Structural Basis of Glyphosate Tolerance Resulting from Mutations of Pro101 in Escherichia coli 5-Enolpyruvylshikimate-3-phosphate Synthase*

Glyphosate, the world's most used herbicide, is a massive success because it enables efficient weed control with minimal animal and environmental toxicity. The molecular target of glyphosate is 5-enolpyruvylshikimate-3-phosphate synthase (EPSPS), which catalyzes the sixth step of the shikimate pathway in plants and microorganisms. Glyphosate-tolerant variants of EPSPS constitute the basis of genetically engineered herbicide-tolerant crops. A single-site mutation of Pro101 in EPSPS (numbering according to the enzyme from Escherichia coli) has been implicated in glyphosate-resistant weeds, but this residue is not directly involved in glyphosate binding, and the basis for this phenomenon has remained unclear in the absence of further kinetic and structural characterization. To probe the effects of mutations at this site, E. coli EPSPS enzymes were produced with glycine, alanine, serine, or leucine substituted for Pro101. These mutant enzymes were analyzed by steady-state kinetics, and the crystal structures of the substrate binary and substrate·glyphosate ternary complexes of P101S and P101L EPSPS were determined to between 1.5- and 1.6-Å resolution. It appears that residues smaller than leucine may be substituted for Pro101 without decreasing catalytic efficiency. Any mutation at this site results in a structural change in the glyphosate-binding site, shifting Thr97 and Gly96 toward the inhibitor molecule. We conclude that the decreased inhibitory potency observed for glyphosate is a result of these mutation-induced long-range structural changes. The implications of our findings concerning the development and spread of glyphosate-resistant weeds are discussed.

Glyphosate (N-phosphonomethylglycine) inhibits the shikimate pathway enzyme 5-enolpyruvylshikimate-3-phosphate synthase (EPSPS 2 ; EC 2.5.1.19) (1), which is essential for the biosynthesis of aromatic compounds in plants, fungi, bacteria, and apicomplexan parasites (2)(3)(4)(5). Glyphosate, the active ingredient in Roundup, exhibits broad-spectrum herbicidal activity, yet is essentially nontoxic to animals and does not persist in the environment. These characteristics have made it the world's most popular herbicide, and usage continues to increase with the adoption of glyphosate-dependent technologies, including herbicide-tolerant crops and minimal tillage (no-till) agriculture. The enormous reliance on glyphosate and the absence of suitably safe alternative herbicides mean that the widespread emergence of glyphosatetolerant weeds would have devastating agricultural and environmental consequences.
EPSPS, the molecular target of glyphosate, catalyzes the transfer of the enolpyruvyl moiety of phosphoenolpyruvate (P-enolpyruvate) to the 5-hydroxy position of shikimate 3-phosphate (S3P) (see Fig. 1). The structure of the glyphosate-inhibited complex shows that glyphosate binds to the P-enolpyruvate-binding site of EPSPS (6 -8), corroborating early kinetic data demonstrating that glyphosate binding is competitive with respect to P-enolpyruvate (1,9,10). Before bacterial enzymes with innate glyphosate tolerance (class II EPSPS) were discovered and used to produce Roundup Ready crops, scientists described several mutations that decreased glyphosate sensitivity in the plant-like EPSPS from Escherichia coli (11,12). Typically, however, these mutant enzymes displayed an increased K m for P-enolpyruvate and a correspondingly decreased catalytic efficiency, indicative of decreased fitness in the absence of glyphosate, and were thus considered unsuitable for the development of glyphosate-tolerant crops.
The structural basis for the Pro 101 mutation-induced glyphosate tolerance was not known. Here, we describe the kinetic and structural characteristics of four E. coli mutant EPSPS enzymes containing single-residue substitutions at Pro 101 . The results indicate that any substitution at this site causes small but significant structural changes in the active site. The implications of our findings concerning the development and spread of glyphosate-resistant weeds are discussed.

EXPERIMENTAL PROCEDURES
S3P was synthesized and purified as described previously (27). P-enolpyruvate was purchased from Sigma. Protein concentration was determined using Coomassie reagent (Pierce) with bovine serum albumin as a standard. The single-site Pro 101 mutations were introduced into wild-type EPSPS from E. coli using the QuikChange mutagenesis kit (Stratagene) and appropriate primers. All primers were synthesized by MWG Biotech (High Point, NC). The pET-24d vector (Novagen) containing the open reading frame of wild-type (WT) EPSPS was used as a template for the mutations. Pro 101 mutant EPSPS enzymes were overexpressed in BL21(DE3) competent cells and purified as described previously (18). After the final purification step, the mutant enzymes were concentrated in 50 mM Tris and 2 mM dithiothreitol using Centricon-30 devices (Millipore Corp., Billerica, MA) at 4°C.
The enzymatic activities of WT and Pro 101 mutant EPSPS were measured spectrophotometrically at 25°C with a Spectra-Max 340PC plate reader (Molecular Devices, Sunnyvale, CA). The assay mixtures (60 l) contained 50 mM HEPES (pH 7.5), 100 mM KCl, 2 mM dithiothreitol, 1 mM S3P, and varied concentrations of P-enolpyruvate and glyphosate. The reaction was initiated by the addition of enzyme (3.4 nM WT; 8.5 nM P101S,       NOVEMBER  P101S and P101L EPSPS enzymes were crystallized at 19°C by the hanging-drop vapor diffusion method in the presence of 5 mM S3P with or without 5 mM glyphosate using the sodium formate crystallization conditions described previously (18). The protein concentration in each case was 37.5 mg/ml, or 810 M, maintaining a ligand-to-receptor molar ratio of ϳ6:1. X-ray diffraction data were recorded at Ϫ180°C using the rotation method on single flash-frozen crystals of Pro 101 mutant EPSPS enzymes (detector, R-AXIS IV ϩϩ imaging plate; x-rays, CuK ␣ , focused by mirror optics; generator, Rigaku RU300 (MSC, The Woodlands, TX)). The collected data were reduced with XDS (28). The program package CNS (29) was employed for phasing and refinement, and model building was performed with program O (30). The structures were solved by molecular replacement using E. coli WT EPSPS (Protein Data Bank code 1g6s) (6) stripped of solvent molecules, ions, and ligands as the starting model. Refinement was performed using data to the highest resolution with no cut-off applied. Solvent molecules were added to the models at reasonable positions, and S3P and glyphosate were modeled according to the clear electron density maps. Several rounds of minimization, simulated annealing (2500 K starting temperature), and restrained individual B-factor refinement were carried out. Data collection and refinement statistics are summarized in Table 1. Figs. 2, 4, and 5 were drawn with MolScript (31), BobScript (32), and Raster3D (33).

RESULTS AND DISCUSSION
Four single-site mutations were introduced into E. coli EPSPS, replacing Pro 101 with glycine, alanine, serine, or leucine. The resulting mutant enzymes were characterized by steady-state kinetics using the forward reaction of EPSPS (Fig. 1). The P101S and P101L mutant enzymes were crystallized in the presence of S3P with or without glyphosate, and the crystal structures were determined to between 1.5and 1.6-Å resolution (Table 1).
Enzyme Kinetics-All four mutant enzymes were found to be catalytically active, with specific activities ranging from 8 to 30 units/mg compared with 50 units/mg for WT EPSPS (Table 2 and supplemental Figs. S1-S5). The enzymatic reactions display normal saturation kinetics. Two distinct trends are discernible in the kinetic data. First, small residues may be substituted for Pro 101 without substantially altering the affinity of the  The torsion angles were calculated with MOLEMAN2 (37),  (Fig.  2). Therefore, the effects of Pro 101 mutations on the inhibition of EPSPS by glyphosate are likely to be due to long-range structural changes in the glyphosate-binding site. X-ray crystal structures at between 1.5-and 1.6-Å resolution revealed that the overall structures of the P101S and P101L mutant enzymes are nearly identical to that of WT EPSPS (Fig. 3). When comparing the glyphosate-bound forms of the mutant enzymes with that of WT EPSPS, the root mean square deviation (r.m.s.d.) values of all 427 C-␣ atoms are 0.066 and 0.097 Å for P101S and P101L, respectively. The largest differences occur in the backbone of Gly 96 , Thr 97 , and Ala 98 , with the C-␣ atom of Thr 97 having 4 -7 times larger r.m.s.d. values than the overall r.m.s.d. (Fig. 3). These structural alterations are also reflected by substantial changes in the backbone torsion angles of residues 96 -98 (Table 3), and the changes are further evident from F o Ϫ F o electron density maps with difference peak heights of 23 and 18 in the backbone around Thr 97 for the P101S and P101L enzymes, respectively (Fig. 4). The altered amino acids Ser 101 and Leu 101 are well defined in the respective electron density maps (Fig. 4). Notably, the side chain of Ser 101 adopts two alternate conformations. Although the hydroxyl group of either Ser 101 rotamer is in hydrogenbonding distance (Ͻ3.2 Å) of the carbonyl oxygen of Thr 97 , its apparent flexibility makes a strong bonding interaction unlikely (Fig. 5). Neither the P101S nor P101L mutation disrupts the ␣-helix present in WT EPSPS. Rather, it seems that the loss of the Pro 101 ring system causes a disruption of hydrophobic interactions that hold in place the carbonyl oxygen of Thr 97 . In the mutant enzymes, the backbone of Thr 97 relaxes and reorients, rotating about its / torsion angles ( Table 3). As a result, Gly 96 and Thr 97 shift slightly toward the glyphosatebinding site, thereby causing repulsive forces (Ͻ3.2 Å) between the C-␣ atom of Gly 96 and the phosphonate moiety of glyphosate. It appears that the altered orientation of Thr 97 and Gly 96 causes a slight narrowing of the binding site for glyphosate. This negatively impacts glyphosate inhibitory potency because this inhibitor interacts with EPSPS efficiently only in its extended conformation (7). Moreover, the binding of the substrate P-enolpyruvate would remain unaltered because the P-enolpyruvate molecule is significantly shorter than glyphosate (18).
Conclusion-The removal of undesired plant species (weeds) to increase nutrient availability and crop yields is an intrinsic component of agriculture. The current enormous (and growing) reliance on glyphosate indicates that the development and spread of glyphosate-resistant weeds would have far-reaching negative consequences (34). To date, reports of target-site mutations suggest that glyphosate-tolerant EPSPS enzymes typically have a substantial fitness cost, particularly in the absence of multiple (compensatory) mutations. Mutations at sites corresponding to Pro 101 appear to incur the least fitness cost, and because point mutations are far more evolutionarily accessible than multiple mutations, target-site glyphosate tolerance seems most likely to arise via mutations of this residue. With continued selective pressure from glyphosate application, the evolution of weedy plants expressing such enzymes appears inevitable. Our study indicates that the structural basis for the glyphosate tolerance of such Pro 101 mutant EPSPS enzymes is due to long-range alterations in the active site of the enzyme, in particular impacting the spatial orientation of Gly 96 and Thr 97 . It has long been known that Gly 96 is critical for the efficient binding of glyphosate. For E. coli EPSPS, mutation of Gly 96 to alanine results in a complete loss of inhibitory potency because of the methyl group protruding into the glyphosate-binding site; however, this glyphosate tolerance comes at the expense of a drastically lowered affinity for P-enolpyruvate and poor catalytic efficiency (18). In general, because the degree of glyphosate tolerance depends on the extent to which the inhibitor-binding site is perturbed and because the catalytic efficiency depends on the extent to which the substrate-binding site is left intact, it appears that the Pro 101 substitution is favorable precisely because the changes in the enzyme active-site structure are so slight. Residues other than proline substituted at position 101 reduce glyphosate binding, whereas residues smaller than leucine at this position essentially do not alter the S3P-and P-enolpyruvate-binding sites, retaining catalytic efficiency.
To prevent the expected deterioration of the herbicidal properties of glyphosate, aggressive strategies for combating glyphosate-tolerant weeds should be implemented. Our data indicate that plants with these target-site mutations likely remain susceptible to glyphosate at high concentrations, but non-target-site mutations may act synergistically (14,22,23), drastically decreasing the effectiveness of glyphosate. On the basis of the extensive structure-activity relationship studies performed with glyphosate analogs (15) and the subtlety of the structural changes observed in glyphosate-tolerant enzymes, glyphosate analogs may not represent suitable replacements for glyphosate. Because plants with target-site glyphosate tolerance mutations remain susceptible to herbi-cides targeting enzymes other than EPSPS, herbicide rotation practices may delay the development and spread of glyphosate-tolerant weeds, and integrated weed management programs should be encouraged (35). The engineering of crops with resistance to other herbicides, such as dicamba (36), holds some promise, but such crops are not yet commercially available, and glyphosate has unique advantages because of its very low toxicity to animals and its broadspectrum activity against plants. In the long run, the development of completely new inhibitors of EPSPS or other shikimate pathway enzymes is desirable, as new nontoxic herbicides will be required.