Chorismate Synthase from the Hyperthermophile Thermotoga maritima Combines Thermostability and Increased Rigidity with Catalytic and Spectral Properties Similar to Mesophilic Counterparts*

Chorismate synthase, the last enzyme in the shikimate pathway, catalyzes the transformation of 5-enolpyruvylshikimate 3-phosphate to chorismate, a biochemically unique reaction in that it requires reduced FMN as a cofactor. Here we report on the cloning, expression, and characterization of the protein for the first time from an extremophilic organism Thermotoga maritima which is also one of the oldest and most slowly evolving eubacteria. The protein is monofunctional in that it does not have an intrinsic ability to reduce the FMN cofactor and thereby reflecting the nature of the ancestral enzyme. Circular dichroism studies indicate that the melting temperature of the T. maritima protein is above 92 °C compared with 54 °C for the homologousEscherichia coli protein while analytical ultracentrifugation showed that both proteins have the same quaternary structure. Interestingly, UV-visible spectral studies revealed that the dissociation constants for both oxidized FMN and 5-enolpyruvylshikimate 3-phosphate decrease 46- and 10-fold, respectively, upon heat treatment of the T. maritima protein. The heat treatment also results in the trapping of the flavin cofactor in an apolar environment, a feature which is enhanced by the presence of the substrate 5-enolpyruvylshikimate 3-phosphate. Nevertheless, stopped-flow spectrophotometric evidence suggests that the mechanism of the T. maritima protein is similar to that of the E. coli protein. In essence, the study shows that T. maritima chorismate synthase exhibits considerably higher rigidity and thermostability while it has conserved features relevant to its catalytic function.

Chorismate synthase catalyzes the final step in the shikimate pathway, which links the metabolism of carbohydrates to the biosynthesis of the three aromatic amino acids and many aromatic secondary metabolites in a series of seven enzymatic steps. The pathway is absent from mammals making it a prime target for the development of antimicrobial and herbicidal agents. Chorismate synthase itself is biochemically unique in nature in that it catalyzes a 1,4-anti-elimination of the 3-phos-phate group and the 6 (pro-R)-hydrogen from 5-enolpyruvylshikimate 3-phosphate (EPSP) 1 to yield chorismate (1,2). There is no other example of this type of catalysis known in nature, thereby making it exclusive. The enzyme has an absolute requirement for reduced FMN (3,4) and can be classified with regard to how it acquires this essential cofactor. The chorismate synthases for which the reduced flavin has to be supplied exogenously are referred to as monofunctional, e.g. those from Escherichia coli and plants (3,(5)(6)(7)(8)(9), while chorismate synthases which possess the intrinsic ability to reduce the flavin at the expense of NADPH are referred to as being bifunctional, e.g. the Neurospora crassa enzyme (4,10,11). From an evolutionary point of view, while it has been concluded from a phylogenetic analysis that all chorismate synthases are of monophyletic origin, it is not clear if the ancestral chorismate synthase was mono-or bifunctional (12). It has been suggested that the ancestral enzyme harbored the intrinsic flavin reductase activity (i.e. was bifunctional) as it is hard to imagine how this activity could have evolved in a framework of what are known to be monofunctional enzymes, i.e. the bacterial and plant chorismate synthases (12). In any case, as reduction of the FMN cofactor could be envisioned as a possible regulatory effector of chorismate synthase it would be intriguing to establish the history of how the enzyme has maintained reduction of the cofactor.
Recently, the genome of the hyperthermophilic bacterium Thermotoga maritima (T max ϭ 90°C, T opt ϭ 80°C) has been completed and 24% of its genes have been reported to be more similar to archaeal genes than to other bacterial genes (13). As this is a much higher percentage than that found in mesophilic bacteria (3-7%), it has been suggested that considerable lateral gene transfer has occurred between archaea and T. maritima (13). In fact, T. maritima is believed to be one of the oldest and most slowly evolving lineages in the eubacteria (14). Therefore, classification of T. maritima chorismate synthase with regard to how it acquires the reduced FMN cofactor can be considered to reflect the nature of the ancestral enzyme. In addition, thermophilic proteins, as a result of residing in an extreme environment, are more stable and tend to display greater conformational rigidity than mesophilic proteins. Therefore, a comparison of the characteristics of a thermophilic chorismate synthase with that of a mesophilic counterpart may provide insights not only into the stability but also on the self-organi-zation of this enzyme. Here we report the cloning, expression, and purification of chorismate synthase from the thermophilic eubacterium T. maritima. This is the first time that this enzyme has been described from an extremophilic organism. We show that the enzyme is monofunctional with respect to its requirement for FMN indicating that the ancestral protein was in fact monofunctional rather than bifunctional as had been proposed previously (12). In addition, we describe the characteristics of the thermophilic enzyme compared with those of the mesophilic E. coli chorismate synthase. In particular, we address the thermal stability as well as the spectrophotometric and electrophoretic properties of both the mesophilic and thermophilic enzymes in the presence and absence of ligands. We also report on the quaternary structure of T. maritima chorismate synthase.

EXPERIMENTAL PROCEDURES
Chemicals-Restriction endonucleases and DNA modification enzymes were purchased from Roche Molecular Biochemicals (Mannheim, Germany) or New England Biolabs (Beverly, MA). Oligonucleotide sequencing and polymerase chain reaction primers were obtained from Microsynth (Balgach, Switzerland). All other chemicals were of the highest grade available.
Cloning and Expression of T. maritima Chorismate Synthase-Cell paste (1 g) of the thermophilic organism T. maritima (strain MSB8) was kindly provided by Dr. Huber (University of Regensburg, Germany). The genomic DNA was isolated using the QIAamp Tissue kit (Qiagen) according to the instructions described by the manufacturer. The DNA sequence of the T. maritima aroC gene which codes for chorismate synthase was obtained from GenBank TM (accession number AE001715.1). The aroC gene was amplified by the polymerase chain reaction (PerkinElmer Life Sciences) (5Ј-primer, GGAATTCCATATG-AAACTCACGATAGCGGGGGATTCC; 3Ј-primer, CCCAAGCTTCTAC-CGATAGTGCTCTTTCCAGAACCC) based on the MSB8 strain sequence and cloned into the NdeI and HindIII restriction sites, respectively, of pET21a (Novagen). This vector allows expression of T. maritima chorismate synthase under control of the IPTG inducible T7 promoter. The expression construct was verified by sequence analysis and transformed into either E. coli BL21(DE3) or E. coli BL21-Cod-onPlus TM (DE3)-RIL cells (Stratagene). For analysis of expression, a single transformant was grown in 5 ml of 2 ϫ YT medium supplemented with 100 mg/liter ampicillin and this was used to inoculate a 50-ml culture. After 1 h at 37°C, expression was induced by addition of IPTG to a final concentration of 0.1 mM. Cells were allowed to grow for another 5 h at 37°C and were then harvested by centrifugation and subsequently stored at Ϫ80°C.
Protein Purification-E. coli cells (80 g) were resuspended in 20 ml of Buffer A (50 mM Tris-HCl, pH 7.5, containing 50 mM potassium chloride, 10% glycerol, 0.4 mM dithiothreitol, 1.3 mM EDTA, and 1.3 mM phenylmethylsulfonyl fluoride). Lysozyme was added to a final concentration of 1 mg/ml and after 20 min at room temperature the cells were further lysed by sonication. The cell debris was removed by centrifugation at 25,000 ϫ g for 30 min at room temperature. The supernatant was heated at 75°C for 15 min after which the precipitated protein was removed by centrifugation (25,000 ϫ g for 15 min at room temperature). The supernatant from the heat treatment was then subjected to anion exchange chromatography on a DEAE-Sephacel column (2.5 ϫ 16 cm) freshly equilibrated in Buffer A. After washing with Buffer A, protein was eluted by a linear gradient of 250 ml of Buffer A and 250 ml of Buffer B (50 mM Tris-HCl, pH 7.5, containing 200 mM potassium chloride, 10% glycerol, 0.4 mM dithiothreitol, 1.3 mM EDTA, and 1.3 mM phenylmethylsulfonyl fluoride). The progress of protein purification was monitored by 10% SDS-PAGE using the buffer system as described by Laemmli (17).
N-terminal Amino Acid Sequencing-Either a crude extract (as outlined above) from IPTG induced expression of E. coli BL21 Codon-Plus(DE3)-RIL cells harboring the constructed pET-T. maritima aroC plasmid or the purified protein was subjected to 10% SDS-PAGE. The electrophoresed sample was blotted onto a polyvinylidene difluoride membrane (Bio-Rad) according to the manufacturer's recommendations. The band of interest was excised and subjected to automated Edman degradation in an Applied Biosystems 477A sequencer.
Immunochemical Methods-For Western blot analysis, the samples were separated by 10% SDS-PAGE and blotted onto a nitrocellulose membrane. The membrane was blocked with TBS (10 mM Tris-HCL, pH 7.5, 0.9% sodium chloride) supplemented with 5% dry skimmed milk and 0.05% Tween 20. The blot was incubated for 90 min with an affinity purified antibody raised against chorismate synthase from the higher plant Corydalis sempervirens at a suitable dilution in the blocking buffer. The membrane was washed five times for 10 min each in TBST (TBS containing 0.05% Tween 20). It was then incubated for 45 min in TBST containing 5% dry skimmed milk and a sheep anti-rabbit peroxidase conjugate (Roche Molecular Biochemicals) diluted 1:3000. After another five washes of 10 min each in TBST, the blot was developed using the chemiluminescent system and according to the protocol supplied by Roche Molecular Biochemicals. Enzyme Assays-Chorismate synthase activity was assayed using forward coupling of the reaction to anthranilate synthase at 30°C essentially as described by Schaller et al. (7), except that the assay mixture used was 0.1 M potassium phosphate, pH 7.6, containing 30 mM ammonium sulfate, 10 mM glutamine, 4 mM magnesium sulfate, 1 mM dithiothreitol, 10 M FMN, 50 picokatal anthranilate synthase (from E. coli), and 80 M EPSP. Either 5 mM sodium dithionite or 1 mM NADPH were used to start the reaction.
CD Spectroscopy-CD measurements were performed with a Jasco-715 spectropolarimeter equipped with a computer controlled water bath, using thermostatted cuvettes of 0.2-mm path length. Thermal unfolding curves were measured by continuously measuring the ellipticity at 222 nm between 3 and 92°C at a scan rate of 1.0 degree min Ϫ1 and with data collection every 20 s. Protein samples were prepared for CD spectroscopy in 10 mM HEPES, pH 7.4, containing 50 mM ammonium sulfate.
PAGE-Native PAGE was performed using the Pharmacia Phast-System with an 8 -25% gel run at pH 8.8 for 240 A⅐V⅐h over a period of 45 min at 15°C. Either oxidized FMN and/or EPSP were added in 5-fold molar excess to samples of enzyme (13 M, in 10 mM HEPES, pH 7.4, containing 50 mM ammonium sulfate) as indicated in the legend to Fig.  3 and 20 min before beginning electrophoresis. Gels were stained with Coomassie Blue. The protein standards used were thyroglobulin (669 kDa), ferritin (440 kDa), catalase (232 kDa), aldolase (158 kDa), and bovine serum albumin (66 and 132 kDa).
Analytical Ultracentrifugation-Conventional sedimentation equilibrium measurements were made according to the method of Liu et al. (18) with a Beckman XL-A analytical ultracentrifuge (Beckman, Palo Alto, CA). The data were collected at 7,000, 9,000, and 12,000 rpm and at 20°C. The sample volume was 120 l at a protein concentration of 0.27 mg/ml in 10 mM Tris, pH 7.4, containing 90 mM potassium chloride. Protein samples were dialyzed exhaustively in the above buffer before use.
UV-Visible Absorbance Spectrophotometry-Absorbance spectra were recorded with a Uvikon 933 spectrophotometer (Kontron Instruments AG, Zü rich, Switzerland) equipped with a Haake D1 waterbath (Digitana AG, Horgen, Switzerland). The spectra were recorded in 10 mM HEPES, pH 7.4, containing 50 mM ammonium sulfate and by systematically varying either the substrate concentration or the temperature degree.
Rapid Reaction Spectrophotometry-Formation and decay of the flavin intermediate was observed using a stopped flow spectrophotometer equipped with a thermostatted 1-cm path length cell and a diode array detector (Spectroscopy Instruments GmbH, D-82205 Gilching) interfaced with a Macintosh IIcx computer. Data were acquired using SPECTRALYS 1.55 software (ZINS ZIEGLER-Instruments GmbH, Nobelstrasse 3-5, D-41189 Mönchengladbach). Rapid reactions were recorded between 300 and 600 nm, the integration time for collecting a spectrum was 10 ms with a resolution of 2 pixels/nm. FMN was reduced with sodium dithionite in anaerobic solutions of the substrate and a stoichiometric concentration of enzyme. The experiments were performed in 10 mM HEPES, pH 7.4, containing 50 mM ammonium sulfate.

RESULTS
PCR of the aroC Gene from T. maritima-Under the conditions used here, the polymerase chain reaction resulted in the amplification of a fragment which was ϳ1150 base pairs in size. Including the engineered restriction sites (19 base pairs), this fragment thus corresponds to the predicted size of the encoding aroC gene from T. maritima (1128 base pairs) based on the DNA sequence and was confirmed by DNA sequencing analysis.
Expression and Purification of Recombinant T. maritima Chorismate Synthase-The PCR product obtained was cloned into the expression vector pET21a to create the construct pET-TmaroC. Expression of T. maritima chorismate synthase (41.7 kDa) could be obtained in E. coli BL21(DE3) cells, as is the conventional protocol for this system (Fig. 1, panel A (Fig. 2, lane 3) was refined in the first instance by a heat precipitation step which was optimal at 75°C. At this temperature the majority of the contaminating proteins precipitated (Fig. 2, lane 4) and could be removed by centrifugation, while the T. maritima chorismate synthase remained in solution (Fig. 2, lane 5). The most prominent contaminating protein remaining at this stage was chloramphenicol acetyltransferase from E. coli (25.7 kDa), as determined by N-terminal amino acid sequencing (15 cycles of 50 pmol of protein resolved the amino acids MEKKITGYTTVDISQ, which are identical to the data base entry for this protein, accession number CAA67774), and could be removed by chromatography using DEAE-Sephacel (Fig. 2, lane 6). The identity of the purified band as T. maritima chorismate synthase (41.7 kDa) was confirmed by N-terminal sequencing analysis (15 cycles of 50 pmol of protein resolved the amino acids MKLTIAGDSHGKYMV, which are identical to the data base entry for this chorismate synthase, accession number Q9WYI2), and the protein was estimated to be greater than 95% pure, as judged from SDS-PAGE (Fig. 2,  lane 6). A typical yield of T. maritima chorismate synthase under these conditions was 30 mg from 80 g of E. coli cell paste.
Requirement of T. maritima Chorismate Synthase for Reduced FMN-The purified recombinant T. maritima choris-mate synthase is catalytically active (Table I) and its specific activity in the presence of dithionite and at 30°C is calculated as 135.5 nmol⅐mg Ϫ1 ⅐min Ϫ1 . The specific activity obtained with T. maritima chorismate synthase (Table I) is lower than that obtained with two mesophilic chorismate synthases (Table I, N. crassa and E. coli), assayed under the same conditions, which is at least partly due to nonoptimal temperature conditions for the thermophilic enzyme. While there was a 2-fold increase in the specific activity of T. maritima chorismate synthase on performing the reaction at 37°C (data not shown), the instability of anthranilate synthase above 40°C (the coupling enzyme used in this assay) did not permit determination of the optimal temperature of T. maritima chorismate synthase using this assay. In order to determine if T. maritima chorismate synthase is mono-or bifunctional, the enzyme assay was performed with either flavin reduced by dithionite exogenously, or adding NADPH to ascertain if the enzyme has the intrinsic ability to reduce the FMN cofactor itself. In this way, the relative percentage of the enzyme activity obtained with the reduced flavin added exogenously to the intrinsic ability of the enzyme to reduce the FMN cofactor itself is a measure of its mono/bifunctionality (where approaching 100% indicates a bifunctional enzyme). As the specific activity obtained for T. maritima chorismate synthase in the presence of NADPH is only 0.5% of that in the presence of dithionite (Table I,    Determination of the Melting Temperature of Chorismate Synthase in the Presence and Absence of Ligands-The stability of both E. coli and T. maritima chorismate synthase in the presence and absence of both oxidized FMN and EPSP was examined using temperature unfolding experiments as measured by CD spectroscopy (Fig. 3). The melting temperatures of E. coli chorismate synthase in the absence and presence of ligands (54.7 Ϯ 0.5°C and 58.9 Ϯ 0.04°C, respectively) indicate a stabilizing effect of the ligands (Fig. 3, A-C). The melting temperature of T. maritima chorismate synthase was estimated to be a minimum of 92°C in both the presence and absence of ligands (Fig. 3, D-F). However, the difference (if any) of the melting temperature for this enzyme with and without ligands could not be determined from these experiments.
Native PAGE of Chorismate Synthase in the Presence and Absence of Ligands-Chorismate synthase from E. coli has been reported to be a homotetramer (156 kDa) and when run on native-PAGE (8 -25%) appears as a diffuse band with a mobility corresponding to an apparent mass of ϳ190 kDa (19). When the enzyme is preincubated in the presence of both oxidized FMN and EPSP and subjected to native-PAGE as for the enzyme alone, there is a marked shift in the mobility toward the anode (140 -150 kDa), which does not occur in the presence of either substrate alone (19). The same analysis was performed here with T. maritima chorismate synthase and its properties were compared with those of the E. coli enzyme (Fig. 4). In contrast to the E. coli enzyme (Fig. 4, E. coli chorismate synthase, lane 1), T. maritima chorismate synthase alone appears as a sharp band with a mobility corresponding to ϳ200 kDa (Fig. 4, T. maritima chorismate synthase, lane 1). However, when T. maritima chorismate synthase is incubated in the presence of both oxidized FMN and EPSP, there is a slight shift in mobility toward the anode akin to what was observed with the E. coli enzyme but not nearly as pronounced (Fig. 4, compare lane 4 of E. coli and T. maritima chorismate synthase). Similar to what was observed with the E. coli enzyme, there is no apparent shift in the mobility of T. maritima chorismate synthase when the enzyme is preincubated with either of the substrates alone (Fig. 4, compare lanes 2 and 3 of E. coli and T. maritima chorismate synthase, respectively).
Determination of the Quaternary Structure of Chorismate Synthase-The molecular mass of T. maritima chorismate synthase was estimated by performing sedimentation equilibrium experiments at a variety of operational speeds in an analytical ultracentrifuge. A representative data set from a sedimentation equilibrium experiment is shown in Fig. 5. The data can be interpreted as resulting from a single ideal species according to the model of Liu et al. (18) under all the conditions used. The molecular mass of the enzyme was calculated to be 168,377 Ϯ 1950 Da under each of the conditions used. The molecular mass of the monomer estimated from the amino acid sequence is 41,754 Da. In addition, the enzyme prepared here gave a single homogenous band on SDS-PAGE with a molecular mass of Ϸ42 kDa. Therefore, it can be concluded that T. maritima chorismate synthase is a tetramer of identical subunits.
UV-Visible Absorption Spectral Properties of T. maritima Chorismate Synthase-Chorismate synthase isolated from T. maritima under the conditions used here, showed a single absorption maximum at 278 nm (Fig. 6, panel A). There were no absorbance maxima characteristic for bound flavin (i.e. Ϸ370 nm and Ϸ450 nm), thus indicating that chorismate synthase is isolated as the apoenzyme form. Oxidized free FMN (25 M in 10 mM HEPES, pH 7.4, containing 50 mM ammonium sulfate) exhibits absorption maxima at 372 and 445 nm, respectively (Fig. 6, panel A). Addition of 28 M chorismate synthase to 25 M oxidized FMN at room temperature caused no perturbation of these maxima (Fig. 6, panel A). Subjecting this solution to ultracentrifugation (Centricon 30) allowed the estimation of a K D of 137 M for binding of oxidized FMN to the enzyme at room temperature (Table II). This result indicates the weak binding of the oxidized co-factor under these conditions and accounts for the isolation of the protein in the apoenzyme form and is akin to what has been observed with E. coli chorismate synthase (20). A gradual increase of the temperature of the chorismate synthase-oxidized FMN solution to 66°C led to a hypsochromic shift of the absorption maximum at 372 to 360 nm and was accompanied by a hypochromic effect on the absorption intensity in this region (Fig. 6, panel B). Albeit to a lesser degree, the increase in temperature also produced an increase in the resolution of the absorption maximum around 450 nm and induced a slight hypsochromic shift of the absorption maximum in this region from 445 to 442 nm (Fig. 6, panel  B). The K D for the binding of oxidized FMN to the chorismate synthase after this heating step was estimated to be 3 M, indicating 46-fold tighter binding.
The addition of EPSP to the heated chorismate synthase/ FMNox solution causes spectral changes which are predominantly manifested in the 450-nm absorption maximum only (Fig. 6, panel B). The main effect is an amplified resolution of the absorption maximum in this region compared with the heating step and a bathochromic shift of the maximum from 442 to 447 nm (Fig. 6, panel B). The spectral changes are highlighted by observing the difference spectra as shown in Fig. 7. From this figure it can be seen that the temperature increase mainly affects the near UV region (Fig. 7A), while the changes observed upon the addition of the substrate are mainly in the visible region of the spectrum, respectively (Fig. 7B). The spectral changes occurring on heating a solution of T. maritima chorismate synthase/FMNox and EPSP (Fig. 7C) were the same as the sum of the changes which occurred upon heating and then adding the EPSP (Fig. 7D). Moreover and importantly, the spectral changes observed due to either temperature and/or adding EPSP were not reversed by cooling. Heating free FMNox from 28 to 66°C produced a slight decrease in the intensity of the absorption maxima at 372 and 445 nm (data not shown), but the changes were insignificant compared with those observed in the presence of the protein and EPSP. The presence of EPSP slightly decreases the K D of oxidized FMN at 28°C and after the heat treatment, respectively (Table II). We also estimated an apparent K D for the binding of EPSP to the enzyme in the presence of oxidized FMN from the absorbance changes at 383 nm as a function of the concentration of EPSP (Fig. 6, panel B, inset). The apparent K D estimated for EPSP (3 M) from these experiments decreases 10-fold upon heat treatment of the enzyme (Table II). Characterization

TABLE II Dissociation constants (K D ) of T. maritima chorismate synthase for oxidised FMN and EPSP at 28°C
The dissociation constants for oxidised FMN were estimated in the presence and absence of EPSP after subjecting the solution to centrifugation in Microcon concentrators (YM-30) at 28°C and with and without heating to 66°C. The dissociation constants for EPSP were estimated from the change in absorbance at 383 nm as a function of the concentration of EPSP under the same conditions as for oxidised FMN using the software Sigma Plot (Jandel Scientific).  (22). Complementary studies showed that in the absence of substrate and under the conditions used, pH 7.0, the reduced flavin is bound to the protein in its deprotonated or monoanionic state (pK a of N 1 -H ϭ 6.7) (20), while in the presence of substrate the reduced flavin is bound in its protonated or neutral form (23). Hence, the occurrence of the flavin intermediate reflects the protonation of the deprotonated reduced flavin upon binding of substrate. As this has been one of the key observations in the investigation of the mechanism of chorismate synthase, we performed the same study here using rapid reaction spectrophotometry and the T. maritima enzyme. The difference in absorption observed between a solution of T. maritima chorismate synthase/FMNred in the presence and absence of EPSP, respectively, are shown in Fig. 8. From the inset in Fig. 8, it can be seen that in the presence of substrate there is a rapid increase in the absorbance at 390 nm which is complete within the dead time of the instrument (10 ms) followed by a quasi-steady state phase before the absorbance decreases to its initial value before adding EPSP. From the series of difference spectra at the indicated times it can be seen that maxima are reached at 319, 390, and 476 nm after 200 ms which decrease with time to their initial value (6 s) with isosbestic points at 359 and 349 nm, thus indicating the homogeneity of this species (Fig. 8). These observations are akin to those reported for the E. coli enzyme (21,22). DISCUSSION To our knowledge, chorismate synthase has not been purified to homogeneity and characterized from any extremophilic or-ganism until now. Characterization of the enzyme was aided by cloning and expressing the enzyme in an E. coli system which resulted in a yield of T. maritima chorismate synthase suitable to enable the observations reported in this study. The purification procedure is quick and relatively easy, the greatest refinement being achieved by the heat treatment step at 75°C which removes almost 80% of contaminating proteins as judged from SDS-PAGE analysis. The monofunctionality of T. maritima chorismate synthase from the evolutionary standpoint of chorismate synthases is very interesting. While it has been concluded from an earlier phylogenetic analysis that chorismate synthases are monophyletic (12), it is not known to date if the ancestral chorismate synthase is mono-or bifunctional. However, it has been suggested that the common ancestor was probably bifunctional given that it is difficult to imagine the evolution of the intrinsic reductase activity in a framework of monofunctional enzymes (12). It was surmised that bifunctionality may have either been maintained only in organisms in which the availability of reduced flavin is limiting or perhaps there was positive selection of monofunctionality (12). T. maritima is thought to be one of (if not) the oldest eubacterium and appears to have undergone considerable lateral gene transfer from the archaea (14). A phylogenetic tree of all chorismate synthases presently known (data not shown) suggests that T. maritima chorismate synthase diverged with the archaea and moreover, considerably before any of the chorismate synthases for which bifunctionality is known (i.e. N. crassa and Saccharomyces cerevisiae). Thus the classification of its chorismate synthase as monofunctional could be considered to be cognate to the ancestral chorismate synthase and would therefore not lend support to bifunctionality being ascendant.
This is the first report on the thermal denaturation behavior of any chorismate synthase. For the mesophilic E. coli enzyme the 4°C increase in the melting temperature in the presence of ligands appears to be related to the major conformational change in E. coli chorismate synthase in the presence of FMN and EPSP to a more compact structure (19). As the melting temperature of T. maritima chorismate synthase could only be estimated under the conditions used here it was not possible to ascertain the effect (if any) on ligand binding. However, obviously and as would be expected, the melting temperature (92°C) is higher than the optimal temperature for growth of the organism (80°C).
The quaternary structure of T. maritima chorismate synthase is represented by a tetramer of identical subunits. This appears to "fit" with the known quaternary structures of other chorismate synthases in that the hierarchy is spread between that of dimer-tetramer (12). Interestingly at the quaternary level, a remarkable feature of hyperthermophiles is the occurrence of anomalous states of association and fused multifunctional proteins (24). However, even though there are many examples for these features (24), T. maritima chorismate synthase does not appear to be one of them. Therefore, additional quaternary interactions cannot account for the higher stability of this protein. This question may be answered in the future by performing specific mutations in the contact surface of the subunits to establish their contribution to the quaternary structure.
From the native-PAGE studies performed here it could be concluded that the apoprotein from T. maritima appears to have a higher structural rigidity compared with that of the E. coli apoprotein which is reflected by the sharp band of the former compared with the rather diffuse band of the latter on native-PAGE. The change in the mobility of E. coli chorismate synthase in the presence of ligands has previously been interpreted to reflect less conformational flexibility than that of the apoprotein (19). In support of this, the decrease in mobility of T. maritima chorismate synthase in the presence of ligands is not as pronounced as that observed with the E. coli enzyme which may reflect the lower flexibility of the thermophilic apoprotein compared with that of the mesophilic enzyme.
In aqueous solution, the near UV-visible absorption spectra of free flavins exhibit two featureless bands at about 450 and 375 nm (25). However, in solvents less polar than water the band at 375 nm shifts to a shorter wavelength and the visible band at about 450 nm shows greater resolution with two pronounced shoulders which are thought to be associated with vibrational transitions in the first electronic absorption at this wavelength (26,27). Additionally, it has been shown that the spectra of many flavoproteins are akin to that observed for free flavin in an apolar solvent indicating such an environment of the flavin cofactor when bound to the protein. Importantly, most flavoproteins show a shift in the 370 nm region to a shorter wavelength and a varying resolution of the band at about 450 nm (25). With T. maritima chorismate synthase, an apolar type flavin spectrum is observed when the temperature of the protein/FMN solution is increased, thus reflecting induced FMN binding. This is borne out by the fact that the K D for FMN binding is ϳ46-fold lower after heating to 66°C and cooling to 28°C compared with that measured at 28°C without heating to 66°C (3-137 M, respectively). Upon FMN binding, the spectral transition is mainly manifested in the 370-nm region which is consistent with solvent-dependent spectral shifts (more apolar) with little development of fine structure. This is accompanied by slightly better resolution in the 450-nm region which indicates that the vibronic transitions in this area are more discrete. This is clearly due to binding of the flavin to the protein which reduces the rotational freedom of the flavin and possibly introduces some strain on the bound cofactor. The phenomenon is further enhanced by the presence of EPSP which substantially increases the band resolution in this area and thus probably the constraint on the flavin. This observation may reflect that T. maritima chorismate synthase has a more rigid/compact structure at room temperature which upon heating becomes more flexible resulting in an increase in the rate of ligand binding and in fact probably allows the cofactor to enter into the active site. This view is supported by the fact that cooling the protein/FMN solution to room temperature after heating does not reverse the process, indicating that the FMN is now "trapped" in the active site.
The flavin-derived transient intermediate previously observed for the E. coli enzyme (21) was also observed in this study with T. maritima chorismate synthase. As for the E. coli enzyme, formation of this flavin species was associated with binding of the substrate, i.e. formation of a ternary complex between enzyme, reduced flavin, and EPSP. Formation of this complex is considered to reflect protonation of the anionic reduced flavin, at least for the E. coli enzyme (20,23). This protonation of the reduced flavin indicates that the flavin experiences a different polarity in the active site when substrate binds to the enzyme and is supported by the apolar-type flavin spectra observed here when EPSP is added to T. maritima chorismate synthase/FMNox after heating. The observation suggests that the catalytic action of T. maritima chorismate synthase proceeds with a mechanism similar to that of E. coli chorismate synthase, i.e. flavin-derived intermediate formation (reduced FMN in neutral form) precedes C-O and C-H bond cleavage of EPSP (22).
Finally, one of the major drawbacks at present in proceeding with studies on this intriguing enzyme is the lack of a threedimensional structure. For x-ray crystallography, it has been suggested that adequately diffracting crystals can be obtained more readily with thermophilic enzymes due to their greater rigidity and stability (28). The apparent rigidity and stability of the protein described here coupled with the characteristics of substrate binding make it an exceptional source with regard to the possibility of elucidating the structure of chorismate synthase.