Spectroscopic and Kinetic Characterization of the Bifunctional Chorismate Synthase from Neurospora crassa

Chorismate synthase catalyzes theanti-1,4-elimination of the phosphate group and the C-(6proR) hydrogen from 5-enolpyruvylshikimate 3-phosphate to yield chorismate, a central building block in aromatic amino acid biosynthesis. The enzyme has an absolute requirement for reduced FMN, which in the case of the fungal chorismate synthases is supplied by an intrinsic FMN:NADPH oxidoreductase activity, i.e. these enzymes have an additional catalytic activity. Therefore, these fungal enzymes have been termed “bifunctional.” We have cloned chorismate synthase from the common bread mold Neurospora crassa,expressed it heterologously in Escherichia coli, and purified it in a three-step purification procedure to homogeneity. Recombinant N. crassa chorismate synthase has a diaphorase activity, i.e. it catalyzes the reduction of oxidized FMN at the expense of NADPH. Using NADPH as a reductant, a reduced flavin intermediate was observed under single and multiple turnover conditions with spectral features similar to those reported for monofunctional chorismate synthases, thus demonstrating that the intermediate is common to the chorismate synthase-catalyzed reaction. Furthermore, multiple turnover experiments in the presence of oxygen have provided evidence that NADPH binds in or near the substrate (5-enolpyruvylshikimate 3-phosphate) binding site, suggesting that NADPH binding to bifunctional chorismate synthases is embedded in the general protein structure and a special NADPH binding domain is not required to generate the intrinsic oxidoreductase activity.

Chorismate synthase catalyzes the seventh step in the shikimate pathway, which utilizes the pentose phosphate metabolite erythrose-4-phosphate and the intermediate of glycolysis, phosphoenolpyruvate, to generate aromatic amino acids and other essential aromatic compounds. The step catalyzed by chorismate synthase is formally an anti-1,4-elimination reaction of the phosphate group and the C-(6proR) hydrogen from 5-enolpyruvylshikimate 3-phosphate (EPSP), 1 introducing a second double bond into the ring Scheme 1 (1,2). Although the reaction does not involve a net change in redox state, the enzyme has an absolute requirement for reduced FMN that is not consumed during substrate turnover. The role of reduced FMN has been subject to intense mechanistic studies (reviewed in Macheroux et al. (3)) that have led to a proposal of a mechanism involving radical chemistry (4,5). Another issue revolving around the requirement for reduced flavin concerns the generation of the reduced cofactor and its sequestration by chorismate synthase. Chorismate synthases from eubacteria and plants have been shown to rely on an external source of reduced FMN, produced for example by a NAD(P)H-dependent FMN oxidoreductase. Because the reduced form of FMN binds much more tightly to the enzyme (6), it is thought that free reduced FMN is sequestered by chorismate synthase either from the cellular environment or is provided by a specific oxidoreductase. Chorismate synthases that depend on an external source of reduced FMN have been termed "monofunctional." On the other hand, chorismate synthase from fungal sources, i.e. Saccharomyces cerevisiae and Neurospora crassa, have been found to possess an additional intrinsic catalytic activity in that they can utilize NADPH to reduce the flavin cofactor (7,8). These enzymes have been termed bifunctional. So far bifunctionality appears to be restricted to chorismate synthases from fungal species, and the physiological implications of this exclusive occurrence in fungi are not known. Although the bifunctional chorismate synthase from N. crassa was among the first chorismate synthases studied (7), the low abundance of the protein has prevented detailed studies of its biochemical properties. The higher molecular mass of the N. crassa enzyme (by 5-8 kDa) as compared with monofunctional bacterial and plant enzymes led to the hypothesis that N. crassa CS contains an additional NADPH binding site (9). However, attempts to identify this domain by construction of deletion mutants have not substantiated this hypothesis (9), and the mode and location of NADPH binding in the protein remain unknown.
All chorismate synthases isolated so far have been shown to bind FMN only weakly (K d (Escherichia coli) ϭ 30 M, (6)), and as a result, the FMN cofactor is entirely lost during purification. However, in presence of EPSP-oxidized FMN, binding is 1000-fold tighter in the case of the E. coli enzyme (6). Therefore, because binding of oxidized FMN to the N. crassa chorismate synthase can be assumed to be a prerequisite for its reduction by NADPH, it is conceivable that formation of a ternary complex of enzyme, FMN, and EPSP is required before reduction can occur. A similar substrate-enhanced reduction of the flavin cofactor by NADPH has been described with flavoprotein-dependent hydroxylases (10,11). In these enzymes, NADPH-dependent reduction of flavin is up to 5 orders of magnitude faster in the presence of the substrate. In contrast to reactions in which redox equivalents are consumed stoichiometrically during substrate turnover, the FMN cofactor of chorismate synthase remains in its active reduced form during EPSP consumption (7,12). However, molecular oxygen inactivates the enzyme as it reacts with reduced flavin to yield the oxidized flavin and hydrogen peroxide. Although exact rate constants for reoxidation of the enzyme-bound reduced FMN by molecular oxygen have not been determined, this process occurs rapidly in vitro (6). Because the redox state of the flavin cofactor is crucial to chorismate synthase activity, it can be envisaged that the redox balance in the cell may play a critical role in regulating the generation of chorismate and perhaps the flow through the shikimate pathway as a whole. However, the parameters that may affect the rate of reoxidation of the chorismate synthase-bound reduced FMN have not been studied experimentally.
The bifunctional chorismate synthase provides an opportunity to directly study the reduction of its flavin cofactor with the physiological reductant NADPH as well as to monitor the reoxidation of the flavin by molecular oxygen with the same enzyme. Therefore, we have inserted the cDNA of the previously cloned gene for N. crassa chorismate synthase (9) into a high copy expression plasmid that can direct the expression of large quantities of the enzyme, thus allowing an extensive biochemical characterization. In this paper we report on the expression and purification of the enzyme and give a detailed analysis of its spectral and kinetic properties.

MATERIALS AND METHODS
Reagents-All chemicals were of the highest grade available and obtained from Sigma or Fluka (Buchs, Switzerland). DEAE-Sephacel was from Amersham Pharmacia Biotech, and cellulose phosphate (P11) was from Whatman. DNA restriction and modification enzymes were obtained from Roche Molecular Biochemicals or New England Biolabs (Beverly, MA). Polymerase chain reaction primers were from Microsynth (Balgach, Switzerland). EPSP was a generous gift of Prof. Dr. John Coggins, University of Glasgow, UK.
Expression and Purification of Anthranilate Synthase-The plasmid pCH11-11B containing the E. coli trpE gene coding for anthranilate synthase component I was a generous gift from Dr. P. Pouwels (University of Rijswijk, The Netherlands). This plasmid is a derivative of the pCH plasmid family containing two copies of trpE. Competent E. coli PC 1562 cells (⌬trpA-E, tryptophan auxotroph) were transformed with pCH11-11B and subsequently cultured according to Hessing et al. (15) overnight at 37°C. The cells were then harvested, and the wet cell paste was stored at Ϫ80°C. Bacterial extracts were analyzed for their expression of anthranilate synthase using 10% SDS-PAGE.
In a typical preparation, cell paste from two liters of culture medium (ca. 20 grams) was resuspended in 20 ml of buffer A (0.1 mM potassium phosphate, pH 7.0, 1 mM dithiothreitol) and lysed by sonication. Cell debris and insoluble components were removed by centrifugation. The resulting supernatant was dialyzed against buffer B (0.01 M potassium phosphate, pH 7.0, 1 mM dithiothreitol) and then loaded onto a DEAE-Sephacel column (2.5 ϫ 16 cm) equilibrated in the same buffer. After washing, the bound protein was eluted with a linear gradient of buffer B and buffer C (0.6 M potassium phosphate, pH 7.0, 1 mM dithiothreitol). Fractions were analyzed for the presence of anthranilate synthase by means of SDS-PAGE, pooled, and dialyzed against buffer B. To remove chorismate synthase from the protein preparation, the dialyzed sample was loaded onto a cellulose phosphate column (2.5 ϫ 16 cm) prepared according to the manufacturer's instructions and subse-quently equilibrated with buffer B. Washing the column under these conditions with buffer B resulted in chorismate synthase binding to the resin, and anthranilate synthase was found in the flow-through. The flow-through was concentrated to ϳ3 mg of protein ml Ϫ1 and stored at Ϫ80°C. The concentration of the anthranilate synthase preparation was determined spectrophotometrically at 280 nm using a molar extinction coefficient of ⑀ 280 ϭ 20720 M Ϫ1 cm Ϫ1 , calculated according to Mach et al. (13) or with Bradford reagent (Pierce) using bovine serum albumin for calibration.
Molecular Techniques-Basic molecular manipulations were performed using standard techniques (28).
Cloning of the aroC Gene from N. crassa-The previously constructed pTrc99a-NcCS vector (9) was used for the amplification of the aroC gene, which codes for chorismate synthase from N. crassa. Two new restriction sites for the endonucleases NdeI at the 5Ј-end and HindIII at the 3Ј-end of the coding sequence were introduced by polymerase chain reaction. The amplified NdeI/HindIII fragment was inserted into the pET21a vector (Novagen, Lucerne, Switzerland), yielding pET21a-NcCS. The construct was verified by DNA sequencing and found to be identical to the sequence published in the N. crassa genome sequencing project (MIPS N. crassa data base at www.mips.biochem.mpg.de/proj/neurospora). The recombinant plasmid was transformed either into competent E. coli BL21(DE3) or BL21CodonPlus(DE3)-RP cells (Stratagene, La Jolla, CA) for analysis of expression.
Purification of N. crassa Chorismate Synthase (NcCS)-All steps were carried out at 4°C. Twenty-five g of bacterial cell paste was resuspended in buffer A (50 mM Tris-HCl, pH 7.5, containing 50 mM KCl, 10% (v/v) glycerol, 1.3 mM EDTA, and 0.4 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride) supplemented with 1 mg/ml lysozyme and 0.1 mg/ml DNase. After 1 h of stirring, cells were further lysed by sonication. Insoluble cell debris was removed by centrifugation (30,000 ϫ g, 20 min). The resulting supernatant was then subjected to a 40-50% ammonium sulfate fractionation. The precipitate from 50% ammonium sulfate saturation, which contained NcCS, was collected by centrifugation and dissolved in buffer A, dialyzed against the same buffer overnight, and then subjected to anion exchange chromatography on a DEAE-Sephacel column (2.5 ϫ 16 cm). After washing with buffer A, proteins were eluted using a linear gradient from 50 to 200 mM KCl (in buffer A). The progress of protein purification was monitored by 10% SDS-PAGE. Fractions containing NcCS were pooled and dialyzed against buffer B (10 mM potassium phosphate, pH 6.6, containing 10% (v/v) glycerol, 0.4 mM dithiothreitol, 1.3 mM EDTA, and 0.45 mM phenylmethylsulfonyl fluoride) and then loaded onto a cellulose phosphate column (2.5 ϫ 16 cm) extensively equilibrated with buffer B. After washing with buffer B, NcCS was eluted with a linear gradient from 10 -500 mM potassium phosphate in buffer B. Fractions containing NcCS were pooled, concentrated by ultrafiltration and transferred into 50 mM MOPS pH 7.5. The concentration of NcCS was determined spectrophotometrically at 280 nm using a molar extinction coefficient ⑀ 280 ϭ 18860 M Ϫ1 cm Ϫ1 , calculated according to Mach et al. (13) or with the Bradford reagent (see above).
Enzyme Assays-Anthranilate synthase activity was determined using the procedure of Ito et al. (14), except that 0.03 M ammonium sulfate was added to the reaction mixture as an amino group donor (15).
Chorismate synthase activity was measured using forward coupling to the reaction catalyzed by AS as described by Schaller et al. (16). The reaction buffer contained 0.1 M potassium phosphate, pH 7.6, 4 mM MgSO 4 , 10 mM glutamine, 0.03 M ammonium sulfate, 1 mM dithiothreitol, 10 M FMN, 80 M EPSP, and 50 picokatal of recombinant anthranilate synthase component I from E. coli. FMN was reduced either by the addition of 5 mM dithionite or 1 mM NADPH.
SDS-PAGE and Western Blot Analysis-Protein samples were separated by SDS-PAGE performed on 10% gels according to Laemmli (17). Gels were stained with Coomassie Brilliant Blue-R250 or alternatively blotted onto nitrocellulose membranes. For the immunodetection of NcCS, an antibody raised against Corydalis sempervirens chorismate synthase (described in 16) and affinity-purified against E. coli chorismate synthase was used.
Site-directed Mutagenesis-To change amino acid residue 98 from a methionine to lysine, the QuikChange site-directed mutagensis kit from Stratagene (La Jolla, CA) was used. The construct pET21a-NcCS served as the template. The following oligonucleotides containing the appropriate codon exchange (ATG 3 AAG) were used for the procedure (the changed codon is underlined): 5Ј-GACTACGGTAACAAGACTAAG-GACATCTACCCCCGCCCT-3Ј and 3Ј-AGGGCGGGGGTAGATGTCCT-TAGTCTTGTTACCGTAGTC-5Ј. All manipulations were performed after the manufacturer's instructions. The mutation was verified with an SCHEME 1 ABI 373 DNA sequencer (Applied Biosystems, Foster City, CA) using the fluorescent dideoxy chain termination method.
N-terminal Amino Acid Sequencing-E. coli crude extracts were separated on 10% SDS-PAGE. Proteins separated by electrophoresis were then transferred onto a polyvinylidene difluoride membrane (Bio-Rad). The band of interest was excised and subjected to automated Edman degradation in an applied Biosystems 477A sequencer.
Fluorescence Spectroscopy-Fluorescence emission spectra were monitored with a Kontron spectrofluorometer (model SFM-25). The excitation wavelength for tryptophan fluorescence was 280 nm and for flavin fluorescence, 365 nm. All spectra were recorded in 50 mM MOPS, pH 7.5, at 25°C.
Stopped-flow Spectrophotometry-Rapid reaction kinetics were performed using a stopped-flow spectrophotometer equipped with a thermostatted 10cm path length cell and a diode array detector (SI Spectroscopy Instruments GmbH, Gutach, Germany). When necessary, enzyme and substrate solutions were made anaerobic by exchanging the dissolved oxygen with argon by several cycles of evacuation and flushing. Analysis of the data was performed using the program Specfit/32 (Dr. R. A. Binstead, Spectrum Software Associates, Marlborough, MA). Presumably the ATG coding for methionine 98 serves as a fortuitous translation start, giving rise to ϳ50% of the total NcCS protein expression. A sequence alignment of all known chorismate synthases (data not shown) indicated that the methionine in this position is not conserved, but the majority of chorismate synthases feature a lysine in the corresponding position. Therefore, the ATG codon was replaced with AAG, introducing a lysine instead of the methionine in position 98. This manipulation of the cDNA resulted in complete abolishment of the expression of the truncated NcCS as is evident from Fig. 1 (panel B, M98Kϩ) and allowed the exclusive expression of the full-length enzyme. The purified M98K mutant protein has a specific activity of 0.8 units/mg, similar to the value of 0.7 units/mg reported by Welch et al. (7) for wild-type enzyme. Because the introduced mutation had no effect on the specific activity of the enzyme, the M98K mutant protein was used in all experiments throughout this study and is referred to as NcCS. Expression of the protein in this manner greatly facilitated its purification, and a homogeneous preparation was obtained in a three-step protocol comprising ammonium sulfate precipitation of NcCS from the crude protein extract followed by DEAE-anion exchange chromatography and phosphocellulose chromatography in an adaptation of the method used by White et al. (12). The progress of purification and the purity of NcCS is documented in Fig. 2. Typically, 70 -80 mg of recombinant NcCS was obtained from 25 g of wet cell paste.

Expression and Purification of N. crassa Chorismate
Determination of Dissociation Constants for 5-Enolpyruvylshikimate 3-Phosphate and FMN-Chorismate synthase was incubated with either EPSP, oxidized FMN, or both, and the equilibration concentration of the free ligands was determined by ultrafiltration through a 10-kDa cut-off membrane (Amicon). The data obtained are summarized in Table I. Importantly, the dissociation constant for EPSP in the absence of oxidized FMN was found to be 33 M and decreased slightly to 17 M in the presence of oxidized FMN under the same conditions (Table I). Using this same method, the dissociation con-  (Table I). The effect of EPSP on the binding of oxidized flavin with the N. crassa chorismate synthase is ca. 100-fold less pronounced than with the E. coli enzyme (6).
UV-Visible Absorbance Spectrophotometry of NcCS-Purified recombinant NcCS was completely devoid of bound FMN (data not shown), indicating weak binding of the cofactor (7). A similar finding has been reported for the enzyme from E. coli (6) and Thermotoga maritima (18). Yet binding of oxidized FMN to NcCS can be directly monitored by the difference UV-visible spectrophotometry, the results of which are shown in Fig. 3, panel A. The observed spectral changes are characterized by a hypochromic effect on the flavin absorbance with a maximum at 378 nm and an isosbestic point at 340 nm. Titration of NcCS with oxidized FMN showed saturation at high FMN concentrations, and a hyperbolic fit to the data produced a dissociation constant of 39 M (see Fig. 3, inset of panel A) for FMN binding. Similarly, stoichiometric amounts of NcCS (31 M) and oxidized FMN (25 M) were titrated with EPSP. As shown in Fig. 3, panel B, the difference spectra are very similar in the near UV range, between 300 and 400 nm, whereas marked differences are found between 400 and 500 nm in comparison with the titration in the absence of EPSP (Fig. 3,  panel A). In this range, a hyperchromic effect at 485 nm is observed with an additional isosbestic point at 475 nm (Fig. 3,  panel B). As for oxidized FMN, titration with EPSP showed saturation behavior with a hyperbolic fit to the data giving a dissociation constant of 9 M (see Fig. 3, inset of panel B). To facilitate direct comparison to binding studies carried out with the monofunctional chorismate synthases from E. coli (6) and T. maritima (18), absolute spectral changes were recorded in a separate experiment shown in Fig. 4. From such a comparison, it is evident that the UV-visible spectral changes are similar for all enzymes studied thus far (6,18), in particular in the near UV range. As noted previously, the observed hypsochromic shift of the near UV absorbance peak at 370 nm and the higher resolution of the peak at 450 nm are indicative of a more apolar flavin environment (6,18). The spectral changes were utilized to determine dissociation constants of oxidized FMN and EPSP in the presence of oxidized FMN as shown in the insets of Figs. 3 and 4 and are summarized in Table I.
Fluorescence Emission Quench Studies-The effect of EPSP on the fluorescence emission of enzyme-bound flavin was also examined. As shown in Fig. 5A, the flavin fluorescence emission was quenched by EPSP and was characterized by a broadening of the emission maximum at 525 nm accompanied with an almost 2-fold decrease in fluorescence intensity. This quench of fluorescence intensity is associated with the emergence of a shoulder at ϳ450 nm (Fig. 5A). To observe changes in the protein directly during binding of flavin and EPSP, tryptophan fluorescence emission was also examined. NcCS possesses a single tryptophan residue at position 109, which is in the vicinity of a highly conserved sequence motif (R 103 PGHAD 108 ). The addition of flavin to the protein results in  quenching of the tryptophan fluorescence emission (Fig. 5B, compare spectra 1 and 2) with further changes occurring during titration with EPSP (Fig. 5B). The observed decrease in fluorescence intensity upon the addition of EPSP is about 20% of the fluorescence emission (Fig. 5B). Both quenching processes showed a dependence on EPSP concentration and exhibited saturation behavior (Fig. 5, A and B, insets). A fit to the observed fluorescence quench data shown in Fig. 5, A and B, yielded dissociation constants for EPSP of 12 and 16 M, respectively, and are comparable with values obtained by other methods (Table I).

Reduction of NcCS-bound FMN with NADPH under Anaerobic Conditions-Although reduction of the NcCS-bound FMN
cofactor by NADPH has been inferred from chorismate synthase activity detected in its presence (7,9,12), direct spectroscopic evidence has not been reported thus far. The data shown in Fig. 6 demonstrate that oxidized FMN is reduced readily as a function of time at the expense of NADPH in the presence of NcCS. The reduction of the flavin is evident from the decrease of absorbance at the absorbance maximum of oxidized FMN at 450 nm. This process is associated with a decrease at 340 nm reflecting consumption of NADPH. Note that reduction of oxidized FMN occurs in the absence of substrate, showing saturation behavior with a K m (NADPH) of 10 M. Upon reoxidation of the reaction mixture with air, the initial absorbance spectrum of FMN is restored (dashed line in Fig. 6).
To obtain more accurate values for the rate of reduction of NcCS-bound FMN, the reduction of FMN was measured in a stopped-flow apparatus as a function of NADPH concentration. In the absence of EPSP, the reduction of FMN appeared to be independent of the NADPH concentration (in the range of 1-20-fold excess of NADPH with an average reduction rate of k red ϭ 70 ϫ 10 Ϫ3 s Ϫ1 (Table II). Similarly, in the presence of a 1.6-and 25-fold excess of EPSP over NcCS and FMN, the rate of reduction appeared to be independent of the NADPH concentration, with an average rate of k red ϭ 50 ϫ 10 Ϫ3 s Ϫ1 (Table II).
Oxidation of NcCS-bound Reduced Flavin by Molecular Oxygen-As mentioned earlier, EPSP turnover does not involve a net redox reaction, and the enzyme remains active as long as FMN is in its two-electron reduced form. Inactivation of chorismate synthase occurs upon reoxidation of the enzyme-bound reduced FMN by molecular oxygen, as was reported in the case of the E. coli enzyme (6); however, the rate of reoxidation was not determined. Therefore, we have studied the rate of reoxidation of fully reduced FMN bound to NcCS in the stopped-flow apparatus. The observed rates of reoxidation were found to be independent of the presence of EPSP, yielding secondary rate constants in the range of 1.25 to 1.92 ϫ 10 4 M Ϫ1 s Ϫ1 (see Table III (18 -20). Several pieces of experimental evidence suggest that this intermediate is the protonated reduced FMN. It returns to the unprotonated reduced flavin once the substrate has been completely consumed (4,19). These studies employed either the chemical reductant  dithionite or photoreduction. Because NcCS utilizes the biologically relevant reductant NADPH (Fig. 6) to reduce the FMN cofactor, this system offers the unique opportunity to study the generation and occurrence of the flavin intermediate under more physiological conditions. Moreover, formation of the reaction intermediate has thus far only been demonstrated for monofunctional (18,20) but not for bifunctional chorismate synthases. When NcCS-bound FMN is reduced by NADPH in the absence of molecular oxygen and mixed in a stopped-flow apparatus with EPSP, the intermediate is formed within the dead time of the instrument (ca. 5 ms) and is characterized by an absorbance maximum at 392 nm and a broad shoulder at longer wavelength (Fig. 7A). The spectral features of the intermediate are very similar to those reported with the enzymes from E. coli and T. maritima (18 -20). The diode array for data acquisition allowed both chorismate formation ( max, chorismate ϭ 285 nm) and the presence of flavin intermediate (at 392 nm) to be monitored simultaneously in a single experiment as is shown in Fig. 7, panel B Fig.  7, panel C. From this plot, it is evident that the flavin intermediate (as represented by the trace at 392 nm) is present during substrate turnover/chorismate formation and decays when chorismate is fully formed (trace at 300 nm). Fig. 6 demonstrate the reduction of FMN by NADPH under anaerobic conditions. In the presence of oxygen (256 M (21)), FMN (25 M) reduction occurs in the presence of NADPH and NcCS; however, the reduced flavin is spontaneously reoxidized by molecular oxygen. Hence, under the conditions used here, NADPH is completely consumed by these two processes (Fig. 8A). Because this consumption of NADPH also occurs in the absence of EPSP, it can be concluded that the oxidoreductase activity is functionally independent of the chorismate synthase activity. Under aerobic conditions, the rate at which NADPH is consumed (Scheme 2) depends strongly on EPSP concentration (Fig. 8A). At very low EPSP concentrations (0 -10 M), there is no observable effect on the   rate of oxidation of NADPH (Fig. 8A). However, under the experimental conditions used, the apparent rate of NADPH oxidation appears maximal at 25 M EPSP (6-fold excess over NcCS) (Fig. 8A). At higher EPSP concentrations, NADPH consumption is characterized by an initial lag phase followed by a fast decline in the NADPH concentration (Fig. 8A). The initial lag phase at higher EPSP concentrations corresponds to EPSP turnover as shown in Fig. 8B, indicating that the rate of NADPH oxidation is diminished during catalysis, i.e. EPSP and NADPH are competing for the same binding site. This competition is gradually relieved by consumption of EPSP, resulting in faster oxidation of NADPH, and gives rise to the complex time course of NADPH oxidation (Fig. 8A). The inhibition of NADPH oxidation (or FMN reduction) is also noticed in the rate of EPSP consumption because reduction of the flavin is required to activate the enzyme. Therefore, at high EPSP concentrations, the initial rate of EPSP consumption is reduced because of a lower concentration of reduced FMN/ NcCS that in turn results from a competition for the common binding site (Fig. 8B, inset). Because the concentration of EPSP decreases, this competition is relieved, leading to more reduced FMN/NcCS and, hence, a faster rate of chorismate formation.

Consumption of NADPH, Formation of Chorismate, and Reduction of NcCS-bound FMN in the Presence of Molecular Oxygen-The data shown in
In parallel studies we have also investigated the reduction of FMN during NADPH and EPSP turnover in the presence of oxygen. As shown in Fig. 9, FMN reduction monitored at 450 nm shows a rapid initial phase followed by an approach to a steady-state level of reduced FMN. This approach to steady state is delayed at higher EPSP concentrations, again indicating that during EPSP turnover access to the FMN binding site is blocked by EPSP, resulting in an inhibitory effect on flavin reduction.
Development of a Convenient Aerobic Chorismate Synthase Assay-In the previous sections it has been demonstrated that chorismate formation takes place in the presence of molecular oxygen (256 M at 25°C), i.e. despite reoxidation (and hence inactivation) of the NcCS-bound reduced FMN cofactor. We have exploited this feature to establish a reproducible aerobic assay for the chorismate synthase reaction as shown in Fig. 10. At low NADPH concentrations, the chorismate synthase reaction cannot be sustained long enough to convert all EPSP to chorismate as monitored at 281 nm (isosbestic point of the NADPH to NADP ϩ conversion). At moderate NADPH concentrations (Ͼ25 M) all EPSP is converted to chorismate, reach- ing a limiting rate at ca. 200 M NADPH. This set-up provides a simple continuous method to measure bifunctional chorismate synthase activity without the requirement for time-consuming operations to exclude oxygen from the assay mixture. DISCUSSION The present study provides a detailed biochemical characterization of a bifunctional chorismate synthase and, in particular, of its intrinsic oxidoreductase activity. The enzyme from the bread mold N. crassa has been heterologously expressed in E. coli (Fig. 1), and a three-step purification protocol has been established to obtain amounts of homogeneous chorismate synthase sufficient for spectroscopic and kinetic studies (Fig. 2). The protein was isolated in its FMN-free form, indicating that oxidized FMN binds only weakly. Similar observations have been reported for chorismate synthases from E. coli (6) and T. maritima (18). In agreement with this observation, titration of the enzyme with FMN revealed a rather high dissociation constant in the range of 40 -90 M (Table I). Again, similar dissociation constants have been obtained for the E. coli and T. maritima enzymes (30 and 140 M, respectively, (6,18)). In the presence of the substrate EPSP, binding of oxidized FMN is ca. 5-10 times tighter (Table I). This effect of EPSP is not as pronounced as has been found for the E. coli enzyme, which showed a 1500-fold tighter binding of oxidized FMN in the presence of EPSP (6). Binding of oxidized FMN alone to NcCS is associated with spectral changes in the UV-visible difference absorbance spectrum of the flavin (Fig. 3A). These changes are characterized by a hypsochromic shift of the near UV band of the flavin absorbance spectrum and have been described for the well characterized E. coli enzyme as well as for the T. maritima enzyme (6,18). Further spectral changes occur upon the addition of EPSP in the long wavelength region around 480 nm (see Fig. 3, panel B), indicating that the flavin environment is altered in the presence of EPSP. These spectral changes have been interpreted in terms of a more hydrophobic environment in the FMN binding pocket (6,22). This general interpretation was confirmed for the enzyme from E. coli (4), which produces an increase of pK a values for the ionizable groups of the flavin analogs 6-hydroxy-and 8-mercapto-FMN. This result has now been substantiated by fluorescence quenching of the NcCS⅐FMN complex during titration with EPSP. The blue shift of the flavin fluorescence maximum at 525 nm and the occurrence of an additional peak at 450 nm also suggest a more hydrophobic environment upon binding of EPSP to the binary NcCS⅐FMN complex (6). The hypothesis has been put forward that the hydrophobicity of the FMN binding pocket renders the reduced flavin a better electron donor for its assumed role as a transient electron transfer agent to EPSP to initiate cleavage of the C-O bond (4,5). However, to date no flavin radical species has been observed spectroscopically during EPSP turnover. The only intermediate in the chorismate synthase-catalyzed reaction that could be observed is the N(1)-protonated form of reduced FMN, which forms rapidly upon binding of EPSP and prior to C-O and C-H bond cleavage (6,19 The most salient property of NcCS is the utilization of NADPH to generate the essential reduced FMN cofactor, a characteristic feature of the bifunctional fungal chorismate synthases (Scheme 2). In the present work, we have shown that NADPH is consumed via reduction of FMN followed by reoxidation of the reduced FMN by molecular oxygen. During NADPH turnover in an aerobic solution, a steady-state concentration of reduced enzyme gives rise to EPSP conversion to chorismate. Therefore, NcCS is an appropriate system to study chorismate formation in an aerobic environment typically encountered in vivo.
The study of the reductive half-reaction (Scheme 2) revealed that the observed rate of flavin reduction appears to be independent of the NADPH concentration and the presence of EPSP. The first observation indicates that binding of NADPH and the transfer of redox equivalents from NADPH to oxidized flavin are not rate-limiting. In view of the weak binding of oxidized FMN, the most likely explanation of the slow and NADPH concentration-independent rate of reduction is that binding of oxidized FMN to NcCS is the rate-determining step in the overall reduction process. Earlier reports on the effect of EPSP on catalytic activity of N. crassa chorismate synthase were inconclusive because the role of FMN as an intrinsic factor of catalysis was poorly understood at that time (7,23). Our studies clearly show that EPSP exerts a negligible effect on the rate of FMN reduction under anaerobic conditions (see Table II). This finding is in contrast to the rate of reduction of FAD-dependent hydroxylases, which was found to be 4 -5 orders of magnitude faster in the presence of the aromatic substrate (10,11). However, it should be pointed out that in this case NADPH is consumed stoichiometrically during the course of the enzyme-catalyzed redox reaction, whereas in the case of chorismate synthase, reduction of FMN is required only once to achieve multiple turnover of EPSP. Therefore, one could argue that substrate-induced rate enhancement of FMN reduction is not required in the case of bifunctional chorismate synthase.
Because reoxidation of NcCS-reduced FMN by molecular oxygen leads to inactivation of the enzyme, it could be limiting to chorismate formation under aerobic conditions. The rate of reoxidation measured for NcCS-bound reduced FMN is in the same range as observed for flavoprotein oxidases (ca. 1.5 ϫ 10 4 M Ϫ1 s Ϫ1 versus 0.9 ϫ 10 4 M Ϫ1 s Ϫ1 for L-lactate oxidase (24), 8.5 ϫ 10 4 M Ϫ1 s Ϫ1 for glycolate oxidase (25), and 6.2 ϫ 10 4 M Ϫ1 s Ϫ1 for L-aspartate oxidase (26)), which have evolved to transfer substrate-derived redox equivalents to molecular oxygen. In comparison with these reoxidation rates, it can be concluded that the NcCS-bound reduced FMN is readily accessible to molecular oxygen, and hence, protection of the reduced FMN does not occur. Nevertheless EPSP turnover occurs in the pres-SCHEME 2 ence of oxygen (see Figs. 8B), indicating that it can effectively compete by molecular oxygen. This is clearly supported by the rates determined for EPSP turnover (1.3 s Ϫ1 under anaerobic conditions) and the rate of reoxidation of the catalytically essential reduced FMN by dissolved dioxygen (1.5 s Ϫ1 ).
Perhaps the most intriguing question revolving around bifunctional chorismate synthases is the nature and location of the NADPH binding site. Initially, it was thought that bifunctional chorismate synthases possess an additional domain that is responsible for NADPH binding (9). This proposal was mainly based on the larger molecular mass of the bifunctional as compared with the monofunctional enzymes (ca. 4 -8 kDa). Based on amino acid sequence alignments, Henstrand et al. (9) identified two regions in bifunctional enzymes that are absent in monofunctional enzymes, suggesting that these may be associated with NADPH binding. To test this hypothesis, several forms of the N. crassa enzyme lacking the areas in question were engineered (9). In all cases studied, utilization of NADPH as a source for redox equivalents was retained, indicating that none of the truncated regions are critical for NADPH binding. Hence, it can be concluded that the ability to bind NADPH is embedded in the chorismate synthase structure. In this context, it is noteworthy that the bifunctional chorismate synthases lack a typical dinucleotide Rossman fold (27), indicating an unusual mode of interaction between these chorismate synthases and NADPH.
In the present study, it is documented that the rate of NADPH oxidation strongly depends on the EPSP concentration; although at low EPSP concentration (equimolar to 10-fold excess over enzyme) the rate of NADPH oxidation is enhanced, this rate decreases significantly at higher EPSP concentrations, giving rise to a pronounced lag phase. This lag phase also corresponds to slower EPSP consumption (Fig. 8B), indicating that EPSP and NADPH compete for a common binding site. Because the rate of flavin reduction and reoxidation by molecular oxygen was found to be independent of the EPSP concentration in single turnover experiments (Table III), it must be concluded that this is not the case under multiple turnover conditions. In fact, under multiple turnover conditions and at low EPSP concentration (equimolar to 2.5-fold excess over enzyme), the initial reduction of FMN is faster (approaching steady state) and reaches a higher level of reduced FMN during steady state (Fig. 9). Hence, it appears that low EPSP concentrations have a stimulating effect on NADPH oxidation/FMN reduction. Because both NADPH and EPSP possess a phosphate group, it is conceivable that this group plays an important role in binding to the active site of chorismate synthase. The lack of a three-dimensional structure of chorismate synthase does not allow a more detailed discussion of this issue, but it should be noted that the fungal bifunctional chorismate synthases (N. crassa, S. cerevisiae, and Schizosaccharomyces pombe) feature several conserved basic amino acid residues absent in monofunctional enzymes. These residues may provide the correct spatial interactions to the phosphate group of NADPH necessary to achieve effective binding to the active site of bifunctional chorismate synthases.