Site-directed mutagenesis of a conserved region of the 5-enolpyruvylshikimate-3-phosphate synthase active site.

The active site of the enzyme 5-enolpyruvylshikimate-3-phosphate synthase (EPSPS) has been probed using site-directed mutagenesis and inhibitor binding techniques. Replacement of a specific glycyl with an alanyl or a prolyl with a seryl residue in a highly conserved region confers glyphosate tolerance to several bacterial and plant EPSPS enzymes, suggesting a high degree of structural conservation between these enzymes. The glycine to alanine substitution corresponding to Escherichia coli EPSPS G96A increases the Ki(app) (glyphosate) of petunia EPSPS 5000-fold while increasing the Km(app)(phosphoenolpyruvate) about 40-fold. Substitution of this glycine with serine, however, abolishes EPSPS activity but results in the elicitation of a novel EPSP hydrolase activity whereby EPSP is converted to shikimate 3-phosphate and pyruvate. This highly conserved region is critical for the interaction of the phosphate moiety of phosphoenolpyruvate with EPSPS.

' The abbreviations used are: EPSPS, 5-enolpyruvylshikimate 3phosphate synthase; S3P, shikimate 3-phosphate; PEP, phosphoenolpyruvate; HEPES, 4-(2-hydroxyethyl)-l-piperazineethanesulfonic acid HPLC, high performance liquid chromatography. of EPSP and phosphate (Pi) from PEP and shikimate 3- in an unusual carboxyvinyl transfer reaction. EPSPS has been extensively studied since it is the target for glyphosate (N-phosphonomethyl glycine) (4), the active ingredient of Roundupm herbicide, widely used for weed and vegetation control (5). Glyphosate is a competitive inhibitor with respect to PEP of EPSPS, and interacts with the E . S3P complex (6). Of the several known PEP-dependent enzymatic reactions, EPSPS is the only enzyme inhibited by glyphosate. The PEP binding region of EPSPS therefore appears to be unique; indeed, EPSPS is the only enzyme which interacts with PEP as an enzyme-substrate complex ( E .S3P) and not as the free enzyme. In this paper, we have probed the PEP binding region of EPSPS enzymes through site-directed mutagenesis and kinetic analysis.
Identification of the active site of EPSPS has been largely based on chemical modification studies. LysZ2 and Lys340 of E. coli EPSPS can be modified by reaction with pyridoxal phosphate (7) and o-phthalaldehyde (8), respectively, resulting in inactivation of the enzyme. Arg" and Arg131 of petunia EPSPS are modified by phenylglyoxal, the former being essential for enzyme activity (9). Cys408 of E. coli EPSPS is highly reactive, but not essential for enzymatic activity (10). This residue is proximal to the active site, since its modification with bulky reagents results in inactivation of the enzyme.
Another useful method of active site characterization is identification of substitutions which impact ligand binding and catalysis. Comai and co-workers have described a glyphosate-tolerant Salmonella typhimurium strain, wherein the tolerance to glyphosate results from a single amino acid substitution of PlOlS in the aroA gene encoding EPSPS (11, 12). We recently reported the isolation of an E. coli B variant, containing a highly glyphosate-tolerant EPSPS (13). Isolation and sequencing of the aroA gene encoding this glyphosatetolerant EPSPS revealed that the altered affinity for glyphosate was the result of a single amino acid substitution of alanine for glycine at residue 96.* As shown in Fig. 1 TAXRX L   16  12  14  14  15  Footnote 3  Footnote 3  Footnote 3  17  18  19 "Based on prediction on start of mature protein (see "Results"). Part of the arom multienzyme complex.
For the consensus sequence, X represents nonconserved amino acids.
amino acids). We now show that the glycine to alanine substitution corresponding to E. coli EPSPS G96A also imparts glyphosate tolerance to five additional EPSPS enzymes. The substitution corresponding to S. typhimurium EPSPS PlOlS (see Table I) (11,12), discussed above, also confers glyphosate tolerance to petunia EPSPS. This region is therefore a critical part of an EPSPS active site highly conserved between plant and bacterial enzymes.

EXPERIMENTAL PROCEDURES
Plasmid Construction-All plasmid constructions were carried out by standard methods (20, 21). Oligonucleotide-directed mutagenesis was carried out according to standard methods (22, 23) with minor modifications. Vectors for expression of the petunia, E. coli, soybean, and maize enzymes in bacteria utilized the PL promoter of phage X (24). The tomato EPSPS bacterial expression vector utilized the E. coli RecA promoter (25,26). In each case the success and fidelity of the mutagenesis was confirmed by sequencing the mutated region (27), using chemically synthesized sequencing primers for an adjacent region.
Enzyme Purification-EPSPSs were purified as previously described (28), with minor modifications. The pG96S EPSPS4 was To avoid confusion, the numbering of all EPSPS amino acid substitutions in this paper will be based on the E. coli EPSPS sequence. To denote different gene sources, a code letter will precede the substitution designation. The following gene source designation codes are used petunia, "p"; tomato, "t"; maize, "m"; Arabidopsis, "a." Therefore, the Gly to Ala substitution corresponding to G96A in followed during purification by HPLC with a Vydac C18 column, using 4545% RP-B in 20 min (RP-A, 0.1% trifuoroacetic acid RP-B, 0.1% trifluoroacetic acid in acetonitrile) at 1 ml/min flow rate; the EPSPS retention time was 14 min.
Equilibrium Dialysis-For binding studies, enzymes were dialyzed at 5 "C into dialysis buffer (50 mM HEPES, 50 mM KC1, 10% glycerol, 5 mM p-mercaptoethanol, pH 7.0) (2 X 2 liters, 4 h). After dialysis, the concentration was adjusted to 1.0-1.8 mg/ml with dialysis buffer (using Ai% = 8.2). The experiments (duplicate) were performed by dialyzing a fixed amount of enzyme against increasing concentrations of [14C]S3P in a microdialyzer apparatus (Hoefer, 250-111 chambers). S3P solutions  were prepared by mixing [l4C]S3p (19.7 mCi/mmol) with increasing amounts of unlabeled S3P. The enzyme (200 pl) was placed into one half-cell and an equal volume of S3P was placed into the other half-cell; dialysis cells were rotated at 20 rpm for approximately 16 h at 5 "C. Aliquots (100 pl) were removed from each half-cell, and the radioactivity present at equilibrium was determined by liquid scintillation counting (Beckman LS). The radioactivity present in the half-cell containing protein was used to calculate the concentration of bound + free S3P. The radioactivity present in the half-cell without protein was used to calculate the concentration of free S3P. The data was then plotted as 1/[S3P]fW, us.

RESULTS AND DISCUSSION
Construction of EPSPS Expression Vectors-We have previously described a system for the production of a mature (minus chloroplast transit peptide) form of petunia EPSPS in E. coli (28). The codon for of petunia preEPSPS (14) that represents the first amino acid of the mature protein was replaced by oligonucleotide-mediated mutagenesis with two codons, Met and Glu, producing a convenient restriction enzyme recognition site that allowed for correct attachment of the coding sequence to a strong E. coli promoter and ribosome binding site (24). For the current study, we have constructed similar vectors for bacterial expression of mature forms of EPSPS from tomato, soybean, and maize. In tomato and soybean the high degree of identity of the EPSPSs to petunia EPSPS allowed for identification of the homologous Lys residue as the putative N terminus of the mature EPSPS, which was altered as described above and inserted into E. coli expression vectors. Our best prediction of the maize CTP cleavage site was between Glufi' and Ala63 of the maize pre-EPSPS.5 In this case mutagenesis was performed to add a codon for a Met residue just upstream of the Ala63 codon. The altered maize sequence was then used to construct a vector for expression in E. coli. The bacterial expression plasmids were introduced into an aroA-deficient E. coli strain, SR481, which does not produce an active EPSPS, as previously described (28). A genomic clone for Arubidopsis EPSPS has previously been used to engineer high level expression of EPSPS in transgenic Arubidopsis plants (15). Plants produced using this system provided a source for the Arabidopsis EPSPSs.
To determine the effect on the plant EPSPSs of the glycine to alanine substitution corresponding to E. coli EPSPS G96A, we introduced the corresponding substitution by site-directed mutagenesis into the EPSPS DNA clones for bacterial expression of the petunia, tomato, soybean, and maize variants, and for plant expression of the Arubidopsis variant (Table I). Because of differences in amino acid sequence numbering, the E. coli EPSPS G96A substitution corresponds to different Gly to Ala substitutions in plant EPSPSs (for instance, GlOlA in E. coli EPSPS will be denoted pG96A in petunia EPSPS. Similarly, specific residues are preceded by their origin code, for instance pGlygfi of petunia EPSPS. Note that the conserved region containing this residue has the same numbering in both E. coli and S. typhimurium EPSPS. ' C. S. Gasser, unpublished data. petunia EPSPS), although as shown in Table I, the residues are homologous. The amino acid numbering system used herein is based on the E. coli EPSPS numbering scheme.* Analysis of Wild-type and Ala Variant EPSPSs-The wildtype and Ala variant EPSPSs were extracted from E. coli cells or Arabidopsis leaf tissue and assayed for EPSPS activity. The wild-type plant EPSPSs had specific activities ranging from 0.18 to 1.6 units/mg, compared to 0.11-0.69 unit/mg for the corresponding Ala variants (Table 11). There does not appear to be any significant alteration in EPSPS specific activity resulting from the engineering of the Ala substitution corresponding to E. coli EPSPS G96A, except perhaps in the tG96A variant, which only exhibits about 25% of the specific activity of the tomato wild-type enzyme. Manipulation of residues immediately around the chloroplast transit peptide cleavage site does not appear to significantly impact the enzyme activity of the mature EPSPSs, probably because this region folds into a domain separate from the mature region of preEPSPS (32). All of the plant EPSPS cDNAs complemented the aroA mutation in E. coli SR481.
Crude extracts containing either the wild-type or Ala-substituted variant EPSPSs corresponding to G96A were also assayed for their glyphosate sensitivity at saturating substrate concentrations. All the wild-type EPSPSs tested were sensitive to inhibition by glyphosate, while the Ala-substituted variants were extremely resistant to inhibition by glyphosate (Table 11). Thus, the substitution corresponding to G96A affects the glyphosate sensitivities of both E. coli and plant EPSPSs to similar degrees, with more than a 500-fold increase in ICso being obtained in all cases. This newly introduced Ala residue therefore interferes with glyphosate binding to EPSPS by a mechanism common to all EPSPSs tested, and furthermore, these results indicate that the active site of EPSPS is highly conserved among the plant and bacterial enzymes.
Steady-state Kinetic Analysis of pG96A Petunia EPSPS-In order to better quantitate the effect of the pG96A substitution on petunia EPSPS, steady-state kinetic constants were determined for purified wild-type and pG96A petunia EPSPSs (Table 111). For the pG96A variant, the K;(.,,)(glyphosate) us. PEP was found to be 2.0 f 0.2 mM, which is 5000 times  (15) or pMON 600 (aG96A variant) wqs frozen in liquid nitrogen and ground into a fine powder with a mortar and pestle, and 1 ml of extraction buffer (with the addition 5 mM ascorbate and 1 mg/ml bovine serum albumin) was added. After further grinding (45 s), the suspension was centrifuged in a microcentrifuge for 5 min. The resulting supernatants were desalted over spin G-50 columns equilibrated with the same buffer and assayed for EPSPS activity by HPLC radioassay (10). greater than that of the wild-type enzyme; the inhibition remained competitive us. PEP. The K,,.,,~(PEP) is elevated from 5.0 f 1.3 p M in the wild-type enzyme to 210 f 20 p~ in the pG96A variant, and the K,(.,)(phosphate) is elevated from 0.59 f 0.09 mM for the wild-type enzyme to 6.7 f 1.2 mM for the pG96A variant. These kinetic constants for the pG96A EPSPS are similar to those reported for the G96A E. coli B EPSPS (Ki(,,,l(glyphosate) = 4.1 mM, K,(,,,(PEP) = 220 p M ) (13). In contrast to the decreases in apparent binding seen for glyphosate, PEP, and Pi in the pG96A variant, the Km(app) values for both S3P and EPSP are essentially unchanged between the wild-type and pG96A EPSPSs (Table  111). The Km(.,,) values determined for the wild-type petunia enzyme are similar to the experimentally determined ligand dissociation constants for E. coli EPSPS (Kd(S3P) = 7 p~, pM, and &(Pi) = 1.4 mM for E. coli EPSPS) (33), indicating that, for EPSPS, measurement of Km(app) values is a valid approximation to the ligand dissociation constants. Taken together, these kinetic data are consistent with a perturbation of a binding interaction common for glyphosate, PEP, and Pi, but not S3P or EPSP. Since glyphosate is a competitive inhibitor of EPSPS with respect to PEP, it is not surprising that the pG96A variant (and the E. coli EPSPS G96A variant) also displays a loss of PEP binding. If one views the K~(a,,~(glyphosate)/K~~~pp~(PEP) as a selectivity factor for PEP over glyphosate binding, the large increase in its value from the petunia wild type EPSPS (K;/K, = 0.08) to the pG96A variant (K;/K, = 9.5) indicates that the introduction of the alanine residue selectively destabilizes the interaction of the enzyme with glyphosate compared to PEP.
In addition to its glyphosate insensitivity, the catalytic efficiency for pG96A EPSPS is also altered; the kcat in the forward reaction direction is 58% of that of the wild-type enzyme. The kCat/K++,,)(PEP) values for the wild-type and pG96A EPSPSs were 7. In order to determine if pG96A EPSPS efficiently interacts with the tetrahedral intermediate analog, we determined the Ki(appl for the R-inhibitor for pG96A and wild-type petunia EPSPS in the forward reaction direction. Interestingly, the analog inhibited pG96A EPSPS with an Ki(app) of 35 f 2 nM, compared to 38 k 5 nM for the wild-type enzyme (S3P = 2 mM; PEP = 5 p M for wild-type and 40 p M for pG96A EPSPS). The petunia EPSPS pG96A substitution therefore does not adversely affect the interaction of the phosphonate moiety of the analog with EPSPS and is in agreement with our observation that the pG96A variant has a V,,, close to that of the wild-type enzyme, presumably due to its high affinity for the reaction intermediate.  constants were determined by HPLC radioassay as previously described (lo), using 5-11 duplicated substrate concentrations. Apparent K,,, values were determined by computer fit of the data to the hyperbolic form of the Michaelis-Menton equation (nonweighted), k the standard error of the value derived from the fit (RS/E, BBN, Cambridge, MA); the value in parentheses is the number of independent experiments. Values for Ki,.,,(glyphosate) were determined using an inverse plot at three glyphosate concentrations. Slope replots gave the K,(.,,,(glyphosate) as the x intercept. The Km(app,(PEP) was determined at 2 mM S3P, the Km(.,)(S3P) was determined at 1 mM PEP, the K,,,,,(EPSP) was determined at 10 mM KPi, and the K,,,,,,,)(Pi) was determined at 100 PM EPSP concentrations. For kcat calculations, the molecular mass of petunia EPSPS used was 47.6 kDa.  However, the pG96S. S3P complex was unable to bind glyphosate, based on fluorescence emission and equilibrium dialysis analyses (data not shown). These features distinguish the pG96S variant from the wild-type enzyme. The intensity of the fluorescence emission spectrum of pG96S EPSPS did not increase upon EPSP addition, whereas a similar treatment of wild-type enzyme did result in a fluorescence emission spectrum intensity increase (36). HPLC analysis of a pG96S/EPSP mixture, after the addition of 10 mM potassium phosphate, indicated that the added EPSP had been hydrolyzed to S3P and an unknown product, while an identical reaction with wild-type petunia EPSPS gave the expected S3P and PEP product peaks. Additional experiments confirmed these results and showed that phosphate was not required for pG96S-catalyzed EPSP hydrolysis (Fig.  1). The most likely fate of the EPSP carboxyvinyl carbon fragment was postulated to be its conversion to pyruvate; EPSPS has been shown to form pyruvate from S3P and P E P as a by-product at an exceedingly low rate (0.00047 s" for E. coli EPSPS) (33). By using lactate dehydrogenase/NADH, we found that pyruvate is indeed the second product of pG96Smediated hydrolysis of EPSP. The EPSP hydrolase activity is not sensitive to inhibition by glyphosate, but is highly sensitive to inhibition by S3P (Table IV). In addition, 0.2 mM of the R-isomer intermediate analog of the EPSPS reaction described above did not inhibit the pG96S hydrolase reaction (at 57 FM EPSP), while inhibiting 94% of the wild-type petunia EPSPS activity at 57 WM EPSP and 50 mM KPi. These results confirm our hypothesis that the pGlyg6 region of the petunia EPSPS active site participates in the recognition of the phosphate moiety of PEP. The phosphate recognition site in the pG96S variant is severely disrupted and results in a loss of binding of the intermediate analog, and by inference, the tetrahedral intermediate itself.
There are three possible mechanisms by which EPSP can FIG. 1. pG96S EPSPS reaction with EPSP. Reactions contained 30 pg of the appropriate EPSPS, 1 mM EPSP, and 50 mM HEPES, pH 7.0, in 100 p1 of total reaction volume. Phosphate (Pi) was used at a concentration of 50 mM. Reactions were run for 25 min, then quenched with 100 pl of 90% ethanol, 0.1 M acetic acid, pH 4.5. Aliquots of 50 p1 were analyzed by HPLC as previously described (lo), using UV detection at 210 nm (0.08 absorbance units at full scale). The EPSPS standard was prepared in water. be hydrolyzed by pG96S EPSPS. The first possibility (Scheme 1, mechanism 1) is that nucleophilic addition of water to the C-2 carbon of the carboxyvinyl moiety of EPSP generates a tetrahedral intermediate I ,which by ketonization of the C2-OH bond yields pyruvate and S3P. The tetrahedral intermediate Z is the dephosphorylated form of tetrahedral intermediate 11, an intermediate of the EPSPS reaction (35). A second possibility is that the phosphorylated tetrahedral intermediate 11 is hydrolyzed by the pG96S enzyme to the intermediate I which then follows the same route suggested for mechanism 1. A third possibility is the direct nucleophilic attack of the pG96S pSerg6 hydroxyl group on the C-2 carbon of the carboxyvinyl moiety of EPSP (mechanism 3), generating an enzyme-bound intermediate which is hydrolyzed to S3P and pyruvate. It is known that the enzyme UDP-N-acetyl mur-

Pyruvate formation by purified pC96S
Reaction mixtures (1.0 ml) contained 0.105 mg of pG96S, 100 PM NADH (Sigma), 80 units of lactate dehydrogenase (Sigma type V-S), and 50 mM HEPES, pH 7.0, with the additional components indicated. After 5-min preincubation at 30 "C, reactions were initiated with 52.5 PM EPSP. Initial rate absorbance measurements at 340 nm were determined at 30 "C on a Hewlett Packard 8452 spectrophotometer, coupling NADH oxidation to pyruvate reduction by LDH. The NADH extinction coefficient used was 6300 M" cm" (39). amic acid synthase reacts with P E P generating an intermediate with covalently bound enol-pyruvate (37). Reaction mechanism 3 is similar except that EPSP replaces PEP as a substrate and water can release the enzyme-bound enol-pyruvate. Mechanisms 1 and 3 can be distinguished from mechanism 2 by the Pi dependence of the EPSP hydrolysis reaction of pG96S EPSPS. Trace amounts of phosphate present in the reaction mix were eliminated by the use of the sucrose phosphorylase/sucrose trap system (30). No differences in reaction rates were seen in the presence or absence of the trap system. There was no change in the rate of EPSP hydrolysis when the Pi concentration was increased from 4 p~ (background concentration)to4m~.Thesefeatures,plusthelackofinhibition by 5-0-[(R)-l-carboxy-l-phosphonoethyl]shikimate S-phosphate, suggest that the EPSP hydrolysis by pG96S operates by the sequence shown in either mechanism 1 or 3. The conversion of intermediate Z to S3P and pyruvate may also occur by two different mechanisms, either direct dissociation into S3P and pyruvate or formation of S3P and enol-pyruvate which tautomerizes to pyruvate. Further distinctions between reaction mechanisms can be made by establishing if enolpyruvate is the product of the reaction and if the enzyme catalyzes EPSP-SSP exchange of the enolpyruvyl moiety.
In order to further characterize the pG96S variant, the K, (,,,,(EPSP) relative to that of the wild-type enzyme was determined using the pyruvate kinase/ADP coupled assay. The K,(,,,)(EPSP) of pG96S was 23 f 6 p~, compared to 20 +-5 p~ for the wild-type EPSPS (10 mM phosphate, spectrophotometric assay at 30 "C (34)). The V,,, for the EPSP hydrolase activity of pG96S was 1.24 unit/mg, yielding a Kcat of 0.98 s-', which is > 2000 times faster than the rate of EPSP hydrolysis catalyzed by wild-type E. coli EPSPS (33). Taken 'SP synthase active site together, these results clearly demonstrate that pSerg6 of pG96S does not interfere with the S3P/EPSP binding site, but perturbs the PEP binding site. Since the carboxyvinyl moiety of EPSP interacts with the enzyme, the inability of pG96S EPSPS to bind PEP may be most likely due to the steric hindrance for interaction of phosphate moiety of P E P at the active site. It is tempting to speculate that the pSerg6 hydroxymethyl group of pG96S petunia EPSPS displaces the phosphate of PEP and functions as a nucleophile, attacking the C-2 carbon of the carboxyvinyl moiety of EPSP as suggested in mechanism 3. If this were to be the case, we would predict that variants of EPSPS containing alanine or valine at Glyg6 should be ineffective in catalyzing this reaction. In accord with this, purified G96A E. coli EPSPS did not have detectable EPSP hydrolase activity. Additional experiments are in progress to determine the hydrolase reaction mechanism of pG96S EPSPS. pP1OlS EPSPSs-Yet another substitution that has been described in the conserved region of EPSPS is that of PlOlS in S. typhirnuriurn EPSPS (11,12). Since this proline is not conserved in all EPSPSs (Table   I), it was of interest to determine if the corresponding proline to serine substitution would result in a glyphosate-tolerant petunia EPSPS. The pPlOlS EPSPS cDNA was therefore constructed by sitedirected mutagenesis and expressed in E. coli SR481. Kinetic analysis of the purified enzyme showed that the Km(app)(PEP) was 44 f 3 p~, the K,(,,)(S3P) was 12 +-2 p M , the K,(,,,)(glyphosate) was 3.0 & 0.8 p~, and the specific activity was 40 units/mg. These results show that for petunia EPSPS, the pPlOlS substitution results in a decrease in glyphosate binding, the magnitude of which is, however, significantly lower than for the pG96A substitution. The results of these mutagenesis experiments further support the view that the active sites of bacterial and plant EPSPSs are highly conserved.
A doubly substituted petunia EPSPS variant of pG96A pPlOlS was constructed in order to determine the effect of combining the two single substitutions which independently confer glyphosate tolerance to EPSPS. The pG96A pPlOlS variant had an K,(,,,)(PEP) of 390 f 40 p~ and an K,(,,,,(glyphosate) of 8.2 & 0.4 mM. Thus, the addition of the pPlOlS substitution to the pG96A variant increases the Ki(.,p)(glyphosate) by a factor of 4.1 and the K,(,,)(PEP) by a factor of 1.8. This increase in K,(,,,(glyphosate) is similar to the increase observed for the pPlOlS single variant relative to wild-type EPSPS (6-fold) and suggests that the pG96A and pPlOlS substitutions impart glyphosate tolerance by affecting distinct interactions between glyphosate and the conserved region. The Km(app)(PEP) may be somewhat obscure since the absolute K,(PEP) has not been obtained. Nevertheless, the results of these studies suggest multiple interactions between PEP/glyphosate and the conserved region of EPSPS. One of these interactions may involve the guanidinium side-chain of Arg'", a conserved residue in all EPSPS enzymes (Table I). Unpublished experiments from our laboratory based on mutagenesis of pArg'" support this hypothesk6 A complete understanding of the nature of these interactions will have to await the x-ray structure elucidation of the enzyme. substrate/ inhibitor complexes.
It is clear that considerable information about the active site of EPSPS has been obtained by site-directed mutagenesis and kinetic characterization of the variant enzymes. Recently, the x-ray struc$ure of wild-type E. coli EPSPS has been elucidated a t 3-A resolution (2). The EPSPS polypeptide has S. R. Padgette, D. B. Re, C. M. Hironaka, and G. M. Kishore, unpublished data. Mutagenesis of the EPSP synthase active site 22369 a two-domain structure with a novel fold that appears to be formed by a 6-fold replication of a protein folding unit comprised of two parallel helices and four stranded sheets. In this structure, the active site appears to be in the interdomain region, held apart by the interdipole repulsions in the absence of the anionic ligands. Since significant conformational changes can be demonstrated upon interaction of EPSPS with either S3P and glyphosate or EPSP and glyphosate, the structure of the native enzyme is insufficient to explain the results obtained with the variant enzymes. However, our knowledge of the variant enzymes can be of great value toward building models for the structure of the enzyme complexes. For a model to be valid, it must account for the perturbations in the kinetic constants of the variants described herein as well as the EPSP hydrolase activity of the pG96S variant. It is interesting to note that the two primary substitutions that confer glyphosate tolerance to EPSPS are localized within the conserved region shown in Table I. Whether this is the only region of EPSPS that can provide differentiation between PEP and glyphosate binding needs to be determined either by elucidation of the x-ray structure of an EPSPS. glyphosate (or PEP) complex or isolating additional glyphosate-tolerant EPSPS variants.
In summary, we have utilized site-directed mutagenesis to identify the active site of EPSPS. These studies have established that the conserved region of EPSPS containing GlyW is critical for interaction with the phosphate moiety of PEP, inorganic phosphate, and the phosphonate of glyphosate. The glyphosate-tolerant EPSPS cDNA clones have been used for conferring in planta RoundupTM tolerance to crop plants (38).
The levels of tolerance are significantly higher than that achieved using the wild-type, glyphosate-sensitive EPSPS enzymes. Understanding the molecular basis of tolerance will further facilitate new inhibitor design and the understanding of the catalytic mechanism of the enzyme.