The side chains of Asn191 and Asn300 constitute a characteristic structural motif of the active site of Pseudomonas fluorescens mannitol 2-dehydrogenase that lacks precedent in known alcohol dehydrogenases and resembles the canonical oxyanion binding pocket of serine proteases. We have used steady-state and transient kinetic studies of the effects of varied pH and deuterium isotopic substitutions in substrates and solvent on the enzymatic rates to delineate catalytic consequences resulting from individual and combined replacements of the two asparagine residues by alanine. The rate constants for the overall hydride transfer to and from C-2 of mannitol, which were estimated as ~ 5×102 s−1 and ~ 1.5×103 s−1 in the wild-type enzyme respectively, were selectively slowed, between 540- and 2700-fold, in single-site mannitol 2-dehydrogenase mutants. These effects were additive in the corresponding doubly mutated enzyme, suggesting independent functioning of the two asparagine residues in catalysis. Partial disruption of the oxyanion hole in single-site mutants caused an upshift, by ≥1.2 pH units, in the kinetic pK of the catalytic acid-base Lys295 in the enzyme–NAD+–mannitol complex. The oxyanion hole of mannitol 2-dehydrogenase is suggested to drive a precatalytic conformational equilibrium at the ternary complex level in which the reactive group of the substrate is ‘activated’ for chemical conversion through its precise alignment with the unprotonated side chain of Lys295 (mannitol oxidation) and C=O bond polarization by the carboxamide moieties of Asn191 and Asn300 (fructose reduction). In the subsequent hydride transfer step, the two asparagine residues provide ~ 40 kJ/mol of electrostatic stabilization.
- alcohol dehydrogenase
- electrostatic stabilization
- kinetic isotope effect
- oxyanion hole
- polyol-specific long-chain dehydrogenases and reductases
- stopped-flow kinetics
Stabilization of the partial negative charge formed on oxygen in intermediates or in the transition state of a chemical reaction is a common source of catalytic power in natural enzymes [1–7]. The particular task fulfilled in catalysis is often reflected by active-site preorganization in the form of a so-called oxyanion hole . A prototypical oxyanion-binding site is that of the serine protease chymotrypsin where two main-chain amide NH groups form hydrogen bonds with the oxygen atom on the substrate that develops the partial negative charge [5,9,10]. Structural characteristics of the chymotrypsin oxyanion hole are remarkably conserved in various enzymes spanning four EC classes [11–17] (see Supplementary Table S1 at http://www.BiochemJ.org/bj/425/bj4250455add.htm). Despite fundamental differences in the chemical transformations catalysed, the reaction co-ordinates for these enzymes are usually united by the occurrence of a distinct, often tetrahedral, oxyanionic intermediate [5,17–20]. Evolution of a common active-site structural motif would therefore seem to reflect the shared catalytic task of stabilizing this intermediate. The proposed mechanism of aldehyde dehydrogenase, for example, proceeds through an anionic thiohemiacetal that is formed upon nucleophilic attack from the thiolate of a catalytic cysteine on the carbonyl carbon of the aldehyde substrate. This thiohemiacetal is subsequently oxidized to a thioester, which decomposes hydrolytically to the acid product . However, unlike aldehyde dehydrogenase and except in Zn-dependent ADHs (alcohol dehydrogenases) [21–23], oxidation of an alcohol substrate by enzyme-bound NAD(P)+ does not involve the clear intermediacy of an oxyanion species. The protease-like oxyanion-binding site in PfM2DH (Pseudomonas fluorescens D-mannitol 2-dehydrogenase) was therefore not anticipated. This site has no precedence in known ADHs and, despite the striking similarity in geometric arrangement of the interacting groups, its role in the catalytic mechanism of PfM2DH cannot be inferred from well-characterized oxyanion holes of other enzymes, some of which are listed in Supplementary Table S1.
Catalytic features of the PfM2DH active site (Figure 1) are conserved in PSLDRs (polyol-specific long-chain dehydrogenases/reductases), a large family of metal-independent ADHs that function in microbial sugar metabolism [24,25]. The three-dimensional structure of a ternary complex of PfM2DH bound to NAD+ and mannitol showed the carboxamide groups of Asn191 and Asn300 within ≤3 Å (1 Å=0.1 nm) distance to the reactive hydroxyl at C-2 of mannitol and the ε-amino group of Lys295 (Figure 1) . In the proposed mechanism of PfM2DH, Lys295 functions as the catalytic base that facilitates hydride transfer to NAD+ by (partial) proton abstraction from alcohol (Scheme 1A) . Precise alignment of Lys295 and the reactive 2-OH of the substrate is achieved through a conformational change that occurs at the level of the ternary complex and is observable as a kinetic isomerization step (k3 and k4 in Scheme 1B) in rapid equilibrium [20,26]. Lys295, which has a pK of 9.2 in enzyme–NAD+, becomes deprotonated, hence primed for catalysis in the precatalytic equilibrium (K2 in Scheme 1B). Conversion of mannitol-bound PfM2DH into the enzyme form that reacts to give product (fructose) is therefore strongly pH-dependent. It is made irreversible above pH 10.0 due to deprotonation of the group comprising the unprotonated side chain of Lys295 and the 2-OH of the substrate [20,26].
Little is known about how the oxyanion holes of PfM2DH and related PSLDRs contribute to catalytic efficiency in enzymatic oxidoreduction of polyol and ketose substrates. Site-directed replacement of Asn300 in PfM2DH by alanine resulted in partial kinetic unmasking of the hydride transfer step, suggesting an auxiliary role for the asparagine in the chemical event of the overall transformation . Using detailed analysis of kinetic consequences in this (N300A) and relevant further variants of PfM2DH, we have explored here the tasks fulfilled by Asn191 and Asn300 in the NAD(H)-dependent interconversion of mannitol and fructose catalysed by the enzyme. Evidence is presented that electrostatic stabilization by the oxyanion hole does not only facilitate the catalytic step of hydride transfer but also ‘activates’ the reactive groups on Lys295 and mannitol in the preceding conformational equilibrium. The triad of residues comprising Asn191, Lys295 and Asn300 appear to fulfill a catalytic function in PfM2DH similar to that of Zn2+ in liver ADH [23,28,29].
Materials and chemicals used, the preparation of enzymes, and details of the kinetic experiments are described in the Supplementary Methods section (at http://www.BiochemJ.org/bj/425/bj4250455add.htm).
Site-directed mutagenesis, protein production and characterization
Mutations leading to substitutions of Asn191 by alanine (N191A) or leucine (N191L) and both Asn191 and Asn300 by alanine (N191A/N300A) were introduced by using the inverse PCR protocol previously employed in constructing the gene for N300A . The following oligonucleotide primers were used (mismatched bases are underlined): 5′-TGCGATGCCCTGCCCCACAAT-3′ for N191A; and 5′-TGCGATCTCCTGCCCCACAAT-3′ for N191L. The universal second primer had the sequence 5′-GGACATCACGGTAAACGC-3′. The primer encoding the substitution N191A and template plasmid carrying the gene for N300A were used in constructing the gene for N191A/N300A. DNA sequencing of entire genes was used to confirm that the desired mutation(s) had been introduced and no substitutions resulting from DNA polymerase errors had occurred.
Mutated genes were expressed in Escherichia coli JM109 and recombinant proteins were purified to apparent homogeneity according to published protocols  (see Supplementary Figure S1 at http://www.BiochemJ.org/bj/425/bj4250455add.htm). CD spectra and fluorescence emission spectra of isolated proteins were recorded using reported methods . Dissociation constants for NAD+ binding (KdNAD+) to wild-type and mutated enzymes were determined by fluorescence titration, as described previously .
Steady-state kinetic analysis
Initial reaction rates were recorded spectrophotometrically at 25 °C, measuring the change in absorbance of NADH at 340 nm (ε340=6.22 mM−1·cm−1) . Apparent enzyme kinetic parameters (kcat, Ksubstrate, Kcoenzyme) were obtained from data recorded at varied substrate (coenzyme) concentration as described elsewhere [20,27,30]. We use subscript O and R on kcat to indicate mannitol oxidation (kcatO) and fructose reduction (kcatR) respectively.
The pH dependencies of kinetic parameters were determined in the pH range 5.2–10.5 for mannitol oxidation and 7.1–10.0 for fructose reduction. A three-component buffer, composed of MES, Tris and glycine, was used that displayed a pH-independent ionic strength of 0.1 M . KIEs (kinetic isotope effects) that result from deuteration of substrate (2-[2H]-mannitol; S-4-[2H]-NADH) or solvent were determined using procedures described in previous studies of wild-type PfM2DH [20,27,30]. A nomenclature is used where superscript D describes the primary deuterium KIE and superscript 2H2O describes the solvent KIE on the respective isotope-sensitive parameter . DKIEs on kinetic constants for mannitol oxidation were obtained in the pH range 6.0–10.5.
Transient kinetic analysis
Stopped-flow kinetic experiments were performed as described previously  using NADH absorbance (at 340 nm) to monitor the progress of the enzymatic reaction. The Applied Photophysics Reaction Analyser (model SX.18 MV) was equipped with a 20 μl flow cell (pathlength=1 cm) and showed a dead time of 1.5 ms under the conditions used. Each reaction was performed in at least seven repetitions. The resulting time courses were analysed as described in detail in the Supplementary Results (including Supplementary Figures S2 and S3 and Table S4 at http://www.BiochemJ.org/bj/425/bj4250455add.htm). Enzyme was used in a concentration equal to or smaller than one-seventh of the concentration of the limiting substrate (for details, see Supplementary Methods). The levels of substrate and coenzyme were chosen to achieve a maximum degree of enzyme saturation (wild-type, ≥75%; mutants, ≥63%) in the steady state. Suitable controls were obtained by mixing reaction solutions lacking the enzyme.
Proton release measurements were done using the pH indicator Phenol Red, whose absorbance at 556 nm was recorded. Wild-type enzyme and N191L were gel-filtered twice to a 0.5 mM Tris/HCl buffer, pH 8.0, containing 3.2 μM Phenol Red, NAD+ (wild-type enzyme: 3.0 mM; N191L: 20 mM) and NaCl such that the final ionic strength of this buffer was that of the reference (100 mM Tris/HCl, pH 8.0). The enzyme solution was rapidly mixed with a mannitol dissolved in the same buffer (wild-type enzyme: 200 mM; N191L: 260 mM). Proton concentrations were determined as described previously .
Sigma Plot 2004 software (version 9.01) was used for non-linear least squares regression analysis. Eqn (1) describes pH dependencies where the activity (Y=kcat, kcat/Ksubstrate, kobs) is constant at high pH and decreases below pK. C is the pH-independent value of Y at the optimum state of protonation.Eqn (2) describes a sigmoidal pH dependence with constant values of Y at high (CH) and low (CL) pH. Eqn (3) describes a decrease of logY at low pH where the curve is flattened in the region of the apparent pK and therefore has a hollow appearance. There are three ionization constants (K) associated with this pH dependence, provided that pK2>pK1>pK3. KIEs on kcat and kcat/Km were obtained from fits of Eqn (4) to the data . EV and EV/K are the isotope effects minus 1 on kcat and kcat/Km respectively, and Fi is the fraction of deuterium in the substrate or solvent.(1) (2) (3) (4)
Properties of site-directed mutants
The spectroscopic signature of isolated PfM2DH mutants obtained by using CD and fluorescence was highly similar to that of the wild-type enzyme (see Supplementary Figure S1). NAD+-binding affinity (KdNAD+) was not as strongly affected as a result of the mutations (Table 1). Therefore, similar to N300A , the new mutants (N191A, N191L, N191A/N300A) appear to be correctly folded and retain a functional coenzyme binding site. Table 1 compares the different enzymes using apparent kinetic parameters determined under optimum pH conditions for oxidation (pH 10.0) and reduction (pH 7.1). Each single-site mutation as well as the combination of the two alanine substitutions in N191A/N300A caused a large decrease in kcat/Ksubstrate for either direction of the reaction, as compared with the corresponding catalytic efficiencies of the wild-type enzyme, by 3 and 7 orders of magnitude respectively. Loss in kcat/Ksubstrate was due to substantial (up to 1000-fold) changes in kcat (decrease) and Ksubstrate (increase). kcat/KM for the coenzyme was also affected by the mutations, however, much less so than kcat/Ksubstrate.
pH-dependencies and kinetic isotope effects
pH profiles of kcat and kcat/Ksubstrate were determined for each of the singly mutated enzymes and are shown in a double-log plot (Figure 2) along with pH dependencies of relevant stopped-flow rate constants (kobs) for wild-type PfM2DH. Determination of kobs is described below. The pH dependencies of kcatO/Kmannitol  and kcatR/Kfructose  for the wild-type enzyme have been reported previously and are therefore not shown in Figure 2. However, relevant pK values from previous studies are shown in Table 2 to facilitate the comparison of native and mutated enzymes. N191A precipitated in the assay for mannitol oxidation at pH ≤6.5, restricting the investigatable pH range for this mutant. No kcatO values are available for N300A at pH ≤6.5 because saturation of the enzyme with mannitol was not achievable under these conditions. Fits of the data are displayed in Figure 2, and the resulting parameters are summarized in Table 2.
Supplementary Table S5 (at http://www.BiochemJ.org/bj/425/bj4250455add.htm) displays DKIEs on kinetic parameters for each direction of reaction catalysed by mutated enzymes. This Table also shows the corresponding DKIEs on kinetic parameters for the wild-type enzyme [20,30]. The 2H2OKIE on kcatO and kcatO/Kmannitol for N191L was determined at pL 10.0, and values of 1.1±0.1 and 3.6±0.4 were obtained respectively.
Figure 3 shows representative stopped-flow progress curves for mannitol oxidation and fructose reduction catalysed by wild-type and mutated enzymes. Reaction time courses for the wild-type PfM2DH in the pH range 7.1–10.0 were characterized by a transient burst of formation or consumption of NADH that was followed by the linear steady-state phase. Corresponding time courses for the mutated enzymes were linear (Figure 3), and the steady-state rate constants (kss) calculated from the data agreed very well with kcat values obtained in conventional initial-rate measurements. We confirmed using wild-type PfM2DH and N191L that transient phase behaviour was independent of whether free enzyme or a pre-formed binary complex of enzyme and coenzyme (for N191L: enzyme–mannitol; see the Discussion section) was rapidly mixed with the corresponding reactant solution (results not shown).
Despite the fact that ~ 90% and ~ 80% of the pre-steady burst in oxidation and reduction took place in the dead-time of the stopped-flow analyser, respectively, we were able to obtain reliable estimates for kobsO and kobsR, as described in detail in the Supplementary Results. kobsO (at 25 °C and pH 10.0) was 1.5±0.1×103 s−1 and kobsR (at 25 °C and pH 7.1) was 4±1×102 s−1. The value for kobsO was confirmed using analysis of the temperature dependence of the stopped-flow rate constant in the range 11–25 °C (see Supplementary Figure S4 at http://www.BiochemJ.org/bj/425/bj4250455add.htm).
Figure 2 (panel A) shows that kobsR was not pH dependent in the studied pH range. This result immediately suggests that the hydride transfer leading to oxidation of NADH by the wild-type enzyme is independent of pH, consistent with implications of a previous analysis of the pH dependence of DkcatR/Kfructose . kobsO, by contrast, displayed a decrease at low pH (Figure 2A). A fit of the pH profile with Eqn (1) yielded a pK of 6.8±0.4. Take note of the change in pK for the pH dependence of kobsO as compared with the pH dependence of kcatO/Kmannitol (Table 2).
Transient kinetic analysis of proton release during enzymatic oxidation of mannitol
Supplementary Figure S5 (at http://www.BiochemJ.org/bj/425/bj4250455add.htm) shows stopped-flow progress curves for the decrease in bulk pH from its initial value of 8.0 during oxidation of mannitol by wild-type and N191L forms of PfM2DH. As expected from the stoichiometry of the overall reaction catalysed, D-mannitol+NAD+↔D-fructose+NADH+H+, one mole proton was formed for each mole mannitol oxidised by either enzyme in the steady state. In the transient kinetic phase for wild-type PfM2DH, the change in total proton concentration corresponded to 1.7±0.2 times the molar equivalent of enzyme present. Proton release was kinetically coupled to formation of 0.6±0.1 moles NADH/mole enzyme in the pre-steady state, indicating that ~1.0 (=1.7−0.6) mole proton/mole enzyme are not accounted for by the chemical reaction. Deprotonation of enzyme–NAD+ as result of binding of mannitol explains the release of ‘extra protons’ in the reaction of the wild-type PfM2DH. N191L, by contrast, released only 0.45±0.05 moles proton/mole of enzyme rapidly. N300A did not release protons prior to formation of NADH in the stopped flow experiment at pH 8.0 (results not shown).
The degree of concertedness in the transfer of hydride and proton during enzymatic oxidation of alcohol by NAD(P)+ determines the development of excess negative charge on oxygen, relative to carbon, in the course of the chemical reaction. The catalytic mechanism of Zn2+-dependent ADH presents an extreme case where electrostatic stabilization by the active-site metal promotes a stepwise transformation in which an alcoholate intermediate undergoes oxidation by hydride abstraction [23,28,29]. General base catalysis from a protein-derived group, as seen in various metal-independent ADHs [35,36], would seem compatible with different levels of concertedness in the hydrogen transfer steps but is unlikely to facilitate a fully stepwise process. Evidence is presented here that establishes the oxyanion hole of PfM2DH as a novel structural motif for electrostatic stabilization in ADH active sites. While we do not propose a completely developed oxyanion in PfM2DH as in Zn2+-dependent ADH , partial abstraction of the alcohol proton prior to hydride transfer presents an important catalytic factor in this enzyme and is facilitated by the residues of the binding pocket, Asn191 and Asn300.
Role of oxyanion hole residues for catalytic hydride transfer by PfM2DH
Observation of a pre-steady state burst of product formation in either direction of the reaction and the characteristic pattern of DKIEs where Dkcat<Dkcat/Ksubstrate while Dkcat/KNAD(H)≈1 supports previous evidence [20,30], suggesting that in wild-type PfM2DH dissociation of the second product, NADH or NAD+, controls the rates (kcat) of mannitol oxidation or fructose reduction at the respective optimum pH of 10.0 and 7.1. Transient kinetic data further show that proton exchange with solvent as result of the chemical reaction occurs prior to the rate-limiting dissociation of coenzyme. Elimination of ‘burst kinetics’ in the mutated enzymes together with the finding that Dkcat was now equal to Dkcat/Ksubstrate clearly locates rate limitation in their catalytic steps. Comparison of kobs (wild-type) and kcat (or kss; single point mutants) reveals that each site-directed replacement resulted in a drastic (5×102–3×103-fold) impaired reaction rate constant. It is interesting that a decrease in kobs as large as this was not accompanied by a significant unmasking of the hydride transfer step, detectable as an increase in Dkcat/Ksubstrate for the mutated enzymes as compared with the corresponding DKIE for wild-type PfM2DH (see Supplementary Table S5). A plausible explanation is that substitution of either Asn in the oxyanion hole not only affects the actual hydride transfer but also causes slowing down of the isomerization step (k3 and k4 in Scheme 1B) that precedes catalysis in the oxidation direction.
Using the relationships ΔΔGT‡=−RT ln[(kcat/Ksubstrate)mutant/(kcat/Ksubstrate)wild-type] and ΔΔG‡=−RT ln[kcatmutant/kobswild-type], catalytic defects resulting from replacement of one or both asparagine residues can be expressed as losses in transition state-stabilising energy (ΔΔGT‡) and activation free energy for the reaction (ΔΔG‡). The calculated ΔΔGT‡ values (see Supplementary Table S6 at http://www.BiochemJ.org/bj/425/bj4250455add.htm) for N→A mutants were similar and additive in the doubly mutated enzyme N191A/N300A, suggesting that Asn191 and Asn300 act independently one from another in promoting NAD(H)-dependent interconversion of mannitol and fructose (Figure 1) (for the general case, see ). The finding that ΔΔGT‡≈ΔΔG‡ (see Supplementary Table S6) implies that the disruptive effect of individual substitutions of Asn191 and Asn300 was predominantly on the catalytic steps, with substrate binding being less strongly affected. The magnitude of the observed changes in Gibbs free energy caused by the mutations is consistent with the expected net effect of removing a hydrogen bond to a charged donor/acceptor group [38,39] and thus supports a function of Asn191 and Asn300 as oxyanion stabilizers.
In the ordered kinetic mechanism of wild-type PfM2DH, where binding of coenzyme precedes the binding of substrate in either direction of the reaction , kcat/Km for NAD+ and NADH are the bimolecular rate constants describing formation of the respective enzyme–coenzyme complex. The pattern of DKIEs associated with this kinetic mechanism is Dkcat/Ksubstrate>1 and Dkcat/KNAD(H)=1 (for the general case, see ). In the mutants (N191A, N191L), this pattern was changed to DkcatO/Kmannitol≈DkcatO/KNAD+≈DkcatO>1, inconsistent with strictly ordered substrate binding, NAD+ prior to mannitol. These DKIEs would suggest a rapid equilibrium mechanism or a steady-state random mechanism for the two mutated enzymes, in which the off rates for both NAD+ and mannitol from enzyme–NAD+–mannitol are equal . Because the binary complex with mannitol is allowed in the reaction of the mutants, kcat/Km for NAD+ determined in the presence of a saturating substrate concentration can be sensitive to site-directed substitution, despite the fact that neither Asn191 nor Asn300 appear to be required for binding of NAD+ to the free enzyme (Kd).
Mechanistic deductions from pH dependencies of rate constants and their DKIEs
Because mannitol and fructose do not ionize in the pH range studied, only protonation/deprotonation equilibria for the enzyme are considered in the interpretation of pH effects on rates and DKIEs. Therefore, pH dependencies of kcatO/Kmannitol and kcatR/Kfructose reveal ionizations in enzyme–NAD+ and enzyme–NADH respectively, whereas pH dependencies of kobs (wild-type enzyme) and kcat (mutants) show ionizations of the respective ternary complexes.
The low pK of 6.8 observed in the pH profile of logkobsO suggests that, upon binding of mannitol to enzyme–NAD+ in wild-type PfM2DH, the pK of Lys295 undergoes a large shift by 2.5 (=9.2−6.8) pH units toward the acidic range. Rapid mixing of enzyme–NAD+ with mannitol at pH 8.0 where Lys295 (pK=9.2) should be largely (94%) protonated was accompanied by the immediate release of one proton from the enzyme, fully consistent with the proposed pK depression in response to binding of alcohol substrate. Using N191L, proton release upon binding of mannitol at pH 8.0 corresponded to only half the molar equivalent of enzyme present, as expected from the pH profile of logkcatO, suggesting a pK of 8.0 for Lys295 in N191L–NAD+–mannitol. The absence of transient proton release by N300A is consistent with the pH–logkcatO dependence, suggesting that at pH 8.0, only a small fraction of N300A–NAD+–mannitol is in the correct protonation state for reaction.
kobsR for wild-type PfM2DH was invariant within the pH span 7.1–10.0, probably indicating that binding of fructose caused a shift in pK for Lys295, from 9.2 in enzyme–NADH to a higher value that was out of the measured range. The large difference in pK for Lys295 in ternary complexes undergoing oxidation (pK=6.8) and reduction (pK>10) is remarkable. An interesting question raised by the high pK of Lys295 in enzyme–fructose–NADH is which enzyme group eventually releases the proton formed during oxidation of mannitol by NAD+. We have preliminary evidence that a mobile Glu292 situated in the water channel that connects Lys295 with bulk solvent  might be responsible (M. Klimacek and B. Nidetzky, unpublished work).
The pH dependencies of kcat/Ksubstrate for the mutated enzymes where (logarithmic) oxidation rates increased linearly as the pH was raised and reduction rates did not vary within the pH range studied are consistent with a substantial increase in pK for Lys295 in enzyme–NAD+ as well as enzyme–NADH, relative to the pK of the lysine residue in the corresponding native complexes, as result of the site-directed replacements. The alternative interpretation that specific Brønsted catalysis by hydroxide and water facilitate oxidation and reduction in the mutants respectively, appears inconsistent with the low 2H2OKIE on kcatO for N191L . Comparison of pK values from pH profiles of logkcatO (mutants) and logkobsO (wild-type enzyme) support the notion of pK elevation for Lys295 in the ternary complex in response to substitution of Asn191 and Asn300. Note: partial retention of activity in N191L at low pH (kcatO, kcatO/Kmannitol) is not explained by the mechanism proposed and was not further pursued in the course of this work.
The pH dependencies of the DKIEs on kcat/Ksubstrate for oxidation and reduction catalysed by native PfM2DH suggested a mechanism (Scheme 1B) where the group comprising the reactive groups of mannitol and Lys295 loses a proton (as H2O) at high pH (≥10.5) . The commitment of the enzyme–substrate complex to catalysis becomes infinite upon deprotonation, causing the DKIE on hydride transfer to be completely masked in the measured value of DkcatO/Kmannitol (=1.00). The equilibrium DKIE (0.85) is seen on kcatR/Kfructose because proton uptake is effectively prevented under these conditions. The absence of a similar pH dependence of Dkcat/Ksubstrate for single-site mutants of PfM2DH in the measured pH range indicates that either deprotonation of the relevant ionizable group does not occur or its pK is shifted to a higher pH once the oxyanion hole has been partially disrupted.
Proposed function of Asn191 and Asn300 in catalysis
This work delineates two important roles fulfilled by the oxyanion hole of PfM2DH in enzymatic catalysis to NAD(H)-dependent interconversion of mannitol and fructose (Scheme 2). The first role is activation of the catalytic base function of Lys295 through suitable pK depression in enzyme–NAD+–mannitol. A pH-dependent conformational equilibrium, also identified in other metal-independent dehydrogenases  and choline oxidase , locks the ternary complex in the ionization state for mannitol oxidation. Asn191 and Asn300 are key to these pre-catalytic steps, and they also determine ionizations observed in binary and ternary enzyme–substrate complexes (Schemes 1B and 2). Mutation of either Asn191 and Asn300 therefore disrupts the enzyme active site of its ability to finely tune the pK of Lys295. Although it stands to reason that binding of fructose results in an increased pK for Lys295, it is not immediately obvious why hydrogen bonding to the carboxamide oxygens of Asn191 and Asn300 (Scheme 2) should stabilize the protonated ε-amino group of the lysine residue in enzyme–NAD+ and enzyme–NADH, as implied by the effects of partial disruption of the oxyanion hole on the pH dependencies of kcat/Ksubstrate. However, the crystal structure of wild-type enzyme bound to NAD+  suggests that the side-chain of Lys295 relays its proton via the active-site water to Glu292, the main candidate residue for proton transfer to bulk solvent. A plausible scenario therefore is that ‘pull’ by the proton shuttle system causes pK depression in Lys295 and that the oxyanion hole is required to position Lys295 for efficient proton relay.
The second role of the oxyanion hole, supported by KIEs and pH dependencies thereof, is activation of the reactive substrate group and stabilization of the emerging negative charge on oxygen in the pre-catalytic conformers for oxidation and reduction (Scheme 2). In the proposed mechanism, partial abstraction of the proton from the 2-OH group of mannitol to Lys295 is aided by the oxyanion hole dipoles and renders the substrate competent for subsequent hydride transfer. In the direction of fructose reduction, the reactive carbonyl group becomes polarised through the interaction of its oxygen with the side-chains of Asn191 and Asn300. Finally, electrostatic contacts are tightened in the hydride transfer step where they contribute approx. 40 kJ/mol of stabilization energy.
In addition to the unprecedented use of a protease-like oxyanion hole for electrostatic stabilization in an ADH active site, structure–function relationships for PfM2DH are also relevant in showing that the task of a pre-catalytic alcohol activation, fulfilled by Lys295, Asn191 and Asn300, is strikingly similar to that of active-site zinc in ADHs of the medium-chain dehydrogenase/reductase type . Family-wide conservation of the PfM2DH triad of residues suggests that this ‘protein-derived functional mimic of zinc’ is a common catalytic feature for PSLDRs [24,25].
Mario Klimacek and Bernd Nidetzky designed research. Mario Klimacek performed experiments and analysed data. Mario Klimacek and Bernd Nidetzky interpreted data and wrote the paper.
This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.
We thank Veronika Milocco and Valentin Pacher for expert technical assistance. Professor Walter Keller (Institute of Molecular Biosciences, University of Graz, Austria) is thanked for assistance during recording CD spectra.
Abbreviations: ADH, alcohol dehydrogenase; kcatO, kcat of mannitol oxidation; kcatR, kcat of fructose reduction; KIE, kinetic isotope effect; DKIE, primary deuterium kinetic isotope effect; 2H2OKIE, solvent kinetic isotope effect; PfM2DH, Pseudomonas fluorescens D-mannitol 2-dehydrogenase; PSLDR, polyol-specific long-chain dehydrogenase/reductase
- © The Authors Journal compilation © 2010 Biochemical Society