Interdomain Conformational Changes Provide Allosteric Regulation en Route to Chorismate*♦

Multifunctional proteins play a variety of roles in metabolism. Here, we examine the catalytic function of the combined 3-deoxy-d-arabino heptulosonate-7-phosphate synthase (DAH7PS) and chorismate mutase (CM) from Geobacillus sp. DAH7PS operates at the start of the biosynthetic pathway for aromatic metabolites, whereas CM operates in a dedicated branch of the pathway for the biosynthesis of amino acids tyrosine and phenylalanine. In line with sequence predictions, the two catalytic functions are located in distinct domains, and these two activities can be separated and retain functionality. For the full-length protein, prephenate, the product of the CM reaction, acts as an allosteric inhibitor for the DAH7PS. The crystal structure of the full-length protein with prephenate bound and the accompanying small angle x-ray scattering data reveal the molecular mechanism of the allostery. Prephenate binding results in the tighter association between the dimeric CM domains and the tetrameric DAH7PS, occluding the active site and therefore disrupting DAH7PS function. Acquisition of a physical gating mechanism to control catalytic function through gene fusion appears to be a general mechanism for providing allostery for this enzyme.

Allostery is the biochemical process whereby binding a remote ligand triggers protein functional changes. Allosteric regulation of key metabolic enzymes is an important control system for many metabolic pathways. The shikimate pathway, which is found in plants and microorganisms, is the biosynthetic route for the formation of the aromatic amino acids and other important aromatic metabolites (1). The first committed step of the pathway, catalyzed by 3-deoxy-D-arabino-heptulosonate-7-phosphate synthase (DAH7PS), 2 is generally con-trolled by allosteric inhibition arising from the remote binding of end products or intermediates of the pathway (2).
DAH7PS catalyzes the metal ion-dependent condensation reaction of phosphoenolpyruvate (PEP) and erythrose 4-phosphate (E4P) to generate 3-deoxy-D-arabino-heptulosonate-7phosphate (DAH7P, Fig. 1). The pathway has six further steps to the common intermediate chorismate. Chorismate is substrate for a variety of enzymes, including those that catalyze steps that ultimately lead to the aromatic amino acids Phe, Tyr, or Trp. Allosteric regulation is observed both at the pathway entry via DAH7PS and at the branch points from chorismate. All known DAH7PS enzymes share a (␤/␣) 8 barrel catalytic domain, which houses the active site. However, there are distinct variations between DAH7PS enzymes from different organisms in the structural extensions to the barrel that support allostery.
DAH7PS enzymes are divided into two distinct types, which differ in sequence, subunit size, and quaternary structure. Type I enzymes are generally smaller than their type II counterparts and have been shown to display allosteric control from a single allosteric effector or to display no allostery. Type I enzymes are further divided into two subtypes. The type I␣ subgroup contains the well characterized Escherichia coli, Saccharomyces cerevisiae, and Neisseria meningitidis enzymes (3)(4)(5). These enzymes have an allosteric site formed by an N-terminal and loop extension and are sensitive to one of the aromatic amino acids. Type I␤ enzymes are more varied and either comprise just the core catalytic barrel and show no allosteric regulation or have a discrete domain found at either the N or C termini of the barrel (6). This addition can be either an ACT domain (7,8) (named after the enzymes in which it was originally identified, aspartokinase, chorismate mutase, and TyrA) or a chorismate mutase (CM) enzyme (9), which catalyzes a branch point step of the pathway (Fig. 1). For the latter, previously characterized DAH7PS enzymes have dual catalytic functionality (both DAH7PS and CM activity), and the DAH7PS activity is inhib-ited by the binding of CM substrate chorismate or its product prephenate. Type II enzymes are the most complex, and the only fully characterized example is from Mycobacterium tuberculosis (10). M. tuberculosis DAH7PS has multiple extensions to the core barrel providing three allosteric binding sites that work synergistically (11). In addition, this protein can form a non-covalent complex with the M. tuberculosis CM allowing for complex allosteric control of both CM and DAH7PS activities (12).
In accordance with the different extensions to the core catalytic barrel, the molecular mechanisms by which allostery is achieved vary between the DAH7PS types. Whereas the type I␤ DAH7PS from Thermotoga maritima undergoes significant ACT domain movement associated with the binding of inhibitor Tyr (7), for other enzymes, such as the type II M. tuberculosis DAH7PS (13) and the type I␣ N. meningitidis DAH7PS (14), allosteric inhibition is conferred by more subtle structural changes and alteration in protein dynamics. The variation in allosteric machinery and mechanism for this enzyme is intriguing both from a mechanistic and evolutionary point of view.
Here, we examine in detail the mechanism of allostery for the DAH7PS from Geobacillus sp. (strain Y412MC61), a thermophilic soil bacterium. From sequence inspection, this DAH7PS is predicted to have an N-terminal CM domain and be a member of the type I␤ group. The independence of the individual catalytic activities of the DAH7PS and CM domains has already been demonstrated for the Bacillus subtilis enzyme (15), and a structure of an uninhibited form of an enzyme of this type has been previously determined for the DAH7PS from Listeria monocytogenes (LmoDAH7PS) (16). Here, we explore the functional properties of this dual activity protein and demonstrate directly the structural changes associated with the allosteric response induced by prephenate binding. Based on our new x-ray crystal structure and small angle x-ray scattering experiments, we reveal that prephenate binding to the CM domain stabilizes a more compact conformation of the protein, limiting the catalytic functionality of the DAH7PS domain active site.

DAH7PS and CM Catalytic Activities of Wild-type GspDAH7PS Can Be Separated into Two Functional Enzymes-
The DAH7PS protein complex from Geobacillus sp. putatively includes two functional units, represented by a DAH7PS and a CM domain. To test this prediction and to investigate the dependence of the catalytic activity of each functional unit of the GspDAH7PS complex, the full-length protein (GspDAH7PS WT ) was expressed, and two truncated forms were generated consisting of either only the predicted DAH7PS (GspDAH7PS D ) or CM (GspDAH7PS C ) domains (Fig. 2). The full-length protein indeed displayed both DAH7PS and CM activity (Table 1), and as for other characterized DAH7PS enzymes, the DAH7PS activity of GspDAH7PS WT was found to be dependent on the presence of a divalent metal ion (Fig. 3A). Consistent with the thermophilic nature of Geobacillus sp., maximum DAH7PS and CM activity was observed at 70°C (Fig. 3, B and C). The DAH7PS from Geobacillus sp. is therefore a bifunctional thermophilic enzyme, which possesses both DAH7PS and CM activity.
Consistent with compartmentalization of activities into the respective domains, in the truncated variants only metal-dependent DAH7PS activity was observed for GspDAH7PS D , and only CM activity was detected for GspDAH7PS C . Intriguingly, whereas the DAH7PS activity of the separated DAH7PS domain was greater than that of the wild-type protein, the efficiency of catalysis of the CM reaction was reduced for the isolated CM domain (Table 1). This change was principally caused by a 5-fold increase in the chorismate K m value for the shorter GspDAH7PS C protein, yet the catalytic turnover, represented by the k cat , was 2-fold higher. The activities of both separated domains were also found to be slightly more sensitive to temperature than was found for the full-length protein (Fig. 3). Thus, the properties of the two functional domains comprising the full DAH7PS-CM protein are affected by its formation, suggesting the possibility of inter-domain communication.
Inhibition of DAH7PS Activity by Prephenate-Structural extensions to DAH7PS core catalytic barrels is a common FIGURE 1. Shikimate pathway for aromatic amino acid biosynthesis. The first step of the pathway is catalyzed by the enzyme DAH7PS. The pathway branches at chorismate, and chorismate mutase catalyzes the formation of prephenate from chorismate, which in turn will form Tyr and Phe. mechanism for elaborating feedback inhibition on DAH7PS activity (4,7,17). The combination of DAH7PS and CM domains may represent another case of this phenomenon. To test this, we assessed whether the product of CM, prephenate, was an inhibitor of the DAH7PS activity, as this would be consistent with the fusion of the CM domain to DAH7PS providing feedback regulation to the pathway. Notably, prephenate inhibited the DAH7PS activity of the full-length GspDAH7PS WT protein but not of separated GspDAH7PS D (Fig. 4). The DAH7PS activity of the full-length enzyme was reduced to ϳ50% with 20 M prephenate and to less than 2% at higher prephenate concentrations compared with the activity in the absence of prephenate. This observation suggests that the inhibitory mechanism directly involves the CM domain, which may provide an allosteric binding site for prephenate. Due to the high CM activity of GspDAH7PS WT , the same inhibition response was observed with chorismate rather than prephenate.
Small Angle X-ray Scattering Shows a Conformational Change in the Presence of Prephenate-To examine the effect on the structure of the complex in the presence of prephenate, experimental SAXS data were obtained for GspDAH7PS in the absence and presence of prephenate to shed light on the conformational changes associated with inhibition ( Fig. 5). Comparison of the SAXS profiles of GspDAH7PS in the absence and presence of prephenate shows clear differences, consistent with a conformational change resulting from the binding of prephenate (Fig. 5A). This change is associated with an increased compactness of the enzyme in the presence of prephenate by the smaller radius of gyration, derived from the Guinier plot (R g ϭ 34.8 Ϯ 0.6 Å with prephenate compared with R g ϭ 36.5 Ϯ 0.4 Å without) (Fig. 5A).
Notably, the bell-shaped curve observed in the Kratky plot ( Fig. 5B) shows that in both the presence and absence of prephenate the protein exhibits a typical folded shape, without random coils. This verifies that both domains remain folded even in the absence of prephenate, although the slightly more flattened curve for the prephenate-free sample may indicate an increased flexibility of the protein. The Porod-Debye plots (Fig.  5C) present two different plateaus in these two cases, suggesting that the protein experiences a discrete conformational change upon ligand binding without any radical change in its  intrinsic flexibility (18,19). This conclusion is supported by the q 3 ⅐I(q) versus q 3 plot (Fig. 5D) that displays a distinct decrease, rather than a plateau as one would expect in the case of a highly flexible enzyme. Prephenate Binds to the CM Domain at the Active Site Located within a Helical Bundle-To further probe the intriguing conformational changes in the presence of prephenate and to identify the prephenate-binding site, we determined the crystal structure of GspDAH7PS in complex with the CM product and DAH7PS inhibitor prephenate. GspDAH7PS was crystallized in the presence of Mn 2ϩ and chorismate, and its structure was determined at a resolution of 2.75 Å in the space group P6 4 . The asymmetric unit contains two adjacent chains, with each comprising a catalytic (␤/␣) 8 barrel common to all DAH7PS enzymes (residues 106 -362) and a CM domain (residues 1-86) connected by a short linker region (residues 86 -105) (Fig. 6A). The two catalytic barrels from each chain in the asymmetric unit are highly similar (C␣ r.m.s.d. ϭ 0.11 Å), whereas the CM domains are slightly less similar (C␣ r.m.s.d. ϭ 0.23 Å), with one of the chains having more unresolved residues than the other in the linker region connecting the CM domain with the catalytic barrel. However, when taken as a whole, the two chains present a C␣ r.m.s.d. of 3.93 Å. This large difference arises from the difference in location of the two CM domains relative to the barrels in each chain (Fig. 6A).
A 2-fold symmetry operation reveals the homotetrameric biological unit of GspDAH7PS (Fig. 6B). The DAH7PS barrel subunits are arranged in a similar fashion to the tetrameric assemblies observed for other structurally characterized type I␤ DAH7PS enzymes (6), with neighboring protomers placed so that adjacent barrels face in opposite directions around the tetramer plane. A manganese ion is modeled at the C-terminal end of each of the barrels and is coordinated to residues Cys-    OCTOBER 14, 2016 • VOLUME 291 • NUMBER 42

JOURNAL OF BIOLOGICAL CHEMISTRY 21839
126, His-296, Glu-322, and Asp-333, identifying the locality of the DAH7PS active site. These metal-binding residues are conserved in all structurally characterized DAH7PS enzymes. The helical CM domains of two diagonally opposed chains associate in an interlaced dimer that resembles a type II or AroQ helical CM (20).
Difference electron density was observed in the area corresponding to the catalytic site of the CM domain. Because of the observed CM activity of GspDAH7PS, prephenate is generated from the added chorismate, and prephenate presence was confirmed by difference density at the CM domain active site (Fig.  6, C and D). The prephenate-binding site reveals that the residues involved in its binding correspond to a subset of the residues responsible for the catalytic activity of the CM domain (  (21), is not conserved and replaced by Ala in GspDAH7PS. These alterations may account for the relatively low CM activity of the GspDAH7PS WT and the truncated GspDAH7PS C protein.
For the structures of LmoDAH7PS, which have no ligand bound in the CM active site, the CM domains are found further from the catalytic DAH7PS barrels in an "open" position when compared with the "closed" position they adopt in the inhibited prephenate-bound GspDAH7PS structure. LmoDAH7PS shares sequence identity of 75.1% over 361 residues (91.1% similarity, Fig. 2) with GspDAH7PS, and structural alignment of FIGURE 6. Crystal structure of GspDAH7PS. A, two chains of the asymmetric unit, chain A (left) and chain B (right). The catalytic TIM barrel is shown in blue, the CM domain in green, and the linker region in purple. Allosteric inhibitor prephenate (Pre) is represented with gray sticks, and the active site is identified with the metal ion, represented with an orange sphere. B, tetrameric assembly with the components of one asymmetric unit colored (TIM barrel is shown in blue, the CM domain in green, and the linker region in purple) and the symmetry-related dimer is colored gray. C and D, binding sites of prephenate in the two unique subunits of the asymmetric unit. Residues important for the binding of prephenate and the catalytic activity of the CM domain are shown with sticks. One CM domain is shown in green, and the adjacent chain forming the catalytically active CM is shown in blue. Prephenate is represented with gray sticks. The A-weighted F o Ϫ F c omit maps for prephenate are represented with a blue mesh and were generated by deleting the prephenate molecules from the n-1 round of refinement. Electron density is contoured at 3 (0.1457e/Å 3 ).
Prephenate Binding Is Associated with Closer Association between the Catalytic and Regulatory Domains-To estimate the likely structural changes of GspDAH7PS on the binding of prephenate, a full model of GspDAH7PS in its active open form, referred to hereafter as the "open model," based on the two known prephenate-free LmoDAH7PS structures was constructed by homology using MODELLER (24). In addition, missing residues and side chains in the crystal structure of GspDAH7PS were reconstructed, and their position optimized, with MODELLER (referred to as the "closed model").
Comparison of these open and closed models sheds light on the structural changes likely associated with the binding of prephenate and mechanism of the allosteric inhibition of GspDAH7PS (Fig. 8). In the absence of prephenate, there is little contact between the CM domain and the DAH7PS barrels (7.6% solvent-accessible surface area (SASA) of the tetrameric barrels buried by the CM domains), and the structure is extended (Fig. 8A), with the entrance to the active sites of the DAH7PS domains accessible for substrate binding (Fig. 8C). In comparison, the closed model with prephenate bound is more compact (Fig. 8B), with the CM domains slightly bent to partly follow the shape of the DAH7PS barrels (15.6% of the SASA of the tetrameric DAH7PS barrels buried by the CM domains). In addition, a number of hydrogen bonds are formed between the DAH7PS and CM domains, in particular between Asp-326 and Arg-52, Tyr-46 and Glu-129, and Arg-175 and Gly-42, occluding in part the entrance to the active sites, suggesting an explanation for the inhibitory effect of prephenate against the DAH7PS activity (Fig. 8C).
Portions of CM domains that are disordered in the open model, namely the inter-helical loops, the N-terminal region, and the beginning of the linker region, are found to be ordered in the closed model and lead to extended CM helices (Fig. 8, A  and B). These parts of the CM domains include residues identified as being responsible for the catalytic activity of the CM domains and for the binding of prephenate, in particular, Ile-34, Lys-38, Asp-47, Arg-50, and Glu-51. We can therefore postulate that the binding of chorismate or prephenate is accompanied by a reorganization of the secondary structure of the CM domains, viz. the formation of longer helices. This reorganization, together with the binding of the allosteric ligand itself, significantly rigidifies the CM domains, favoring a conformation in which these domains lean on the DAH7PS catalytic barrels.

SAXS Data Confirm That the Conformational Compaction upon Prephenate Binding Observed Crystallographically Also
Occurs in Solution-To confirm the conformational changes upon prephenate binding observed in the crystal structure of GspDAH7PS and to gain insights into the possible allosteric regulation mechanism adopted by this enzyme, the SAXS data recorded in the absence and presence of prephenate were compared with the open and closed (prephenate-bound) models of the protein (Fig. 9). The theoretical scattering curves calculated for each model present a good fit with the corresponding experimental curves ( 2 ϭ 0.70 and 0.37 for the closed and open forms, respectively). In contrast, the comparisons of the theoretical curve of the open structure to the data for prephenatebound GspDAH7PS, and of the closed form to the apoenzyme data, give poor fits ( 2 ϭ 8.02 and 3.13, respectively).
The experimental SAXS data were further fitted using a mixture of open and closed models of GspDAH7PS. In the absence of prephenate, the analysis shows a better fit of the theoretical scattering from a combination of both open and closed GspDAH7PS models ( 2 ϭ 0.22) to the experimental data obtained in the absence of prephenate. This combination predicts that Ϸ73% of the protein adopts the open conformation, and Ϸ27% of the protein adopts the closed form conformation (Fig. 9A)  conformationally similar structures, is the sole form adopted by the prephenate-bound protein (Fig. 9B). These data are consistent with the full loss of DAH7PS activity of GspDAH7PS by prephenate binding observed experimentally.

Discussion
Multifunctional enzymes are observed in many living systems, and they play important roles in substrate channeling, enzyme compartmentalization, and metabolic control (25). In the thermophilic soil bacterium Geobacillus sp., a CM domain is found fused to the N terminus of the first enzyme dedicated to aromatic amino acid biosynthesis, DAH7PS. We have shown here that this fusion gives rise to a bifunctional enzyme presenting both DAH7PS and CM catalytic activity. Additionally, we have shown that the smaller enzyme, CM, can bind its product prephenate, and this inhibits the activity of the fusion partner DAH7PS.
The helical AroQ form of CM is frequently observed as a fusion partner to enzymes involved in Tyr and Phe biosynthesis. For example, not only is it attached to the DAH7PS enzyme studied herein, it is found fused to both prephenate dehydratase and prephenate dehydrogenase enzymes (26 -30). Geobacillus sp. has both the AroQ form (fused) and the other form of CM, AroH, encoded by its genome. Sequence inspection reveals that this AroH variant does not seem to be fused to another protein.
For the B. subtilis DAH7PS enzyme, the CM catalytic activity of the AroQ in the fusion protein was shown to be extremely low, and it was predicted that the AroH protein was the major contributor to CM activity, and the primary role of the CM domain of the equivalent CM-DAH7PS fusion was for allosteric control (15). In contrast, the AroQ CM domain of the hyperthermophilic Geobacillus sp. fusion protein has moderate activity, suggesting it may play a dual role in both providing catalytic function in tandem with the AroH protein as well as providing a feedback mechanism to control entry into the pathway. Notably, the CM catalytic efficiency is enhanced by its association with DAH7PS, principally due to its significantly lower Michaelis constant.
Through a combination of enzyme kinetics, x-ray crystallography, SAXS, and modeling, we can now build a picture of the molecular details of how the feedback inhibition of DAH7PS is achieved in this class of DAH7PS enzymes. In the absence of the allosteric ligand prephenate, the CM domains remain at a distance from the DAH7PS catalytic barrels, permitting the substrates to enter the DAH7PS active site. The binding of prephenate to the CM domain is then accompanied by a conformational change, evident from SAXS data and comparison of our prephenate-bound crystal structure with the homologous unbound structure of LmoDAH7PS. The latter structure shows a far more extended structure in the absence of ligands within the CM domain binding site. In the presence of prephenate, the protein adopts a more compact and more ordered form, with the extension of helices in the CM domain, which bear residues forming binding contacts with prephenate. Furthermore, the SAXS data suggest that the CM domains remain close to the DAH7PS catalytic barrels in the presence of prephenate, because of the very good fit obtained for the data in the presence of inhibitor, using the closed form model. The compactness of the structure impacts on substrate access to the DAH7PS active site and thus inhibits the enzyme. Conformational transitions and the occasional sampling of conformations resembling the closed form may account for the observed slightly higher activity of the truncated DAH7PS mutant lacking the CM domain.
This mechanism of allostery involving physical gating of the active site has been employed for other DAH7PS enzymes, albeit with structurally distinct regulatory domains. The T. maritima DAH7PS bears an ACT regulatory domain, which on inhibitor Tyr binding also restricts access to the active site (7). These two quite distinct allosteric solutions suggest that adopting this structural approach for the control DAH7PS repre-sents a general mechanism, albeit that both of these allosteric control mechanisms require the same basic tetrameric structure for the DAH7PS modules to operate. Other DAH7PS enzymes have been shown to adopt allosteric regulation mechanisms based on changes in protein dynamics (13,14) rather than using a physical gating mechanism, which reinforces the versatility of both the protein fold and the approaches to allosteric control.
As noted above, the acquisition of the CM domain by DAH7PS comes with a slight burden to its catalytic function, as evidenced by the moderate increase in DAH7PS catalytic efficiency observed for the truncated protein of GspDAH7PS D . In contrast, the CM activity benefits from its association with its DAH7PS partner, with its independent activity following excision being reduced. This reduction suggests that the transient contacts the CM dimer forms with its DAH7PS partner in the full-length protein are important for its catalytic function as well as for the delivery of allostery. Although it is difficult to unravel the individual contributions to binding and catalysis from the steady-state kinetic parameters, it is notable that the CM activity of the full-length protein has a lower chorismate K m value (by a factor of almost 5) and a lower k cat value (almost a 3-fold decrease). These changes reveal the impact domain interaction observed in the x-ray structure has on CM catalytic function. The decrease in K m value for the fused CM may reflect the contribution that conformational restriction affords by the interaction with DAH7PS, whereas the lower k cat value may reflect the tailoring of the CM active site for prephenate binding and the provision of allostery for the DAH7PS activity, rather than for catalyzing the CM reaction.
The connected and overlapping functions between enzymatic activities demonstrate the complexity of cellular mechanisms for both providing allosteric control and delivering cellular biosynthetic functionality. DAH7PS, through its associations and varied mechanisms for providing allosteric control, provides insight into the evolutionary processes that provide solutions to biosynthetic control.

Experimental Procedures
Construction of Plasmids-The gene sequence encoding GspDAH7PS WT (strain Y412MC61) was codon-optimized for expression in E. coli, synthesized by Geneart (Life Technologies, Inc.), and ligated into the expression vector pET-28b between the NcoI and XhoI sites to give a protein with a noncleavable C-terminal His 6 tag. This plasmid was used as the template for the preparation of constructs for expressing GspDAH7PS D (the DAH7PS portion of GspDAH7PS WT ) and GspDAH7PS C (the CM portion of GspDAH7PS WT ) variants, created by truncating between Lys-92 and Ala-93. Genes  encoding the truncated proteins were amplified using genespecific primers (Table 2) for cloning using the Gateway system (Life Technologies, Inc.). The products from the first round of PCR were extended using primers that incorporated the attb1 site and attb2 site, and the PCR products were gel-purified. The purified products were cloned into the donor vector pDONR-221 using BP clonase enzyme mix (Life Technologies, Inc.), and the product was subsequently transformed into E. coli One Shot TOP10 cells (Life Technologies, Inc.). Purified plasmids from transformants were sequence-verified. LR clonase enzyme mix (Life Technologies, Inc.) was used to sub-clone into the destination vector pDEST-17 (for the construct encoding GspDAH7PS D ), which encodes an N-terminal His tag and TEV protease cleavage site, or pDEST-15 (for the construct encoding GspDAH7PS C ), which encodes an N-terminal GST tag, a His tag, and a TEV protease cleavage site.
Expression and Purification of GspDAH7PS WT , GspDAH7PS D , and GspDAH7PS C -The vectors bearing the genes of GspDAH7PS WT and GspDAH7PS C were transformed into E. coli BL21 (DE3) cells, whereas the vector bearing the gene coding for GspDAH7PS D was transformed into E. coli BL21 (DE3) pBB540/pBB542 cells for expression. Cultures were grown at 37°C in 1 liter of Luria-Bertani (LB) medium supplemented with 100 g/ml kanamycin for GspDAH7PS WT , 100 g/ml ampicillin for GspDAH7PS C , and a combination of 100 g/ml ampicillin, 25 g/ml chloramphenicol, and 10 g/ml spectinomycin for GspDAH7PS D . Expression of target protein was induced through the addition of 0.5 mM isopropyl ␤-Dthiogalactopyranoside, and the cultures were grown for a further 4 h at 37°C (for expression of GspDAH7PS WT ) or overnight at 23°C (for the expression of GspDAH7PS D and GspDAH7PS C ). Cells were harvested by centrifugation and resuspended in lysis buffer (50 mM 1,3-bis(tris(hydroxymethyl) methylamino) propane (BTP), 200 mM KCl, 2 mM dithiothreitol (DTT), 200 M PEP, pH 7.2, EDTA-free protease inhibitor) prior to lysis by sonication.
Purification of GspDAH7PS WT and GspDAH7PS D -The crude lysate was heat-treated (30 min at 70°C for GspDAH7PS WT and 55°C for GspDAH7PS D ) and then clarified by centrifugation at 21,000 ϫ g for 35 min. The supernatant was recovered and filtered, and after addition of 50 mM imidazole, it was loaded onto a HisTrap HP column (GE Healthcare) that had been pre-equilibrated with the binding buffer (20 mM sodium phosphate, 500 mM NaCl, 50 mM imidazole, pH 7.4). The column was washed with binding buffer, and protein was eluted with elution buffer (containing 500 mM imidazole). Fractions containing enzyme were pooled together and run through a desalting column eluting with binding buffer to lower the imidazole concentration, before overnight treatment with TEV protease at a 1% mol/mol concentration (for GspDAH7PS D only). TEV-treated enzyme was subsequently run through a second HisTrap HP column to separate protease and His tag from cleaved enzyme using the same elution conditions as used previously. Untagged enzyme was pooled together for size-exclusion chromatography (SEC).
Purification of GspDAH7PS C -The crude lysate was centrifuged at 21,000 ϫ g for 35 min, and the supernatant was recovered, filtered, and loaded onto a GSTrap HP column (GE Healthcare) that had been pre-equilibrated with the binding buffer (140 mM NaCl, 2.7 mM KCl, 10 mM sodium phosphate buffer, pH 7.3). The column was washed with binding buffer, and protein was eluted with elution buffer (50 mM Tris-HCl containing 10 mM reduced glutathione, pH 8.0). Fractions containing enzyme were pooled together and run through a desalting column eluting with binding buffer, before overnight treatment with TEV protease at a 1% mol/mol concentration. TEV-treated enzyme was subsequently run through a second GSTrap HP column to separate protease and GST tag from cleaved enzyme. Untagged enzyme eluted in the flow-through of the second column, and fractions were pooled for SEC. Determination of Enzymatic Activity-Assays were performed in 50 mM BTP buffer, pH 6.8, at the stated temperature in a total volume of 1 ml. Initial rates were fitted to the Michaelis-Menten equation to determine kinetic parameters using GraFit (Erithacus software).
DAH7PS Assay-A modified continuous assay of Schoner and Herrmann (31) was used as described previously by Schofield et al. (32) for use at elevated temperatures. Enzyme activity was monitored by following the loss of absorbance at 232 nm due to consumption of PEP at 60°C in the presence of The drop sizes were 2 l, and the reservoir volume was 500 l. The crystallization trays were incubated at 20°C until immediately before loading the crystals into the loop for flash cooling prior to data collection. The reservoir solution containing 20% PEG 400 was used as a cryoprotectant. Intensity data were collected at the Australian Synchrotron using the MX2 beamline with an ADSC Quantum 315r detector (Area Detector Systems Corp.) at a temperature of 110 K. The collection and refinement data are provided in Table 3.
Phasing and Model Refinement-Diffraction data sets were processed and scaled using XDS (33) and Aimless (34). Initial phase estimates were calculated by molecular replacement using PHASER (35) with the monomeric core barrel (with CM domains removed) from the previously determined structure of LmoDAH7PS (PDB code 3TFC) as the search model. Two core barrels were placed in the asymmetric unit, and after rigid body refinement using REFMAC5 (36), this solution was further used to place the CM domains in the difference Fourier map using phased molecular replacement in MOLREP (37). Further structural refinement was performed using REFMAC5, and electron density maps were analyzed using Coot (38). Prephenate was modeled after the penultimate round of refinement. Structures were validated using the MolProbity server (39) and the tools of Coot.
Reconstruction of the Missing Residues and Side Chains of GspDAH7PS-Residues 1-2 and 355-362 in chain A as well as residues 88 -101 and 355-362 in chain B are absent in the crystal structure model of GspDAH7PS obtained after refinement due to the weak electron density in these regions. These residues, as well as some unmodeled side chains, were reconstructed using MODELLER (24) version 9.13. First, the tetrameric form was generated from the asymmetric dimeric unit in PyMOL (40), and the topologies and parameters for prephenate and Mn 2ϩ , obtained from the SwissParam server (41) and the MM3 force field (42)(43)(44), respectively, were added to the MODELLER library. Second, 100 models of GspDAH7PS were generated using the "automodel" routine of MODELLER tuned to improve model accuracy. More specifically, the variable target function method was set to slow, and the maximum number of conjugate gradient iterations was set to 300, although the level of molecular dynamics refinement was adjusted to slow and the complete optimization-refinement process was repeated twice. The "env.io.hetatm" instruction was defined as true to include the metal ion and prephenate in the model. During the model generation, only the missing residues and residues with missing atoms were allowed to move, and the chorismate mutase domains were restrained to keep their secondary structure intact, and a symmetry restraint was applied on the ␣-carbons of the symmetric chains of the tetramer. The best model was selected among the models presenting the lowest MODELLER objective function value and the lowest global DOPE score, following visual inspection and residue-by-residue DOPE score analysis. The C-terminal tail of this model was further refined using the "loopmodel" class of MODELLER (45) using the same protocol as described above, and the best of the 100 models generated was assessed using the same evaluation procedure.
Homology Model Construction of GspDAH7PS in Its Open Form-Homology modeling of GspDAH7PS in its open, active form was performed using the two previously determined structures of DAH7PS from L. monocytogenes EGD-e (PDB codes 3NVT and 3TFC) as templates. For each structure of LmoDAH7PS, the tetrameric form was generated in PyMOL, and the topology and parameters of PEP obtained with SwissParam as well as those of Mn 2ϩ were included into MODELLER libraries. The missing residues and side chains were then added with MODELLER using automodel to create 25 models using the same parameters and the same assessment procedure as described above. The best model for each of the structures was aligned, and their sequences were aligned against the full sequence of GspDAH7PS using T-COFFEE (46). The homology model of GspDAH7PS in its open form was performed with MODELLER using the automodel routine as before with the env.io.hetatm instruction set to true to include PEP and Mn 2ϩ in the model. Secondary structure restraints were applied to the chorismate mutase domains, and symmetry restraints were used for the symmetric chains of the tetramer. A total of 100 models were generated, as above, and the best model was selected among the models presenting the lowest MODELLER objective function value and the lowest global DOPE score, following visual inspection and residue-by-residue DOPE score analysis. SAXS-Measurements were performed at the Australian Synchrotron SAXS/WAXS beamline equipped with a Pilatus detector (1M, 170 ϫ 170 mm, effective pixel size, 172 ϫ 172 m). The wavelength of the x-rays was 1.0332 Å. The sample detector distance was 2.6 m, which provided an s range of 0.01-0.35 Å Ϫ1 (where q is the magnitude of the scattering vector, which is related to the scattering angle (2) and the wavelength () as follows: s ϭ (4/)sin(). Data were collected from samples in a 1.5-mm thin-walled glass capillary at 25°C at 2-s intervals.
Raw scattering data were processed with Scatterbrain developed at the Australian Synchrotron. Scattered intensity (I) was plotted versus s using Primus (47) and Origin 8. All samples were devoid of an increase in intensity at low s (indicative of aggregation). Guinier plots were linear for sR g Ͻ1. 3. The data sets for structural analysis were recorded with 447 data points over the range 0.005 Յ s Յ 0.35 Å Ϫ1 . Theoretical scattering curves were generated from the models of open form (active) and closed form (inactive) GspDAH7PS generated as described above and compared with experimental scattering curves using OLIGOMER (47). Theoretical scattering data of two distinct GspDAH7PS forms were further mixed and fitted to the individual experimental scattering profile in the presence or absence of prephenate with OLIGOMER to investigate the volume fraction of each form in the solution with and without ligand (47).
Additional Computational Methods-SASA for the open and closed models were calculated in VMD (48) using the "measure sasa" command and using a solvent probe radius of 1.4 Å. Sequence identities were determined using LALIGN (49) using the BLOSUM50 scoring matrix with an opening gap penalty of Ϫ12 and an extending gap penalty of Ϫ2. Sequence alignments were generated with T-Coffee (50) and rendered with ESPript 3 (51). Structural figures were generated with PyMOL.