme activity was determined by measuring the change in absorbance at 340 nm with a spectrophotometer and a FORTRAN program to calculate initial velocities with a linear or parabolic fit of the data. Some data were fitted to the Michaelis-Menten equation with the HYPER program [45]. The kinetic constants for the sequential bi mechanism were determined by initial velocity studies with concentrations of both substrates varied over a 9-fold range in a systematic manner (5 x 5 matrix) and fitting the data to Eq. 1 with the SEQUEN program, where concentrations of reactants (A = NAD⁺, B = alcohol) are given in brackets, and the kinetic constants for the respective substrates have the corresponding subscripts. For the reverse reaction, P = aldehyde is substituted for B, and Q = NADH is substituted for A [46]. Product and dead-end inhibition data were fitted with the general equation for noncompetitive inhibition (NONCOMP) where the slope and intercept inhibition constants were evaluated (Eq. 2) or the equations for competitive (COMP) or uncompetitive (UNCOMP) inhibition where only $K_{\text{is}}$ or $K_{\text{ii}}$ were evaluated [45]. Standard errors for kinetic constants were usually <10–20 % of the values, indicating good estimates.

$$\mathrm{v}=V_1[\mathrm{A}][\mathrm{B}]∕(K_{\text{ia}}{K}_{\mathrm{b}}+K_{\mathrm{a}}[\mathrm{B}]+K_{\mathrm{b}}[\mathrm{A}]+1)$$ Eq. 1:

$$\mathrm{v}=V[\mathrm{S}]∕((K_{\mathrm{m}}(1+[\mathrm{I}]∕K_{\text{is}})+[\mathrm{S}](1+[\mathrm{I}]∕K_{\text{ii}}))$$ Eq. 2:

The pH dependencies of kinetic constants were determined at 30 °C by initial velocity studies. The pH dependence was usually studied in 10 mM Na₄P₂O₇ buffers adjusted to the desired pH and a final ionic strength of 0.1 with appropriate concentrations of NaH₂PO₄/Na₂HPO₄ for the pH range from 5.5 to 9.0 or with the appropriate ratio of 5 mM NaHCO₃/Na₂CO₃ for the pH range 9.5 to 10.5, at 30 °C. (See Supplementary Data for the preparation.) Other buffers are described in the Results. In general, the coenzymes were dissolved in water, to avoid degradation by acidic or basic conditions, and diluted into 2X buffer at the desired pH just before the reaction was started. For assays, the enzymes were diluted into a neutral buffer containing 1 mg/ml of bovine serum albumin and 0.25 mM EDTA and 1 mM 2-mercaptoethanol. The pH profiles can be described by various models (Scheme 2). Some pH profiles are described by a single ionizing group, such as WAVL [45] given in Eq. 3. A more general model includes two ionizations with rate constants for two of the three states of protonated enzyme species, Eq. 4 and 5 [47]. The pK values for the pH dependencies were determined by fitting the kinetic constants with a nonlinear least-squares program (NONLIN, C. M. Metzler, The Upjohn Co.) to the equations derived for various mechanisms.

Scheme 2.

Models and equations for describing pH dependencies

3. Results

3.1. Protein properties

The proteins were prepared in good yield. From 140 g of wet cells, 45 mg of H48S enzyme was obtained. From 120 g of cells, 44 mg of purified H48E enzyme was obtained. The enzymes appeared to be homogenous on polyacrylamide electrophoresis. The H48S enzyme migrated slightly faster and the H48E enzyme clearly faster than the wild-type enzyme in nondenaturing discontinuous electrophoresis at pH 9.5. (For reference, see Fig. 2 in Ref. [48] for ADHs with substitutions that add one basic residue or two acidic residues.) Titration of active sites with NAD⁺ in the presence of pyrazole showed the typical difference absorption peaks at 284 and 292 nm, with extinction coefficients at 292 nm of 12 mM⁻¹cm⁻¹ for the H48S enzyme and 10 mM⁻¹cm⁻¹ for the H48E enzyme. As compared to the protein concentration determined from A₂₈₀ (1.26 A/cm per mg/ml), about 70 % of the protein subunits had functional binding sites [40]. The A₂₈₀/A₂₆₀ ratio is 1.8, typical for a clean protein.

3.2. Enzyme kinetic mechanism

Kinetic constants were determined by initial velocity studies in which the concentrations of both coenzyme and substrate were varied in a systematic way for both forward and reverse reactions. (Table 1). All of the enzymes have the general sequential bi bi mechanism where both substrates bind before the redox chemistry and release of products. Product inhibition studies were used to determine some of the constants and provide evidence about the kinetic mechanism (Table 2). Coenzymes were mutually competitive inhibitors, and the $K_{\mathrm{i}}$ values agreed with those determined by the initial velocity studies. Ethanol inhibited noncompetitively against varied concentrations of acetaldehyde with the substituted enzymes, providing $K_{\text{ib}}$ values, but acetaldehyde appears to a competitive inhibitor against varied concentrations of ethanol for the H48Q, H48E, and H48S ADHs. (H48Q ADH is not a focus of this study, but the kinetic constants are included for comparison.) The agreement between the $K_{\text{eq}}$ calculated from the Haldane relationship and the directly determined value of 10 pM shows that the kinetic constants are internally consistent.

Table 1.

Kinetic constants for yeast alcohol dehydrogenase I with substitutions of His-48^a

Constant	Wild-typeb	H48Qc	H48Ed	H48Se
$K_{\mathrm{a}}$ (μM), NAD+	160	96	8.5	140
$K_{\mathrm{b}}$ (mM), ethanol	21	18	8.0	19
$K_{\mathrm{p}}$ (mM), acetaldehyde	0.74	15	0.52	3.7
$K_{\mathrm{q}}$ (μM), NADH	94	150	25	110
$K_{\text{ia}}$ (μM), NAD+	950	460	56	510
$K_{\text{ib}}$ (mM)	120	100	50	100
$K_{\text{ip}}$ (mM)	1.5	f	g	f
$K_{\text{iq}}$ (μM), NADH	31	6.0	12	11
$K_{\text{ia}}{K}_{\mathrm{b}}/K_{\mathrm{a}}$ (mM)	96	27	76	58
$K_{\text{iq}}{K}_{\mathrm{p}}/K_{\mathrm{q}}$ (mM)	0.21	0.55	g	0.43
$V_1/{\mathrm{E}}_{\mathrm{t}}$ (s−1)	360	27	2.0	120
$V_2/{\mathrm{E}}_{\mathrm{t}}$ (s−1)	1800	2800	220	2700
$V_1/{\mathrm{E}}_{\mathrm{t}}{K}_{\mathrm{b}}$ (mM−1 s−1)	17	1.5	0.25	6.3
$V_2/{\mathrm{E}}_{\mathrm{t}}{K}_{\mathrm{p}}$ (mM−1 s−1)	2400	190	390	730
$V_1/{\mathrm{E}}_{\mathrm{t}}{K}_{\mathrm{b}}{K}_{\text{ia}}$ (mM−2 s−1)	18	3.3	4.5	12
$V_2/{\mathrm{E}}_{\mathrm{t}}{K}_{\mathrm{b}}{K}_{\text{iq}}$ (mM−2 s−1)	77000	32000	32000	66000
$K_{\text{eq}}$ (pM)	12	5.3	6.4	9.3
Activity ($V/{\mathrm{E}}_{\mathrm{t}}$, s−1)	400	100	9.2	360
$K_{\mathrm{i}}$ CF3CH2OH (mM)h	2.5	33	130	26

^a$K_{\mathrm{a}}$, $K_{\mathrm{b}}$, $K_{\mathrm{p}}$, and $K_{\mathrm{q}}$ are the Michaelis constants for NAD⁺, ethanol, NADH and acetaldehyde respectively. $K_{\mathrm{i}}$ values are the dissociation (inhibition) constants. $V_1/{\mathrm{E}}_{\mathrm{t}}$ is the turnover number for ethanol oxidation and $V_2/{\mathrm{E}}_{\mathrm{t}}$ the turnover number for acetaldehyde reduction, in 83 mM potassium phosphate buffer 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. $V$ is the turnover number in standard assay [43] at 30 °C, based on titration of the active sites. The equilibrium constant is calculated from the Haldane equation, $K_{\text{eq}}=V_1{K}_{\mathrm{p}}{K}_{\text{iq}}$[H⁺]/$V_2{K}_{\mathrm{b}}{K}_{\text{ia}}$, where [H⁺] = 5 x 10⁻⁸ M at pH 7.3. The experimental value is 9.8 pM [49].

^bFrom Ref. [19].

^cFrom Ref. [50].

^dData from initial velocity studies where the concentrations ranged from 9 to 78 μM NAD⁺ and 3.3 to 30 mM ethanol in the forward reaction, and from 3.4 to 60 μM NADH and 0.15 to 2.5 mM acetaldehyde for the reverse reaction. Inhibition constants for the coenzymes were also determined as described previously [40].

^eIn the forward reaction the concentrations ranged from 55 to 500 μM NAD⁺ and 8.3 to 75 mM ethanol, and for the reverse reaction from 22 to 200 μM NADH and 1.7 to 15 mM acetaldehyde.

^fNot determined because product inhibition by acetaldehyde against varied concentrations of ethanol appeared to be competitive, but the slope inhibition constant, $K_{\text{iq}}{K}_{\mathrm{p}}/K_{\mathrm{q}}$, was determined.

^gNot determined.

^hDetermined by competitive inhibition ($K_{\text{is}}$) against varied concentrations of ethanol at a fixed, saturating concentration of NAD⁺.

Table 2.

Product and dead-end inhibition kinetics for H48E and H48S ADHs^a

Name	Varied substrate, mM	Fixed substrate. mM	Inhibitor mM	$K_{\mathrm{m}}$ mM	$K_{\text{is}}$ mM	$K_{\text{ii}}$ mM	$V/{\mathrm{E}}_{\mathrm{t}}$ s−1	Fit
H48E
$K_{\mathrm{a}}{K}_{\text{iq}}$	NAD+ 0.014–0.20	CH3CH2OH 5	NADH 0–0.03	0.022 ± 0.004	0.010 ± 0.002	NDb	0.74 ± 0.03	C
$K_{\mathrm{q}}{K}_{\text{ia}}$	NADH 0.017–0.12	CH3CHO 10	NAD+ 0–0.5	0.013 ± 0.001	0.076 ± 0.009	ND	184 ± 4	C
$K_{\mathrm{p}}{K}_{\text{ib}}$	CH3CHO 0.14–1.0	NADH 0.1	CH3CH2OH 0–40	0.54 ± 0.05	76 ± 26	62 ± 10	110 ± 4	NC
$K_{\mathrm{b}}{K}_{\text{TFE}}$	CH3CH2OH 7–50	NAD+ 1.0	CF3CH2OH 0–400	6.1 ± 0.4	130 ± 10	ND	1.6 ± 0.03	C
$K_{\mathrm{a}}{K}_{\text{TFE}}$	NAD+ 0.014–0.10	CH3CH2OH 5	CF3CH2OH 0–100	0.028 ± 0.003	89 ± 15	ND	0.46 ± 0.02	C
H48S
$K_{\mathrm{a}}{K}_{\text{iq}}$	NAD+ 0.07–0.49	CH3CH2OH 10	NADH 0–0.03	0.37 ± 0.03	0.011 ± 0.001	NDb	46 ±2	C
$K_{\mathrm{q}}{K}_{\text{ia}}$	NADH 0.023–0.16	CH3CHO 5.0	NAD+ 0–3.4	0.044 ± 0.009	0.52 ± 0.01	ND	160 ± 90	C
$K_{\mathrm{b}}{K}_{\text{ip}}$	CH3CH2OH 14–100	NAD+ 2.0	CH3CHO 0–5	24. ± 1	0.43 ± 0.02	ND	110 ± 3	C
$K_{\mathrm{p}}{K}_{\text{ib}}$	CH3CHO 1.2–14	NADH 0.2	CH3CH2OH 0–120	2.5 ± 0.1	58 ± 6	155 ± 14	1720 ± 40	NC
$K_{\mathrm{b}}{K}_{\text{iq}}1$	CH3CH2OH 14–100	NAD+ 2.0	NADH 0–0.02	28 ± 1	0.07 ± 0.01	0.26 ± 0.06	140 ± 3	NC
$K_{\mathrm{a}}{K}_{\text{iq}}2$	CH3CH2OH 14–100	NAD+ 0.12	NADH 0–0.03	34 ± 1	0.013 ± 0.001	0.064 ± 0.008	41 ± 1	NC
$K_{\mathrm{a}}{K}_{\text{ip}}$	NAD+ 0.084–0.60	CH3CH2OH 20	CH3CHO 0–0.25	0.31 ± 0.02	0.26 0.02	ND	56 ± 2	C
$K_{\mathrm{q}}{K}_{\text{ib}}$	NADH 0.024–0.16	CH3CHO 5	CH3CH2OH 0–120	0.073 ± 0.003	ND	63 ± 3	1710 ± 40	UC
$K_{\mathrm{b}}{K}_{\text{TFE}}$	CH3CH2OH 14–100	NAD+ 2.0	CF3CH2OH 0–96	24 ± 1	26 ± 1	ND	110 ± 5	C
$K_{\mathrm{a}}{K}_{\text{TFE}}$	NAD+ 0.071–0.50	CH3CH2OH 20	CF3CH2OH 0–100	0.26 ± 0.01	100 ± 20	47 ± 4	51 ± 1	NC

^aThe “name” puts the $K_{\mathrm{m}}$ for the varied substrate first, followed by the $K_{\mathrm{i}}$ for the varied inhibitor. Letters a, b, p, and q represent the substrates. Concentrations and $K$ values are in mM units. The buffer was pH 7.3, at 30 °C. The kinetic constants are not corrected for the concentration of the fixed substrate. The best fit for the pattern of inhibition are C, competitive; NC, noncompetitive; UC, uncompetitive.

^bND, the $K_{\mathrm{i}}$ value had high error and was not determined well.

The H48E substitution significantly changed several kinetic constants as compared to the wild-type ADH1 (Table 1). The $K_{\mathrm{m}}$ and $K_{\mathrm{i}}$ values for coenzymes and substrates are decreased by factors of 2 to 20, and the turnover numbers for ethanol oxidation ($V_1$/${\mathrm{E}}_{\mathrm{t}}$) and acetaldehyde reduction ($V_2$/${\mathrm{E}}_{\mathrm{t}}$) are decreased 180-fold and 9-fold, respectively, compared to those for wild-type enzyme. Tighter binding of coenzymes ($K_{\text{ia}}$ and $K_{\text{iq}}$) may result from better hydrogen bonding between the 3'-hydroxyl group of the coenzyme and an oxygen of the glutamate carboxylate as compared to a histidine imidazole group. Trifluoroethanol binds 50-fold less tightly to the H48E enzyme as compared to wild-type ADH, which might reflect electrostatic repulsion by Glu-48 on the ionization of the zinc-ligated alcohol.

Inhibition studies can provide data to distinguish between ordered and random mechanisms. Wild-type enzyme has a preferred ordered mechanism with coenzyme binding first [40, 51-53]. Critical evidence consistent with an ordered mechanism is the mutual noncompetitive inhibition by ethanol or acetaldehyde against each other at fixed coenzyme concentration as shown in published figures [31, 51]. Furthermore, trifluoroethanol, a dead-end inhibitor, is competitive against ethanol and uncompetitive against varied NAD⁺ concentrations.

Table 2 reports some inhibition studies on the H48E and H48S ADHs. Standard errors of the fits of the data to the appropriate equations are included in order to validate the conclusion about the type of inhibition. For H48E enzyme, ethanol is noncompetitive against acetaldehyde (“$K_{\mathrm{p}}{K}_{\text{ib}}$”), consistent with an ordered addition of NADH before acetaldehyde. However, trifluoroethanol is competitive against varied concentrations of ethanol or NAD⁺, consistent with a random binding of NAD⁺ and ethanol.

Studies on the H48S enzyme were more extensive, with all eight combinations of a product inhibiting against a substrate, and dead-end inhibition by trifluoroethanol against NAD⁺ or ethanol (Table 2). The patterns do not totally fit the predictions for Ordered Bi Bi, Rapid Equilibrium Random Bi Bi, a hybrid Rapid Equilibrium AB Ordered PQ, or those same mechanisms with a dead-end complex, such as EBQ (E-NADH-alcohol). The predictions are described well in a textbook [54].

Some critical results are presented in Fig. 3. Ethanol is a noncompetitive inhibitor against varied concentrations of acetaldehyde (Fig. 3A), consistent with ordered addition of NADH and acetaldehyde and not with a simple rapid equilibrium mechanism, which is expected to show competitive inhibition. Acetaldehyde is a competitive inhibitor against varied concentrations of NAD⁺ (Fig. 3B), consistent with a rapid equilibrium mechanism and not with an ordered mechanism, which should show noncompetitive inhibition. The inconsistencies can indicate more complex mechanisms or rate constants in a mechanism that differentially change the relative slope ($K_{\text{is}}$) or intercept ($K_{\text{ii}}$) inhibition constants. A 5-fold difference can produce patterns that trend from noncompetitive to competitive or uncompetitive. The dead-end inhibitor, trifluoroethanol is competitive against varied ethanol concentrations, but noncompetitive against varied concentrations of NAD⁺ (Fig. 3C), consistent with a random mechanism. For wild-type yeast and liver ADHs, trifluoroethanol is uncompetitive against NAD⁺, strong evidence for an ordered addition of NAD⁺ before binding of ethanol. However, if trifluoroethanol binds to the enzyme-NAD⁺ complex (EAI) and forms a dead-end complex with enzyme-NADH (EQI), the inhibition against NAD+ could be noncompetitive with an Ordered Bi Bi mechanism. It is significant that ethanol inhibits uncompetitively against varied concentrations of NADH (Fig. 3D), consistent with formation of an abortive dead-end enzyme-NADH-ethanol complex (EBQ), which has been observed crystallographically for horse liver alcohol dehydrogenase [55] and proposed by kinetic studies for various alcohol dehydrogenases [51, 56-58]. The steady-state kinetic patterns for both H48E and H48S enzymes suggest that the mechanisms have some random pathways.

Fig. 3.

Kinetic inhibition studies with the H48S enzyme. Conditions are reported in Table 2. Inhibitor concentrations increase from bottom line to top line in each plot. The initial velocities represent the observed $\Delta {\mathrm{A}}_{340}∕\text{min}$. (A) $K_{\mathrm{p}}{K}_{\text{ib}}$. (B) $K_{\mathrm{a}}{K}_{\text{ip}}$. (C) $K_{\mathrm{a}}{K}_{\text{TFE}}$. (D) $K_{\mathrm{q}}{K}_{\text{ib}}$.

3.3. Deuterium isotope effects

Substituting the hydrogen (H with D) that is transferred from ethanol to the nicotinamide ring of NAD⁺ or from NADH to acetaldehyde provides information on rate-limiting steps and the extent of randomness in the mechanism. Large substrate isotope effects are observed on the turnover number and catalytic efficiency for H48E ADH with ethanol ($V_1$ and $V_1∕K_{\mathrm{b}}$), which is consistent with a mechanism in which alcohol oxidation in the ternary complex is at least partly rate-limiting in catalysis and that the mechanism tends toward rapid equilibrium random binding of NAD⁺ and ethanol (Table 3). The isotope effects at pH 7.3 are larger than those observed with the wild-type enzyme (${}^{\mathrm{D}}{V}_1{=}^{\mathrm{D}}{V}_1∕K_{\mathrm{a}}=1.8$ and ${}^{\mathrm{D}}{V}_1∕K_{\mathrm{b}}=3.2$) [32]. The larger isotope effects for H48E ADH indicate a lower commitment to catalysis, probably due to both weaker alcohol binding and slower hydride transfer [59]. In the reverse reaction, the isotope effects are smaller, in particular, ${}^{\mathrm{D}}{V}_2$, suggesting that hydrogen transfer is not a major rate-limiting step and that the commitment to catalysis is higher, which is also reflected in the turnover numbers. The isotope effect on ${}^{\mathrm{D}}V∕K_{\mathrm{q}}$ was 1.0 ± 0.1 (lower than for ${}^{\mathrm{D}}V∕K_{\mathrm{p}}$), which is consistent with an ordered binding of NADH first, followed by acetaldehyde binding. The isotope effects with the H48E enzyme are not significantly dependent on pH.

Table 3.

Deuterium isotope effects for forward and reverse reactions with H48E ADH^a

pH	Buffer	$V_1/{\mathrm{E}}_{\mathrm{t}}$, s−1	${}^{\mathrm{D}}{V}_1$	${}^{\mathrm{D}}V/K_{\mathrm{b}}$	$V_2/{\mathrm{E}}_{\mathrm{t}}$, s−1	${}^{\mathrm{D}}{V}_2$	${}^{\mathrm{D}}V/K_{\mathrm{p}}$
7.3	Pi/KCl	1.8 ± 0.03	3.8 ± 0.1	5.1 ± 0.2	216 ± 7	1.3 ± 0.1	2.5 ± 0.2
6.0	MES	0.24 ± 0.01	4.7 ± 0.3	2.8 ± 0.4	188 ± 8	1.0 ± 0.1	3.0 ± 0.6
8.0	TAPS	1.8 ± 0.05	4.3 ± 0.2	3.8 ± 0.4	122 ± 5	1.1 ± 0.1	2.3 ± 0.3
10.0	DMG	23.8 ± 0.4	3.8 ± 0.1	4.2 ± 0.2	27 ± 1	1.2 ± 0.1	2.1 ± 0.1

^aThe superscript D indicates the ratio of rates with H and D substrates. Initial velocity data were determined in different buffers: 83 mM sodium phosphate, 40 mM KCl, pH 7.3; 0.1 M sodium 2-(N-morpholino)ethanesulfonate, pH 6.0; 0.1 M sodium N-tris(hydroxymethyl)methyl-3-aminopropanesulfonate, pH 8.0; and 0.1 M sodium N,N-dimethylglycinate, pH 10.0; all at 30 °C. For the forward reaction, the concentration of NAD⁺ was fixed at 2 mM (approximately saturating), and concentrations of ethanol or ethanol-d₅ were varied from 3.9 to 100 mM. (Since fully deuterated ethanol was used, the magnitudes of observed isotope effects also include secondary effects.) For the reverse reaction the concentrations of NADH or (4R)[4-²H]NADH were fixed at 0.1 mM, and concentration of acetaldehyde was varied from 0.12 to 12 mM. Errors were propagated from the fits of the data at each saturation experiment to HYPER. At pH 7.3, concentrations of NADH(D) were varied from 3.5 to 60 μM, and acetaldehyde from 0.15 to 2.5 mM, and the data were fitted to the equation for a sequential reaction (SEQUEN). In this experiment, ${}^{\mathrm{D}}V∕K_{\mathrm{q}}$ was 1.0 ± 0.1.

The H/D isotope effect for H48E ADH acting on butanol was determined at pH 7.3, 30 °C with 1 mM NAD⁺ and 6 concentrations of 1-butanol or 1-butanol-d₉ between 18 and 250 mM. The $K_{\mathrm{m}}$ for each butanol is about 47 (± 7) mM, $V_1∕{\mathrm{E}}_{\mathrm{t}}$ is 0.35 ± 0.03 s⁻¹ for protio butanol, and ${}^{\mathrm{D}}{V}_1=4.2\pm 0.4$ and ${}^{\mathrm{D}}{V}_1∕K_{\mathrm{b}}=4.5\pm 0.6$, suggesting that hydride transfer is a major rate-limiting step for the oxidation of this alcohol.

The H/D isotope effects for the H48S enzyme were determined by initial velocity studies where both the coenzyme and substrate concentrations were varied (Table 4). There is not a significant isotope effect on the dissociation constants for NAD⁺ or NADH. These results are consistent with a random mechanism for ethanol oxidation and a preferentially ordered mechanism for acetaldehyde reduction. For this enzyme, it appears that increasing the pH decreases the commitment to catalysis in the reverse direction (Table 5).

Table 4.

Deuterium isotope effects for forward and reverse reactions with H48S ADH^a

${}^{\mathrm{D}}{V}_1$	${}^{\mathrm{D}}{V}_1/K_{\mathrm{b}}$	${}^{\mathrm{D}}{V}_1/K_{\mathrm{a}}$	${}^{\mathrm{D}}{V}_2$	${}^{\mathrm{D}}{V}_2/K_{\mathrm{p}}$	${}^{\mathrm{D}}{V}_2/K_{\mathrm{q}}$
4.0 ± 0.2	4.1 ± 0.4	3.6 ± 0.3	1.07 ± 0.05	2.4 ± 0.1	0.92 ± 0.04

^aThe buffer was 83 mM sodium phosphate, 40 mM KCl, pH 7.3, at 30 °C. Determined with initial velocity experiments with both substrates varied and fitted to SEQUEN.

Table 5.

Deuterium isotope effects for the reverse reaction of H48S ADH^a

pH	Buffer	$V_2/{\mathrm{E}}_{\mathrm{t}}$, s−1	$K_{\mathrm{p}}$, mM	${}^{\mathrm{D}}{V}_2$	${}^{\mathrm{D}}{V}_2/K_{\mathrm{p}}$
6.0	MES	1900 ± 50	1.4 ± 0.2	0.85 ± 0.03	3.5 ± 0.4
8.0	TAPS	1400 ± 30	2.3 ± 0.2	1.10 ± 0.03	2.5 ± 0.2
10.0	AMPSO	170 ± 10	2.3 ± 0.6	1.4 ± 0.1	1.3 ± 0.5

^aFixed, saturating concentrations of 0.20 mM NADH and 0.185 mM NADD were used, with acetaldehyde varied from 0.25 to 50 mM, at 30 °C. The buffers are 0.1 M as described in Table 3, except that 0.1 M sodium N-(1,1-dimethyl-2-hydroxyethyl)-3-amino-2-hydroxypropanesulfonate was used at pH 10.

3.4. pH dependence of kinetic constants

The contribution of His-48 to catalysis is altered by the H48E and H48S substitutions. The pH profiles for ethanol oxidation and acetaldehyde reduction by H48E ADH are shown in Fig. 4. The parameters for ethanol oxidation, $V_1∕{\mathrm{E}}_{\mathrm{t}}$ and $V_1∕K_{\mathrm{b}}{\mathrm{E}}_t$, have similar dependencies since the $K_{\mathrm{m}}$ for ethanol ($K_{\mathrm{b}}$) varies less than 2-fold over the pH range. In the reverse reaction, the $K_{\mathrm{m}}$ for acetaldehyde ($K_{\mathrm{p}}$) increases from 0.45 mM at low pH to 5.3 mM at high pH, and therefore the pH dependencies for $V_2∕{\mathrm{E}}_{\mathrm{t}}$ and $V_2∕K_{\mathrm{p}}{\mathrm{E}}_{\mathrm{t}}$ are somewhat different.

Fig. 4.

pH dependence of forward and reverse reactions of H48E enzyme. For the forward reaction, 1 mM NAD⁺ and 4 and 30 mM ethanol (with duplicate assays) in a constant ionic strength buffer [60] at varied pH were studied. This buffer is 100 mM 2-(N-morpholino)ethanesulfonic acid, 50 mM tris(hydroxymethyl)aminomethane, and 50 mM 2-amino-2-methyl-1-propanol adjusted to the desired pH with NaOH or CH₃COOH, for an ionic strength of 0.1. In separate experiments, the stability of the enzyme was determined; it had a half-life of 1 hr at pH 5 and of 50 hr at pH 10.5, at 30 °C, without substrates, at a concentration of 0.1 mg/ml in 1 mg/ml of bovine serum albumin. The assays required 1 to 2 min, so that the enzyme was not inactivated at the extremes of pH. For the reverse reaction, 0.1 mM NADH with five concentrations of acetaldehyde (0.25 to 2.25 mM at low pH and 1.8 to 5.3 mM above pH 8.5) were used. Buffers without primary amino groups were used, all being 0.1 M, brought to the desired pH with NaOH: 2-(N-morpholino)ethanesulfonic acid (pHs 5.5-6.5), 3-(N-morpholino)propanesulphonic acid (pHs 7.0-7.5), N-tris(hydroxymethyl)-3-aminopropanesulfonic acid (pHs 8.0-8.5), and 3-[(1,1-dimethyl-2-hydroxyethyl)amino]-2-hydroxypropanesulfonic acid (pHs 9.0-10.5). Initial velocities at each pH were fitted to the HYPER program for estimation of $V$ and $V∕K$ [45]. The curves are computer fits to Eq. 4 or Eq. 5 with the parameters given in Table 6.

The data for pH dependencies of $V∕{\mathrm{E}}_{\mathrm{t}}$ and $V∕K_{\mathrm{m}}{\mathrm{E}}_{\mathrm{t}}$ can be described by various mechanisms as shown in Scheme 2. For H48E ADH, Eq. 4 describes the data for the forward reaction and Eq. 5 describes the results for the reverse reaction. Usually the simplest mechanism is chosen. Such equations have given good fits for pH dependence data (pK values with errors < ± 0.2) for wild-type enzyme and other substituted ADHs [19, 32]. Table 6 shows that the data for the forward reaction of H48E ADH can be described with two ionizations, but the lower pK is not well defined for the reverse reaction. The pH dependence for wild-type enzyme is described well with just one ionization (Eq. 3). Graphs of the pH dependence for wild-type enzyme were presented earlier [19], and graphs for the forward reaction are shown below. The significant result is that the pK value of 7.3 to 7.8 for wild-type enzyme are replaced by two pK values for H48E ADH.

Table 6.

pH dependence of kinetic constants for forward and reverse reactions of H48E and wild-type ADHs^a

	H48E		Wild-typeb
Kinetic constant	pK values	limiting constants	pK values	limiting constants
ethanol	${\text{pK}}_{\mathrm{b}}$, ${\text{pK}}_{\mathrm{a}}$	Eq. 4, $k_2$, $k_3$	${\text{pK}}_{\mathrm{a}}$	Eq. 3, $k_2$, $k_3$
$V_1∕{\mathrm{E}}_{\mathrm{t}}({\mathrm{s}}^{-1})$	6.2 ± 0.1	$k_2$, 1.0 ± 0.1	7.3 ± 0.1	$k_2$, 190 ± 14
	9.5 ± 0.1	$k_3$, 43 ± 6	-	$k_3$, 530 ± 16
$V_1∕{\mathrm{E}}_{\mathrm{t}}{K}_{\mathrm{b}}$ (mM−1s−1)	6.5 ± 0.1	$k_2$, 20 ± 0.03	7.8 ± 0.2	$k_2$, 5 ± 0.9
	8.9 ± 0.1	$k_3$, 1.7 ± 0.2	-	$k_3$, 41 ± 5
acetaldehyde	${\text{pK}}_{\mathrm{a}}$, ${\text{pK}}_{\mathrm{b}}$	Eq. 5, $k_1$, $k_2$	${\text{pK}}_{\mathrm{a}}$	Eq. 3, $k_3$, $k_2$
$V_2∕{\mathrm{E}}_{\mathrm{t}}$ (s−1)	6.8 ± 0.3	$k_1$, 200 ± 16	7.8 ± 0.2	$k_2$, 3600 ± 300
	9.4 ± 0.1	$k_2$, 98 ± 10	-	$k_3$, 1200 ± 200
$V_2∕{\mathrm{E}}_{\mathrm{t}}{K}_{\mathrm{p}}$	7.5 ± 0.1	$k_1$, 420 ± 17	7.7 ± 0.1	$k_2$, 3200 ± 200
	9.5 ± 0.2	$k_2$, 14 ± 5	-	$k_3$, 500 ± 80
1-butanolc	${\text{pK}}_{\mathrm{b}}$, ${\text{pK}}_{\mathrm{a}}$	Eq. 4, $k_2$, $k_3$	${\text{pK}}_{\mathrm{b}}$, ${\text{pK}}_{\mathrm{a}}$	Eq. 4, $k_2$, $k_3$
$V_1∕{\mathrm{E}}_{\mathrm{t}}$, (s−1)	6.1 ± 0.1	$k_2$, 0.43 ± 0.04	7.0 ± 0.2	$k_2$, 63 ± 5
	10.0 ± 0.3	$k_3$, 4.5 ± 1.4	10.4 ± 0.8	$k_3$, 245 ± 210
$V_1∕{\mathrm{E}}_{\mathrm{t}}{K}_{\mathrm{b}}$ (M−1s−1)	6.4 ± 0.2	$k_2$, 5.3 ± 1.4	6.9 ± 0.1	$k_2$, 500 ± 110
	8.9 ± 0.2	$k_3$, 92 ± 12	9.0 ± 0.3	$k_3$, 1800 ± 260

^aThe data for H48E ADH are shown in Fig. 4. The values were obtained from fits to the logarithmic forms of Eq. 4 or 5. For the forward reaction (with ethanol), $k_1$ is not significant, and for the reverse reaction, $k_3$ is not.

^bThe data for wild-type enzyme were obtained by initial velocity studies where both the NAD⁺ and ethanol or NADH and acetaldehyde concentrations were varied, and fitted to SEQUEN (Eq. 2), and the kinetic constants were fitted with “WAVL” (Eq. 3) [45], which describes limiting activities at low and high pH for protonated or unprotonated enzyme forms. Graphs of the pH dependence and fits to the equations for the wild-type enzyme reported previously [19] were confirmed and extended in the present study, obtaining slightly different constants, especially for binding of the coenzymes. (See Supplemental Data Table 1.) Fitting the data for log $(1∕K_{\text{ia}})$ produced a pK value for free enzyme of 8.1 ± 0.2, and a pK of 9.1 for the E-NAD⁺ complex, with a limiting $K_{\text{ia}}$ value at low pH of 0.54 ± 0.06 mM and a limiting $K_{\text{ia}}$ at high pH of 4.9 ± 0.9 mM. Likewise, fitting log $(1∕K_{\text{iq}})$ gave a pK value of 7.4 ± 0.2 for free enzyme and a pK of 8.9 for the E-NADH complex, with a limiting $K_{\text{iq}}$ at low pH of 15 ± 2 μM and at high pH of 410 ± 80 μM.

^cThe pH dependence was studied in 10 mM Na₄P₂O₇ buffer adjusted to the desired pH and a final ionic strength of 0.1 with sodium phosphate [19] with 5 mM NAD⁺ and concentrations of 1-butanol varied at 10, 50 and 100 mM for H48E ADH, and 20, 50 and 100 mM butanol for wild-type ADH. The data (Supplemental Data, Tables 2 and 3) were fitted to Eq.4.

The pH dependence for oxidation of 1-butanol was also studied, because butanol reacts more slowly than ethanol, and transient and steady kinetics, and substrate deuterium isotope effects (${}^{\mathrm{D}}{V}_1∕{\mathrm{E}}_{\mathrm{t}}=4$), indicate that hydrogen transfer is a major rate-limiting step in the overall reaction with wild-type enzyme [22, 52, 61]. The dependence of $V_1∕{\mathrm{E}}_{\mathrm{t}}$ on pH for wild-type enzyme is not simply described by one pK value [22]. In the present study, the dependence of log $V_1∕{\mathrm{E}}_{\mathrm{t}}$ and log $V_1∕K_{\mathrm{b}}{\mathrm{E}}_{\mathrm{t}}$ on pH for both H48E and wild-type ADHs were also complex, described by Eq. 4 (Table 6).

The pH dependencies for ethanol oxidation catalyzed by the H48S enzyme are shown in Fig. 5, with a comparison to the dependence for wild-type ADH. The pH dependence for wild-type enzyme suggests that deprotonation of a group with a pK of ~7 increases both $V_1$ and $V_1∕K_{\mathrm{b}}$, which has been associated with a histidine residue [14, 62]. The pH dependence with H48S ADH is essentially linear, certainly different from that of wild-type ADH. The linear fits are given in the legend to Fig. 5, but the data could also be fitted well to Eq. 4 as reported in Table 7. The pH dependence for oxidation of butanol (varied at 30, 40, 50, 85, 125, and 200 mM, with 5 mM NAD⁺ in the pyrophosphate buffer system were also fitted well to a straight line (log $k$ versus pH) with a slope of 0.36 ± 0.01 for $V_1∕{\mathrm{E}}_{\mathrm{t}}$ and 0.43 ± 0.01 for $V_1∕{\mathrm{E}}_{\mathrm{t}}{K}_{\mathrm{b}}$, but the data also fit well to Eq. 4 (Table 7).

Fig. 5.

pH dependence for ethanol oxidation by wild-type and H48S ADHs. The circles are for wild-type ADH and the squares are for H48S enzyme. The units are log s⁻¹ for $V_1∕{\mathrm{E}}_{\mathrm{t}}$, filled symbols, and log mM⁻¹s⁻¹ for $V_1∕K_{\mathrm{b}}{\mathrm{E}}_{\mathrm{t}}$, open symbols. The data were obtained with a buffer that was 20 mM Na₄P₂O₇ adjusted to the desired pH and a final ionic strength of 0.1 with sodium phosphate [19]. For the H48S ADH, the concentration of NAD⁺ was fixed at 5 mM, and the concentrations of ethanol were varied from 7.7 to 100 or 250 mM. The $K_{\mathrm{m}}$ decreased from 29 mM at low pH to 16 mM at high pH. For the H48S enzyme, the data were fitted well with a straight line, slope of 0.32 ± 0.01 (R² = 1.0) for log $V_1∕{\mathrm{E}}_{\mathrm{t}}$ and 0.37 ± 0.01 (R² = 0.999) for log $V_1∕K_{\mathrm{b}}{\mathrm{E}}_{\mathrm{t}}$.

Table 7.

pH dependence of kinetic constants for alcohol oxidation by H48S ADH^a

Kinetic constant	pK values	limiting constants	pK values	limiting constants
	ethanol		1-butanolb
	${\text{pK}}_{\mathrm{a}}$, ${\text{pK}}_{\mathrm{b}}$	Eq. 4, $k_2$, $k_3$	${\text{pK}}_{\mathrm{a}}$, ${\text{pK}}_{\mathrm{b}}$	Eq. 4, $k_2$, $k_3$
$V_1∕{\mathrm{E}}_{\mathrm{t}}$ (s−1)	6.1 ± 0.1	$k_2$, 86 ± 13	6.0 ± 0.2	$k_2$, 6.9 ± 0.9
	8.6 ± 0.1	$k_3$, 650 ± 50	9.1 ± 0.1	$k_3$, 88 ± 11
$V_1∕{\mathrm{E}}_{\mathrm{t}}{K}_{\mathrm{b}}$ (mM−1 s−1)	6.2 ± 0.1	$k_2$, 4.1 ± 0.6	6.3 ± 0.2	$k_2$, 0.14 ± 0.03
	8.8 ± 0.1	$K_3$, 41 ± 4	9.4 ± 0.2	$k_3$, 2.9 ± 0.6

^aThe data for ethanol oxidation are presented in Fig. 5. The constants for wild-type enzyme are in Table 6.

^bThe data for 1-butanol oxidation are given in Supplementary Data Table 4.

4. Discussion

4.1. Background studies on yeast alcohol dehydrogenase

X-ray crystallography and cryo-electron microscopy have provided six structures of yeast ADH1, including apoenzyme, binary and ternary complexes, and two different conformational states, which provide a basis for understanding the general enzymatic mechanism (Scheme 3) [11-13]. The open (E) conformation binds coenzyme between the coenzyme and catalytic domains of the protein, and the closed (F) conformation traps the bound coenzyme. The open conformation is characterized by an alternative coordination of the catalytic zinc with Cys-43, Cys-153, His-66, and Glu-67, and the closed conformation displaces Glu-67 with water or substrates (Scheme 4, based on 5env.pdb) [12]. The isomerization (E ↔ F) of the open conformations of the complexes with coenzyme is coupled to the change in coordination of the catalytic zinc. The interchange of water and substrate in the ternary complexes may also involve a transient change in zinc coordination through a trigonal bipyramidal pentacoordinate intermediate [11, 13, 63, 64]. At some step, the alcohol (Alc) is deprotonated to effectively form an alkoxide (Alk), and a hydride ion is transferred to NAD⁺ to form NADH and aldehyde (Ald).

Scheme 3.

General mechanism of alcohol dehydrogenases

Scheme 4.

Change in zinc coordination for open to closed conformations in yeast ADH when trifluoroethanol (Alc) binds to the enzyme-NAD+ complex.

Inspection of the structure of the active site shows that several ionizable amino acid residues could directly or indirectly affect the pH dependencies observed for wild-type and enzymes with amino acid substitutions. His-48 in the proposed proton relay system also interacts with Asp-53 (Fig. 1), which could also promote the relay of protons. His-15 and His-51 are nearby, but have no direct interactions with substrates, and the H15R substitution has small effects on the kinetics [48]. His-44 binds the pyrophosphate of the coenzyme, and the H44R substitution modestly affects coenzyme binding, kinetic constants, and pH dependencies [19]. Asp-201 and Lys-206 interact with the adenosine ribose, and the D201G substitution allows NADP⁺ to bind, but substantially decreases coenzyme binding and activity [65]. Asp-46 and Glu-67 participate in binding of the catalytic zinc, and the D46N and E67Q substitutions substantially decrease catalytic activity and alter pH dependencies, attributed to effects on the isomerization of enzyme-substrate complexes [32]. The classical coordination of the catalytic zinc can bind water or alcohol, which are prime candidates for ionizable groups as shown in Scheme 1 [13].

The pH dependencies of the wild-type enzyme have been extensively studied, but further explanations are needed. The binding of NADH (1/$K_{\text{iq}}$) depends on a group on the free enzyme with a pK of 7.1 that is shifted to 8.7 in the E-NADH complex, resulting in the uptake of ~ 0.7 protons per subunit at pH 7.9, which is consistent with protonation of His-44, His-48 or the catalytic zinc water; NADH binds more tightly at low pH than at high pH [15]. We find similar pK values of 7.4 and 8.9 for wild-type ADH. The H44R substitution produces tighter binding of NADH, with pK values of 7.8 and 8.6 that could be due to His-48 or zinc-water where NADH binds more tightly when one of these groups is protonated [19]. Binding of NAD⁺ (1/$K_{\text{ia}}$) has a small, but similar pH dependence, and the calculated pK values are 8.1 and 9.1 [15, 19]. The H44R substitution inverts the pH profile and improves binding of NAD⁺ at high pH, which can be ascribed to better interaction of the introduced arginine with the pyrophosphate of the coenzyme, and tighter binding above a pK value of 9.8 for the complex, perhaps due to ionization of the catalytic zinc-water interacting with the positively-charged nicotinamide ring[19].

By comparison, the horse liver ADH1E has Arg-47 at the position homologous to His-44, and binding of NAD⁺ is also tighter at high pH than at low pH with pK values of 8.5-9.2 for free enzyme and 6.9-7.6 for the enzyme-NAD⁺ complex; these pK values are ascribed to the catalytic zinc-water [3, 5, 7, 18, 66, 67]. Binding of NAD⁺ also is associated with an isomerization where the conformation of the enzyme changes [68]. Binding of NADH to the horse liver ADH1E is relatively pH independent up to pH 9, but affinity decreases above that, ascribed to deprotonation of the zinc-water and Lys-228 [3, 18, 69].

The information from studies on the pH dependence of coenzyme binding are relevant for the overall kinetic mechanism, but the chemical events are reflected in the catalytic efficiency that is measured by $k_{\text{cat}}∕K_{\mathrm{m}}$ for the second substrate, which includes the rate constants for binding, chemical reaction and release of the first product. The reactions of the yeast ADH-NAD⁺ complex with ethanol ($V_1∕{\mathrm{E}}_{\mathrm{t}}{K}_{\mathrm{b}}$) and of E-NADH with acetaldehyde ($V_2∕{\mathrm{E}}_{\mathrm{t}}{K}_{\mathrm{p}}$) are described by a similar pK of 7.7, consistent with a histidine (i.e., His-48) acting as a base for alcohol oxidation and the same histidine acting as an acid for aldehyde reduction (Table 6) [15, 16, 19]. This histidine could also be involved in the interchange of water and substrates on the catalytic zinc. However, this pK value also could be due to ionization of the water or alcohol bound to the zinc in the proton relay (Scheme 1). Interactions in the proton relay system and kinetic complexities (commitment factors) can also shift pK values [70]. Therefore, it is customary to determine the binding (equilibrium) of an unreactive alcohol analog, such as trifluoroethanol, to get intrinsic pK values for the enzyme-NAD⁺ complex [18, 71]. Significantly, the pH dependence for trifluoroethanol binding (log $1∕K_{\mathrm{i}}$) shows no inflection at pH 7.7, but rather a simply linear increase in affinity with pH from 6 to 10 with a slope of 0.38 ± 0.03 [53]. We confirmed this finding; the data are in Supplemental Data, Table 5. This dependence is not expected when the ionization of one group controls the binding, and it indicates some complexity in the mechanism.

4.2. Mechanistic effects of the H48E substitution

The substitution of His-48 with a glutamate residue substantially decreases enzyme activities for both forward and reverse reactions, decreases affinity for trifluoroethanol by 50-fold (Table 1), and alters the pH dependencies of catalytic reactions (Fig. 4, Table 6). These results show that His-48 is important for catalysis. The deuterium isotope effects (Table 3) suggest that the oxidation of ethanol by H48E ADH is substantially limited by the rate of transfer of hydrogen, and because the $K_{\mathrm{m}}$ is also pH independent (averaging about 10 mM over the pH range), the $K_{\mathrm{m}}$ for ethanol might approximate the dissociation constant for the alcohol binding to the enzyme-NAD⁺ complex. Thus the mechanism approximates rapid equilibrium random for ethanol oxidation [59]. Nevertheless, the noncompetitive inhibition by ethanol against acetaldehyde suggests that acetaldehyde reduction may be steady-state ordered. Such a mechanism, with random binding of NAD⁺ and ethanol and ordered binding of NADH and acetaldehyde, was suggested for D49N ADH [32]. For the purposes of the present study, it is sufficient to know that the mechanism is sequential, with coenzymes binding to free enzyme, so that the reactions of the enzyme-coenzyme complexes with ethanol and acetaldehyde can be studied when the concentrations of coenzymes are saturating.

The H48E substitution apparently does not cause significant structural changes in the enzyme since the glutamate can replace histidine in hydrogen bonding to the 3’-hydroxyl group of the nicotinamide ribose (Fig. 2), and affinities for coenzymes increase (Table 1). The H48E substitution decreases catalytic efficiency ($V_1∕{\mathrm{E}}_{\mathrm{t}}{K}_{\mathrm{b}}$) on ethanol by 68-fold, but the overall termolecular rate constant, $V_1∕{\mathrm{E}}_{\mathrm{t}}{K}_{\mathrm{b}}{K}_{\text{ia}}$, only decreases 4-fold as compared to wild-type enzyme. This kinetic constant includes the rate constants for binding of NAD⁺ and ethanol, hydrogen transfer and release of acetaldehyde (the first irreversible step), so it is in some respects a better measure of catalytic power. For the reverse reaction, $V_2∕{\mathrm{E}}_{\mathrm{t}}{K}_{\mathrm{p}}{K}_{\text{iq}}$ only decreases 2.4-fold. The large decreases in turnover numbers can result from slower rate constants for dissociation of coenzymes and reactions in the ternary complexes.

The pH dependency results shown in Fig. 4 and Table 6 might be explained by two ionizable groups with pK values of 6.2-7.5 and 8.9-9.5 for the binary complexes with coenzymes and the ternary complexes. It appears that the forms of enzyme in two different states of ionization are active, as in Scheme 2. We focus here on the oxidation of ethanol because the isotope effects (Table 3) suggest that hydride transfer is a major rate-limiting step, which can be affected by the state of protonation of the enzyme or the alcohol. It appears that ethanol reacts most rapidly with the unprotonated forms of the enzyme-NAD⁺ complex, and the singly protonated form has intermediate reactivity.

It is not clear which groups on the enzyme might be responsible for these pK values. It might seem that ionization of the water or alcohol bound to zinc is responsible for one of the pK values, but the $K_{\mathrm{b}}$ for ethanol is relatively pH independent. What groups would control the hydride transfer? Glu-48 could participate, but it is not clear what structural effects would raise its intrinsic pK from ~ 4.5 upward to a value of ~ 6.3 [72]. By the principle of microscopic reversibility, a protonated glutamic acid residue would then donate a proton to the aldehyde, and the pK value of 7.7 for the reverse reaction could be assigned to Glu-48. In any case, it appears that Glu-48 is not as catalytically effective as His-48. His-44 and Lys-209 interact with coenzyme, binding to the pyrophosphate and adenosine ribose, respectively, but their deprotonation should decrease activity by weakening the ionic attractions. Substitution of His-44 with arginine did not abolish the pK of 7.7 [19]. Perhaps when His-48 is substituted, His-44 can have some weak role, and the effects of these secondary ionizations would become visible. Perhaps the pH dependencies result from ionizable groups that indirectly affect activity through conformational changes of the enzyme and the reorganization of the zinc ligands.

4.3. Mechanistic effects of the H48S substitution

The results with the H48S ADH are notable because the kinetic constants are very similar to those for wild-type enzyme (Table 1). The lack of a direct interaction of an amino acid residue at position 48 has not severely compromised the activity (Fig. 2). However, at pH 7.3, trifluoroethanol binds 10-fold less well than to the wild-type enzyme, perhaps because Ser-48 does not facilitate deprotonation of the alcohol. The inhibition pattern (Table 2, Fig. 3) and isotope effects (Table 4) studies suggest that the kinetic mechanism approximates rapid equilibrium binding of NAD⁺ and ethanol in the forward reaction and ordered with NADH binding first in the reverse reaction. The deuterium isotope effects (Table 4) suggest that hydride transfer is a major rate-limiting step for ethanol oxidation. The pH dependencies for the reactions with ethanol are clearly different from those of the wild-type enzyme, and are described most simply by a straight line, with slopes of 0.32 and 0.37 for log $V_1$ or log $V_1∕K_{\mathrm{b}}$ versus pH (Fig. 5). The oxidation of butanol shows a similar dependence. For both alcohols, the $K_{\mathrm{b}}$ values show a weak dependence on pH, reflected in the differences in the slopes for log $V_1$ or log $V_1∕K_{\mathrm{b}}$. However, the pH dependencies can also be described by a mechanism with ionization of two groups, according to Scheme 2, Eq. 4 (Table 7). As with the H48E enzyme, it is clear that the H48S substitution has removed the contribution of His-48 to catalysis, but it is difficult to assign groups that could explain the pH dependencies. Further analyses of different mechanisms are required.

4.4. Conventional interpretation of pH profiles for yeast alcohol dehydrogenase with substitutions of His-48

When multiple ionizable groups participate in the catalytic mechanism, the pH dependencies can be bell-shaped, indicating one group needs to be unprotonated and another group needs to be protonated for maximum activity. In general, as the pH is raised or lowered away the pK values, the slopes of the pH profiles will approximate +1 or −1 [54]. If activity is modulated by two groups, each of which contribute to activity as the pH changes, the complicated pH dependencies can be described by equations with multiple pK values, such as shown in Scheme 2. Some interesting examples are found for other enzymes [73-76].

For the three yeast ADHs with substitutions of His-48, the pH dependencies for oxidation of ethanol ($V_1∕K_{\mathrm{b}}$) can be fitted to Eq. 4, where successive ionization of two groups leads to increasing activity (Fig. 6, Tables 6 and 7) with pK values of 6.2-6.8 and 8.7-8.9 for the three substituted enzymes. Thus, the pK value of 7.8 for wild-type enzyme is missing, and it would seem reasonable to assign that pK to His-48. However, which groups in the ADHs with the His-48 substitutions now contribute to catalysis? Reasonable candidates are a water or alcohol ligated to the catalytic zinc, where deprotonation should improve catalysis by producing the zinc-alkoxide. Deprotonation of His-44 might also provide base catalysis, but His-44 binds the pyrophosphate moiety of the coenzyme, and some rotation would be required to break the electrostatic interaction and put the imidazole group in proximity to the nicotinamide ribose 2’-hydroxyl group. A simple explanation for the pH dependencies then could assign the pK values of 6.3-6.8 to ionization of the zinc-water or zinc-alcohol, and the pK values of 8.7-8.9 to His-44, as discussed above in section 4.2.

Fig. 6.

pH Dependence for ethanol oxidation by wild-type ADH with His-48 and three ADHs with substitutions of His-48. The data for wild-type, H48S and H48Q enzymes were obtained with buffer that was 20 mM Na₄P₂O₇ adjusted to the desired pH and a final ionic strength of 0.1 with sodium phosphate or sodium carbonate [19]. The data for wild-type enzyme were fitted to Eq. 3. The data for the substituted enzymes were fitted to Eq. 4; see Tables 6 and 7. The graphs for the H48Q and H48E enzymes were presented in a preliminary report [31].

An alternative explanation for the pH profiles for $V∕K_{\mathrm{m}}$ such as those in Fig. 6 is that they are flattened (exhibiting a “hollow”), when “proton movement into and out of the active site is restricted, so that the protonation of the enzyme substrate complex is not equilibrated rapidly with respect to the rates of reaction to produce or release products” [71]. (See also Refs. [54, 70].) The flattening is associated with “sticky” substrates that are “committed” to catalysis, that is, the rate constants for the forward reaction are faster than for the reverse reaction. In such a case, the midpoint of the pH profile (e.g., ~ pH 7.5) could reflect ionization of one group on the enzyme, and at low pH the slope should approach a value of +1, which, however, was not observed here, even for the H48E enzyme. Typically, the substrate with the largest $V∕K_{\mathrm{m}}$ is considered to be sticky, and for ADH, this would be the aldehyde, not the alcohol (Table 1). Furthermore, the kinetic isotope effects are generally larger for ethanol oxidation than for acetaldehyde reduction (Tables 3 and 4), which also suggests that ethanol is not sticky in the mechanism. Oxidation of the non-sticky 2-propanol by wild-type ADH is described by a simple pK of 7.6, which can be assigned to His-48 [17].

4.5. Specific base catalysis by yeast alcohol dehydrogenase with substitutions of His-48

A more reasonable explanation is based on the facts that the data (Fig. 6) for the H48S and H48Q enzymes are fitted just as well to the equation for a straight line, with slopes of 0.37 for H48S ADH (Fig. 5) and 0.52 for H48Q ADH [30]. Specific base catalysis (by hydroxide) could be responsible for the pH dependence, if the expected slope of +1.0 were lowered because the activity coefficient for hydroxide is decreased in the local protein environment [77]. (To rephrase Sir Doyle, “After you have dismissed the conventional suspects, a novel explanation must be considered”.) Hydroxide ion could accept a proton from the hydroxyl group of water or alcohol bound to the catalytic zinc, relayed to the 2’-hydroxyl group of the ribose through the hydroxyl group of Thr-45. The constrained environment of the active site could decrease the activity coefficients for solvated protons or hydroxide ions. Activity coefficients in aqueous solutions are affected by ionic strength and solutes, but we do not know of any estimates for coefficients in enzymatic reactions. At pH 7.3, $V_1∕K_{\mathrm{b}}$ decreases ~3-fold for H48S ADH and ~10-fold for H48Q ADH as compared to wild-type ADH, which could reflect steric hindrance for hydroxide attack. Although the dependence for H48E ADH does not fit well to a straight line, $V_1∕K_{\mathrm{b}}$ is decreased 68-fold, and the negatively-charged Glu-48 could electrostatically and sterically decrease hydroxide attack. In contrast, the wild-type enzyme with His-48 shows a 10-fold “modulation” of activity, and does not show evidence for hydroxide catalysis, which might be diminished by steric effects. The ratio of the finite activities of wild-type ADH on the excellent substrate, ethanol, at high and low pH values suggests that His-48 contributes a factor of about 10 to catalysis.

However, when and how the proton is transferred from the alcohol is not known. As discussed above, when wild-type yeast ADH forms the ternary complex with NAD⁺ and an alcohol, the conformation of the protein and the coordination of the catalytic zinc change as the alcohol displaces the carboxyl group of Glu-67 (Scheme 4). A structure for the open conformation of the enzyme-NAD⁺ complex has been determined (5env.pdb, subunit B [12]), but a closed conformation with NAD⁺ and a water bound to the zinc is only inferred from a structure of the homologous E. coli ADH (5gvk.pdb [78]). It is not clear if water binds first to the zinc and then alcohol replaces it by an interchange mechanism (Scheme 1 in Ref. [29] and Fig. 4 in Ref. [63], also Refs. [25, 64]). At high concentrations of alcohol, it is possible that alcohol binds directly to the zinc and displaces Glu-67. It is also not known if an interchange could involve protonation/deprotonation of the zinc ligand. Transient kinetics showed no pre-steady state burst in proton release or production of NADH, and it was suggested that proton release (at least with the “slow” substrate butanol) occurs at the “same rate as production of NADH” [61]. Thus, mechanistic features could also moderate the kinetics of proton release.

A relevant precedent for suggesting specific hydroxide catalysis comes from the studies on human ADH1B1 with the H51Q substitution where general base catalysis by buffers and hydroxide could compensate for some loss of enzymatic activity, giving a Brønsted relationship with a slope of 0.53 ± 0.15 [8]. The dependence on pH from 7.0 to 7.5 provided a slope of ~0.3. It appears that external bases could interact through the proton relay system. However, the pK values for binding of caprate and trifluoroethanol shifted only from 7.8 to 8.5 by the H51Q substitution. We did not study possible buffer catalysis with the substituted yeast ADHs. The H51Q substitution in horse liver ADH1E showed a simple pH dependence with a pK of 8.4 for $V∕K_{\mathrm{b}}$ with ethanol and also shifted the pK for binding of trifluoroethanol from 7.6 to 8.5 [10].

The pH profiles for $V_1∕K_{\mathrm{b}}$ for ethanol oxidation are also altered by other substitutions in yeast ADH1. Glu-67 binds to the catalytic zinc in the apoenzyme and is displaced by water or alcohol in complexes with coenzyme; the E67Q substitution decreases $V_1∕K_{\mathrm{b}}$ by 80-fold at pH 7.3 and produces a pH profile similar to that for H48E ADH with a pK value of 6.7 and an upward swing at higher pH that could be hydroxide catalysis [32]. Substitution of Asp-49, which interacts with zinc-ligand His-66, with asparagine decreases $V_1∕K_{\mathrm{b}}$ by 1100-fold and produces an enzyme with a simple pK of 6.3 [32]. The effects of the E67Q and D49N substitutions were attributed to alterations in the isomerization of enzyme-substrate complexes. Substitution of Cys-153, a ligand to the zinc, with aspartate decreases $V_1∕K_{\mathrm{b}}$ by 20-fold and produces a complex pH profile with pK values of 7.0 and 9.0 [29]. The T45G substitution disrupts the proton relay system, decreases $V_1∕K_{\mathrm{b}}$ by 4000-fold, and produces a simple pH profile with a pK of 6.3 [63]. For these enzymes, the pK of 6–7 could be due to His-48 or to water or alcohol ligated to the zinc [63].

In contrast to yeast ADH1, the apoenzyme of horse liver ADH1E has the classical coordination of the catalytic zinc with water, and in the ternary complexes with coenzymes the water is replaced by the oxygens of substrate analogues [79-81]. The structure of the binary complex with NADH shows evidence of alternative positions of the catalytic zinc that can resemble an intermediate state in the exchange of ligands on the catalytic zinc [81]. Transient kinetic studies show that a proton is released when NAD⁺ binds to wild-type ADH1E, because the pK of a group, probably water bound to the zinc, shifts from 9.3 to 7.5 [7]. Although the horse and yeast ADHs are homologous, and have similar magnitudes of $V_1∕{\mathrm{E}}_{\mathrm{t}}{K}_{\mathrm{b}}$, the fundamental differences in the binding of NAD⁺, the zinc coordination, and the pH dependencies indicate that the mechanisms differ in some details.

Specific acid/base catalysis may occur in other enzymes [77]. Substitution of the catalytic aspartic acid in trypsin with asparagine decreased $k_{\text{cat}}∕K_{\mathrm{m}}$ and produced a complex pH profile with a significant linear contribution of hydroxide above a pH of 8 [75]. Substitution of Arg-301, a putative base in phosphite dehydrogenase, with alanine produced a pH profile for $k_{\text{cat}}$ with a slope of 0.6, perhaps consistent with hydroxide catalysis [82]. The R143K substitution in superoxide dismutase produced a pH profile with a linear decrease in activity from pH 8 to 12 with a slope of 0.43 [76]. Substituting Asp-27, apparently a proton donor for catalysis by dihydrofolate reductase, with serine or asparagine, substantially decreased activities, and changed the bell-shaped pH profiles to strictly linear ones with slopes of 0.54 and 0.66 for $k_{\text{cat}}$, and slopes of ~0.7 for $k_{\text{cat}}∕K_{\mathrm{m}}$ over a pH range of 3 units [74].

4.6. Related studies on alcohol dehydrogenases

Several enzymes with substitutions of the residue homologous to His-48 (His-51 in horse liver ADH) have been studied. For the enzymes with tyrosine, the proton relay could proceed via an intervening water molecule or by a histidine residue that interacts with the pyrophosphate of NAD (1cdo.pdb, 1mc5.pdb) and might intermittently swing to form a hydrogen bond with a hydroxyl group of the nicotinamide ribose [24, 25]. The substitution of Tyr-52 in maize ADH1-1F with aspartic acid in ADH1-C^m is apparently associated with an 95% decrease in ADH activity [83]. Some medium chain glucose dehydrogenases appear to have a glutamate (2cdb.pdb, 2vwh.pdb) that may act via intervening water molecules [27-29]. Human ADH2 (π) has the homologous Thr-52, which would not act as a base, but its hydroxyl group appears to be connected by a water molecule to the ribose hydroxyl group (3cos.pdb), and perhaps His-48 rotates from its interaction with the pyrophosphate of NAD to serve as a base [26]. Substitutions of Thr-52 with histidine or serine have small effects on catalytic efficiency for ethanol oxidation, leading to the conclusion that a “base at position 51 may not be a universal requirement for a functional alcohol dehydrogenase” [26]. Mouse ADH2 (1e3e.pdb) has Asn-51 and very low enzyme activity, but the P47H substitution (homologous to His-44 in yeast ADH) increases the activity, and modeling of the X-ray structure suggests that His-47 could rotate to function as a base [84, 85]. In the benzyl alcohol dehydrogenase encoded by the TOL plasmid pWW0 of Ps. putida, the residues homologous to Arg-47 and His-51 in horse ADH1 are His and Val, and the H47Q substitution decreased catalytic efficiency for benzyl alcohol oxidation by 125-fold, but the double substitution, H47Q/V51H, restored activity by 12-fold, suggesting that a His-47 could participate [86, 87]. Structural studies have not demonstrated such altered conformational states for this residue, but transient species may have low occupancies.

Enzymes are flexible and dynamic protein molecules, which require various experimental approaches to determine the contributions of the amino acid residues to the catalytic events [77]. Many amino acid residues participate in catalysis, and each residue can contribute to different steps in the mechanism. Even a small change of one amino acid residue can have complex effects on the overall mechanism. Replacement of a single amino acid in the active site does not abolish all catalytic activity, but can alter local interactions and conformational changes.

5. Conclusions

His-48 contributes to catalysis by yeast ADH. The H48E and H48S substitutions decrease catalytic efficiencies ($V∕K_{\mathrm{m}}$) at pH 7.3 and alter the pH dependencies. The substitutions apparently only cause small local structural changes, as coenzymes bind more tightly, whereas trifluoroethanol inhibits less well. In the oxidation of ethanol to form acetaldehyde, hydride transfer is partially rate-limiting, but when the proton is released from the alcohol is not known. The pH dependence of log $V_1∕{\mathrm{E}}_{\mathrm{t}}{K}_{\mathrm{b}}$ for ethanol oxidation by H48S ADH is linear over the pH range from 5.8 to 10.0, with a slope of 0.37. A simple explanation is that hydroxide ion accepts the proton, which is relayed through the hydrogen bonded network with the hydroxyl groups of Thr-45 and the 2’-hydroxyl group of the nicotinamide ribose. A slope of 1.0 is generally expected for specific hydroxide catalysis, but the activity coefficient in the protein environment could be decreased. The proton release may involve the change of coordination of the catalytic zinc. The pH dependencies for other enzymes may be explained by specific hydronium or hydroxide catalysis.

His-48 contributes a factor of 10 toward catalytic efficiency of ethanol oxidation

The H48E and H48S substitutions increase affinity for coenzymes

These substitutions also decrease catalytic efficiencies and alter pH dependencies

Oxidation of alcohols by the Ser-48 enzyme has a linear dependence on pH The oxidation of alcohols appears to use base catalysis by hydroxide ion

Abbreviations:

ADH

alcohol dehydrogenase

H48S

represents the substitution of His-48 with Ser

H48E

of His-48 with Glu

H48Q

of His-48 with Gln