On the active site of Old Yellow Enzyme. Role of histidine 191 and asparagine 194.

Old Yellow Enzyme (OYE) binds phenolic ligands forming long wavelength (500-800 nm) charge-transfer complexes. The enzyme is reduced by NADPH, and oxygen, quinones, and alpha,beta-unsaturated aldehydes and ketones can act as electron acceptors to complete catalytic turnover. Solution of the crystal structure of OYE1 from brewer's bottom yeast (Fox, K. M., and Karplus, P. A. (1994) Structure 2, 1089-1105) made it possible to identify histidine 191 and asparagine 194 as amino acid residues that hydrogen-bond with the phenolic ligands, stabilizing the anionic form involved in charge-transfer interaction with the FMN prosthetic group. His-191 and Asn-194 are also predicted to interact with the nicotinamide ring of NADPH in the active site. Mutations of His-191 to Asn, Asn-194 to His, and a double mutation, H191N/N194H, were made of OYE1. It was not possible to isolate the N191H mutant enzyme, but the other two mutant forms had the expected effect on phenolic ligand binding, i.e. decreased binding affinity and decreased charge-transfer absorbance. Reduction of the H191N mutant enzyme by NADPH was similar to that of OYE1, but the reduction rate constant for NADH was greatly decreased. The double mutant enzyme had an increased rate constant for reduction by NADPH, but the reduction rate constant with NADH was lower by a factor of 15. The reactivity of OYE1 and the mutant enzymes with oxygen was similar, but the reactivity of 2-cyclohexenone was greatly decreased by the mutations. The crystal structures of the two mutant forms showed only minor changes from that of the wild type enzyme.

protein was able to oxidize NADPH and, somewhat less efficiently, NADH (5,6). The physiological oxidant of OYE has yet to be determined, but oxygen and a number of quinones can act in this capacity to complete the redox cycle (7). More recently it was found that 2-cyclohexenone (8) and a large number of other ␣,␤-unsaturated aldehydes and ketones are able to act as effective electron acceptors (9), suggesting that the physiological function of the enzyme might be the reduction of such a compound.
Phenolic ligands bind to Old Yellow Enzyme with perturbation of the flavoprotein spectra and formation of striking long wavelength (500 -800 nm) absorbance bands (6,10). This binding is affected by substituents on the phenol and a correlation between the energy of the charge-transfer absorption, and the Hammett para constant has been demonstrated. This correlation and an associated one with the redox potential of the flavin (11) have been used as evidence that the phenolate ion is the charge-transfer donor in the enzyme-phenol complex and the isoalloxazine ring of FMN is the acceptor (5,12).
OYE is prepared from brewer's bottom yeast using a phenol affinity column (13). This protein preparation ran as one 45-kDa band on SDS-PAGE. Subsequent experiments demonstrated that this enzyme preparation could be separated into multiple peaks using high performance liquid chromatography (14) and FPLC (8). Amino terminal sequence determination of the FPLC peaks suggested that more than one isozyme of OYE exists and that the enzyme as prepared was a mixture of homoand heterodimers derived from separate genes. OYE1 was cloned from brewer's bottom yeast, and the expressed protein coeluted with one of the bands from the FPLC separation of the isozymes. Southern blotting confirmed at least one more gene for OYE in brewer's bottom yeast (15). Another clone of OYE was isolated from a CenA library of Saccharomyces cerevisiae. The expressed protein had 91% sequence identity with OYE1 and was somewhat displaced from any of the FPLC peaks seen with the protein from brewer's bottom yeast (8). Deletion of the coding region for OYE2 from the genome of S. cerevisiae had no readily detectable phenotype (16), and it was possible to isolate a second OYE by affinity chromatography, which was more anionic than OYE2. This second isozyme from S. cerevisiae was cloned and designated OYE3 (17).
The cloning of the isozymes of OYE resulted in the expression and isolation of large quantities of homodimeric protein, making it possible to crystallize OYE in a form suitable for structural determination. The crystal structure of OYE1 was solved at 2-Å resolution (18), and amino acid residues that influence the binding of FMN to OYE were identified. The crystal structure was also solved with p-iodophenol and phydroxybenzaldehyde bound to the enzyme. The phenolic ligand was positioned parallel to the si face of the isoalloxazine ring of FMN, and the phenolate oxygen was within hydrogenbonding distance of His-191 and Asn-194. Solution of the crys-tal structure with an NADPH analogue showed the amide oxygen of the nicotinamide ring was positioned approximately the same as the phenolate oxygen with the C-4 position of the nicotinamide ring close to N-5 of FMN, suggesting optimal positioning of NADPH for hydride transfer to the flavin (19).
In order to evaluate the effect of Asn-194 and His-191 on ligand binding and enzyme function, these residues were mutated in OYE1. His-191 was changed to Asn, Asn-194 was changed to His, and the two residues were exchanged to yield OYE1 H191N/N194H. These mutated proteins have been expressed in Escherichia coli and procedures for isolation developed. The ligand binding properties, the enzymatic properties, and the crystal structures of the variants of the mutations have been studied and compared with those of native OYE1.
Mutation of OYE1-Sequence alignments and restriction endonuclease mapping was done using the GCG (Genetic Computer Group) program maintained by the Clinical Research Core of the University of Michigan. Oligonucleotides were designed to introduce a restriction endonuclease site along with a point mutation and were synthesized by the University of Michigan Molecular Biology Core Facility. The Oligonucleotide primers had the following sequence: H191N (GGTGTTGA-AATTAATAGTGCTAACGGT), N194H (GAAATTCACAGTGCACAC-GGT TACTTG), and H191N/N194H (GTTGAAATTAATAGTGCACAC-GGTTACTTG). A plasmid containing a 3.5-kilobase pair insert with the open reading frame for OYE1 in pGEM3Zf(Ϫ) was used for the generation of single-stranded DNA. This plasmid was used to transform E. coli JM109, and helper phage R408 was used for the rescue of singlestranded DNA. Mutagenesis was accomplished with the Sculptor in vitro mutagenesis system (Amersham Pharmacia Biotech), which is based on the phosphorothioate technique of Eckstein (20). Success of the introduction of the mutation was monitored by establishing the introduction of the appropriate restriction endonuclease site.
Expression of OYE1 and OYE1 Mutants-The expression plasmid for OYE1 contained a 1.7-kilobase pair insert in pET3b and was maintained in E. coli HMS174. The expression plasmid for the mutants was constructed by replacing the 750-base pair KpnI-ApaI fragment from the native OYE1 expression plasmid with the equivalent fragment from the mutated pGEM plasmid. The sequence of the mutant plasmid was confirmed by the University of Michigan Molecular Biology Core Facility using the T7 primer and three oligonucleotides designed to span the total coding region of the OYE.
Calcium-competent E. coli BL21DE(3) cells were transformed with the OYE plasmid for expression of the protein. Growth was in Luria Broth (1% tryptone, 0.5% yeast extract, 0.5% NaCl) containing ampicillin (50 g/ml) at 30°C. After 10 -12 h, IPTG was added to a concentration of 0.4 mM and growth allowed to proceed for an additional 10 -12 h.
Native OYE1 was isolated using a phenol affinity column as described previously (13). It was not possible to isolate the mutated proteins with this procedure, and a combination of anion exchange and gel exclusion chromatography was used. The first anion exchange column was DEAE-Sepharose, which was loaded in 40 mM Tris, pH 8.0, and eluted with a Tris, pH 8.0, gradient of 40 -250 mM (10). Fractions were evaluated by flavoprotein spectra and by SDS-PAGE. Fractions containing OYE were combined, dialyzed overnight in 10 mM Tris, pH 8.0, containing 10 M phenylmethylsulfonyl fluoride, 10 M FMN. The dialysate was loaded on a Q-Sepharose column and eluted with a sodium chloride gradient (0 -250 mM) in 10 mM Tris, pH 8.0 (8). The Q-Sepharose fractions containing OYE were combined, concentrated with a Centriprep 30 (Amicon), and dialyzed overnight against 10 mM potassium P i , pH 7.0 containing phenylmethylsulfonyl fluoride. The concentrated, dialyzed OYE sample was then applied to an S-200 Sephadex column (0.9 ϫ 180 cm) that had been equilibrated in 10 mM potassium P i , pH 7.0, and eluted with the same buffer. An OYE fraction, which ran as a single 45-kDa band on SDS-PAGE and had a flavoprotein spectrum with a 278 nm/460 nm ratio of approximately 10, was isolated from the column.
Spectral Properties of Old Yellow Enzyme-Spectral properties of OYE and OYE mutants were measured using a Cary 219, Hewlett-Packard diode array, or a Cary 3 spectrophotometer. The spectra of oxidized OYE1 and the H191N and H191N/N194H mutants were recorded in 0.1 M potassium P i , pH 7.0 buffer. OYE1 and OYE mutants were titrated with phenolic ligands to determine dissociation constants for the protein-ligand interaction. Dissociation constants were determined by the change in absorbance noted on the addition of the ligand in 0.1 M potassium P i , pH 7.0 at 25°C. The concentration of enzymeligand complex was determined from the extent of spectral change (A 460 Ϫ A long wavelength max ) compared with that at saturating ligand concentration and K d calculated from the relation Cuvettes with a 1 cm path length were used for most determinations, but 10-cm path length cuvettes were used for determination of phenolic ligand binding to OYE1 in order to get a more accurate measurement of K d for tight binding ligands.
The extinction coefficient of protein-bound FMN was evaluated. In all cases, the spectrum of protein-bound FMN was recorded, the protein was denatured by the addition of 0.1 volume of 50% trichloroacetic acid, and the precipitated protein removed by centrifugation. The supernatant was neutralized with solid sodium bicarbonate and the spectrum recorded against a reagent blank containing everything but the flavoprotein. In an alternate method, the protein was denatured by the addition of 0.05 volume of 10% SDS, with spectra recorded until no further changes occurred due to release of FMN. The extinction coefficient of the enzyme bound flavin at its wavelength maximum was calculated from the known extinction coefficient of 12,500 M Ϫ1 cm Ϫ1 of FMN (21).
Reaction Kinetics-The kinetics of the reduction of the enzyme and mutants by NADPH were measured using a stopped flow spectrophotometer as described previously (17). Reduction of OYE1 and the double mutant H191N/N194H by NADH was determined using the stopped flow spectrophotometer, but the reduction of the H191N mutant was determined under anaerobic conditions using the diode array spectrophotometer. In all cases data were acquired at pH 7.0 in 0.1 M potassium phosphate at 25°C.
The oxidation of the OYE1, H191N, and the double mutant H191N/ N194H by oxygen and cyclohexenone were studied by stopped flow spectrophotometry. After reduction under anaerobic conditions by xanthine/xanthine oxidase (22) or an NAD(P)H generating system with glucose-6-phosphate dehydrogenase and glucose 6-phosphate in which the concentration of NAD(P)H was adjusted so that the reduction rate would not interfere with the subsequent reoxidation, the enzyme was mixed with 0.1 M potassium P i , pH 7.0, 25°C, bubbled for 15 min with various concentrations of oxygen or anaerobically with various concentrations of cyclohexenone.
Steady State Turnover-Turnover of OYE1 and the mutants by NADPH and oxygen was determined by stopped flow spectrophotometry using initial rates of reaction monitoring absorbance change at 340 nm. The turnover of OYE1 and the double mutant H191N/N194H by NADPH and cyclohexenone was also determined with the stopped flow instrument in potassium P i , pH 7.0, 25°C. Turnover of the H191N mutant with NADPH and cyclohexenone was determined under anaerobic conditions by monitoring the absorbance change at 340 nm at various concentrations of cyclohexenone and NADPH with the Cary 219 spectrophotometer.
X-ray Crystallographic Methods-Single crystals of mutants H191N and H191N/N194H were grown from the same conditions used for wild type OYE1 (18), except the precipitant concentrations were: 30% PEG400 (v/v) for H191N and 28% PEG400 (v/v) for H191N/N194H. These crystals were isomorphous, belonging to space group P4 3 2 1 2, with unit cell parameters a ϭ b ϭ 143.3 Å, c ϭ 43.0 Å for H191N, and a ϭ b ϭ 143.6 Å, c ϭ 42.8 Å for the double mutant. X-ray diffraction data were collected on a San Diego Multiwire System area detector (23) with a Rigaku RU-200 rotating anode (Cu-K␣) as described previously (18). The data were then merged using SCALEPACK (24), and their qualities were evaluated with program Rmeas (25). All crystallographic refinements were carried out using program X-PLOR version 3.1 (26). The wild type OYE1 structure (Ref. 18; PDB entry 1OYA) was used as the starting model, with residues 191 and 194 changed appropriately, as guided by the electron density maps. These initial models were then refined against their corresponding diffraction data between 8 Å and the highest resolution limits, using conventional positional refinement and then restrained individual B-factor refinement, followed by manual adjustments guided by 2F o Ϫ F c (␣c) and F o Ϫ F c (␣c) maps. Several rounds of refinement yielded the final models. The data collection statistics, crystallographic refinement statistics, and the final model quality are shown in Table I. Soaking trials with various ligands (including p-hydroxybenzaldehyde) were also carried out, but crystals were unstable in the x-ray beam and no usable data were collected.

Mutagenesis-
The mutagenesis primers were designed to introduce restriction endonuclease sites concomitantly with the amino acid substitution. The oligonucleotide primers introduce the mutations H191N and N194H and the double mutation H191N/N194H, as well as the silent mutation, which introduced a VspI site with the H191N mutation and an Alw44I site with the N194H mutation. Restriction endonuclease digestion followed by agarose gel electrophoresis verified that the H191N mutation has a new VspI site and N194H mutation has a new Alw44I site. The double mutant, H191N/N194H, has both mutations and both restriction endonuclease sites.
The expression plasmid for the mutant proteins was constructed in pET3b, which was the expression vector for wild type OYE1 (15). The mutation was introduced by replacing the 750-base pair KpnI-ApaI segment of the coding region with the mutated coding region. The complete plasmid was sequenced, and it was established that the DNA sequence contained the expected mutation in an otherwise unaltered coding region.
Expression of Mutants-E. coli BL21(DE3), E. coli BL21(DE3)pLysS, and E. coli BL21(DE3)pLysE were transformed with plasmid DNA for the expression of OYE1 and the mutants of OYE1. The expression of OYEI and the mutant enzymes was highest when E. coli BL21(DE3) was the host and this strain was used for the expression of all OYE clones. At 37°C some of the expressed protein was associated with the pellet of the cell lysate, but the yield of soluble protein increased at 30°C. Time of induction with IPTG was 10 -12 h with growth allowed to proceed for an additional 12 h.
There was a large fluctuation in the expression level of the different OYE mutants. H191N was expressed at levels that appeared to be equivalent to that of wild type OYE1. The double mutant H191N/N194H was expressed less efficiently, and the N194H mutant was expressed only at low levels. It is not clear if this was a result of failure to express the mutant protein or an increased lability of the expressed protein.
Isolation of the Expressed Proteins-Old Yellow Enzyme from brewer's bottom yeast and the cloned isozymes of OYE are isolated using a phenol affinity column (13). Oxidized OYE binds to the phenolic ligand attached to a Sepharose support. After thorough washing to remove other proteins in the cell extract, the purified protein is eluted with a reducing buffer containing sodium dithionite. When this procedure was used with the expressed mutant proteins, they did not bind to the phenol affinity column, which was a first reflection of the fact that phenol ligand binding affinity was severely decreased.
The isolation procedure used for the mutants was a modification of the one which had been used originally to isolate OYE (10). The cell extract was first applied to a DEAE-Sepharose column and eluted with a Tris, pH 8.0 gradient (40 -250 mM). The second anion exchange resin was Q-Sepharose eluted with a sodium chloride gradient (0 -125 mM) as described for the separation of the isozymes of OYE from brewer's bottom yeast on FPLC (8). The final step in the isolation procedure was size exclusion chromatography with a S-200 Sephacryl column (0.9 ϫ 180 cm) eluted with potassium phosphate buffer (pH 7.0, 0.1 M). The yield of OYE1 using the phenol affinity column is approximately 60 mg of protein/liter of culture. Using the anion exchange columns and gel exclusion chromatography, the yield of H191N was 20 -25 mg/liter of culture and the yield of the double mutant was 6 -8 mg/liter. In all cases the resulting product ran as a single band on SDS-PAGE and had a ratio A 280 /A 460 of less than 10. The yield of OYE1 N194H was so low that it was not possible to isolate pure material.
Spectral Properties of OYE1, OYE1 H191N, and OYE1 H191N/N194H-The spectra of FMN, OYE1, and the mutant enzymes are slightly different. Free FMN has wavelength maxima at 370 and 445 nm. These maxima were shifted to higher wavelengths when bound to all enzyme forms (Fig. 1). The extinction coefficient of the protein-bound FMN was determined for OYE1 and the mutant enzymes. Even though SDS binds to OYE, when added to a final concentration of 0.5% denaturation with complete liberation of FMN occurs over a period of 10 -30 min, allowing a direct calculation of the extinction coefficient of the enzyme-bound flavin at its wavelength maximum using the known extinction of 12,500 M Ϫ1 cm Ϫ1 for free FMN (21). In a second method, trichloroacetic acid was used to denature the protein and release FMN. The denatured protein was removed by centrifugation and the supernatant neutralized with sodium bicarbonate. The sample was read against a blank containing everything except the enzyme, and the concentration of free FMN was determined. The wavelengths of the absorbance maxima as well as the extinction coefficients for the protein bound FMN are summarized in Table II.
The extinction coefficient used for the mixture of OYE isozymes from brewer's bottom yeast has been reported previously as 10,600 M Ϫ1 cm Ϫ1 (4, 10). The value determined for the cloned isozyme, OYE1, is higher. This higher value for the extinction coefficient of FMN bound to OYE1 was also confirmed by titrating FMN with apo-OYE1 and following the change in absorption spectrum as the holoenzyme was formed (results not shown).
Binding of Phenolate Ligands-Mutation of His-191 and Asn-194 was expected to alter phenol binding and development of long wavelength charge-transfer absorbance. Both amino acid residues are within H-bonding distance of the bound phenol oxygen and would be expected to stabilize the phenolate anion involved in charge-transfer complex formation. From the crystal structure of OYE1, it is evident that His-191 and Asn-194 could stabilize the phenolate anion form of phenols by c R meas is the corrected R-factor for describing data quality, as defined by Diederichs and Karplus (25). d The crystallographic R-factor is defined as R ϭ ¥ hkl ͉F obs Ϫ F calc ͉/¥ hkl F obs . e R free is the cross validation R-factor, calculated against 7% of the reflections that were not used in the refinements (27).
Old Yellow Enzyme:  hydrogen-bonding (18), effectively lowering the pK a of the bound phenol by several pH units (5). Thus it was not surprising that the H191N and double mutant forms failed to give the normal intense charge-transfer bands at pH 7.0 when the pK a of the free phenol is 8.0 or greater. However, with phenols possessing a low pK a , such as pentafluorophenol (pK a ϭ 5.3), charge-transfer absorbance bands are still observed with the mutant enzymes. In Fig. 2 the binding of p-hydroxybenzaldehyde to OYE1 is compared with the binding to H191N and H191N/N194H. There is a marked decrease in the extinction of the charge-transfer complex and a decreased perturbation of the flavoprotein spectra for the mutant enzymes compared with OYE1. The absorbance maxima for the charge-transfer bands of H191N and the double mutant with p-hydroxybenzaldehyde are shifted to longer wavelengths with a decrease in extinction and a decreased phenol binding affinity. Comparison of dissociation constants for OYE1 and the two mutant forms with p-chlorophenol, p-cyanophenol, p-hydroxybenzaldehyde, and pentafluorophenol are summarized in Table III. As expected, the mutant enzymes bind all phenols tested much more weakly that OYE1, consistent with the proposed hydrogenbonding to His-191 and Asn-194. Catalytic Turnover with NADPH and Oxygen-Turnover measurements of OYE1, H191N, and H191N/N194H with NADPH and oxygen were carried out by measuring initial rates of NADPH oxidation at 340 nm using the stopped flow spectrophotometer as a convenient tool for monitoring the reaction at different oxygen concentrations. Enzyme at approximately 100 nM concentration was mixed with NADPH at various concentrations between 50 and 500 M and oxygen at 10, 21, 50, and 100% saturation, and the oxidation of NADPH was followed at 340 nm. Plots of 1/turnover number versus 1/[NADPH] resulted in a series of parallel lines with a slope close to zero for OYE1 and the double mutant and an increased slope with H191N. Secondary plots of primary intercepts versus the reciprocal of the oxygen concentration resulted in finite values of k cat for OYE1 of 165 min Ϫ1 and H191N of 105 min Ϫ1 with similar K m values for both NADPH and oxygen. The kinetics with the double mutant (H191N/N194H) were almost second order in oxygen dependence. These results are summarized in Table IV.
Catalytic Turnover with NADPH and Cyclohexenone-Turnover of OYE1 with NADPH and cyclohexenone was determined under anaerobic conditions in the stopped flow spectrophotometer using 50, 500, and 1500 M starting concentrations of cyclohexenone and 60, 120, 250, and 400 M NADPH. The turnover number at saturating concentrations of both substrates was 250 min Ϫ1 . At the lowest concentration of cyclohexenone, turnover was maximal even at the lowest concentration of NADPH tested, and increasing concentrations of  NADPH did not result in an increased rate of turnover. At the higher concentrations of cyclohexenone, saturation with NADPH was reached at the same limiting turnover rate, but the apparent K m of NADPH was higher, indicating inhibition by high concentrations of cyclohexenone. This is consistent with the fact that OYE has one active site in which turnover occurs by a ping-pong mechanism and that cyclohexenone also binds tightly to oxidized enzyme effectively competing with NADPH. Turnover of H191N with NADPH and cyclohexenone was too slow for determination in the stopped flow spectrophotometer and was measured under anaerobic conditions with a Cary 219 spectrophotometer. Turnover was followed at 340 nm using concentrations of cyclohexenone of 1, 5, 10, and 20 mM and NADPH concentrations of 45-180 M. A double-reciprocal plot resulted in a series of parallel lines. A secondary plot of intercepts versus the reciprocal of the cyclohexenone concentration gave a V max for the reaction of 82 min Ϫ1 with a K m for cyclohexenone of 10 mM and a K m for NADPH of 55 M. While the H191N mutant has a V max that is approximately one-third that of OYE1, the K m for cyclohexenone is dramatically increased and is 2-3 orders of magnitude higher than the concentration of cyclohexenone that results in maximum turnover of NADPH by OYE1. The dramatically decreased ability of the double mutant to catalyze NADPH-cyclohexenone reductase activity is illustrated in the stopped flow experiment of Fig. 3. In this experiment, 3 equivalents of NADPH were mixed with the enzyme in the presence of 100 M cyclohexenone. The enzyme flavin was reduced rapidly and stayed essentially completely in the reduced state during turnover. The rate of NADPH oxidation was almost completely independent of the NADPH concentration, indicating a very low K m for NADPH. In the experiment shown in Fig. 3, it took 10 min before the NADPH depletion was complete and another 10 min before the flavin was completely reoxidized. The overall rate of turnover and that of flavin reoxidation was second order in cyclohexenone concentration over the range studied (Fig. 3, inset).
Reductive Half-reaction with NADPH-The kinetics of reduction of OYE1, H191N, and the double mutant, H191N/ N194H, by NADPH were measured by stopped flow spectrophotometry. In previous studies with the mixture of isozymes from brewer's bottom yeast, it was possible to demonstrate at least two oxidized enzyme intermediates before the reduction of enzyme-bound flavin, as shown in Scheme 1 (6,7).
The binding of OYE and NADPH took place in the 3-ms dead time of the stopped-flow instrument and could be followed by the change in flavin absorbance. The initial binding step was followed by the rapid appearance of a long wavelength absorbing charge-transfer complex of NADPH as charge-transfer donor and oxidized flavin as acceptor (steps k 3 and k 4 in Scheme   Old Yellow Enzyme: His-191 and Asn-194 1). This step was followed by the reduction of the enzyme flavin (steps k 5 and k 6 ) and the rapid equilibrium release of the NADP product from the reduced enzyme. Similar results were found for OYE1 and the two mutant forms, except that the intensity of the charge-transfer complexes with the latter was only about 50% that found with OYE1. The spectra of OYE1 are shown in Fig. 4. The rapid change of absorbance measured at 460 nm and in the 520 -600 nm range for OYE1 (k 3 ϩ k 4 ) demonstrated that the binding of NADPH to the enzyme is concentration-dependent while the rate of reduction of the enzyme (k 5 ϩ k 6 ) appears to be independent of NADPH concentration at 5. Reductive Half-reaction with NADH-It had been established previously for the mixture of OYE isozymes from brewer's bottom yeast that reduction of the enzyme by ␤-NADH proceeded with a limiting rate constant similar to that of OYE by ␤-NADPH (7). This was only marginally the case with cloned OYE1, one of the isozymes from brewer's bottom yeast. The reduction phase with NADPH was independent of pyridine nucleotide concentration at all concentrations studied with a rate constant of 5.1 s Ϫ1 . With NADH as the reducing agent, the observed rate constant of reduction was quite dependent on pyridine nucleotide concentration and a double-reciprocal plot gave a saturating k red of 0.9 s Ϫ1 , one-fifth that for NADPH reduction. In the case of the H191N mutant, reduction of the enzyme by NADPH was dependent on the NADPH concentration, but the limiting value of k red was similar to that of OYE1. The NADH reduction of the H191N mutant was so slow that it was more convenient to determine it under anaerobic conditions without the use of a stopped-flow spectrophotometer. The resulting k red was 0.03 s Ϫ1 , a decrease of approximately 150fold from that of NADPH reduction of H191N. The K d for NADH is of the same order of magnitude as that for NADPH. The reduction of the double mutant by NADH was studied in the stopped-flow spectrophotometer and yielded k obs values that did not reach saturation up to 8 ϫ 10 Ϫ4 M NADH. A double-reciprocal plot of the data resulted in a K d of approximately 5 ϫ 10 Ϫ3 M and k red of approximately 4 s Ϫ1 . This value is higher than that for reduction of OYE1 by NADH, but k red for NADPH with this enzyme form was also surprisingly higher than that found with OYE1. Results for all three enzyme forms are summarized in Table V.
Oxidative Half-reaction with Oxygen-Reduced OYE is oxidized by oxygen to yield H 2 O 2 . The kinetics of oxidation of OYE1, H191N, and the double mutant by oxygen were determined using the stopped flow spectrophotometer and in each case found to be second order in oxygen concentration. The rate constants are similar in magnitude and are not markedly affected by alterations of the amino acid residues His-191 and Asn-194. The rate constants for OYE1, H191N, and H191N/ N194H were found to be 3.8 ϫ 10 3 M Ϫ1 s Ϫ1 , 7.2 ϫ 10 3 M Ϫ1 s Ϫ1 , and 6.4 ϫ 10 3 M Ϫ1 s Ϫ1 respectively.
This oxidation reaction was dramatically less efficient with the mutant enzymes than with the wild type enzyme. Oxidation of OYE1 by cyclohexenone reaches saturation at micromolar concentrations of cyclohexenone as indicated in Fig. 5. From the double-reciprocal plot, a k ox of 10 2 s Ϫ1 and a K d of 3 ϫ 10 Ϫ5 M for cyclohexenone was determined. It was necessary to use millimolar concentrations of cyclohexenone to approach saturation in the oxidation of the H191N mutant, giving a K d of 5.6 mM and a limiting rate constant of 1.1 s Ϫ1 . The rate of oxidation of the double mutant by 2-cyclohexenone was also very much lower than that of OYE1 and was directly proportional to the REACTION 1 FIG. 4. Reaction of OYE1 with NADPH. OYE1 was mixed in the stopped-flow spectrophotometer with NADPH to give an enzyme concentration of 14.7 M and an NADPH concentration of 150 M. The initial spectrum of OYE1 is given by the continuous line. Reduction was followed at the wavelengths shown, and the spectra for the EFMN-NADPH charge-transfer complex (filled circles) and the reduced enzyme (open circles) were calculated from reaction traces. cyclohexenone at all concentrations of cyclohexenone tested (0.1-1.5 mM), indicating a dramatically decreased binding affinity for cyclohexenone. The rate constants determined for the oxidative and reductive half-reactions are summarized in Table V.
Structures of H191N and H191N/N194H-Crystal structures of the H191N and H191N/N194H variants have been determined at 2.3-and 2.7-Å resolutions, respectively. In each case the mutated residues are clearly defined by electron density maps. Neither variant shows global perturbation of the protein structure; however, small but significant shifts are observed in the active site (Fig. 6). Aside from the mutations themselves, the most notable difference in each variant structure is the absence of the chloride ion that in the structure of wild type OYE1 interacts with His-191 (3.4 Å), Asn-194 (3.2 Å), and Tyr-196 (3.5 Å) and is displaced in a charge-transfer interaction of phenols with the flavin (18).
In the H191N structure, the Asn-191 side chain is oriented such that the Asn-191 O␦ atom is equivalent to the wild type His-191 N␦ atom (ϳ0.6 Å), to conserve the hydrogen bond donated by the main chain nitrogen of Ala-193. This places the Asn-191 N␦ atom at the position occupied by His-191 C␦2, where it has no hydrogen bond partners. The chloride ion present in wild type OYE1 is replaced by an ordered water molecule (WAT661) in a nearby position. The shift is such that the water molecule can form reasonable hydrogen bonds with the side chain of Tyr-196 (3.1 Å), but makes no interaction with either Asn-191 (6.3 Å) or . In addition to these changes, the isoalloxazine ring slightly relaxes toward Asn-191, with the flavin N3 atom moving as much as 0.6 Å. This movement fills some of the void left by the loss of the chloride and the His-191 side chain. Interestingly, at the other end of the flavin ring, the side chain Tyr-375 shifts slightly (ϳ0.6 Å) in the same direction as it shifts upon ligand binding (Fig. 6). Finally, the loop between ␤6 and ␣6 (residues 293-306) has very weak electron density and has presumably become more disordered in the mutant structure: the average main chain B factor of this segment is 43 Å 2 for H191N, compared with 20 Å 2 for the wild type.
The double mutant, H191N/N194H, has Asn-191 situated as in the H191N variant, and His-194 is positioned similar to Asn-194, but with two additional atoms extending its reach. The conformation of the His-194 side chain appears to place its N␦ atom at the position occupied by the wild type Asn-194 O␦, where it can hydrogen-bond to WAT522 (2.5 Å), which is present in wild type OYE1. The C⑀ atom is situated only about 2 Å from the wild type WAT615 position and that water shifts 1.0 Å to a more reasonable distance of 2.9 Å. Since the additional length of the His-194 side chain points away from the Asn-191 side chain, the N194H mutation cannot compensate for the H191N mutation to recreate the ligand binding site unless it is reoriented upon ligand binding. Aside from the mutated residues, most of the structural changes are similar but in general larger than those seen in the H191N variant: the loss of the chloride, a shifting of the hydroxyl of Tyr-375 by ϳ1 Å, and the increased disorder of the loop between ␤6 and ␣6 (residues 293-306), which has even weaker electron density than that of the H191N variant. The average main chain B factor of this section is 51 Å 2, , despite the fact that its overall main chain atom B factor for the rest of the structure is even smaller than the wild type. As in H191N, the FMN cofactor also shifts, but in the double mutant the shift is more significant, with the dimethylbenzene ring, the ribityl group and the phosphate all further away from the substrate binding pocket. The C-7 atom of the isoalloxazine ring moves by as much as 0.75 Å. The shift of the FMN appears most directly related to the 1 Å displacement of WAT615, which in turn pushes the O-3 hydroxyl of the ribityl by 0.50 Å.

DISCUSSION
Old Yellow Enzyme is historically significant as the enzyme in which it was first demonstrated that a vitamin-derived cofactor is required for the catalytic function and thus is also the first discovered member of the large family of flavoprotein enzymes. In the last few years, many NAD(P)H oxidoreductase FIG. 5. Oxidation of OYE1 by 2-cyclohexenone. OYE1 was reduced by an NADPH-generating system consisting of glucose-6-phosphate (0.625 mM), glucose-6-phosphate dehydrogenase (5 l), and NADP (0.625 M). Reduction of the enzyme was followed spectrally and was complete at 84 min. The reduced enzyme was mixed in the stopped flow apparatus with the concentrations of cyclohexenone shown, and the reaction traces at 460 nm were used to calculate the values of k obs shown.
FIG. 6. Active site of OYE1. Overlay of four structures around the active sites: PHB bound wild type OYE1 as the thickest lines (PDB entry code 1OYB), empty wild type structure (PDB entry code 1OYA) as medium thickness lines, H191N as dashed lines, and H191N/N194H as thin lines.
Old Yellow Enzyme:  homologues of OYE have been cloned from plant, fungal, and bacterial sources. The sequences around His-191 and Asn-194, the residues involved in orientation of the active site ligand, are compared in Fig. 7. All of the homologues have a histidine at position 191 and an asparagine or histidine at position 194. Of the homologous proteins for which function has been measured, all, like OYEI, are able to reduce a double bond ␣,␤ to an aldehyde or ketone. Two of the OYE homologues for which function is known, morphinone reductase from Pseudomonas putida and 12-oxophytodienoate reductase from Arabidopsis thaliana, have the physiological function of reducing the olefinic bond of an ␣,␤ unsaturated ketone. The conservation of His-191 and Asn-194 throughout the large group of proteins indicates the importance of these residues to the protein function.
Solution of the crystal structure of Old Yellow Enzyme (18) made it possible to explain the positioning of the phenol ligands, which are known to act as electron donors in forming long wavelength (500 -800 nm) charge-transfer complexes with OYE (5). Fig. 6 depicts OYE1 with p-hydroxybenzaldehyde positioned parallel at 3.6 Å from the plane of the FMN ring overlaid with the structures of OYE1, H191N, and H191N/ N194H without ligand present. His-191 and Asn-194 are within hydrogen-bonding distance of the phenolate oxygen. It would be expected that mutation of these residues, even if the substitutions are conservative, would affect the ligand binding properties of OYE. The first evidence for compromised phenolic ligand binding by the OYE mutants came when it was found that the phenol affinity column that had been developed for the isolation of wild type OYE could not be used for isolation of the mutants, since they did not bind to the phenolic resin. It was necessary to use ion exchange columns and gel exclusion chromatography to purify the proteins from the E. coli extract.
Replacement of His-191 with Asn, resulting in two asparagine residues in the vicinity of the phenolic oxygen, reduced the binding affinities of the substituted phenols that were tested by 145-fold for p-hydroxybenzaldehyde and 2400 -4000-fold for p-chlorophenol, p-cyanophenol, and pentafluorophenol. This corresponds to a decrease in binding energy of 4.6 -4.9 kcal mol Ϫ1 , consistent with the removal of a strong hydrogen-bonding interaction. The extinction of the long wavelength absorbance band, if one was formed, was lower than that of the native OYE1, and the peak of the charge-transfer band was shifted to higher wavelength, i.e. lower energy. Since it has been estab-lished that the oxygen anion of the phenol ligand acts as the donor in the formation of the charge-transfer complex, a shift in the local pH of the ligand binding pocket would be expected to have the greatest effect on substituted phenols with high pK a values, such as p-chlorophenol, which has a pK a of 9.4. As expected, neither the H191N-OYE1 mutant nor the double mutant form a charge-transfer complex with p-chlorophenol. Substituted phenols with lower pK a values still form chargetransfer complexes at pH 7, presumably because of binding of the predominant phenolate ion.
The crystal structure of OYE1 shows hydrogen-bonding of the aldehyde group in p-hydroxybenzaldehyde to tyrosine 375. This could explain the less severe alteration in binding affinity of p-hydroxybenzaldehyde to H191N-OYE1, since there is still one unaltered residue to position the ligand. The K d for phydroxybenzaldehyde binding to H191N-OYE1 is 25 M, a 125fold decrease in binding strength compared with native OYE1, while the binding of p-cyanophenol, p-chlorophenol, and pentafluorophenol is more severely compromised.
The asparagine 194 residue also is positioned to affect the pyrimidine moiety of the isoalloxazine ring. The amide nitrogen of Asn-194 is 3.3 Å from N-1 and 3.5 Å from C-2-O of the flavin. These distances are not altered greatly by p-hydroxybenzaldehyde binding (18). It is a surprising result that replacement of Asn-194 with His results in the expression of a protein that is difficult to isolate, especially since some of the OYE homologues are able to accommodate two histidine residues in equivalent positions and since the double mutant H191N/N194H could be isolated. It appears that FMN is not bound well to the N194H variant. The structure of the double mutant clearly shows that the N194H mutation disturbs FMN binding by pushing the ribityl side chain. We speculate that the impairment is partially offset by the H191N mutation, which provides room for the FMN to shift in a way that can relax the strained interaction between His-194 and the ribityl. Other factors such as electrostatic effects and the binding of water (38) may also contribute to the instability of N194H, but no clear evidence for this is visible in the available structures.
The general trend toward longer wavelength charge-transfer complex absorbance is not seen with the double mutant (H191N/N194H). The effect of this mutation is to exchange the two amino acid residues involved in phenolate anion stabilization, resulting in the same amino acid residues being present. However, as the crystal structures show, the residues are in FIG. 7. Sequence of OYE and OYE homologues. The sequence of OYE1 is compared with that of OYE2 and OYE3 and other homologous proteins identified by the Basic Local Alignment Search Tool (28). The proteins are OYE1 from brewer's bottom yeast (15), OYE2 from S. cerevisiae (8), OYE3 from S. cerevisiae (17), KYE1 from Kluveromyces lactis (29), estrogen-binding protein (EBP I) from Candida albicans (30); glycerol trinitrate reductase (NER A) from Agrobacterium radiobacter (31); N-ethylmaleimide reductase (NEM A) from E. coli (32), 12oxophytodienoate reductase (OPDA) from A. thaliana (33), pentaerythritol tetranitrate reductase (PETN) from Enterobacter cloacae (34), MOR B (morphinone reductase (MOR B) from P. putida (35), bile acid-inducible operon (BAI H) from Eubacterium sp. (36), and NADH oxidase (NADH Ox) from Thermobacillus brockii (37). different orientations so that they do not recreate the original binding site. Formation of the charge-transfer complex between p-hydroxybenzaldehyde and the double mutant results in a decreased binding affinity similar to that seen with the H191N mutant and very little shift in the wavelength maximum of the charge-transfer transition. The absorbance seen on binding of pentafluorophenol appears as a broad shoulder with a low extinction on the flavoprotein peak. Binding of pentafluorophenol to OYE1 resulted in a long wavelength absorbance band at 550 nm. Exchange of the amino acid residues that affect phenolate binding could alter the electron accepting properties of the flavin resulting in collapse of this band to a shoulder. Again, there is little or no shift in the flavoprotein peaks at 380 and 460 nm upon binding of the phenolic ligands.
The long wavelength extinction coefficient is a measure of the probability that a charge-transfer transition will occur. For both the H191N and the double mutant, this value is decreased for all the ligands that form charge-transfer complexes. This is probably an indication that the highest occupied molecular orbital of the donor and the lowest unoccupied molecular orbital of the acceptor are not as favorably oriented in the mutants as in the native OYE1, again consistent with the importance of these residues in the interaction of the ligand with the enzyme-bound flavin.
The steady state turnover of NADPH with oxygen was not greatly affected by the H191N mutation, a result consistent with the similar values of k red and k ox (O 2 ). The k cat for the NADPH/oxygen turnover was 64% that of OYE1, and the effect of the mutations was chiefly characterized by the higher K m values for both NADPH and oxygen. At various oxygen concentrations, a series of parallel Lineweaver-Burk plots resulted similar to that previously found for all OYE preparations, showing a small dependence on NADPH concentration at all experimentally available NADPH concentrations. The double mutant also had a small dependence on NADPH concentration similar to that of native OYE1, but for this mutant enzyme the secondary plot of intercept versus the reciprocal of the oxygen concentration of the parallel Lineweaver-Burk plots was second order with respect to oxygen concentration over the experimentally available range of oxygen concentrations. Again, this result is consistent with the values of k red and k ox (Table V), since the K m for O 2 is given by k red /k ox ϭ 1.18 ϫ 10 Ϫ2 M (17), a concentration greatly in excess of the solubility of oxygen in water.
Turnover of NADPH with cyclohexenone was strongly compromised by the H191N mutation. While the NADPH dependence of the reaction was not greatly altered, the effectiveness of cyclohexenone as an electron acceptor was greatly reduced. Proper alignment of cyclohexenone to receive a hydride from N-5 of the flavin at the ␤-position of the unsaturated carbonyl compound requires that His-191 interact with the carbonyl oxygen in much the same way as the p-hydroxybenzaldehyde phenolate oxygen interacts with His-191 (9,18). If the mutation to Asn results in the lack of a hydrogen-bonding entity to interact with the oxygen, orientation of cyclohexenone would not be as well controlled by the mutant and loss of optimal hydrogen bonding would result in weaker binding, as found experimentally. NADPH/cyclohexenone turnover with the double mutant is also decreased seriously at most concentrations of NADPH and cyclohexenone. Over the range of cyclohexenone concentrations tested, the steady state turnover of NADPH by the double mutant was directly proportional to cyclohexenone concentration.
Study of the oxidative and reductive half-reactions was done independently by rapid reaction techniques. Reduction of native OYE1 and of the H191N mutant enzyme by NADPH were similar. NADPH is thought to align parallel to the isoalloxazine ring of FMN with the amide oxygen positioned similarly to the oxygen atom of the phenolate ligand and C-4 of the nicotinamide ring properly orientated for hydride transfer to N-5 of FMN (18,19). It would be expected that His-191 would have an orienting function and its replacement by another amino acid residue would alter the efficiency of hydride transfer, but this does not seem to be the case. Other forces such asinteractions between the isoalloxazine ring and the nicotinamide ring and the contribution of other amino acid residues to the binding of NADPH must be adequate enough that a noticeable difference in reduction rate does not occur. In addition, as proposed by Fox and Karplus (18), the binding of NADPH involves significant rearrangement of the loop between B6 and A6 (residues 290 -310), which makes extensive interactions with NADPH. Interestingly, in both mutant structures this loop becomes more flexible, and the change in its mobility may have an enhancing effect on NADPH binding, compensating for the loss of other interactions. The formation of the transient FMN-NADPH charge-transfer complex was observed during the reduction of the H191N mutant as it has been for all wild type forms of OYE. The formation of the charge-transfer complex of NADPH with OYE1 occurred with a rate constant of 350 s Ϫ1 and was dependent on the concentration of NADPH (k 2 /k 1 ϳ 100 M) while the reduction of the enzyme-bound FMN was independent of NADPH concentration (k 3 ϩ k 4 ϭ 5.1 s Ϫ1 ; k 4 small but indeterminate). The rate constant for formation of the charge-transfer transition between the oxidized H191N mutant and NADPH was lower, approximately 70 s Ϫ1 , and there was a dependence on NADPH concentration for both formation of the charge-transfer complex and subsequent reduction of the enzyme. The loss of an orienting residue in the active site of OYE could result in a decrease in the rate of the forward reaction, resulting in a situation where k 3 and k 4 in Scheme I are approximately equal. Because of the resulting two-step mobile equilibrium preceding flavin reduction, the observed dependence of the reduction of the H191N mutant on the NADPH concentration would be explained (K d overall ϭ k 2 k 4 /k 1 (k 3 ϩ k 4 ).
It is surprising that the double mutant, H191N/N194H, is reduced by NADPH with a rate constant that is so much greater than that for native OYE1 or H191N. The reduction is so fast that no formation of a charge-transfer complex between the oxidized enzyme and NADPH was discernible in the time frame of the stopped-flow experiments. To explain this, we suggest that the exchange of the two amino acid residues alters the orientation of the flavin (as seen in the structure) and the NADPH to more effectively align the C-4 of the NADPH nicotinamide and the N-5 of the FMN for hydride transfer. This surprising 15-fold increase in the rate of FMN reduction suggests that wild type OYE1 is not optimized for this process alone. In retrospect, this conclusion makes sense because the single active site pocket of OYE is designed to carry out two different redox reactions; the NADPH-dependent reduction of the FMN and the FMNH 2 -dependent reduction of an enone (for which 2-cyclohenexone is an analog), and thus it cannot be fully optimized for either reaction. Indeed, the mutations enhancing the rate of the first redox reaction have highly detrimental effects on the second redox reaction so that overall function is diminished.
Another surprising result of the comparison of OYE1 with the two mutant enzymes was that the rate of reduction of the mutant enzymes by NADH was so much lower than that by NADPH. For the mixture of isozymes from brewer's bottom yeast, the rate constant for reduction of the enzyme by NADH was approximately the same as that for NADPH. For the OYE1 isozyme the rate constant for reduction by NADH is approximately one-fifth that with NADPH, but reduction rate constants of the mutants are decreased more significantly by replacing NADPH with NADH. For the H191N mutant the decrease in rate constant is approximately 180-fold, and for the double mutant the rate decrease is approximately 20-fold. Although there is no apparent binding pocket for the additional phosphate on NADPH, as is usual for enzymes that are specific for NADPH (18), it is probable that the additional phosphate of NADPH interacts with the protein. The crystal structure with an analogue of NADPH in the active site showed the nicotinamide ring positioned such that the amide oxygen was oriented toward His-191 and Asn-194 with C-4 positioned adjacent to N-5 to FMN, but no fixed location for the remainder of the molecule was determined (18). The additional phosphate on NADPH is some distance from the functional nicotinamide ring and the N-5 position of FMN which receives the hydride from NADPH, so the functional difference could be due to the cumulative effect of mutation of the His-191 and Asn-194 amino acid residues and the loss of the additional phosphate on binding of the NADH cofactor to the enzyme.
Oxidation of the reduced OYE by oxygen was unaffected by the mutations, as demonstrated by oxidation rate constants that are approximately equivalent. This is not surprising since accessibility to the active site for the oxygen molecule should be unaffected by the mutations. Oxidation of reduced OYE by cyclohexenone would be affected by the orientation of this electron acceptor in the active site. Presumably the carbonyl oxygen of cyclohexenone is positioned much like the phenolate oxygen of the phenolic ligands and would therefore be greatly affected by a change in the amino acid residues which would orient the molecule in the active site. This orientation positions the ␤-carbon of the double bond of the substrate adjacent to the N-5 position of FMN from which hydride transfer can occur. Hydride addition to the ␤-carbon would result in an enolate anion form, which could be stabilized by hydrogen bonding to . Evidence that the ␣-hydrogen is derived from solvent via the adjacent tyrosine residue, Tyr-196, is presented in the accompanying paper (39). If the orienting residues are changed, the oxidation reaction would be expected to be altered, as is found. The oxidation of the single mutant, H191N, by cyclohexenone is more strongly compromised than is the oxidation of the double mutant enzyme. This effect is also reflected in the decreased affinity of H191N for binding of cyclohexenone.