Mechanism of Proton Transfer in the 3α-Hydroxysteroid Dehydrogenase/Carbonyl Reductase from Comamonas testosteroni*

3α-Hydroxysteroid dehydrogenase/carbonyl reductase from Comamonas testosteroni catalyzes the oxidation of androsterone with NAD+ to form androstanedione and NADH with a concomitant releasing of protons to bulk solvent. To probe the proton transfer during the enzyme reaction, we used mutagenesis, chemical rescue, and kinetic isotope effects to investigate the release of protons. The kinetic isotope effects of DV and D2OV for wild-type enzyme are 1 and 2.1 at pL 10.4 (where L represents H, 2H), respectively, and suggest a rate-limiting step in the intramolecular proton transfer. Substitution of alanine for Lys159 changes the rate-limiting step to the hydride transfer, evidenced by an equal deuterium isotope effect of 1.8 on Vmax and V/Kandrosterone and no solvent kinetic isotope effect at saturating 3-(cyclohexylamino)propanesulfonic acid (CAPS). However, a value of 4.4 on Vmax is observed at 10 mm CAPS at pL 10.4, indicating a rate-limiting proton transfer. The rate of the proton transfer is blocked in the K159A and K159M mutants but can be rescued using exogenous proton acceptors, such as buffers, small primary amines, and azide. The Brønsted relationship between the log(V/Kd-baseEt) of the external amine (corrected for molecular size effects) and pKa is linear for the K159A mutant-catalyzed reaction at pH 10.4 (β = 0.85 ± 0.09) at 5 mm CAPS. These results show that proton transfer to the external base with a late transition state occurred in a rate-limiting step. Furthermore, a proton inventory on V/Et is bowl-shaped for both the wild-type and K159A mutant enzymes and indicates a two-proton transfer in the transition state from Tyr155 to Lys159 via 2′-OH of ribose.

3␣-Hydroxysteroid dehydrogenase (3␣-HSD/CR 2 ; EC 1.1.1.50) from Comamonas testosteroni is a member of the protein superfamily of short-chain dehydrogenases/reductases (SDR) (1). Structurally, 3␣-HSD/CR has an ␣/␤ folding pattern very similar to a Rossmann fold, which consists of a central ␤-sheet flanked by ␣-helices (2,3). In the binary complex, the NAD ϩ cofactor is bound at the carboxyl-terminal ends of the ␤-strands in the 3␣-HSD/CR from C. testosteroni. 3␣-HSD/CR reversibly catalyzes the oxidation of the steroid alcohol using NAD ϩ as the oxidant. The reaction catalyzed by 3␣-HSD/CR shows an ordered Bi Bi kinetic mechanism with oxidized dinucleotide added first and the reduced dinucleotide released last (4). Studies of the pH profile and structural determinations elucidated the chemical mechanism of the 3␣-HSD/CR-catalyzed reaction (2,3,5). The triad of Ser 114 , Tyr 155 , and Lys 159 in the active site participates in the enzyme catalysis. The role of Lys 159 is to lower the pK a of Tyr 155 through electrostatic interaction between the protonated ⑀-amino of Lys 159 and the hydroxy group of Tyr 155 , whereas the unprotonated form of Tyr 155 with an apparent pK a of 7.2 acts as a catalytic base that abstracts a proton from the hydroxyl group of the substrate. Ser 114 is in hydrogen bonding with the substrate during the reaction catalyzed by the WT enzyme and alternatively acts as the general base to rescue catalysis of proton transfer in the Y155F mutant enzyme. NAD ϩ cofactor bound in the syn conformation accepts a hydride from the 3␤-position of androsterone through the "si face" of the nicotinamide ring to form androstanedione and NADH with the concomitant release of protons in the forward reaction direction.
The overall oxidoreductive reaction catalyzed by 3␣-HSD/CR is composed of the deprotonation of tyrosine, proton abstraction by the tyrosinate anion, and hydride transfer from the hydroxysteroid to NAD ϩ , followed by the release of a proton from the hydroxy group of tyrosine to the solution (Fig. 1). A proton relay system is proposed in the SDR family through which protons are shuttled to bulk solvent. Analysis of the structures of 3␣/17␤-hydroxysteroid dehydrogenase and R-specific alcohol dehydrogenase suggests a proton transfer from the hydroxyl group of Tyr, via the 2Ј-OH of the nicotinamide ribose, the Lys side chain, and a water molecule hydrogenbonded to the backbone carbonyl of Asn (6,7). The theoretical calculations of the ionization properties of the hydroxyl group of Tyr 155 , the O2Ј ribose hydroxyl, and the Lys 159 ⑀-amino group in the active site of Drosophila alcohol dehydrogenase are consistent with the proton relay occurring in the SDR family (8). SDR is a large protein family with highly diverse functions in pro-and eukaryotes and is composed of a majority of oxidoreductases with NAD(P) ϩ as cofactor (9,10). Although sequence alignment between different SDR enzymes typically shows 15-30% identity, the conserved sequences include an N-terminal Gly-X 3 -Gly-X-Gly cofactor binding motif and a tetrad of catalytically important Ser, Tyr, Lys, and Asn residues, of which Tyr is the most conserved in the SDR family. The mechanistic roles of a catalytic tetrad have been elucidated through chemical modification, site-directed mutagenesis, sequence alignment, and structural comparison (5,6,9). Mutation of tyrosine, serine, and lysine residue results in the loss of most activity in the SDR family. Mutation of Asn 111 to Leu in 3␤/17␤ hydroxysteroid dehydrogenase results in inactivation of the enzyme. The Asn residue is also important in maintaining the active site configuration and is involved in the proton relay during the catalysis. Hence, the catalytic tetrad of Asn, Ser, Tyr, and Lys is involved in catalysis in the SDR family.
Proton transfer plays many roles in the enzyme-catalyzed reaction. It participates in the stabilization of the transition state to enhance the rate through the acid-base catalysis and the formation of a low barrier hydrogen bond (11)(12)(13) and is important in bioenergetics (14,15). The electron transfer is coupled with the movement of protons in many respiratory oxidases and results in a proton gradient across the membrane to drive ATP synthesis. The movement of protons through a protein scaffold is often essential for enzyme-catalyzed reactions. Protons may move inside proteins along pathways provided by a network of hydrogen-bonded amino acid side chains and water molecules (16,17). The intramolecular proton transfer involved in the enzyme reaction is generally characterized by the Brønsted acid-base analysis (12,18), solvent deuterium isotope effect (19,20), and chemical model study (21). Here, we studied proton transfer in the 3␣-HSD/CR-catalyzed reaction, examined the number of proton transfers in the transition state in the rate-limiting step, and evaluated the roles of the Lys 159 residue in the active site. We studied deuterium and solvent kinetic isotope effects to determine the ratelimiting step in the overall reaction. We then substituted alanine or methionine for Lys 159 to block the pathway for proton transfer, used small molecules to restore the proton shuttle, and carried out a proton inventory to investigate intramolecular proton transfer. Buffers and exogenous proton acceptors are capable of participating in catalysis when proton transfer is blocked in the K159A and K159M mutant enzymes. Furthermore, the bowlshaped proton inventory gives two protons participating in proton transfer in the rate-limiting step. In addition to lowering the pK a of the general base Tyr 155 , the results further demonstrated the role of Lys 159 in shuttling the protons to bulk solvent in the 3␣-HSD/CR-catalyzed reaction.

EXPERIMENTAL PROCEDURES
We used the QuikChange site-directed mutagenesis kit from Stratagene. Androsterone and androstanedione were purchased from Steraloids, Inc., and NAD ϩ was from Roche Applied Science. Glucose-6-phosphate dehydrogenase from Leuconostoc mesenteroids and CAPS were from Sigma. Ammonia was obtained from Showa (Japan). Methylamine, ethanolamine, 2,2,2-trifluoroethylamine, and azide were purchased from Lancaster. Ethylamine and propylamine were from Acros, and ethylenediamine was from Tedia (Japan  (22). Glucose-6-phosphate dehydrogenase from L. mesenteroids reversibly catalyzes the stereospecific reduction of NAD ϩ from 1-deutero-glucose to form B-side-labeled NADH and gluconic acid-␥-lactone. The reaction essentially goes to completion due to the unstable gluconic acid-␥-lactone at pH 8.0, which is hydrolyzed to give gluconate. [4S-2 H]NADH was separated from NAD ϩ by gradient elution of ammonium bicarbonate (10 -500 mM) at pH 9.3 through DEAE-cellulose ionic exchange column. Fractions of [4S-2 H]NADH were pooled and lyophilized. The labeled 3␣-androsterone was further prepared by stereospecific oxidation of the resulting [4S-2 H] NADH from 5␣-androstan-3,17dione as reversibly catalyzed by 3␣-HSD/CR. The progress for reverse reduction of androstanedione with [4S-2 H]NADH catalyzed by 3␣-HSD/CR was followed by the disappearance of [4S-2 H]NADH at 340 nm spectrophotometrically at pH 7.0. The reaction was quenched by the addition of CHCl 3 to denature the enzyme after it reached equilibrium. Deuterated androsterone was extracted by ethyl acetate and further purified by reverse phase HPLC.
Expression and Purification of Wild-type and Mutant Enzymes-Both wild-type and mutant genes were expressed in the Escherichia coli strain BL21(DE3). Mutagenic replacements were performed using the Quikchange site-directed mutagenesis kit and Pfu polymerase with pET-15b-3␣-HSD/CR plasmid as template. The synthetic oligonucleotide primers used to create the cDNA for K159M were 5Ј-CTGGCCTATGCGGGCA-GCATGAATGCTTTGACGGTGGC-3Ј (sense) and 5Ј-GCC-ACCGTCAAAGCATTCATGCTGCCCGCATAGGCCAG-3Ј (antisense). The boldface codons indicate the mutation on the amino acid residue by the replacement of the underlined codon. The mutant vector thus obtained was transformed into competent E. coli BL21(DE3) cells. The gene was sequenced for K159M mutation and compared with that of the wild-type 3␣-HSD/CR using BLAST. The K159A mutant enzyme was prepared as described previously (5). In brief, recombinant proteins were overexpressed in BL21(DE3) cells and grown at 37°C to an optical density of 0.6 -1 at 600 nm in LB medium containing 50 g/ml ampicillin. Isopropyl ␤-D-thiogalactopyranoside (0.5 mM) was added to the culture to induce protein expression. Growth was continued for an additional 4 h at 37°C. The cells were then harvested and lysed by sonication. The overexpressed proteins were purified via an Ni 2ϩ -nitrilotriacetic acid affinity column making use of the enzyme's His tag. SDS-PAGE was used to analyze protein purity. The protein concentrations were determined by a Bradford assay with bovine serum albumin as a standard (23).
Kinetic Studies-The oxidation of androsterone catalyzed by 3␣-HSD/CR was monitored by the formation of NADH spectrophotometrically at 340 nm. A typical assay for the enzymatic reaction included 1 mM NAD ϩ and varied concentrations of androsterone in 0.1 M CAPS at pH 10.4 at 25°C. All reactions were initiated by the addition of enzyme. Initial velocities for wild type and mutants were measured at varying concentrations of androsterone at several fixed concentrations of NAD ϩ . Data were fitted using Sigmaplot software for appropriate rate equations to obtain kinetic parameters. Data for substrate saturation curves at a fixed concentration of the second substrate were fitted using Equation 1. In Equation 1, v and V represent the initial and maximum velocity, respectively, and K m is the Michaelis constant. Data for a sequential and a rapid equilibrium order kinetic mechanism were fitted to Equation 2 and 3, respectively, where A and B are the varied substrates, K ia is the inhibition constant for A, and K A and K B are the Michaelis constants for substrate A and B, respectively.

Chemical Rescue of the K159A and K159M Mutant Enzymes-
The effects on the K159A and K159M mutant enzyme-catalyzed reaction of exogenous small molecular proton acceptors were determined by measuring the initial velocity of the oxidation of androsterone at 25°C as a function of the added amine. Primary amines included ammonia, methylamine, ethylamine, propylamine, ethanolamine, and 2,2,2-trifluoroethylamine. The assay mixtures contained 1 mM NAD ϩ , 46.6 M androsterone, and various concentrations of the rescue agents in 5 mM CAPS, pH 10.4, at 25°C. The reactions were initiated by the addition of the mutant enzyme (0.19 g for K159A mutant and 0.37 g for K159M mutant). The initial rate was measured in the presence of different concentrations of the rescue agents ([amine]). Data were then fitted to Equation 4 to obtain the chemical rescue parameters, V/E t , V/K d-base Et, and K d-base of chemical rescue reagents.
In Equation 4, V/K d-base Et is the second order rate constant for the chemical rescuer. K d-base is the concentration of the rescue agent for which half of V was observed. v amine is the initial rate obtained by subtracting the rate in the absence of the external amine from the rate in the presence of amine. The concentration of conjugate base of the amine, [amine], was calculated based on Equation 5, where K a is the acid dissociation constant of the amine, and [amine] total is the total concentration of the added amine. Contributions from the electronic properties (pK a ) and molecular volume of the primary amine to the efficiency of chemical rescue (V/K d-base Et) were analyzed by fitting data to Equation 6, where ␤ is the Brønsted coefficient, V is the molecular volume correction coefficient, and C is the constant term for the specific reaction.
Isotope Effects and Proton Inventory-The primary kinetic isotope effects, D V and D (V/K androsterone ), were determined by a direct comparison of the kinetic parameters V max and V/K androsterone with unlabeled and deuterated androsterone. V max and V/K androsterone were obtained by measuring the initial rate as a function of either the deuterated or unlabeled androsterone concentration at 1 mM NAD ϩ . The concentrations of deuterated and unlabeled androsterone were determined in triplicate by end point assay. The solvent kinetic isotope effects (SKIE) were measured according to the method of Quinn and Sutton (24). In the case of reactions measured in 2 H 2 O, all reactants were prepared and lyophilized twice in 2 H 2 O. Buffers were titrated to the desired p 2 H (where p 2 H is equal to the pH meter reading plus 0.4) using NaO 2 H. Initial rates (v) were measured at varied concentrations of androsterone and 1 mM NAD ϩ in the presence of H 2 O or 2 H 2 O. Data from the experiments were fitted to Equations 7-10, when an isotope effect was observed on both V and V/K, an equal isotope effect on V and V/K, V only, and V/K only, respectively.

Proton Transfer in 3␣-HSD/CR
In Equations 7-10, F i is the fraction of the deuterium label in the substrate, E V,V/K is the isotope effect minus 1 for an equal isotope effect on V and V/K, and E V and E V/K are the isotope effect minus 1 on V and V/K, respectively (25).
To obtain information on the number and fractionation factors for proton(s) being transferred at the transition state(s), the proton inventory method was utilized (24). The kinetic parameters V and V/K were obtained by measuring the initial rate at different fractional concentrations of 2 H 2 O in the reaction mixture by combining appropriate volumes of H 2 O or 2 H 2 O. Data were then fitted to the Gross-Butler equation as follows, where n k is the ratio of the rate constants (V or V/K) measured in different fractional concentrations of 2 H 2 O compared with 100% 2 H 2 O, D2O k is the solvent deuterium isotope effect (i.e. the ratio of the rate constants (V or V/K) in H 2 O and 2 H 2 O), n is the fractional concentration of 2 H 2 O, T is the corresponding deuterium fractionation factor for the exchangeable protonic sites relative to bulk water, and Z represents a medium effect. Data for the linear and bowl-shaped proton inventory were fitted to Equation 12 and 13, respectively. In Equation 13, two exchangeable protonic sites with identical fractionation factors are involved in the transition state and a unit fractionation factor of reactant state.

Dependence of the Activity of the K159A Mutant Enzyme on
Buffers-Substitution of alanine for lysine is expected to block the proton shuttle from the active site to the solvent. In an attempt to restore the proton transfer, an external proton acceptor was added to rescue the reaction catalyzed by the K159A mutant enzyme. We observed that the activity of the K159A mutant enzyme was dependent on the concentrations of CAPS. In comparison, the activity of the wild-type enzyme was not affected at either 5 or 100 mM CAPS at pH 10.4, indicating that the buffers participate in the reaction only when the Lys 159 side chain is eliminated ( Table 1). The initial rates of the K159A enzyme-catalyzed reaction exhibited a hyperbolic dependence on the concentration of CAPS varying from 5 to 200 mM at 72 M androsterone and 1 mM NAD ϩ , pH 10.4. To elucidate the role of CAPS in the kinetic mechanism, an initial velocity pattern was obtained by varying the buffer concentrations at different fixed concentrations of androsterone at 1 mM NAD ϩ , pH 10.4. The double-reciprocal plot intersected to the left of the ordinate, suggesting a sequential kinetic mechanism (see supplemental Fig. S1). Data were fitted to Equation 2; values for V/E t were 13 Ϯ 2 s Ϫ1 , and K m for androsterone and CAPS in the mutant enzyme was 34 Ϯ 11 M and 8 Ϯ 4 mM, respectively.
Chemical Rescue of the Activity of the K159A Mutant Enzyme by Exogenous Proton Acceptors-To further characterize the contribution of the mutated side chain on the catalytic rate enhancement provided by wild-type enzyme, noncovalent chemical rescue reagents were used to restore the activity of the mutant enzyme. The small molecules used were the primary amines, ammonia, methylamine, ethylamine, propylamine, ethanolamine, and 2,2,2-trifluoroethylamine, as well as azide. Those molecules have different pK a values and different molecular volumes (26). An increase in the initial rate of the mutant enzyme was observed in the presence of all of the chemical rescue agents with the exception of 2,2,2-trifluoroethylamine (up to 0.4 M, pK a ϭ 5.7) in the presence of 5 mM CAPS, pH 10.4. The activity of wild-type 3␣-HSD/CR was unaffected by 0.4 M methylamine within the range of experimental error ( Table 1). The activity of the K159A mutant enzyme in the presence of methylamine obeyed Michaelis-Menten kinetics with respect to androsterone as the varied substrate. The kinetic parameters we obtained for the K159A mutant-catalyzed reaction were compared in the absence and presence of 0.4 M methylamine at 5 mM CAPS, pH 10.4. No significant change in affinity for androsterone was observed in response to external amine. The V/Et and V/K androsterone Et for the K159A mutant enzyme at 5 mM CAPS was increased 6-fold with 0.4 M methylamine, which was up to 30% in V/Et and 5% in V/K androsterone Et of the wildtype level. The rescue of activity in K159A by external bases is shown in supplemental Fig. S2. The second order rate constant (V/KEt) for methylamine increased as the pH increased, indicating that the basic form of amine is an active species (data not shown). The concentration of the conjugated base form of the chemical rescuers was then calculated based on the pK a of the bases and the pH value in the solution according to the Henderson-Hasselbach Equation 5. Data were fitted to Equation 4 to obtain the second order rate constant, V/K d-base Et. The kinetic parameters V/Et, V/K d-base Et, and K d-base that we obtained are shown in Table 2. To evaluate the steric effect caused by the external amine, the kinetic parameters V/K d-base Et for methylamine, ethylamine, and propylamine with similar pK a values were compared. A plot of log V/K d-base Et versus molecular volume showed a linear relationship with a slope of Ϫ0.0085 and intercept of 3.38. The decrease in log V/K d-base Et as the side chain increased in length indicates a steric discrimination in the active site of the K159A mutant enzyme. The result clearly indicated that the size of the external amine affected the catalytic efficiency in proton transfer for the K159A mutant enzyme. In contrast, the plot comparing log V/K d-base Et and pK a did not show a clear correlation. Therefore, data for rescue efficiency were fitted to Equation 6, resulting in a Brønsted coefficient ␤ of 0.85 Ϯ 0.09, a volume coefficient V of Ϫ0.011 Ϯ 0.003, and a constant value C of Ϫ5.5 Ϯ 0.9. Corrected by the steric effect, the plot comparing rescue efficiencies (log V/K d-base Et Ϫ V ϫ molecular volume) and pK a of the primary amines showed a linear relationship during the reaction catalyzed by K159A mutant enzyme (Fig. 2). Similarly, the external amines were capable of restoring the activity of the K159M mutant enzyme (Table 2 and supplemen-tal Fig. S3). Data for rescue efficiency were fitted to Equation 6, resulting in a Brønsted coefficient ␤ of 1.2 Ϯ 0.2, a volume coefficient V of Ϫ0.020 Ϯ 0.004, and a constant value C of Ϫ9 Ϯ 2 (Fig. 2).
Since CAPS also participates in binding and catalysis, the steric discrimination in the active site of the K159A mutant enzyme may exclude CAPS from binding at the active site. To distinguish between the binding sites of CAPS and the small amines, the initial rate pattern was studied by varying the concentrations of CAPS at different fixed concentrations of methylamine in the presence of 46.6 M androsterone and 1 mM NAD ϩ at pH 10.4. A plot of the reciprocal rate versus the concentrations of CAPS is shown in Fig. 3. The initial rate observed at the saturated concentration of CAPS increases as the concentration of methylamine increases, suggesting that CAPS and methylamine do not compete for the same binding site.
Isotope Effect Study-To elucidate protonic participation in the rate-determining transition state for catalysis by 3␣-HSD/ CR, we studied the primary deuterium kinetic isotope effect, solvent deuterium kinetic isotope effect, and proton inventory. A stereospecific labeled [3-2 H]androsterone was prepared via  Table 2. By correcting the volume effect of the external amine, the log (V/K d-base Et) is linearly dependent on the pK a of the external amine. The lines represent a fit of data to Equation 6, and give the Brønsted coefficient ␤ ϭ 0.85 Ϯ 0.09, volume coefficient (V) ϭ Ϫ0.011 Ϯ 0.003, and C ϭ Ϫ5.5 Ϯ 0.9 for the K159A mutant enzyme (solid line) and the Brønsted coefficient ␤ ϭ 1.2 Ϯ 0.2, volume coefficient (V) ϭ Ϫ0.020 Ϯ 0.004, and C ϭ Ϫ9 Ϯ 2 for the K159M mutant enzyme (dashed line), respectively.

Proton Transfer in 3␣-HSD/CR
the reduction of androstanedione with B-side [4S-2 H]NADH catalyzed by 3␣-HSD/CR. [4S-2 H] NADH was prepared by the oxidation of 1-deutero-glucose with NAD ϩ catalyzed by glucose 6-phosphate dehydrogenase. The primary kinetic isotope effect was obtained by direct comparison of the kinetic parameters of V max and V/K androsterone for labeled and unlabeled androsterone . The initial velocity patterns obtained by varying the concentrations of androsterone at several fixed concentrations of NAD ϩ indicated a sequential order and a rapid equilibrium order kinetic mechanism for the wild-type and K159A mutant enzymes, respectively (data not shown). Therefore, data obtained from the isotope effect study were fitted to Equation 10 for the wild-type enzyme and Equation 8 for the K159A mutant enzyme, which gives an equal isotope effect on V max and V/K androsterone . The primary deuterium KIE for wild-type and K159A mutant enzymes is shown in Table 3. We measured the solvent deuterium kinetic isotope effect to further explore the intramolecular proton transfer. Previously, the pK a of wildtype and K159A mutant enzymes were obtained as 7.2 and 9.1, respectively (5 (Table 3).
Proton Inventory Study-To obtain information on the number of protons involved in the proton shuttle and their fractionation factors in the transition state, a proton inventory of the wild-type and mutant enzymes was carried out. A dependence of n V max versus n, the mole fraction of solvent deuterium, was observed for both the wild-type enzyme at 100 mM CAPS and K159A mutant enzyme at 10 mM CAPS when the concentration of androsterone was varied at a fixed concentration of NAD ϩ (1 mM) (Fig. 4). Data obtained from the wild-type and K159A mutant enzymes were fitted to Equation 12 or Equation 13 for one-and two-transition state protons that contributed to the isotope effect, respectively. The latter regressed to less variance for both wild-type ( 2 ϭ 0.0106 and 0.0068, respectively) and K159A mutant enzymes ( 2 ϭ 0.0662 and 0.0154, respectively). Hence, the results gave a better fit for two protons with D2O V ϭ 2.19 Ϯ 0.06 and T ϭ 0.64 Ϯ 0.02 for wild-type enzyme, and

Chemical Rescue in the K159A Mutant Enzyme by Proton
Acceptors-In a previous study, the pH profile of the K159A mutant enzyme showed similar pH dependence compared with wild-type enzyme with a shift of the pK a in V/K androsterone Et from 7.2 to 9.1, suggesting that Lys 159 lowers the pK a of the Tyr 155 , which acted as a general base. Although the phenol group of Tyr 155 was deprotonated for an optimal reaction in 100 mM CAPS at pH 10.4, the activity of K159A mutant enzyme was 50-fold less than that of the wild-type in V/K androsterone Et and 10-fold less in V/Et (Table 1). In the present study, a series of small amines and buffers were added to investigate the proton transfer performed by the amino group. The activity of wild-type enzyme was not affected by in the presence of either CAPS or methylamine at pH 10.4, but that of the K159A mutant enzyme was found to depend on the concentration level of CAPS and the external amine. The initial velocities exhibit a saturated kinetics with respect to the base rescuers, implying that CAPS and the external amine participate in the binding and catalysis of this enzyme (Table 1 and supplemental Fig. S1). Buffer has been proposed to act as a proton acceptor assisting in proton transfer. A proton relay system for human liver alcohol dehydrogenase is composed of the proton transfer from the Zn 2ϩ -bound alcohol substrate, via the hydroxyl group of Ser 48 , the 2Ј-OH of the nicotinamide ribose, and the His 51 side chain to the solvent. Replacement of His 51 with Gln decreases the V/K alcohol 6-fold at pH 7, but it can be restored by buffers (28). Furthermore, carbonic anhydrase II has evolved a proton shuttle to allow buffer components to participate in the reaction  from solution. The rate of proton abstraction from Zn 2ϩbound H 2 O is dependent on the external buffer, imidazole, which exhibits saturated kinetics (29 -33). We argue that buffer is capable of binding with K159A mutant enzyme to facilitate a proton release into solvent. This phenomenon is similar to that for human liver alcohol dehydrogenase (28). The initial rate pattern indicates a sequential kinetic mechanism for the 3␣-HSD/CR-catalyzed reaction (i.e. all reactants of NAD ϩ , androsterone, and CAPS must combine with the enzyme before reaction can take place and any product can be released). This kinetic mechanism is different from the ping-pong mechanism of the carbonic anhydrase-catalyzed reaction in which imidazole is bound with the intermediate of Zn 2ϩ -bound H 2 O and facilitates proton transfer to solution in the second half of the reaction (29). It has been demonstrated that catalytic activity can be restored to the inactive K258A mutant of E. coli aspartate aminotransferase by exogenous amines (26,34) and that the observed rate constants are linearly dependent on the concentration of the free base form of the amine catalyst in the K258A mutant, suggesting that K258 acts as the general base to facilitate the reaction catalyzed by aspartate aminotransferase. This methodology has been used to study many enzyme reactions to confirm the role of lysine and arginine in the general acid-base catalysis and the reaction mechanisms. The enzymes studied included mannitol 2-dehydrogenase (35), ketopantoate reductase (36), dihydroorotate dehydrogenase (37), and inosine 5Ј-monophosphate dehydrogenase (38). We extended this method to investigate the proton relay system for the reaction catalyzed by 3␣-HSD/CR. We aimed to further clarify the contribution of the ⑀-amine side chain of Lys 159 to catalysis by restoring the activity conferred upon the K159A mutant enzyme by the exogenous proton acceptor. We added external proton acceptors with different pK a of the primary amine and azide to the reaction. The fact that by adding a series of primary amines, we could restore up to 30% of wild-type activity in K159A mutant enzyme strongly suggests that exogenous amino groups can compensate for the loss of the mutated side chain. However, the chemical rescue appears to be an incomplete restoration of activity to the K159A mutant enzyme, suggesting that a proper orientation of ⑀-amine on Lys 159 is essential for catalysis. The observed second order rate constant log V/K d-base Et is linearly dependent on the pK a value of the base form of the rescue agents for the K159A mutant enzyme (Fig.  2). The sensitivity to base strength observed on the Brønsted plot suggests that the chemical reaction by the external base is rate-limiting in the K159A mutant enzyme. The Brønsted correlation with ␤ ϭ 0.85 further indicates a late transition state for proton transfer in the reaction coordinate, where the proton has almost completely (85%) transferred to the external base. Consistent with these findings, a large SKIE of 4.4 was measured at the low concentration of CAPS.
Although both CAPS and small amines can restore the enzyme activity caused by the substitution of Lys 159 with Ala, their binding site in the K159A mutant enzyme is different due to the size of the rescue reagent. The increase in rate at saturated CAPS as the concentration of methylamine increases supports the idea that CAPS and methylamine do not compete for the same site on the K159A mutant enzyme (Fig. 3). Hence, there are two mechanisms for activation of the proton transfer in the K159A mutant enzyme by the external amine, either through the active site or external to the active site. The log V/K d-base Et for the external base on the K159A mutant enzyme was linearly dependent on the size of the exogenous amine with the same pK a value ( Table 2), suggesting that the amine binds at the active site of the K159A mutant enzyme and occupies the vacancy created by the substitution of alanine for lysine (a difference of 80 Å 3 in volume (39)), replacing the function of lysine in the proton transfer. To confirm the steric effect exerted by the binding of the external amine at the active site of the K159A mutant, we investigated the amine volume dependence of the rate constant in the K159M mutant, since the methionine side chain should fill any cavity created in K159A. Surprisingly, the log V/K d-base Et on the K159M mutant enzyme is also related to the size and base strength of the external amine. The rescued efficiency for K159M mutant enzyme is more sensitive to the size of the external amine than that of K159A mutant enzyme as a result of the smaller space at the active site in the K159M mutant enzyme (volume coefficient of Ϫ0.020 versus Ϫ0.011 for K159M mutant and K159A mutant enzymes, respectively). Although the methionine side chain could fill any cavity created in K159A, the external amine may bind with K159M mutant enzyme at the active site to facilitate proton transfer. Conversely, the larger CAPS will be bound with K159A mutant enzyme at the surface near the 2Ј-OH of the nicotinamide ribose to facilitate proton transfer to solvent. Comparing the kinetic parameter, V/K androsterone Et for the K159A mutant enzyme in 5 mM CAPS, 400 mM methylamine (1.15 ϫ 10 6 M Ϫ1 s Ϫ1 ) with 100 mM CAPS (0.36 ϫ 10 6 M Ϫ1 s Ϫ1 ), the rescue efficiency of the small external amine binding at the active site appears to be higher than activation external to the active site by CAPS.
Isotope Effect Study-The primary KIE and SKIE on V max and V/K androsterone for the wild-type 3␣-HSD/CR were studied to evaluate the rate-limiting steps in the enzyme-catalyzed reaction. The kinetic parameter V max combines all the steps from the ternary complex of 3␣-HSD/CR-NAD ϩ -androsterone to the release of the last product (NADH) in the reaction, whereas V/K androsterone includes the steps from the binding of androsterone to the first irreversible step (i.e. the releasing of androstanedione). The size of deuterium KIE reflects the step of hydride transfer, whereas that of the SKIE indicates the steps on the transfer of exchangeable protons. Our results show no isotope effect for the wild-type enzyme on the D V and D2O (V/ K androsterone ) and a finite value of 2 on the D (V/K androsterone ) and D2O V. This differential effect on V max and V/K androsterone obtained from SKIE and KIE suggests that the hydride transfer and proton transfer occur in different steps. Any ratelimiting steps outside the hydride transfer step could mask the intrinsic KIE, resulting in observed isotope effects equal to 1. Hence, the D V of 1 indicates a rate-limiting step on the release of NADH once substrates are bound, whereas the finite isotope effect of D (V/K androsterone ) for the wild-type enzyme suggests that a rate-limiting step occurs with hydride transfer from the addition of androsterone until the releasing of androstanedione. Furthermore, the D2O V of 2.1 we observed suggests that the proton transfer limits the overall reaction. This may include the proton transfer from Tyr 155 via Lys 159 to solvent. Combined with the results from the SKIE and primary KIE, we conclude that the release of NADH product coupled with proton transfer is the most rate-limiting step in the overall reaction catalyzed by 3␣-HSD/CR. Evidence for a rate-limiting dissociation of NADH was demonstrated in the reaction catalyzed by the 3␣-HSD from Pseudomonas sp. The binding of NADH with the 3␣-HSD from Pseudomonas sp. is a two-step mechanism with an initial loosely bound form followed by a tightly bound isomerized form in a transient phase kinetic study (40). The structures of the apo-and holo-3␣-HSD with NADH on the 3␣-HSD from Pseudomonas sp. further demonstrated that binding with the coenzyme NADH induces a significant conformation change, resulting in a shift from the substrate-binding loop to a helix conformation (41). Hence, proton transfer could be accompanied by a conformational change during the product-releasing step and contribute to the rate-limiting step(s). The observed SKIE will be due to a decrease in the rate of conformational rearrangement of the enzyme once the protons are substituted by deuterons. This result implies that enzyme flexibility and enzyme activity are tightly coupled in 3␣-HSD/CR-catalyzed reactions.
The rate-limiting step in the reaction catalyzed by K159A mutant enzyme occurs upon hydride transfer at saturated CAPS, evidenced by the equal deuterium isotope effect on V max and V/K androsterone , no SKIE, and rapid equilibrium order kinetic mechanism. It changes to a proton transfer at limited amounts of CAPS. Hence, proton transfer to the saturated CAPS is rapid, whereas limiting the amount of CAPS probably results in a decrease in the rate of proton transfer, resulting in a larger observed SKIE of 4.
Proton Inventory Study-Proton inventory can determine the number of protons undergoing changes in a rate-limiting step and allow a breakdown of the overall SKIE into its reactant state and transition state components (20,24). A linear relationship between the observed rate constant and the mole fraction of the deuterium fraction indicates that one proton is involved in the transition state, whereas a curve indicates either the involvement of more than one proton or a contribution effect from the medium. A bowl-shaped proton inventory indicates multiple hydrogen sites in either the reactant or transition state, and a dome-shaped proton inventory arises from offsetting normal and inverse contributions to the solvent isotope effect. Our proton inventory study of wild-type and K159A mutant enzymes showed a bowl shape between the observed rate constant and the mole fraction of the deuterium fraction, suggesting that multiple exchangeable hydrogenic sites were involved in the proton transfer. The bowl-shaped proton inventory of V max on the wild-type and K159A mutant enzyme is best fitted for two protons in flight in the transition state. The values of T ϭ 0.64 and 0.44 were obtained for the wild-type and K159A mutant enzyme, respectively, which could be assigned to the proton bridge involving the hydroxyl group of Tyr 155 , 2Ј-OH of ribose, and the NH 2 group from either Lys 159 or an external amine. The difference in the transition state fractionation factor for the wild-type versus the K159A mutant enzyme suggests a different mechanism of pro-ton transfer. The proton transfer for 3␣-HSD/CR from C. testosteroni, a member of the SDR family, is shown in Fig. 5. The Lys 159 residue is involved in the proton transfer. The proton transfer includes the deprotonation of Tyr 155 to form tyrosinate anion, the abstraction of the proton from the 3-hydroxyl group of the substrate androsterone, followed by hydride transfer to NAD ϩ and relay of the proton to solvent through the 2Ј-OH of the nicotinamide ribose, the amino group of Lys 159 , and to the bulk solvent. A proton transfer from the 2Ј-OH of ribose to Lys 159 must be a rate-limiting step and a late transition step with a transition state fractionation factor of 0.64 for the wildtype enzyme. In the case of the mutant K159A-catalyzed reaction shown in Fig. 6, the proton transfer is blocked by the replacement of Lys 159 with alanine. The activity is restored by the external amine, and the rate of proton transfer is dependent on the strength and size of the external base. The external small amine is bound at the active site and is involved in proton transfer with a late transition state in the rate-limiting step. In addition, a proton is alternately transferred from the 2Ј-OH of shows a similar proton relay system composed of the residues Tyr 155 , 2Ј-OH of ribose, Lys 159 , and Asn 86 and water in a 3␣-HSD/CR-catalyzed reaction. The hydrogen bonding distances (in Å) within the residues in the proton relay system are shown. The plot was generated using the program PyMOL. Water is shown as a red sphere. FIGURE 6. The proton relay system is assisted by the external small amine in the K159A mutant-catalyzed reaction. The external small amine is bound with enzyme at the position created by the substitution of Lys 159 with alanine to facilitate proton transfer and restore activity. The proton is transferred to the external amine in a rate-limiting step with a Brønsted coefficient of 0.85. ribose to CAPS at the surface with a transition state fractionation factor of 0.44. The smaller fractionation factor of 0.44 in K159A mutant versus 0.64 in wild-type enzyme suggests that the proton transfer to the external CAPS in K159A mutant enzyme has a looser transition state structure compared with Lys 159 in the active site of the wild-type enzyme.
In summary, the proton relay system is involved in the 3␣-HSD/CR-catalyzed reaction through Tyr, 2Ј-OH of the nicotinamide ribose, Lys, Asn, and waters at the active site. The observed normal solvent kinetic isotope effect on V max indicates that proton transfer is mostly rate-limiting. The Lys 159 residue is important for shuttling protons to solvent during the enzyme catalysis. The transfer of proton was blocked when Ala was substituted for Lys. The activity can be restored by exogenous proton acceptors, such as small amines, and buffer through the active site or external to the active site. These findings, combined with the results showing a Brønsted coefficient of 0.85 for the external amines, the large solvent isotope effect of 4.4, and a bowl-shaped proton inventory for the K159A enzyme-catalyzed reaction at 10 mM CAPS, pH 10.4, has allowed us to elucidate the proton transfer mechanism, whereby the release of proton to the bulk solvent occurs through two protons being transferred to the external amine at a late transition state in a rate-liming step. These findings are important because they demonstrate the role of Lys in the catalytic tetrad of SDR: 1) to lower the pK a of the general base and 2) to serve as the proton shuttle in the enzyme catalysis, establishing the mechanistic basis for general insights into SDR catalysis proton transfer.