Contribution of buried distal amino acid residues in horse liver alcohol dehydrogenase to structure and catalysis

Abstract

The dynamics of enzyme catalysis range from the slow time scale (∼ms) for substrate binding and conformational changes to the fast time (∼ps) scale for reorganization of substrates in the chemical step. The contribution of global dynamics to catalysis by alcohol dehydrogenase was tested by substituting five different, conserved amino acid residues that are distal from the active site and located in the hinge region for the conformational change or in hydrophobic clusters. X‐ray crystallography shows that the structures for the G173A, V197I, I220 (V, L, or F), V222I, and F322L enzymes complexed with NAD⁺ and an analogue of benzyl alcohol are almost identical, except for small perturbations at the sites of substitution. The enzymes have very similar kinetic constants for the oxidation of benzyl alcohol and reduction of benzaldehyde as compared to the wild‐type enzyme, and the rates of conformational changes are not altered. Less conservative substitutions of these amino acid residues, such as G173(V, E, K, or R), V197(G, S, or T), I220(G, S, T, or N), and V222(G, S, or T) produced unstable or poorly expressed proteins, indicating that the residues are critical for global stability. The enzyme scaffold accommodates conservative substitutions of distal residues, and there is no evidence that fast, global dynamics significantly affect the rate constants for hydride transfers. In contrast, other studies show that proximal residues significantly participate in catalysis.

Short abstract

PDB Code(s): 5CDT; 5CDS; 5CDG; 5CDU; 5KJ1; 5KJ6; 5KJC; 5KJE; 5KJF

Abbreviations

ADH

alcohol dehydrogenase

PFB

2,3,4,5,6‐pentafluorobenzyl alcohol

RMSD

root mean square deviation

G173A

Gly‐173 substituted with Ala, etc.

Introduction

Enzymes are dynamic macromolecules that undergo local and global conformational changes that control substrate binding, positioning of catalytic residues, chemical transformations, and release of products. Catalysis can be affected by local changes in the active site geometry, as well as by alterations in the conformational states of the protein, through allosteric effects and global protein dynamics.1, 2, 3, 4 Amino acid residues distal from enzyme active sites can contribute to catalysis.5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27 However, not all protein motions are coupled to catalysis, and biochemical studies are required to distinguish between slow conformational changes that preorganize the active enzyme scaffold and the fast reorganization during the chemical transformations in the central complexes.28, 29, 30, 31, 32 Global dynamics on a fast time scale (fs to ps) may contribute to catalysis through the equilibrium motions reflected in X‐ray crystallographic B‐factors and molecular dynamics computations, or by non‐statistical “rate‐promoting vibrations.”31, 32, 33 Our studies focused on distal amino acid residues that could be important for enzyme activity, protein structure, and global dynamics.

Horse liver alcohol dehydrogenase (ADH1E, EC 1.1.1.1) is a good subject for these studies.34, 35 ADH catalyzes the direct transfer of hydride from the alcohol to C4N of the nicotinamide ring of NAD⁺ with quantum mechanical tunneling that involves coupled motions.31, 36, 37, 38, 39, 40, 41, 42 The enzyme has an Ordered Bi Bi mechanism, including the isomerization of the enzyme–NAD⁺ complex (Scheme 1).43 When horse liver ADH binds the coenzymes, the open conformation changes to a closed conformation by a rotation of about 10° of the catalytic domain relative to the coenzyme binding domain, and the active site is constricted.44, 45 In the closed conformation, domain and local motions in the active site may facilitate a decrease in the distance between the reacting carbon of the alcohol substrate and the C4N atom of the nicotinamide ring of NAD⁺ from 3.4 Å to 2.7 Å where tunneling would more likely occur.46 Molecular dynamics studies of horse liver ADH have identified anti‐correlated motions between the catalytic domain and the coenzyme binding domain that may contribute to catalysis.47, 48

Schwartz and co‐workers proposed that rate‐promoting vibrations of the protein drive substrates toward each other, modulating the height and the width of the activation barrier.42, 49, 50 They identified a conserved evolutionary motif in horse liver ADH that spans the catalytic and coenzyme binding domains (including Ser‐144, Gly‐181, Val‐203, Gly‐204, Val‐207, Glu‐267, Ile‐269, and Val‐292) and could participate in catalysis. They used transition path sampling to conclude that “protein compression” in the homologous yeast ADH51 “mediates a near barrier‐less hydride transfer.”52, 53 They suggested that “slight changes in promoting vibrations (in lactate dehydrogenase) result in dramatic changes in enzyme chemistry.”54, 55 The contribution of such fast dynamics to catalysis of hydride transfer by ADH needs experimental evaluation.

Val‐203, Val‐207, Val‐292, and Ile‐269 are close to the active site, and substitutions modestly (usually less than 10‐fold) decrease the rate constants for hydride transfer from benzyl alcohol and catalytic efficiencies.56, 57, 58 The V203A enzyme has diminished hydrogen tunnelling.37 The V292S enzyme has altered conformational equilibria, but retains temperature‐independent isotope effects consistent with vibrationally assisted tunneling.39 Residues distal from the active site also contribute to catalysis as chimeric yeast and liver alcohol dehydrogenase with distal substitutions differ in catalytic efficiencies by ∼10‐fold on certain substrates.59, 60 Distal substitutions can affect the slow formation of the Michaelis complex or the fast reorganization that occurs during hydride transfer.3, 31, 35, 61, 62 Fast, global dynamics, including rigid body rotations and translations, local displacements and vibrations, are reflected by X‐ray crystallographic anisotropic displacement parameters (B‐factors), even at cryogenic temperatures where major conformational changes are prevented.63, 64, 65, 66, 67

This study focused on conserved, buried, distal amino acids that should affect protein stability and dynamics, and potentially catalysis. Substitutions of amino acid residues that are buried (not accessible to bulk solvent) in hydrophobic clusters can be deleterious,68 and changing the sizes of side chains can alter protein stability by 1–2 kcal/mol without changing the global protein structure.69, 70, 71, 72, 73 Protein stabilities correlate with peptide backbone hydrogen exchange rates and NMR relaxation, measures of global dynamics.74, 75, 76

Five amino acid residues were chosen (Fig. 1). These residues are connected to the conserved evolutionary motif (residues 144–292) identified by Mincer and Schwartz.49 All of the residues are buried and generally conserved in dimeric ADHs.77 All of the residues are in regions with low B‐factors, suggesting that the local structure is relatively rigid. Because ADH is a dimer, substitution of one residue changes two per molecule.

Figure 1.

Substitutions of buried distal residues. The view is down β‐sheet of the Rossmann fold of the coenzyme binding domain of one subunit of the dimer. The residues are shown in blue. The figure shows wild‐type ADH complexed with NAD⁺ and 2,3,4,5,6‐pentafluorobenzyl alcohol, which is ligated to the catalytic zinc in the middle of the structure (4DWV.pdb). Val‐207 was studied previously. The figure was prepared with PyMOL.

Scheme 1.

Alcohol dehydrogenase mechanism.

Two residues (Gly‐173 and Phe‐322) are in the hinge region that has two peptide chains connecting the catalytic and coenzyme binding domains. Substitutions of these amino acids could alter the rates of the conformational change that occurs when coenzymes bind and the fast, rigid body motions of the domains in the enzyme‐substrate complex. Gly‐173 is in a helix next to Cys‐174, which is ligated to the catalytic zinc. Phe‐322 is located in a hydrophobic cluster in the catalytic domain “behind” the hinge region, and this “wax‐like” region could control the rates of the conformational change. Phe‐322 is close to Phe‐319, which forms a hydrogen bond from its peptide NH to the nicotinamide O7N. Mutations in hinge regions can produce significant effects on stability and activity of proteins.78, 79

Three residues (Val‐197, Ile‐220, and Val‐222) are in hydrophobic clusters within the coenzyme binding domain. Branched chain residues contribute to the global stability of proteins.75 Val‐197 is located in the coenzyme binding domain in a hydrophobic cluster between the β‐sheet and a flanking α‐helix. It is close to Val‐207, which is in turn close to Val‐203, which contacts the nicotinamide ring, and both valine residues were suggested to be part of the sequence contributing to protein promoting vibrations.80 Ile‐220 is in a hydrophobic cluster between a β‐strand and an α‐helix in the coenzyme binding domain, with its CB about 21 Å away from C4N and 12 Å from C2A of the bound NAD. Val‐222 is in a cluster near C2A of the adenine ring of the NAD.

The buried, distal residues were substituted by partially random mutagenesis, and the effects on the structure and kinetics were determined. The structural changes should affect protein stability and thus the global dynamics. Changes in the relatively slow conformational changes and ligand binding were ascertained from the enzyme kinetics and structural studies. Fast, global motions are proposed to affect the catalysis of hydride transfer. If the structures of the enzyme‐substrate complexes are not affected by the substitutions, but the rates of hydrogen transfer are altered, it would appear that fast, global protein dynamics are important. Conversely, no changes in kinetics and structure would be evidence against a role for global dynamics.

Results

X‐ray crystallography

High resolution structures were determined to examine how the substitutions affect the local structure, overall conformation, the orientation of the substrate analogue and coenzyme in the active site, and the temperature factors that reflect the fast dynamics. The complexes of all of the substituted ADHs with NAD⁺ and 2,3,4,5,6‐pentafluorobenzyl alcohol are in the same P ₁ space‐group (i.e., isomorphous) as the wild‐type ternary complex (4DWV.pdb81), with one dimeric molecule in the asymmetric unit, and both subunits are in the closed conformation. No significant differences in overall structure relative to the wild type enzyme are observed. Although crystal lattice contacts could affect the observed structure, other substituted ADHs crystallize in the open conformation found for the apoenzyme, and it appears that the distal substitutions do not affect the conformational equilibrium. The electron density maps clearly show the substitutions. The positions of the NAD⁺ and PFB in the different ternary complexes are clearly defined in the electron density maps and are essentially the same as for the wild‐type enzyme. Data collection and refinement statistics are summarized in Table 1.

Table 1.

X‐ray Data and Refinement Statistics for Substituted Liver Alcohol Dehydrogenases^a

Enzyme	G173A	V197I	I220V	I220L	I220F	V222I	F322L
PDB entry	5KJ1	5KJ6	5CDT	5CDS	5CDG	5KJC	5KJE
cell dimensions (Å)	44.38, 51.50, 92.42	41.47, 51.65, 92.68	44.30, 51.50, 92.35	44.27, 51.53, 92.44	44.26, 51.38, 92.39	44.23, 51.25, 92.55	44.37, 51.28, 92.50
cell angles (deg)	91.82, 103.1, 110.1	91.92, 103.0, 110.2	91.92, 103.1, 110.2	91.75, 103.1, 109.3	91.75, 103.1, 109.4	92.02, 103.1, 109.9	92.02, 102.9, 110.0
mosaicity	0.62	0.80	0.97	1.05	1.6	0.94	1.08
resolution range (Å) (shell)	20.0–1.2 (1.24)	19.5–1.14 (1.18)	20.0–1.70 (1.76)	20.0–1.40 (1.45)	20.0–1.40 (1.45)	20.0–1.2 (1.24)	20.0–1.26 (1.31)
no. of reflections (total, unique)	868947, 218364	805594, 234800	253953, 77459	486777, 132451	423650, 119994	1488789, 215195	554468, 164103
redundancy (shell)	3.95 (3.87)	3.43 (3.20)	3.23 (3.15)	3.64 (3.61)	3.48 (3.52)	6.88 (6.30)	3.34 (3.11)
completeness (%) (outer shell)	94.6 (91.1)	86.0 (51.1)	96.4 (94.4)	91.0 (86.7)	82.9 (83.3)	93.6 (72.0)	82.8 (51.8)
R pim (%) (outer shell)b	3.3 (24.7)	2.5 (21.5)	5.1 (22.8)	3.1 (22.8)	3.5 (55.2)	2.9 (21.5)	4.2 (20.7)
average <I>/σ<I> (outer shell)	9.2 (2.1)	14.8 (2.3)	8.2 (2.4)	11.3 (2.3)	7.9 (1.8)	11.9 (2.3)	9.4 (2.4)
R value, R free, (%) (test %, no.)c	12.4, 17.0 (0.5, 1099)	13.6, 16.4 (0.5, 1172)	16.3, 20.6 (1.4, 1139)	13.2, 18.4 (1.0, 1340)	14.2, 18.7 (1.5, 1803)	12.9, 16.8 (0.5, 1067)	12.4, 14.4 (1.0, 1589)
RMSD for bond distances (Å)d	0.017	0.021	0.020	0.014	0.015	0.016	0.015
RMSD for bond angles (deg)d	1.97	2.20	2.06	1.74	1.78	1.93	1.87
estimated errors in coordinates (Å)	0.027	0.026	0.084	0.047	0.050	0.031	0.024
mean B‐factor (Å2), (Wilson, Refmac)e	10.2, 17.0	9.9, 17.2	21.2, 23.1	14.2, 23.4	18.3, 22.5	11.9, 19.3	12.4, 18.1
non H atoms fitted	6996	6935	6687	6717	6735	6896	6873
protein (B‐factor)	5799 (15.5)	5783 (15.6)	5697 (23.1)	5692 (22.3)	5678 (21.5)	5808 (17.5)	5726 (16.5)
4 Zn, 2 NAD+, 2 PFB, 2–5 MRDf	150 (18.3)	150 (19.9)	134 (23.7)	134 (22.5)	134 (21.9)	158 (23.6)	150 (20.7)
waters	1047 (30.8)	1002 (30.6)	853 (35.0)	891 (35.0)	923 (34.4)	930 (36.2)	997 (33.8)
Ramachandran (%) (favored, outliers)	97.24, 0	97.8, 0	96.95, 0.13	97.45, 0.13	97.19, 0.13	97.12, 0	97.06, 0
MolProbity (clash, score, rank)	1.4, 98th, 1.07, 98th	1.24, 98th, 0.97, 98th	1.44, 99th, 1.08, 100th	1.35, 99th, 0.97, 99th	1.1, 99th, 0.99, 99th	0.95, 99th, 0.91, 99th	1.34, 99th, 1.09, 98th
relative B‐factorg	0.97	1.04	1.12	1.27	1.27	0.98	1.02

^aThe space group is P ₁, with one dimeric molecule of 748 amino acid residues, two NAD⁺ and two 2,3,4,5,6‐pentafluorobenzyl alcohols (PFB) in the asymmetric unit.

^b R _pim, redundancy independent merging, = R _meas/(n ^½), where n = redundancy.

^c R _value = (Σ|F _o – kF _c|)/Σ|F _o|, where k is a scale factor. R _free was calculated with the indicated percentage of reflections not used in the refinement.

^dDeviations from ideal geometry.

^eThe data in the following three lines were calculated with the PARVATI server.

^fMRD, (4R)‐2‐methyl‐2,4‐pentanediol.

^gThe average of the B‐factors for the α‐carbons of the new residue in chains A and B divided by the average B‐factor for the protein relative to the adjusted B‐value for the corresponding residue in the wild‐type enzyme.

The structure of the G173A ADH complex is almost identical to the wild‐type complex (4DWV.pdb), with an overall RMSD for superpositioning of α‐carbons of 0.08 Å. The methyl group of Ala‐173 is inserted into a cavity occupied by a water molecule in the wild‐type enzyme, and the water is shifted by 1.3 Å, retaining the hydrogen bonds to His‐67 O, Ala‐69 N, and Pro‐91 O (Fig. 2). The CB atom of Ala‐173 makes contacts of 3.1 Å to Cys‐170 O, 3.4 Å to His‐67 O, and 3.4 Å to Ala‐69 N, and the cavity expands slightly as nearby residues are shifted by ∼0.1 Å. Gly‐173 is common in dimeric ADHs, but Ala is not present, whereas Ser is found in many plant enzymes, usually associated with the I90L substitution.77 In those enzymes with Ser‐173, the OG atom could essentially replace the water that is present in the Gly‐173 enzymes and is shifted in the G173A enzyme.

Figure 2.

Structure and electron density map for the G173A enzyme. The G173A enzyme is shown in ball and stick with atom coloring and superpositioned onto the structure for the wild‐type enzyme (4DWV.pdb, stick in green). The 2|F _o| – |F _c| map is contoured at about 1.5 e^–/ų. The label “G173A” shows the electron density for the inserted methyl group. The hydrogen bonds to the shifted water are shown in black dashed lines, and the position of the corresponding water in the wild‐type enzyme is shown in green just above the labeled water (“Wat”).

The structure of the V197I ADH is also almost identical (RMSD for α‐carbons, 0.14 Å) to the wild‐type enzyme complex. The additional methyl group is inserted into a cavity that has no evidence of a water molecule in the wild‐type enzyme, and there are only small local shifts in the structure (Fig. 3). Van der Waals contacts are formed with Cys‐195, Ala‐196, Ile‐208, and Phe‐266, and other interactions are with Ile‐220 and Cys‐211. Ile‐197 is present in many dimeric plant ADHs, often accompanied by C195V, C211A, I208M or I208A, and F266V substitutions, relative to horse ADH1E.77

Figure 3.

Structure and electron density map for the V197I enzyme. The V197I enzyme is shown in ball and stick with atom coloring and is superpositioned onto the structure for the wild‐type enzyme (4DWV.pdb, stick in green). The 2|F _o| – |F _c| map is contoured at about 1.4 e^–/ų. The label “V197I” shows the electron density for the inserted methyl group.

The electron density maps for the three substitutions of Ile‐220 clearly define the changed structures (Fig. 4). The α‐carbon atoms of all three structures are superpositioned with an RMSD of less than 0.5 Å onto the wild‐type enzyme. As compared to the wild‐type enzyme, the substitutions at residue 220 produce only small local changes in the hydrophobic cluster. In dimeric ADHs, this residue is usually Ile, but not Val, Leu, or Phe. In the I220V enzyme, the deletion of the methyl group introduces a small cavity that does not appear to allow a water molecule to bind [Fig. 4(A)]. There are no ordered water molecules in the hydrophobic cluster, but several waters are hydrogen‐bonded to nearby residues on the surface of the protein. Previous results showed that cavity‐forming substitutions in ADH may or may not allow a water to bind, consistent with studies of other proteins.58, 82 Both the wild‐type and I220V enzymes have two alternative conformations for Glu‐239, but only one conformation is shown, with its interaction with Arg‐218. (A structure for the I220V ADH complexed with NAD⁺ and pyrazole was also determined (5CDU.pdb).) For the I220L ADH, the rearranged side chain places a methyl group in steric conflict with one alternative conformation of Glu‐239 so that only one conformation is found, where it interacts with a water, Arg‐218 and Thr‐238 [Fig. 4(B)]. In the I220F ADH, the introduction of the bulky benzene ring also results in the displacement of an alternative conformation of Glu‐239, and there are also small shifts (few tenths of Å) of nearby side chains [Fig. 4(C)].

Figure 4.

Structures and electron density 2|F _o| – |F _c| maps for the three enzymes with substitutions of Ile‐220. The wild‐type enzyme (4DWV.pdb) is shown in green stick and the substituted enzymes are shown in ball and stick with atom coloring. (A) The I220V enzyme (map at ∼0.65 e^–/ų) lacks the methyl group of Ile‐220 (“–Me”), and the structures are closely superimposable. (B) The I220L enzyme (map at ∼0.6 e^–/ų) lacks the CG2 methyl group of Ile and inserts the CD2 methyl group of Leu, which displaces the major conformation of Glu‐239 and accommodates a water that is hydrogen bonded (black dotted lines) to Arg‐218 NE, Thr‐238 OG1, and Glu‐239 OE1, similar to the minor alternative conformation for Glu‐239 observed in the wild‐type enzyme. (C) The I220F enzyme (map at ∼0.7 e^–/ų) also has Glu‐239 displaced and the additional water, and the hydrophobic pocket is expanded slightly as Arg‐218, Leu‐254, and Met‐257 are shifted relative to the wild‐type enzyme.

The structure of the V222I enzyme is almost identical (RMSD for all α‐carbons, 0.13 Å) to the wild‐type enzyme (Fig. 5). (A structure for the V222I ADH complexed with NAD⁺ and 2,2,2‐trifluoroethanol was also determined (5KJF.pdb)). The inserted methyl group fits into an empty cavity with minimal changes in the cluster. Val‐222 is the typical residue in dimeric ADHs, but sometimes it is replaced by Ile, which may be accompanied by changes in the nearby residues, Ile‐250 or Leu‐254.77

Figure 5.

The structure and electron density map for the V222I enzyme. The V222I enzyme is shown in ball and stick with atom coloring and is superpositioned onto the structure for the wild‐type enzyme (4DWV.pdb, stick, in green). The 2|F _o| – |F _c| map is contoured at ∼1.2 e^–/ų. The inserted methyl group (CD1) is by the label “V222I.” The CG2 methyl of Ile‐222 is 4.5 Å to C2A of the adenine ring of NAD⁺.

The structure of the F322L enzyme is almost identical (RMSD for α‐carbons, 0.13 Å) to the wild‐type enzyme. There is a small distortion of the pocket that contains residue 322, but no major changes (Fig. 6). Residue 322 is almost equally conserved as Phe, Tyr, and Trp, but Leu is found in the dimeric human ADH4, for which most kinetic constants are 10–100‐fold higher than those for class I ADHs.77, 83, 84 The F322Y substitution is accompanied by the L331V substitution, but enzymes with the F322W or F322L substitutions retain the Leu‐331 found in other ADHs.

Figure 6.

The structure and electron density map for the F322I enzyme. The F322I enzyme is shown in ball and stick with atom coloring and is superpositioned onto the structure for the wild‐type enzyme (4DWV.pdb, stick, in green). The 2|F _o| – |F _c| map is contoured at ∼1.1 e^–/ų. The substitution is labeled “F322I.”

Local and global dynamics

Although the various substitutions did not affect the overall three‐dimensional structures of the enzymes, the dynamics of the protein could be affected. X‐ray crystallography provides temperature (B) factors, which include information about fast dynamics and flexible structural elements. Because the B‐factors vary with the nature of the crystals and the data collection, we cannot determine if the substitutions changed the overall, global B‐factors. Nevertheless, the B‐factors for the α‐carbon atoms of the 748 residues of the substituted and the wild‐type enzymes have very similar patterns, indicating that various regions of the structures have not been differentially perturbed. Each of the five residues studied is in a region with relatively low B‐factors. As an example, Figure 7 compares normalized B‐factors for the I220X and wild‐type enzymes. Because B‐factors at the sites of substitution could be altered by the structural changes, the observed B‐factor was divided by the average B‐factor for all of the residues in the chain and the ratio of the B‐factor for the substituted residue with the corresponding B‐factor for the wild‐type enzyme was calculated (Table 1, last row). Only for the I220L and I220F enzymes is the B‐factor for the substituted residue noticeably larger (about 25%) than for the wild‐type enzyme. The B‐factors for Gly‐221 in the I220L and I220F enzymes are also about 25% higher than for the I220V or wild‐type enzymes, but B‐factors for some other nearby residues are altered less than 25%. It appears that the enzyme scaffold accommodates the conservative substitutions of distal residues and disperses the effects on the global dynamics.

Figure 7.

Comparison of the isotropic B‐factors for the I220X and wild‐type (4DWV.pdb) enzymes. The observed values were normalized by subtracting the average B‐factor for all residues in the chain for each enzyme from the value for each residue. The symbols are wild‐type (◊, green line), I220V (o, red line), I220L (Δ, blue line), I220F (□, black line). The figure shows that the overall pattern of B‐factors is very similar for the A chain for the four enzymes. The B chains also have similar patterns.

Steady‐state kinetics

The steady‐state kinetic constants are not significantly changed as compared to those for recombinant wild‐type ADH58, showing that the mechanisms are not altered (Table 2). Changes of 2–3‐fold may be significant experimentally, but they are trivial with respect to the magnitude of enzyme catalysis. Michaelis (K _a and K _q) and dissociation constants (K _ia and K _iq) for the coenzymes are similar to those for wild‐type ADH. The turnover numbers (k _cat = V ₁/E_t and V ₂/E_t), which are controlled by the release of product coenzymes in the Ordered Bi Bi mechanism,85 are essentially the same for the substituted and wild‐type enzymes. The rate constants for binding and dissociation of NAD⁺, which can be estimated for the ordered reaction (from k _on = V ₁/E_t K _a and k _off = V ₁ K _ia/E_t K _a),35, 86, 87 are also unchanged by the substitutions. As for the wild‐type enzyme, the calculated rate constants suggest that the enzyme–NAD⁺ complex isomerizes (step 2 in Scheme 1, related to deprotonation of the zinc‐water and the global conformational change) because the calculated rate constant for binding (k _on = k ₁ k ₂/(k _–1 + k ₂)) of NAD⁺ is smaller than diffusion‐controlled, and the calculated dissociation (k _off = k _–1 k ₂ k _–2/(k _–1 +k ₂)(k ₂ + k _–2)) is smaller than the observed turnover in the reverse reaction (V ₂/E_t), when the isomerization step is included).35, 85, 88 The binding of NADH also is unchanged for the various enzymes (see below). Catalytic efficiencies (V ₁/K _bE_t and V ₂/K _pE_t), which relate to the binding of substrate, hydride transfer and release of the product, are essentially unchanged. The binding constants for substrates are not equivalent to the Michaelis constants (K _b or K _p), but the dissociation constant of the dead‐end inhibitor 2,3,4,5,6‐pentafluorobenzyl alcohol, determined by competitive inhibition against benzyl alcohol, is 0.56 μM for the V197I enzyme, similar to the value for the K _i of 0.52 μM for wild‐type ADH. The overall equilibrium constants calculated from the Haldane relationship (K _eq) agree with the thermodynamic values of 35–70 pM reported previously,89, 90, 91 indicating internal consistency of the kinetic constants.

Table 2.

Steady‐state Kinetic Constants for Liver Alcohol Dehydrogenases^a

Kinetic constant	WTb	G173A	V197I	I220V	I220L	I220F	V222I	F322L
K a (μM)	3.9	5.2	5.8	6.8	6.1	5.1	4.2	5.0
K b (μM)	14	13	11	22	17	10	8.6	14
K p (μM)	32	63	30	33	32	33	24	29
K q (μM)	1.5	6.2	1.5	3.1	2.8	2.5	1.3	2.0
K ia (μM)	20	38	28	28	42	39	21	23
K iq (μM)	0.31	0.36	0.25	0.61	0.56	0.57	0.25	0.29
V 1/Et(s−1)	3.0	1.8	2.1	2.6	2.0	2.7	4.5	2.6
V 2/Et (s−1)	21	35	29	24	20	21	35	25
V 1/Et K b (mM−1s−1)	210	140	210	120	120	270	520	190
V 2/Et K p (mM−1s−1)	660	550	880	730	620	640	1400	870
K eq (pM)c	51	24	19	35	25	62	42	29
turnover no. (s−1)d	1.6	0.73	1.5	1.4	1.3	1.2	1.6	1.0

^aKinetic constants were determined in 33mM sodium phosphate, 0.25mM EDTA buffer (pH 8) at 25 °C by varying the concentrations of both substrates and coenzymes in initial velocity studies. K _a, K _b, K _p, and K _q are the Michaelis constants for NAD⁺, benzyl alcohol, benzaldehyde, and NADH, respectively. K _ia and K _iq are the dissociation constants for NAD⁺ and NADH, respectively, and were determined by product inhibition studies. V ₁/E_t is the turnover number for alcohol oxidation and V ₂/E_t is the turnover for aldehyde reduction. Standard errors for all fitted parameters were ≤25% of the values, indicating good estimates.

^bRecombinant wild‐type enzyme.

^c K _eq is the Haldane relationship calculated from V ₁ K _p K _iq/V ₂ K _b K _ia[H⁺]

^dTurnover number determined in the standard assay at 25°C based on titration of active sites.

Transient kinetics

Stopped‐flow kinetic experiments were used to determine directly rate constants for binding of coenzymes and hydride transfer for the G173A, V197I, and I220F enzymes, which were obtained in sufficient amounts. Rate constants for association and dissociation of NADH determined from transient experiments are comparable to the rate constants calculated from the steady‐state data and to those for wild‐type enzyme (Table 3). Rate constants for hydride transfer are also similar for these enzymes, and isotope effects support the assignment of rate‐limiting steps in the mechanisms.

Table 3.

Rate Constants for NADH Binding and Hydride Transfer for Liver ADHs^a

Kinetic constant	Wild‐typeb	G173A	V197I	I220F
k on, NADH (μM−1s−1)	11, 14c	12, 5.6c	12, 19c	8.4c
k off, NADH (s−1)	5.5, 4.3c	2.0c	7.2, 4.8c	4.8c
k H, oxidation (s−1)d	24 (38)	21	19 (40)	17 (33)
k H, reduction (s−1)d	320 (310)		480 (460)	(310)

^aExperiments were carried out in 33 mM sodium phosphate, 0.25 mM EDTA buffer, pH 8, at 25 °C. Oxidation of benzyl alcohol or reduction of benzaldehyde was studied. Estimated errors for the parameters were ≤ 20%.

^bNatural enzyme.

^cValues calculated from steady‐state data: k _on = V ₂/E_t/K _q = k _–6, k _off = V ₂ K _iq/E_t K _q = k ₆, Scheme 1.

^dRate constants for hydride transfer. The first value is the observed relaxation rate constant for the exponential burst phase extrapolated to saturating concentrations of substrate, and the values in parentheses are the microscopic constants obtained by simulation of the complete mechanism. See text below and Table 4.

For the G173A ADH, the rate constant for binding of NAD⁺ is 1.0 × 10⁶ M⁻¹s⁻¹, and the rate constant for binding of NADH is 12 × 10⁶ M⁻¹s⁻¹, which are comparable to the constants determined for the wild‐type enzyme.43 The apparent rate constant for hydride transfer for oxidation of benzyl alcohol determined from the transient, exponential burst phase is 21 s⁻¹ at saturating concentrations of alcohol, followed by a linear steady‐state, turnover phase with a rate constant of 0.77 s⁻¹, controlled mostly by release of NADH. For oxidation of α,α‐d ₂‐benzyl alcohol, the burst phase has a rate constant of 5.6 s⁻¹, giving an H/D isotope effect of 3.8, consistent with the value for the wild‐type enzyme of 3.6,43 indicating that hydride transfer is limiting the exponential burst phase, while the steady‐state phase has a turnover rate constant of 0.64 s⁻¹. Overall, these kinetic data show that the G173A enzyme is very similar to the wild‐type enzyme and provide no indication that the rate constants for binding of the coenzymes, conformational changes, and hydride transfer steps are altered significantly by the substitution.

The rate constants for association and dissociation of NADH for the V197I enzyme (Table 3) are very similar to those determined for wild‐type ADH.43 Coenzyme release is the major rate‐limiting step for alcohol oxidation by V197I ADH, as it is for the wild‐type enzyme.43 Deuterium isotope effects on the steady‐state reaction with 2 mM NAD⁺ and 5–50 mM protio or α,α‐d ₂‐benzyl alcohol were 1.7 ± 0.1 on V _1/E_t and 2.2 ± 0.3 on V ₁/E_t K _b, also indicating that hydrogen transfer is only partially limiting for steady‐state turnover. For the transient oxidation of benzyl alcohol by V197I ADH, an exponential burst phase with a rate constant of 19 s⁻¹ at saturation by alcohol is followed by a linear steady‐state, turnover phase with a rate constant of 1.5 s⁻¹. For α,α‐d ₂‐benzyl alcohol, the burst phase was slower, consistent with a substantial H/D isotope effect, but the burst and steady‐state phases were not readily separated. The reduction of benzaldehyde has a maximal rate constant for the burst phase of 480 s⁻¹ and a steady‐state phase with a rate constant of 22 s⁻¹, similar to the values for the wild‐type enzyme.43

Oxidation of benzyl alcohol by I220F ADH has an exponential burst phase with a rate constant at saturating protio benzyl alcohol of 17 s⁻¹, followed by a steady‐state turnover phase with a rate constant of 1.3 s⁻¹. For α,α‐d ₂‐benzyl alcohol, the exponential phase was slower, consistent with a substantial H/D isotope effect, and not readily separated from the steady‐state phase. It appears that hydride transfer is mostly rate‐limiting for the transient burst phase of alcohol oxidation by these enzymes and that the chemical step is not affected significantly by the substitutions of the amino acids in the enzyme.

Kinetic simulation of the mechanisms for the V197I and I220F ADHs

The rate constants for hydride transfer estimated from the transient studies are apparent relaxation constants, which are starting values for simulations that can provide estimates of the microscopic rate constants for individual steps in the mechanism. The transient data for the V197I enzyme (Fig. 8) are fitted very well by the Ordered Bi Bi mechanism with the rate constants in Table 4. The V197I and wild‐type enzymes have very similar rate constants.43 The simulation shows that the microscopic rate constant for hydride transfer during benzyl alcohol oxidation (k ₃) is somewhat larger (40 s⁻¹) than the apparent relaxation constant (19 s⁻¹) estimated from the observed data. The results support the conclusions from the steady‐state kinetics and indicate that the V197I substitution has small effects on the catalytic mechanism.

Figure 8.

Simulation of transient data for the forward and reverse reactions catalyzed by V197I ADH. Reactions were carried out in 33 mM sodium phosphate buffer, 0.25 mM EDTA, pH 8, at 25°C. Oxidation of benzyl alcohol was observed with NAD⁺ at 1 mM with 50 μM (▼), 67 μM (▲), 100 μM (♦), 200 μM (▪), and 500 μM (•) of the alcohol. Reduction of benzaldehyde was observed with NADH at 60 μM with 75 μM (○), 150 μM (□), 300 μM (◊), and 750 μM (Δ) benzaldehyde. Symbols represent the transient reaction data and lines represent the simulated curves. The mean square deviation of the data from the fitted values is 2.2 × 10⁻⁵.

Table 4.

Estimated Rate Constants for Wild‐type and V197I Alcohol Dehydrogenases^a

Kinetic constant	Wild‐typeb	V197I
k 1 (M−1 s−1)c	1.2 × 106	2.3 × 106
k –1 (s−1)	56	64
k 3 (M−1 s−1)	3.7 × 106	3.5 × 106
k –2 (s−1)	58	110
k 3 (s−1)	38	40
k –3 (s−1)	310	460
k 4 (s−1)	66	69
k –4 (M−1 s−1)	0.83 × 105	0.54 × 106
k 5 (s−1)	5.5	3.6
k –5 (M−1 s−1)c	1.1 × 107	1.0 × 107
k eq (pM)	66	45

^aRate constants were determined by fitting with FITSIM the stopped‐flow data in Figure 8 to the mechanism given above. Nine progress curves were used with varied concentrations of substrates (1 mM NAD⁺ and 0.05–0.5 mM benzyl alcohol, or 60 μM NADH and 0.075–0.75 mM benzaldehyde). Errors for all fitted parameters were ≤14% and the R ² coefficient was 0.9975, which indicates a good fit, but the range of values within the confidence limits is larger as some of the values are correlated. Kinetic constants for the isomerization of the enzyme‐NAD⁺ complex for the V197I enzyme were not fitted because the binding of NAD⁺ was not studied in detail.

^bNatural enzyme.

^cBecause the concentrations of coenzymes were not varied, these values were determined separately by transient kinetics or from steady‐state kinetic constants (K _ia and K _iq) and fixed. The isomerization of the E‐NAD⁺ complex is included within the net rate constants for step 1.

The microscopic rate constants for hydride transfer were also estimated by kinetic simulation for the I220F ADH by fixing the rate constants for binding steps to those used for the wild‐type enzyme, and allowing the rate constants for the hydride transfer step to vary.43 The rate constant for the hydride transfer for benzyl alcohol oxidation was fitted to be 33 s⁻¹, as compared to the value of 38 s⁻¹ determined for the wild‐type enzyme, and the rate constant for reduction of benzaldehyde is the same as for the wild‐type ADH, at 310 s⁻¹.43 The data for α,α‐d ₂‐benzyl alcohol oxidation were fitted to give forward and reverse rate constants of 3.8 and 71 s⁻¹, consistent with the H/D isotope effect. Thus, the I220F substitution has no significant effect on the rate constants for hydride transfer.

The steady‐state and transient kinetic parameters for turnover and coenzyme binding indicate that all of these enzymes have rates of conformational changes that are similar to those for wild‐type enzyme.35

Discussion

Conservative substitutions of distal amino acid residues have small effects on structure, catalysis, and global dynamics of alcohol dehydrogenase

The overall structures and active site geometries of the ternary complexes with NAD⁺ and a substrate analogue for the wild‐type and the seven substituted enzymes are essentially the same, attesting to the stability and adaptability of the protein. The five studied residues are associated with the Rossmann coenzyme‐binding domain, but the substitutions do not alter binding of NAD⁺ or NADH, conformational changes, or any of the other kinetic constants for catalysis significantly. The substitutions do not affect the slow preorganization leading to the active ternary complex structure. Because the enzyme kinetics are not changed by the substitutions, it is not surprising that these enzymes have structures isomorphous to the wild‐type enzyme. In contrast, the V292S and G293A/P295T enzymes have altered kinetics and crystallize with coenzymes in open conformations and different space groups (see 1JU9.pdb and 1QLH.pdb). However, the distal substitutions should have affected global stability and fast, global dynamics. Computational studies suggested that rate‐promoting vibrations of a conserved evolutionary motif that includes the substituted residues are associated with catalysis, but magnitudes of the contribution were not estimated.49, 50 Our biochemical studies provide no evidence to support a significant contribution of the distal residues to the proposed rate‐promoting vibrations, or in general, of fast global dynamics, to catalysis of hydrogen transfer by ADH. These conclusions are discussed further below.

Substitutions of other distal residues in horse liver ADH also have small effects on structure and catalysis. Val‐207 is buried near Val‐203, which interacts with the nicotinamide ring of bound NAD, and both residues were suggested to participate in promoting vibrations.49 However, the V207A substitution has only small effects on structure and catalysis.58 Ile‐224 and Ile‐269 are “remote” from the active site, but interact with the adenine ring of NAD. The I224G substitution decreases affinities for coenzymes by ∼60‐fold, and the I269S substitution decreases affinities by 300–700‐fold, but neither substitution alters catalytic efficiencies for ethanol oxidation and acetaldehyde reduction, indicating that the enzyme‐NAD(H) complexes are fully competent.56 Substitutions of Ile‐224 cause only small changes in hydrogen isotope effects or tunneling.92

A triple substitution (M330F:L308I:W314L) in a hydrophobic cluster in the dimer interface in ADH increases turnover numbers, dissociation constants for coenzymes, rate constant for dissociation of NADH, and catalytic efficiencies for reactions with ethanol and acetaldehyde by 2–4‐fold.93 The apparent rate constant for hydride transfer from ethanol is also increased 1.3‐fold. (The enzymes with the single W314L or W314F substitutions were poorly expressed and could not be studied, but with the three changes, as found in the frog ADH, the enzyme was stable.) The increases in kinetic constants suggest that the conformational change and global dynamics are modestly affected by the substitutions.35, 93 The conformational change in wild‐type enzyme is relatively slow (∼ms),43, 88 but fast anti‐correlated motions of the domains of the ternary complex were suggested to contribute to the chemistry.47, 48 In the ternary complexes, the fast, pseudo‐rigid body motions of catalytic and coenzyme binding domains contribute ∼50% to the overall B‐factors for residues in the active site, as calculated from TLS analysis of the anisotropic temperature factors.58 X‐ray structures for the triply substituted enzyme are needed to distinguish between structural and dynamic effects. Nevertheless, the G173A and the F322L substitutions in the hinge region for the conformational change have small effects on structure and catalysis. A glycine to alanine substitution can affect global dynamics, as for example, the G23A substitution destabilizes RNase A T1 by ∼1.2 kcal/mol and produces NMR chemical shifts throughout the protein, without changing the structure determined by X‐ray crystallography.73

Conservative substitutions of a buried amino acid residue can alter global protein stability by perhaps 1 kcal/mol,69, 70, 71, 72 which should have an effect on global dynamics. The global stability of the mesophilic horse liver ADH has not been experimentally determined because the dimeric enzyme is not reversibly denatured, and it requires Zn(II) for renaturation.94, 95 The enzyme is moderately stable, surviving heat treatment at 52°C for 15 min,96 but not for long at higher temperatures. Enzyme activity has a half‐life of 22 min in 6 M urea at 25°C and pH 8.6.97 The amino acid residues studied here seem to be important because less conservative substitutions destabilized the protein and decreased expression. Deleterious substitutions included the following: Gly‐173 with Val, Glu, Lys and Arg; Val‐197 with Gly, Ser, and Thr; Ile‐220 with Gly, Ser, Thr, Asn, and Asp; and Val‐222 with Gly, Ser, and Thr. Substitutions with the hydrophilic residues are expected to destabilize hydrophobic clusters.98 The F322V enzyme was poorly expressed. We conclude that the buried, distal amino acid residues contribute to stablizing the appropriate scaffold for the active enzyme, and the conservative substitutions should affect global dynamics.

If the substitutions altered global stability by 1 kcal/mol, or perhaps 1/10th of the total energetics, and if this change is dynamically connected to the active site, enzyme activity might change by 5‐fold. However, the changes in enzyme kinetics were typically less than 2‐fold, and we should explain why these substitutions have only small effects on catalysis. The major reason is that the enzyme seems to accommodate conservative substitutions so that the active site scaffold is not altered. Even if a substituted protein is more flexible, most of the enzyme forms the native, active conformation. Perhaps the distal residues are not in a network connected to the reaction coordinate. However, they are in the conserved evolutionary motif that is proposed to contribute to promoting vibrations, and the “entire macromolecule is involved in the catalytic process.”2 We suggest that the results from substitutions at nine sites (five studied here and V207A, I224G, I279S and M330F:L308I:W314L), in ADH is a reasonable sample to conclude that distal residues make small contributions via fast, global motions to catalysis of hydrogen transfer in the ternary complex. Our results are consistent with other studies that show that substitutions of amino acid residues (mostly polar, however) in the second or third shell removed from the active sites in several enzymes generally have “insignificant” (< 3‐fold) effects on catalysis.5 Of course, multiple substitutions in ADH might synergistically cause larger effects on catalysis and dynamics.

Substitutions of distal amino acid residues can affect slow global dynamics, preorganization, and catalysis

Our results may be surprising because studies of other enzymes suggest that substitution of a distal residue can alter catalytic efficiencies and protein dynamics significantly, but alteration of slow conformational equilibria, structural states, and the fraction of enzyme in a productive form can explain those effects. Studies on two enzymes are informative for this discussion.

The G121V substitution in dihydrofolate reductase decreases the rate of hydride transfer by 200‐fold, binding of NADPH by 40‐fold, and the estimated rate of a conformational change in the central complex by 570‐fold.12 Computational results suggest that Gly‐121 is a distal member of a dynamic network that participates in coupled motions.11, 15 Gly‐121 is ∼15 Å from the catalytic center in a mobile loop as shown by X‐ray crystallography and NMR, and multiple conformational states have been identified for the wild‐type enzyme and its complexes.19, 99 Computations suggest that the G121V substitution decreases the fraction of enzyme in the catalytically productive conformation, probably by affecting the “M20 loop’ that is close to the nicotinamide binding site.100, 101 NMR studies suggest that less than 1% the G121V enzyme‐substrate complex is in the closed, productive conformation.19 No X‐ray structure of the G121V enzyme is available, but molecular modeling shows that the G121V substitution causes severe steric clashes that would affect a network extending to the active site.27 Nevertheless, the G121V enzyme exhibits intrinsic kinetic isotope effects with only a small temperature dependence, and thus it appears that when a productive enzyme‐substrate complex is formed, the nature of the hydride transfer step is not altered significantly.23, 102 The kinetic isotope effects are “consistent with models in which the motion in the active site is coupled to the hydride transfer coordinate”, but some calculations suggest that rate‐promoting vibrations are not coupled to the reaction coordinate.100, 103 The general conclusion is that the G121V substitution alters the slow global dynamics required for preorganization of the active site. Similarly, a “dynamic knockout” produced with the N23PP/S148A substitutions can alter conformational states without affecting the dynamics of hydride transfer.104, 105

A “remote” Y209W substitution in E. coli thymidylate synthase substantially decreases catalytic efficiency (7000‐fold), with only small changes in the structure of a relevant ternary complex, by affecting the correlated, global dynamics of the enzyme as indicated by X‐ray crystallographic anisotropic B‐factors.66 The hydroxyl group of Tyr‐209 interacts with the hydroxyl group of dUMP, which is “remote” from the site for methyl transfer, but not strictly a “distal” effect.20 The Y209W substitution apparently affects the preorganization enough so that small thiols can trap a reaction intermediate. The dUMP phosphate binding loop is shifted ∼1.0 Å, and the increased mobility indicated by the B‐factors is propagated to the loop containing Cys‐146, which participates in the chemical reaction. It appears that the fraction of the enzyme in the reactive complex is decreased, which can account for some of the loss of activity. Nevertheless, the temperature‐independent kinetic isotope effects show that the substitution has small effects on the reorganization dynamics of hydrogen transfer for the fraction of the enzyme that is in the productive form. This study shows that structural changes at the active site can affect subtle features of slow preorganization and fast protein dynamics.

Substitutions of proximal residues often change catalysis

Although substitutions of distal residues might have small effects on the activity of ADH, substitution of residues at or near the active site can affect fast protein dynamics and catalysis involved in the “reorganization” that leads to hydrogen transfer in the Michaelis complex. Schwartz and coworkers suggested that Val‐203 and Val‐292 participate in promoting vibrations that contribute to catalysis.49, 50 Val‐203 contacts the nicotinamide ring, and substitution with Ala decreases the rate constant for hydride transfer by 16‐fold and diminishes the contribution of hydrogen tunneling, which is attributed to an increased donor‐acceptor distance for hydride transfer and altered dynamics.31, 37, 46, 47, 48, 106, 107, 108 However, X‐ray crystallography of complexes of V203A ADH with NAD⁺ and fluoro alcohols show that the active site geometry is almost identical to that for the wild‐type enzyme.58 The decrease in catalysis can be attributed to altered dynamics (equilibrium or non‐equilibrium) because of increased flexibility and motion resulting from the cavity introduced behind the nicotinamide ring. However, the V203A substitution allows a water molecule to come in contact with the nicotinamide ring, and this interaction might alter the chemistry.58 The carbonyl oxygen of Val‐292 is hydrogen bonded to the carboxamido group of the nicotinamide ring, and the V292S substitution decreases the rate of hydride transfer by about 4‐fold, decreases the affinity for coenzymes by about 80‐fold, and shifts the conformational equilibrium to favor an open conformation, as characterized by X‐ray crystallography (1JU9.pdb). Nevertheless, the temperature‐independent isotope effects for benzyl alcohol oxidation and benzaldehyde reduction for the V292S ADH are consistent with hydride transfer proceeding with vibrationally assisted tunneling for the fraction of enzyme that is in the productive conformation.35, 39 The V292A and V292T substitutions also decrease affinity for coenzymes and decrease the apparent rate constant for hydride transfer from benzyl alcohol by about 4‐fold, while the V292T ADH–NAD–pyrazole complex crystallizes in the closed conformation (but accompanied by the introduction of new water molecule, 1N8K.pdb), indicating that the enzyme can form a ternary complex similar to that of the wild‐type enzyme.57 The side chain of Thr‐178 makes van der Waals contacts with the nicotinamide ring, and the T178S substitution decreases the rate of hydride transfer by 9‐fold without affecting coenzyme binding significantly, whereas the T178V substitution substantially decreases affinity for coenzymes without affecting the rate of hydride transfer from benzyl alcohol.57

When the enzyme binds coenzyme, a flexible loop involving residues 293–298 changes conformation, and the peptide NH of Val‐295 forms a hydrogen bond with O3D of the nicotinamide ribose. The double substitution, G293A/P295T, keeps the enzyme in the open conformation when coenzyme binds (1QLH.pdb), and Michaelis and dissociation constants increase 40–2000‐fold, while catalytic efficiencies (V/K _m for ethanol and acetaldehyde) decrease by 50–1400‐fold and hydride transfer rates decrease by about 30‐fold.63 These effects are similar to those observed when Lys‐228, which interacts with the O3B hydroxyl group of the adenosine ribose moiety, is modified by picolinimidylation.35, 109

Substitutions of residues that are near the nicotinamide ring in other enzymes also can affect catalytic dynamics. Substituting Ile‐14 in dihydrofolate reductase with smaller residues, Val, Ala, and Gly, substantially decreases hydride transfer rate constants (to 0.14, 0.025, and 0.001 of that of wild‐type, respectively) and gives rise to temperature‐dependent kinetic isotope effects, which are explained by QM/MM simulations and X‐ray crystallographic studies to indicate increased flexibility in the active site, even though the average hydride donor‐acceptor distances are similar.110, 111

Fast local dynamics in catalysis

Computational studies suggest that Val‐203 and Val‐292 in horse liver ADH and Val‐136 in lactate dehydrogenase participate via “rate‐promoting vibrations” (“nonequilibrium” or nonstatistical dynamics) that are on the same time scale as the hydrogen transfer.42, 54 However, other studies show that “equilibrium” motions can explain the dynamics.28, 32, 46, 112 In either case, the changes in hydrogen transfer dynamics might be explained by alteration of fast local motions involved in reorganization to form the tunneling ready state.58, 62, 113 These motions include the vibrations, rotations and translations of proximal residues that result in a favorable configuration for hydride transfer.

X‐ray crystallographic B‐factors provide information about these local motions and vibrations on a fast (∼ps) time scale, which are still evident at 100 K. After removing contributions to the observed B‐factors from crystal lattice disorder and translations and rotations of pseudo‐rigid domains (TLS refinement), and extrapolating from 100 K to 300 K, “residual”, local displacements of about 0.5 Å at 300 K can be estimated for atoms at the active site.58, 64, 114 In ADH, these equilibrium motions can account for the decrease in donor‐acceptor distance (C4N of NAD⁺ to C7 of benzyl alcohol) from 3.4 Å in the ground state to 3.2 Å in the tunneling ready state62 and to ∼2.7 Å for the hydrogen transfer.46 The relative motions of the substrates and residues involved in catalysis are illustrated in Figure 9 as the sizes of the atoms. The amplitudes are about the same as those calculated by molecular dynamics simulations.46, 112, 115

Figure 9.

Model of a Michaelis complex for liver alcohol dehydrogenase with NAD⁺ and benzyl alcohol showing the pro‐R hydrogen in position for hydride transfer (H^–) with atomic radii approximating the local equilibrium motions derived from X‐ray crystallographic atomic displacement factors. Atom colors: C, gray; O, red; N, blue; H, cyan, P, green; Zn, violet. The oxygen of benzyl alcohol is bound to the zinc and to OG of Ser48 in the hydrogen bond network including O2D of the ribose and NE2 of His51, which acts as the proton acceptor. The model is derived from 4DWV.pdb.

Although the data for this study were collected at cryogenic temperatures (85 or 100 K), a structure of the enzyme–NAD⁺–pentafluorobenzyl alcohol complex determined to 2.1 Å at 278 K (1HLD.pdb116) is essentially the same as the structure determined to 1.14 Å at 100 K (4DWV.pdb81), with an RMSD for superposition of all α‐carbon atoms of 0.37 Å. The data at high resolution and low temperature permit modeling of many alternative amino acid side chain conformations, which are likely to be present at 25°C or in vivo. Most of the alternative conformations are for hydrophilic residues on the surface of the protein, but leucine residues 57 and 309 in the substrate binding site have alternative conformations (See Fig. 2 in Ref. 81). Combining the dynamics for conformational changes of proximal residues in the ensemble of structures involved in catalysis with the fast internal dynamics would lead to average displacements even larger than those estimated from the data collected at cryogenic temperatures.

The critical question is how much does a distal amino acid residue contribute to catalysis by ADH? Conservative substitutions of a distal residue that do not alter the global structure or conformational changes of the enzyme might change activity by 2‐fold, and the present studies show some effects of this magnitude. Combining three or four such changes could result in the 10‐fold effects observed for isoenzymes of yeast and horse ADHs. In contrast, conservative substitutions of one amino acid residue in the active site can change the rate constant for hydride transfer by 10‐fold or more, and combining the contributions of several amino acid residues in the active site might account for the enormous catalytic enhancement as compared to the non‐enzymatic reaction. We conclude that fast global motions have small effects on catalysis, whereas local dynamics within the active site can be substantial.

Materials and Methods

Reagents

LiNAD⁺ and Na₂NADH were purchased from Roche Molecular Biochemicals (Indianapolis, IN). Benzyl alcohol‐α,α‐d ₂ (98.6% D) was purchased from MSD Isotopes (Woburn, MA). Benzyl alcohol and benzaldehyde were redistilled before use. The fluoro alcohols (98–99%) and pyrazole were purchased from Sigma‐Aldrich (St. Louis, MO) and used without further purification. 2‐Methyl‐2,4‐pentanediol (MPD) was obtained from Kodak (Rochester, NY) and treated with activated charcoal before use.

Enzyme preparation

The plasmid pBPP/EqADH117 containing the cDNA for horse liver ADH1E (Equus caballus, NCBI taxonomy ID 9796, GenBank accession number M64864) was used to create the substitutions, generally using the Stratagene (La Jolla, CA) Quick Change method. After the PCR reaction and Dpn I digestion, the dsDNA was transfected into XL1‐Blue super‐competent cells. The candidate clones were selected for ampicillin and tetracycline resistance. Plasmids were isolated, and the mutations were identified by sequencing the DNA at The University of Iowa DNA Facility. Expression and activity of the recombinant proteins were judged by Coomassie blue and enzyme activity staining of a nondenaturing 1% agarose gel after electrophoresis at pH 8.0, where the ADH migrates toward the cathode, and the E. coli proteins migrate mostly to the anode.117 The enzymes were purified to apparent homogeneity according to the published procedure.117

The G173A substitution was prepared by partially random site‐directed mutagenesis of the cDNA by D. H. Park with the Amersham kit118, 119 and a degenerate primer synthesized in The University of Iowa DNA Facility with the sequence: C TGT CTC ATT (A/C/G)(A/T/C)G TGT GGA TTT T (where the underlines mark the sites of mutation), which produced the substitutions coding for Ala, Glu, and Val, but could also have produced the Thr, Gln, Pro, Leu, Met, and Lys enzymes. Use of another primer produced the Arg and Lys substitutions. Only the G173A enzyme was expressed well.

The V197I substitution was prepared by partially random mutagenesis with a degenerate primer from Integrated DNA Technologies, Inc. (IDT, Coralville, IA) for the forward direction (complementary, reverse primers implied): CC CAG GGC TCC ACC TGT GCC (A/G)(C/G/T)C TTT GGC CTT GGA GGA G. The mutations could have resulted in enzymes with Gly, Ala, Ser, Thr, Val, or Ile substitutions, but the Gly, Ser, and Thr enzymes were poorly expressed, and the Ile enzyme was moderately expressed as compared to the wild‐type Val enzyme.

The I220X substitutions were prepared with degenerate oligonucleotides synthesized by IDT with two different sets of primers with the following sequences for the forward primers (complementary, reverse primers implied): 5′‐CG GCC AGG ATC (A/G)(A/C/G)C GGG GTG GAC ATC AAC A‐3′, and 5′‐CG GCC AGG ATC (A/C/T/G)(T/G)C GGG GTG GAC ATC AAC A‐3′. The site‐directed mutagenesis provided eight of the 11 possible substitutions. The proteins with Gly, Ser, Thr, Asn, and Asp substitutions were poorly expressed. The I220F, I220I, and I220V ADHs were expressed and purified.

The V222I substitution was prepared with a degenerate primer (IDT) with the sequence: GGA GCG GCC AGG ATC ATT GGG (A/G)(C/G/T)C GAC ATC AAC AAA GAC. Enzyme with the same six substitutions as for Val‐197 could have been obtained. The Gly, Ser, and Thr constructs were obtained, but were not expressed well, whereas the V222I enzyme had good expression.

The F322L substitution was prepared with a degenerate primer (IDT) with the sequence: GCT ATT TTT GGC (A/C/G)TC AAG AGT AAA G, and the complementary reverse primer, which could produce Ile, Leu, and Val substitutions. The F322I and F322L enzymes were expressed reasonably well, whereas the F322V enzyme was expressed poorly and had less activity.

Crystallization

Crystals of the recombinant enzymes with NAD⁺ and 2,3,4,5,6‐pentafluorobenzyl were prepared by the procedure used for the wild‐type enzyme.81, 116 The enzyme was first dialyzed extensively against 50 mM ammonium N‐[tris(hydroxymethyl)methyl]‐2‐aminomethane sulfonate, 0.25 mM EDTA, pH 7.0 buffer at 5°C (pH 6.7 at 25°C), and any precipitate was removed by centrifugation. The enzyme was adjusted to 10 mg/ml (A₂₈₀ = 4.55/cm, but diluted for the measurement), and 0.5–1.0 ml was put into ¼ in. diameter dialysis tubing (washed by boiling in dilute sodium bicarbonate and EDTA solution and rinsed with water) and put into 10 ml of the buffer, to which 2,3,4,5,6‐pentafluorobenzyl alcohol (20 mg to make ∼10 mM) and LiNAD⁺ (8 mg to yield ∼1 mM) was added. After an hour (time for the fluorobenzyl alcohol and NAD⁺ to bind to the enzyme), 1.1 ml 2‐methyl‐2,4‐pentanediol (MPD) was added. The concentration of MPD was further increased over some days to about 12%, when crystals formed, and slowly raised to a final concentration of 25%, which is sufficient for cryoprotection at 100 K. The concentrations of NAD⁺ and pentafluorobenzyl alcohol saturate these enzymes, as the K _d for NAD⁺ is < 40 μM, and the K _i for pentafluorobenzyl alcohol is < 1 μM. For the V197I enzyme 19 mM pentafluorobenzyl was used, and for the V222I enzyme the concentration of MPD was 35%. The crystals were mounted on fiber loops (Hampton Research) and flash‐vitrified by being plunged into liquid nitrogen.

X‐ray crystallography

The data for the G173A enzyme ternary complex were collected at the Advanced Photon Source at Argonne National Laboratories on beamline 19‐ID with wavelength of 0.9184 Å at 85 K with the helium cryostat (with special thanks to Dr. Stephan Ginell). The data for the V197I and V222I enzyme ternary complexes were collected at the Advanced Light Source on the Molecular Biology Consortium beamline (ALS 4.2.2) in Berkeley, CA at 100 K with a wavelength of 0.827 Å. Data were also collected at ALS beamline 4.2.2 for I220L and I220F, and I220V ADHs at 100 K with a wavelength of 0.800 Å. The data for the F322L enzyme ternary complex were collected at ALS at 100 K with a wavelength of 0.8865 Å.

Collection of data over 360° for the triclinic crystals provided 3–4‐fold redundancy for one crystal. Data were processed with d*TREK.120 The high resolution shell included data with <I>/σ<I> ≥ 2 and ≥ 50% completion. Three data sets had 83–86% completeness over the full range of resolution, but the data redundancy was good, and the electron density maps provided high quality structures. The structures were solved by molecular replacement with the isomorphous wild‐type enzyme (4DWV.pdb) and refined with REFMAC5.121 R _free was calculated with at least 1000 reflections.122 Model bias was avoided by changing the wild‐type residue to alanine (except for the G173A enzyme) and removing the pentafluorobenzyl alcohol from the model. The dictionary for NAD⁺ was modified to relax the restraints on bond angles and distances so that the geometry of the nicotinamide ring in the complex could be properly fitted, and thus the name in the coordinate files is “NAJ.”57, 81 Riding hydrogens were added, and anisotropic displacement parameters were refined for all structures, except for the I220V enzyme. After each refinement cycle, the models were rebuilt manually using the program “O” using 2|F _o| – |F _c| and |F _o| – |F _c| difference maps.123 Several residues were modeled with alternative conformations where electron density indicated. Water molecules were included where there was well‐formed electron density, B‐factors were less than about 50 Ų, and hydrogen‐bonds connected them to the protein.

The quality of the refinement was monitored with PROCHECK and SFCHECK,121 MolProbity,124 and PARVATI.125 Stereo figures were made with the Molray web interface server in Uppsala, Sweden.126

Steady‐state kinetics

The concentration of active sites was determined by titration with NAD⁺ in the presence of pyrazole.127 A standard assay was used to determine the specific activity for the enzymes and is reported as the turnover number based on the titration of active sites.109 The concentrations of the coenzymes NAD⁺ and NADH were determined by absorbance at 260 and 340 nm, respectively. Substrate concentrations of benzyl alcohol and benzaldehyde were determined by absorbance at 257 nm (ɛ = 200 M⁻¹cm⁻¹) and 249 nm (ɛ = 12,500 M⁻¹cm⁻¹), respectively. All assays were performed in 33 mM sodium phosphate, 0.25 mM EDTA, buffer at pH 8 and 25°C. Kinetics were determined on a Cary 118C spectrophotometer or a Fluorolog FL‐3 fluorometer (λ_ex = 340 nm, λ_em = 460 nm) with computer fitting of the progress curves to a linear or parabolic function to obtain initial velocities. Steady‐state initial velocity experiments for the forward and reverse reactions used a systematically varied 5X5 matrix of concentrations of both substrates and coenzymes (duplicate assays) in order to obtain true K _m and V _max values, and data were fitted to Cleland's programs for a sequential mechanism (for NAD⁺ and benzyl alcohol) or a ping pong mechanism (for NADH and benzaldehyde, because K _iq is small relative to K _q).128 Product inhibition was also used to obtain the inhibition (dissociation) constants for coenzymes, and data were fitted to the equations for competitive or noncompetitive inhibition. The fits gave errors of the estimates of < 25%, indicating good fits. Kinetic constants usually were determined two to four times and averaged. Differences between the wild‐type and substituted enzymes of more than two‐fold may be significant at a P of < 0.01, but are not remarkable for the overall enhancement of the reaction due to enzyme catalysis. The kinetics for all enzymes fit an Ordered Bi Bi mechanism.87

Transient kinetics

A BioLogic SFM3 stopped‐flow instrument (dead time = 2.3 ms) was used. The rate constant for NAD⁺ binding to the G173A enzyme was determined by mixing 2 μN enzyme (active site concentration) with 17–50 μM NAD⁺ and 10 mM pyrazole (to form the tight ternary complex) and monitoring the decrease in tryptophan protein fluorescence with excitation at 295 nm and emission in the range of 310–385 nm. The transient data were fitted to a first order reaction, and the bimolecular rate constant was calculated from the dependence of observed rate constants on NAD⁺ concentration. Similarly, the rate constants for NADH binding to the G173A and V197I enzymes were determined with 2 μN enzyme mixed with 1.7–12 μM (5–20 μM for V197I ADH) NADH and 10 mM isobutyramide (to form the tight ternary complex) by monitoring protein fluorescence quenching and fitting the observed rate constants against NADH concentration. The rate constant for dissociation of NADH from the enzyme‐NADH complex for V197I ADH was determined by mixing 9.4 μN enzyme and 15 μM NADH with 0.1–0.5 mM NAD⁺ and 10 mM pyrazole, which traps free enzyme as the enzyme‐NAD⁺‐pyrazole complex that absorbs at 294 nm, after NADH dissociates. The concentration of NAD⁺ was varied to ensure that the dissociation is irreversible.

Transient reactions of benzyl alcohol oxidation and benzaldehyde reduction were measured by the change in absorption at 332 nm, which is the isosbestic point for free and enzyme bound NADH (ɛ = 5,500 M⁻¹cm⁻¹). Enzyme concentrations were 10–15 μN with 1 mM NAD⁺ and varied concentrations of benzyl alcohol in the range of 0.05–0.5 mM benzyl or 60 μM NADH and 0.075–0.75 mM benzaldehyde. The transient “burst” data were fitted to an equation for first order kinetics followed by the steady‐state phase by the BioKine software, and the maximum rate constants were determined by fitting to the Michaelis–Menten equation. KINSIM and FITSIM129 were used to simulate and fit selected progress curves simultaneously to obtain rate constants for each step in the Ordered Bi mechanism.