A single amino acid substitution uncouples catalysis and allostery in an essential biosynthetic enzyme in Mycobacterium tuberculosis

Allostery exploits the conformational dynamics of enzymes by triggering a shift in population ensembles toward functionally distinct conformational or dynamic states. Allostery extensively regulates the activities of key enzymes within biosynthetic pathways to meet metabolic demand for their end products. Here, we have examined a critical enzyme, 3-deoxy-d-arabino-heptulosonate 7-phosphate synthase (DAH7PS), at the gateway to aromatic amino acid biosynthesis in Mycobacterium tuberculosis, which shows extremely complex dynamic allostery: three distinct aromatic amino acids jointly communicate occupancy to the active site via subtle changes in dynamics, enabling exquisite fine-tuning of delivery of these essential metabolites. Furthermore, this allosteric mechanism is co-opted by pathway branchpoint enzyme chorismate mutase upon complex formation. In this study, using statistical coupling analysis, site-directed mutagenesis, isothermal calorimetry, small-angle X-ray scattering, and X-ray crystallography analyses, we have pinpointed a critical node within the complex dynamic communication network responsible for this sophisticated allosteric machinery. Through a facile Gly to Pro substitution, we have altered backbone dynamics, completely severing the allosteric signal yet remarkably, generating a nonallosteric enzyme that retains full catalytic activity. We also identified a second residue of prime importance to the inter-enzyme communication with chorismate mutase. Our results reveal that highly complex dynamic allostery is surprisingly vulnerable and provide further insights into the intimate link between catalysis and allostery.

Proteins are dynamic molecules that exert functionally important motions over both long (milliseconds to seconds) and short (femptoseconds to microseconds) time scales. Protein structures and conformational dynamics are tuned for the delivery of complex protein roles. Allostery, where a signal initiated by ligand binding is communicated to elicit a remote functional response, exploits conformational dynamics (1). Such ability to communicate between sites and deliver control on activity is essential for biosynthetic enzymes. Most of these enzymes utilize allostery as a mechanism to control metabolic flux and enable delivery of metabolites in response to metabolic demand.
The enzyme 3-deoxy-D-arabino-heptulosonate 7-phosphate synthase (DAH7PS) 3 catalyzes the first step in the shikimate pathway and is feedback regulated by the pathway end products, the aromatic amino acids Phe, Tyr, and Trp. The DAH7PS family is categorized into subfamilies including type I␣, I␤, and type II DAH7PSs (2), which display a wide range of mechanisms for allosteric regulation. The common structural feature among all subfamilies is the core catalytic unit, which adopts a (␤/␣) 8 barrel-fold and hosts the active site. The subfamilies differ in both quaternary structure and in the types of structural elements decorating the catalytic core, and these decorations deliver the binding sites for allosteric ligands. Type I␣ DAH7PS, including those from Escherichia coli (3)(4)(5), Saccharomyces cerevisiae (6,7), and Neisseria meningitides (8) contains an N-terminal extension and extended loops to the core barrel, which host the binding site for a single allosteric ligand (Fig.  1A). Whereas type I␤ contains a group of DAH7PS that has a discrete allosteric domain covalently linked to the catalytic barrel, such as those from Thermotoga maritima (9) or Geobacillus sp. (Fig. 1B) (10).
The first fully characterized type II DAH7PS was from Mycobacterium tuberculosis (MtuDAH7PS, Fig. 1C) (11,12). MtuDAH7PS contains both an N-terminal extension (three helices) and two other inserted helices to the core catalytic barrel, which constitute distinct allosteric binding sites specific to three different allosteric ligands (Trp, Phe, and Tyr) (13,14). The N-terminal extension (helices ␣0a-␣0c) contributes to binding sites for Phe and Tyr, whereas the inserted helices (␣2a-␣2b) contribute to the Trp-binding site (Fig. 1C). The additional structural elements are also associated with the formation of the dimer and tetramer interfaces to form the tetrameric quaternary structure of MtuDAH7PS (Fig. 1D).
MtuDAH7PS exhibits a complex internal dynamic communication network resulting in a highly sophisticated regulation mechanism (13,15). Binary combinations of aromatic amino acids that include Trp significantly inhibit catalytic activity of DAH7PS, and the presence of all three aromatic amino acids completely abolishes activity (13,14). Communication exists between allosteric sites as well, as binding of one ligand favors subsequent binding of other ligands (15). Binding of the allosteric ligands is not associated with any significant conformational change and the mechanism of the allosteric regulation in MtuDAH7PS is mediated by internal changes in protein dynamics (15).
The complexity of this dynamic allosteric mechanism is not limited to within the MtuDAH7PS enzyme. It has been shown that the allosteric machinery of MtuDAH7PS can be utilized by another small protein partner, chorismate mutase (MtuCM), by forming a noncovalent complex (16 -18). CM catalyzes a downstream reaction in the shikimate pathway, converting chorismate to prephenate, and is at the start of the branchpoint of the shikimate pathway that leads to the production of Phe and Tyr. It has been shown that complex formation between MtuDAH7PS and MtuCM significantly enhances the catalytic activity of MtuCM (16 -18). Furthermore, the normally unregulated MtuCM becomes sensitive to inhibition by Phe, which binds to the dimer interface on MtuDAH7PS (16 -18). A similar scenario was also observed in another type II enzyme, i.e. the DAH7PS from Corynebacterium glutamicum (19), indicating this inter-enzyme communication may be a more common feature among the type II DAH7PS enzymes.
In this study, we aimed to pinpoint the precise molecular mechanism that governs the communication networks that allow for such sophisticated allosteric control of the gatekeeper enzyme for aromatic biosynthesis. Using type II MtuDAH7PS as a reference, sequence correlations and key residue roles were explored, illuminating a crucial point in the internal dynamic pathway that is responsible for propagation of allosteric signal. We report a variant enzyme with interrupted signal at this crucial point that is completely lacking any allosteric response yet maintains full catalytic capacity.

Sequence analysis and protein sectors in type II DAH7PS
Type II DAH7PS utilizes internal dynamic networks to facilitate allostery, thus it is not possible to pinpoint residues involved in the network purely based on structural inspection. Due to the functional importance of allostery, such networks are expected to be conserved and thus to have coevolved, so that information regarding the internal dynamic communication pathway lies in the correlations between residues in the sequences of type II DAH7PS. Statistical coupling analysis (SCA) has been demonstrated to identify successfully groups of residues that have coevolved to preserve function or structural integrity of enzyme families (20 -22). We employed SCA to investigate sequence co-variations within the type II DAH7PS family that may reflect important communication pathways EDITORS' PICK: Uncoupling dynamic allostery in DAH7PS that are preserved due to functional importance, such as allostery, through the course of evolution. In SCA, the correlation matrix of residue conservations is computed from a multiple sequence alignment of the protein family, and analyzed by spectral decomposition using independent component analysis (20). Independent components (IC) are then identified, which represent conserved, differentially evolving functional units in protein families. The ICs that show inter-correlations are further grouped into protein sectors, which in other enzyme families, have been shown to correspond to distinct functional roles (20).
A total of 4217 amino acid sequences of type II DAH7PS were included in the initial multiple sequence alignment (MSA). Pairwise identities between the sequences in the alignment were analyzed and showed a distribution with an average identity value of 50% (Fig. S1). The initial MSA was then processed by removing highly gapped positions and sequences, and mapping position numbers to the reference sequence (MtuDAH7PS in this case) (20 -23). The processed MSA contained 3345 sequences with 431 positions. A weighting factor was then applied to correct for the biasing effects of sequences with high identities, and the effective number of sequences in this processed MSA was calculated to be 510, which indicates the sequences in the processed MSA are sufficient to give good estimates of amino acid frequencies that are representative of the protein family (20).
The correlations of positional conservations in the processed MSA of type II DAH7PS were analyzed by SCA. Eleven ICs were identified, which were further grouped into three protein sectors (Fig. S2). To examine possible functional roles for these protein sectors, residues identified in each protein sector were mapped onto the structure of MtuDAH7PS, and the locations of protein sectors and ICs were examined. Sector 1 contains 119 residues, and consists of ICs 1, 7, and 8 ( Fig. 2A). Residues in ICs 1 and 8 are mostly located near the center of the barrel hosting the active site (Fig. S3). Interestingly, IC 1 contains a number of residues in the active site that are known to play a role in substrate binding, such as Lys-306, Arg-337, Arg-126, Arg-284 for PEP binding, and Arg-135, Ser-136, Tro-280, Glu-248, and Lys-133 for E4P binding (14). IC 1 also contains residues of the Phe-specific binding site, including Phe-91, Pro-56, and Tyr-173. Residues in IC 7 appear to form a potential pathway between known allosteric binding sites for Trp and Phe (Fig. 2B), and contains residues of the Trp-binding site (Ala-192, Leu-107, and Leu-194). The locations and functional roles of residues of Sector 1 suggest that it likely contributes to the communication between the allosteric ligand sites, and to the active sites within the DAH7PS tetramer.

EDITORS' PICK: Uncoupling dynamic allostery in DAH7PS
the exterior of the catalytic barrel. Sector 2 consists of ICs 2, 3, 4, 5, and 11. Residues in ICs 3, 4, 5, and 11 do not appear to form a linked pathway between any known functional sites of type II DAH7PS (Fig. S4). However, residues in IC 2 appear to connect the allosteric ligand sites to the interface that is formed between DAH7PS and CM upon complexation (Fig. 3B). It is possible that residues in Sector 2 play a role to facilitate the communication between type II DAH7PS and CM upon complex formation, with major contributions from IC 2.
Sector 3 contains 49 residues and consists of ICs 6, 9, and 10 ( Fig. S5), which are mostly located between helices and ␤-sheets. Most of the residues identified in Sector 3 do not belong to any known functional sites on DAH7PS, thus the functional role of Sector 3 is yet unclear. Previous SCA studies on other protein families have also identified protein sectors with unknown apparent functional roles (7). In addition, coevolved residues could be identified for structural or functional importance (20,24). Based on the observed locations of Sector 3, it likely contributes to maintaining of structural-fold of type II DAH7PS family.

A single amino acid substitution within Sector 1 completely annihilates allostery
SCA identified groups of residues that may play an important role in the communication pathways between the different functional sites in type II DAH7PS. In particular, both IC 7 in Sector 1 and IC 2 in Sector 2 form potential communication pathways connecting different sites, and interestingly, strong cross-sector correlations are also observed between IC 7 and IC 2 ( Fig. S6), indicating the cross-talk between different internal residue networks and the level of complexity of the whole signal propagation mechanism.
To further interrogate the communication pathways in MtuDAH7PS without physically changing any binding sites, we substituted Gly-190 from IC 7 with a proline residue using sitedirected mutagenesis, with the intention to alter the dynamic properties within the predicted allosteric communication pathway (Fig. 4A). Gly-190 is located in close proximity to the bound Trp ligand (ϳ13 Å) but is not directly involved in the binding of Trp. Gly-190 is remote from the active site with a distance of ϳ30 Å to the active site Mn 2ϩ ion (Fig. 4B).
The G190P substitution did not alter the catalytic features of the enzyme. Kinetic assay results indicated MtuDAH7PS G190P displayed highly similar kinetic characteristics compared with the WT enzyme, with comparable K m and k cat values (Table 1). Furthermore, MtuDAH7PS G190P retained the ability to activate MtuCM, demonstrating that the inter-enzyme interaction was intact (Table 1) (16 -18). Both the WT enzyme and substituted MtuDAH7PS enhanced the activity of MtuCM over 100-fold in similar ways, by lowering the K m value for chorismate and enhancing turnover.
Whereas this G190P substitution had no effect on catalysis, the impact on allostery was far more striking. In marked contrast to the WT enzyme, which is synergistically inhibited by combination of aromatic amino acids involving Trp, activity of the MtuDAH7PS G190P variant was completely unaffected by the presence of aromatic amino acids either alone or in combinations (Fig. 5A). Full activity was retained even in the presence of all three amino acids, which was shown to completely inhibit the WT enzyme. These results indicate that this single amino acid substitution at Gly-190 has completely abolished the allosteric response in the MtuDAH7PS enzyme, while maintaining its full catalytic capacity.
The lack of allosteric response in the MtuDAH7PS G190P variant is due to severed communication pathways as a result of the G190P substitution, rather than to the attenuation of ligand binding. This was confirmed using differential scanning fluorimetry (DSF) experiments in the presence and absence of allosteric ligands, results of which showed large changes in T m associated with ligand binding. These changes in thermal stability, precisely paralleled those observed for MtuDAH7PS WT (Table 2) (25). As the G190P substitution is in close proximity to the Trp-binding site, we further probed the binding of Trp to the MtuDAH7PS G190P variant using isothermal titration calorimetry experiments. A K d value of 11.7 Ϯ 0.7 M was obtained (Fig. S7). Albeit this measurement shows a slight increase compared with the WT enzyme (K d ϭ 4.7 Ϯ 0.1 M) (15), the result confirms binding of Trp to MtuDAH7PS G190P with comparable affinity, consistent with the DSF findings.
We also examined the effect of the G190P substitution on the allosteric response of MtuCM. Similar to that observed for the  (Fig. 5B). However, different responses were observed in the presence of all three aromatic amino acids. The WT MtuDAH7PS-MtuCM complex exhibits a synergistic inhibitory response to ternary combination of aromatic amino acids resulting in only ϳ5% activity remaining. In contrast, combinations of aromatic amino acids do not deliver a greater inhibitory response on MtuCM activity in the presence of the MtuDAH7PS G190P variant. The lack of synergy between Phe/Tyr and Trp in the presence of MtuDAH7PS G190P implies that the synergistic communication between allosteric sites has been severed by the G190P substitution. The inter-enzyme catalytic enhancement and allosteric regulation between MtuDAH7PS and MtuCM appear to be otherwise unaffected by the G190P substitution.
We then further interrogated the effect of the G190P substitution on the structure of the DAH7PS enzyme. The solution structure of the MtuDAH7PS G190P variant was analyzed by small angle X-ray scattering (SAXS) and compared with WT enzyme (Fig. 5C). SAXS data confirmed that the quaternary structure of the MtuDAH7PS G190P variant was maintained ( Table 3). As for the WT enzyme, MtuDAH7PS G190P adopts a tetrameric subunit arrangement, with comparable R g , D max , and porod volume values. MtuDAH7PS G190P experimental scattering profile agrees well with the theoretical scattering generated from the MtuDAH7PS WT crystal structure (PDB 3NV8). We also obtained the crystal structure of the MtuDAH7PS G190P variant (PDB 6PBJ), which showed high similarity to that of the WT structure (Fig. 5D), with a rootmean-square deviation value of 0.35 Å (matching 1556 C ␣ atoms of the tetramer) compared with the WT crystal structure (PDB 3NV8), further confirming that the overall structure is undisrupted by the G190P substitution. Several regions are disordered in the structure of MtuDAH7PS G190P , including residues 264 -266, 372-380, and 429 -436 in chain A, and residues 11-15, 264 -265, 372-381, and 413-443 in chain B. Clear electron density is observed around the Pro residue at the site of mutation (Fig. S8A), with minimal change in conformations of local residues in close proximity to Pro-190 (Fig. S8B). Although a subtle shift in the inserted helix ␣2a consisting of residues 194 -209 can be observed (Fig. S8C), the residues involved in the Trp-binding site are mostly unaffected (Fig.  S8D), with the exception of Leu-194, which shows a small change in side chain position.

A Sector 2 substitution affects inter-enzyme communications
Sector 2, identified by SCA, identifies residues with a likely role in inter-enzyme communication between DAH7PS and CM. To interrogate the functional connections within this sector, a second target, Tyr-131, was selected (Fig. 6A). Tyr-131 is located between the Phe-specific binding site and the active site and is not directly involved in any interactions with the Phe  EDITORS' PICK: Uncoupling dynamic allostery in DAH7PS ligands (Fig. 6B). Tyr-131 is ϳ11 Å away from the active site metal ion Mn 2ϩ and ϳ12 Å away from the bound Phe ligand. We generated the variant MtuDAH7PS Y131A to examine the effect of this residue on the communication pathways. MtuDAH7PS Y131A variant was found to be catalytically active and demonstrates overall similar catalytic efficiencies compared with the WT enzyme (Table 1). Although the K m values of MtuDAH7PS Y131A decreased compared with the WT enzyme, its k cat value was also reduced by around 75%. The allosteric response of MtuDAH7PS Y131A by aromatic amino acids is somewhat impaired, in which the presence of all three aromatic amino acids caused moderate levels of inhibition on the MtuDAH7PS Y131A activity with ϳ30% activity still remaining. The response of MtuDAH7PS Y131A to Phe alone or in binary combinations is attenuated and appears to enhance the DAH7PS activity (Fig. 7A).
Surprisingly, MtuDAH7PS Y131A is capable of enhancing the overall catalytic efficiency of MtuCM much more significantly, by over 400-fold, compared with the 100-fold activation observed under the same conditions with MtuDAH7PS WT or MtuDAH7PS G190P (Table 1). This significant rate enhancement for MtuCM observed in the presence of the MtuDAH7PS Y131A variant was largely contributed by the increased turnover number. The inhibitory effect of Phe on MtuCM upon complexation with MtuDAH7PS Y131A was largely reduced compared with that observed in the WT enzyme, with over 60% remaining MtuCM activity in the presence of Phe alone (Fig. 7B). Whereas the inhibitory responses of MtuCM-DAH7PS Y131A to other aromatic amino acid combinations are similar to those observed in MtuCM-DAH7PS G190P .
The Y131A substitution also resulted in some modifications in the structure of the enzyme as revealed by SAXS analysis (Fig.  7C). The structural parameters generated from the SAXS scattering indicate that MtuDAH7PS Y131A may adopt a slightly larger and more loosely-packed overall shape compared with the WT enzyme (Table 3). We generated a rigid body model of MtuDAH7PS Y131A ( 2 ϭ 1.9, Fig. 7F), which shows a flattened tetramer interface with an altered relative position between the tight dimers compared with the WT enzyme (which does not fit well with the MtuDAH7PS Y131A SAXS data with 2 value of 5.6, Fig. 7E). The predicted loose tetramer interface in MtuDAH7PS Y131A is echoed by the observation of a much lower melting temperature measured in DSF experiments, in which a T m of 44°C was measured for apo MtuDAH7PS Y131A ( Table 2) compared with that of ϳ53°C in WT enzyme and

EDITORS' PICK: Uncoupling dynamic allostery in DAH7PS
MtuDAH7PS G190P variant. Interestingly, the flattened tetramer interface also impacts the regions that interact with MtuCM upon complex formation (Fig. 7D), therefore, it is possible that the improved enhancement of MtuCM activity by the MtuDAH7PS Y131A variant is partially contributed by its changed structure.

Discussion
Allosteric regulation of known type II DAH7PS enzymes is dynamically driven (15,19,26). Among these, MtuDAH7PS shows the most complex dynamic allosteric response. MtuDAH7PS utilizes four distinct functional sites, including one active site and three allosteric sites that are specific to each of the three aromatic amino acids (Trp, Phe, and Tyr). Binding of allosteric ligands does not cause any observable structural changes on the enzyme. Instead, MtuDAH7PS exploits complex internal dynamic communication pathways to deliver the allosteric signal.
By examining co-evolved residues across the type II DAH7PS subfamily, we identified a critical node, Gly-190, in the internal dynamic network within MtuDAH7PS. A single G190P substitution abolishes the allosteric response in MtuDAH7PS, but unexpectedly leaves catalysis and ligand binding intact. Deciphering dynamic allosteric networks that deliver functional change can be challenging, and recent studies showed that amino acid substitutions of key residues in the allosteric networks usually also result in altered catalysis (27)(28)(29)(30). For example, substitutions of two hydrophobic residues involved in the dynamically driven allosteric communication in protein kinase A lead to inactivation of the enzyme (27) in ATP-phosphoribo-syltransferase, substitution of an Arg important for allosteric signal transmission to either Gln or Ala resulted in severely impaired catalysis (31). In contrast to the mutagenesis experiments in other reports, which modify the properties of the amino acid side chains, it should be noted that our G190P substitution only alters the dynamic property of the protein backbone. Given that no overall structural changes are observed and that allosteric ligand binding is fully maintained in MtuDAH7PS G190P , it appears that loss of the allosteric signal here is due to altered dynamic motions of the protein backbone. Gly-190 is located at the start of the two inserted helices ␣2a and ␣2b, which are responsible for the formation of Trp-binding sites and the tetramer interface in MtuDAH7PS. Interestingly, at the end of these two inserted helices is another Gly residue (Gly-232), and the substitution of G232P changed the quaternary structure of MtuDAH7PS, which altered catalysis and broke allostery (32). The drastic changes caused by these Gly to Pro mutations clearly indicate the functional importance of helices ␣2a and ␣2b in communicating the allosteric signal within MtuDAH7PS. To the best of our knowledge, the uncoupling of allostery and catalysis caused by G190P substitution is the most clear-cut example of dynamic communication network disruption to date.
Another complex feature of the allosteric regulation mechanism of MtuDAH7PS is the inter-enzyme communication to MtuCM (16 -18). This small enzyme partner forms a noncovalent complex with MtuDAH7PS, which enables it to adopt the Phe-binding sites of MtuDAH7PS to deliver allosteric inhibition of MtuCM by Phe binding. SCA high-

EDITORS' PICK: Uncoupling dynamic allostery in DAH7PS
lighted Tyr-131 as a part of the dynamic network that may be important for the inter-enzyme communications between MtuDAH7PS and MtuCM. Here we showed that modifying the side chain through Y131A substitution led to impaired inhibitory responses of MtuCM and MtuDAH7PS to Phe, and Phe alone appears to enhance activity of MtuDAH7PS.
Unexpectedly MtuCM activity was significantly more enhanced in the presence of the MtuDAH7PS Y131A variant, compared with that in the presence of the WT enzyme or MtuDAH7PS G190P . The Y131A substitution causes structural changes in DAH7PS, which likely resulted in the enhanced activation of MtuCM. Dynamic allostery or entropically-driven allostery has been extensively studied in recent years, with its intriguing functional responses accompanied with the apparent lack of observable structural changes (33)(34)(35)(36)(37)(38)(39)(40)(41)(42). Dynamic allostery is an intrinsic property of enzymes and is intimately linked with catalysis (34,39,42,43). In this study, we identified a key node embedded in an intricate and extensive allosteric network, which is remote to any ligand-binding site. We have broken the allosteric links through a simple change in backbone dynamics. Our results reveal that the highly sophisticated and complex dynamic allostery can be surprisingly vulnerable. This vulnerability may represent a glimpse of the evolution of such an important biosynthetic enzyme en route to more advanced regulations and provides further insight on the nature of the intimate link between catalysis and allosteric regulation.

Statistical coupling analysis
Full sequence alignment of type II DAH7PS was obtained from PFAM database (44,45) (release 32.0, accession code PF01474, total of 4217 sequences) and was subject to pre-processing with default settings as described (20). After processing, the alignment contains 3345 sequences and 431 positions. The number of effective sequences was 510. According to original publication of SCA (20), an effective sequence number of greater than 100 indicates the sequence alignment is large enough to give good estimates of amino acid frequencies. The pre-processed sequence alignment was then subject to the full SCA analysis using the pySCA toolbox, which consists of five steps as outlined (20).

Site-directed mutagenesis of MtuDAH7PS mutants
QuikChange site-directed mutagenesis was used to generate protein variants MtuDAH7PS G190P and MtuDAH7PS Y131A using WT pProExHTa-MtuDAH7PS (Rv2178c) plasmid as template (12). Primers for mutagenesis were designed with the point of mutation in the center and sufficient overlaps (ϳ15 base pairs) on both ends. Primers (5Ј-3Ј) GCTGACT-TCGTCGCCCCTGGCGTCGCTG and CAGCGACGCCA-GGGGCGACGAAGTCAGC were used for mutagenesis of MtuDAH7PS G190P , CATCGCCGGTCAGGCGGCGAAGC-CTCGG and CCGAGGCTTCGCCGCCTGACCGGC-GATG were used for MtuDAH7PS Y131A . Mutagenesis was performed using a QuikChange II Site-directed Mutagenesis Kit (Agilent) with reaction (50 l) and cycling protocols recommended by the manufacturer. PCR products were treated with DpnI (New England BioLabs) for removal of methylated templates. The product was then transformed into Stellar TM (Clontech) cells and sequence was verified before transforming into E. coli BL21(DE3) pGroESL cells for expression.

Expression and purification of MtuCM and MtuDAH7PS variants
MtuCM and MtuDAH7PS variants were expressed and purified following the previously described protocols for the WT MtuCM and MtuDAH7PS, respectively (12,17). Briefly, transformants containing the recombinant plasmids were preincubated in Luria-Bertani medium at 37°C with the appropriate antibiotics with shaking until mid-logarithmic phase (OD 600 Ӎ 0.4 -0.6) before the protein expression was induced by addition of isopropyl ␤-D-thiogalactopyranoside (Roche) (0.5 mM) and incubated overnight at 23°C. Following cell harvest and lysis, MtuCM was purified by using a GSTrap HP column (GE Healthcare) before and after the tobacco etch virus protease treatment (17). MtuDAH7PS variants were purified by using a HiTrap HP column (GE Healthcare) before and after tobacco etch virus protease treatment. The proteins were then further purified using size-exclusion chromatography (SEC).

Kinetic measurements
Steady-state kinetics were measured for DAH7PS activity by monitoring the consumption of PEP at 232 nm (⑀ ϭ 2.8 ϫ 10 3 M Ϫ1 cm Ϫ1 at 303 K) and for CM activity by monitoring the consumption of chorismate at 274 nm (⑀ ϭ 2.63 ϫ 10 3 M Ϫ1 cm Ϫ1 at 303 K) using a Varian Cary 100 UV-visible spectrophotometer (12,17). For the DAH7PS assays, initial velocity values were determined at varying concentrations of each substrate, when the concentration of the other substrate was fixed at saturating concentrations (defined as at least 5-fold higher than the K m value). The reaction mixture contained 0.05 M protein, varying or saturating concentrations of PEP (for determination of in the same buffer environment as for the DAH7PS assays. Each reaction was incubated for 10 min prior to initiation with chorismate to allow formation of the CM-DAH7PS complex. Apparent kinetic parameters were determined by fitting triplicate data to the Michaelis-Menten equation using GraphPad Prism 7.

Inhibition studies
Inhibition studies for DAH7PS and CM activity in response to aromatic amino acids alone or in combinations were performed using the same method as described for kinetic measurements. The l of protein at 0.9 mg/ml, and 1 l of ϫ20 SYPRO Orange dye. The plate was subjected to a thermal cycling program that heated the samples from 20 to 85°C in 0.02°C/s increments. Each protein sample was measured in triplicate and compared with a negative control in which no protein sample was present. Melting temperatures were calculated using Protein Thermal Shift Software 1.3 at the temperature at which the maximum inflection occurred along the generated melting curve of each protein sample. Errors were calculated based on standard deviation from triplicate measurements.

SAXS data collection, analysis, and modeling
SEC-SAXS co-flow setup was used for collection of SAXS data at the Australian Synchrotron SAXS/WAXS beamline equipped with a Pilatus detector (1M, 170 ϫ 170 mm, effective pixel size, 172 ϫ 172 m) (46). The X-ray wavelength was 1.0332 Å. The sample-detector distance was 1.6 m, which delivered a s range of 0.01-0.5 Å Ϫ1 (where s is the magnitude of the scattering vector, which is related to the scattering angle (2) and the wavelength () as follows: s ϭ (4/)⅐sin). Scattering data were collected at 25°C following elution of the protein samples (6 mg ml Ϫ1 ) from a SEC column (Superdex 200_Increase 5/150), pre-equilibrated with buffer containing 10 mM BTP (pH 7.5), 150 mM NaCl, 200 M PEP, 200 M TCEP, and 3% (v/v) glycerol. Raw data were processed and background-subtracted using Scatterbrain (9). Scatterings from A 280 nm peaks were summed and averaged for SEC experiments. Scattering intensity (I) versus s of each protein was generated with Guinier plots representing a linear range for R g Ͻ 1.3, and plots were scaled for comparison using Primus (47).

EDITORS' PICK: Uncoupling dynamic allostery in DAH7PS
Theoretical scattering profile of MtuDAH7PS was generated from the model coordinates (PDB 3NV8), compared, and fitted with corresponding experimental scatterings with Crysol (48).
Rigid body modeling was performed for MtuDAH7PS Y131A using SASREF (49). Monomer coordinates were extracted from PDB 3NV8 and used for SASREF along with the SAXS scattering data for MtuDAH7PS Y131A with no symmetry applied. The following constrains were applied: dist 25

Crystallization and refinement
Crystal structure of the MtuDAH7PS G190P variant was obtained in conditions containing 20 mM BTP, 150 mM NaCl, 0.5 mM TCEP, 0.2 mM PEP, 0.1 mM MnCl 2 , 2 M Li 2 SO 4 , and 2% PEG 400. Diffraction data were collected at the Australian Synchrotron (50). Data were processed in Imosflm (51) and the initial model was obtained by molecular replacement using the ligand-free crystal structure of the MtuDAH7PS WT enzyme as the search model (PDB 3NV8). The structure was further refined using REFMAC5 (52). Details of the data collection and refinement statistics are in Table S1.

Data availability
The coordinates and structure factors for MtuDAH7PS G190P are deposited in the Protein Data Bank with accession code 6BPJ. All other data are contained in this manuscript.