Conservation of the structure and function of bacterial tryptophan synthases

The tryptophan synthases from three human pathogens show remarkable structural conservation, but at the same time display local differences in both their catalytic and allosteric sites that may be responsible for the observed differences in catalysis and inhibitor binding. This functional dissimilarity may be exploited in the design of species-specific enzyme inhibitors.

Tryptophan biosynthesis is one of the most characterized processes in bacteria, in which the enzymes from Salmonella typhimurium and Escherichia coli serve as model systems. Tryptophan synthase (TrpAB) catalyzes the final two steps of tryptophan biosynthesis in plants, fungi and bacteria. This pyridoxal 5 0 -phosphate (PLP)-dependent enzyme consists of two protein chains, (TrpA) and (TrpB), functioning as a linear heterotetrameric complex containing two TrpAB units. The reaction has a complicated, multistep mechanism resulting in the -replacement of the hydroxyl group of l-serine with an indole moiety. Recent studies have shown that functional TrpAB is required for the survival of pathogenic bacteria in macrophages and for evading host defense. Therefore, TrpAB is a promising target for drug discovery, as its orthologs include enzymes from the important human pathogens Streptococcus pneumoniae, Legionella pneumophila and Francisella tularensis, the causative agents of pneumonia, legionnaires' disease and tularemia, respectively. However, specific biochemical and structural properties of the TrpABs from these organisms have not been investigated. To fill the important phylogenetic gaps in the understanding of TrpABs and to uncover unique features of TrpAB orthologs to spearhead future drug-discovery efforts, the TrpABs from L. pneumophila, F. tularensis and S. pneumoniae have been characterized. In addition to kinetic properties and inhibitor-sensitivity data, structural information gathered using X-ray crystallography is presented. The enzymes show remarkable structural conservation, but at the same time display local differences in both their catalytic and allosteric sites that may be responsible for the observed differences in catalysis and inhibitor binding. This functional dissimilarity may be exploited in the design of species-specific enzyme inhibitors.

Introduction
Tryptophan synthase (TrpAB) is a pyridoxal 5 0 -phosphate (PLP)-dependent enzyme that participates in the final two steps of tryptophan synthesis in plants, fungi and bacteria (reviewed in Dunn, 2012;Raboni et al., 2003Raboni et al., , 2009Dunn et al., 2008). The enzyme consists of two protein chains, (TrpA) and (TrpB) (Crawford & Yanofsky, 1958), that operate as a linear heterotetrameric complex containing two functional TrpAB units (Fig. 1). In bacteria, TrpA and TrpB are encoded by usually adjacent trpA and trpB genes that belong to the highly regulated tryptophan-biosynthesis operon (reviewed in Merino et al., 2008). The TrpA subunit converts indole-3-glycerol phosphate (IGP) into glyceraldehyde 3-phosphate (G3P) and indole (IND) (Fig. 2). Subsequently, the latter product is utilized by TrpB, where it reacts with the l-serine (l-Ser) substrate to generate l-tryptophan (l-Trp). The reaction has a complicated, multistep mechanism involving enzyme-cofactor and substrate covalent adducts and results in the -replacement of the hydroxyl group of l-Ser with the indole moiety ( Fig. 2) (reviewed in Raboni et al., 2009).
As originally shown for TrpAB from the Gram-negative Salmonella typhimurium (StTrpAB), TrpA adopts a canonical (/) 8 -barrel fold (also known as a TIM barrel) with numerous additional elements (Hyde et al., 1988;Figs. 1 and 3). The active site is located at the top of the central -barrel, with two acidic residues involved in catalysis: StGlu49 belonging to the S2 strand and StAsp60 originating from loop L2. Another structural element, loop L6, serves as a lid closing over the binding pocket. TrpB represents a type II PLP-dependent enzyme with two domains, the N-and C-terminal domains, with the active site located in a cleft between them and carrying the covalently attached PLP cofactor. The N-terminal domain encompasses the so-called communication domain (COMM) that plays a key role in coordinating the activity of the two active sites (Schneider et al., 1998). In the tetrameric arrangement, the TrpA and TrpB catalytic sites of the adjoining subunits are connected by a 25 Å long hydrophobic channel that facilities indole transport from TrpA to TrpB.
The TrpA-and TrpB-catalyzed chemical transformations are highly controlled by allosteric effects and other factors, for instance the binding of monovalent cations to TrpB, linked to substrate channeling. These molecular measures, together with other bacterial regulatory mechanisms (Merino et al., 2008), are in place to ensure that the cellular resources are efficiently utilized to produce l-Trp, which is a scarce and most energetically expensive amino acid to biosynthesize (Akashi & Gojobori, 2002). The well documented ligand-induced reciprocal communication between subunits leading to the mutual activation involves conformational rearrangements. During the catalytic process, both TrpA and TrpB cycle between a low-activity open conformation ( O or O ) and a high-activity closed state ( C or C ) (Dunn, 2012), depending on the reaction state. The formation of the aminoacrylate Schiff-base intermediate, E AA , from l-Ser and PLP in TrpB triggers movement of the TrpB COMM domain towards a closed state ( C ), which subsequently activates TrpA by closure of the L6 loop ( C ). In a reciprocal process, IGP substrate binding to TrpA promotes an C state, which in turn activates TrpB ( C ). The two protein chains convert back to their open states when the l-Trp external aldimine, E A,ex2 , is produced.
The availability of l-Trp, either supplied by the environment or synthesized in cellulo, is a prerequisite for bacterial survival. Some species rely heavily on external sources and maintain either no or only limited functionality of the l-Trp operon, while others preserve the complete system for de novo biosynthesis. The absence of the l-Trp biosynthetic pathway in animals and humans makes it a potentially attractive drug target for the treatment of bacterial diseases, even though the enzymes involved are only essential under certain conditions; that is, when exogenous l-Trp becomes depleted. Recent studies exploring these avenues showed that anthranilate synthase component I, TrpE (Zhang et al., 2013)  Overall structure of the tryptophan synthase heterotetramer from S. pneumoniae. TrpA is shown in yellow and TrpB is shown in cyan, with the COMM domain shown in orange and the PLP cofactor depicted in a sphere representation.
gers the expression of host indoleamine 2,3-dioxygenase (IDO-1), an enzyme responsible for l-Trp breakdown, or possibly even before this defense mechanism is mounted (Wellington et al., 2017). Similar mechanisms inducing l-Trp starvation also function in lung-specific mouse infections with Streptococcus pneumoniae and Francisella tularensis, which are Gram-positive and Gram-negative bacteria, respectively. Under such conditions, the latter organism also requires TrpAB for growth (Peng & Monack, 2010). Other pathogens that utilize tryptophan biosynthesis to evade host defenses or even to highjack it for their own purposes include urogenital serovars of Chlamydia trachomatis (a Gram-negative obligate intracellular parasite), which employ a partly dysfunctional TrpAB to produce l-Trp from external sources of indole provided by coexisting bacteria (Caldwell et al., 2003;Bonner et al., 2014). The growing list of human pathogens in which the l-Trp biosynthetic pathway plays an important role extends beyond prokaryotes. For example, Cryptosporidium species (parasitic protozoa) inhabiting intestines encode bacteriaderived TrpB, which potentially acts in a similar fashion as it does in C. trachomatis (Sateriale & Striepen, 2016). Specific biochemical and structural traits of the tryptophan synthases from these organisms have not been explored, with the recent exception of the M. tuberculosis ortholog. The structural and functional information gathered over the past 60 years has helped to explain the roles of individual residues in catalysis and allosteric regulation of the two active sites. Research has focused primarily on a prototypic tryptophan synthase from S. typhimurium (StTrpAB) and to a lesser extent those from E. coli (Heilmann, 1978;Lane & Kirschner, 1983;Drewe & Dunn, 1985, 1986Houben & Dunn, 1990;Lim et al., 1991) and Pyrococcus furiosus (Yamagata et al., 2001;Ogasahara et al., 2003;Hioki et al., 2004;Lee et al., 2005;Buller et al., 2015;Heilmann, 1978;Lane & Kirschner, 1983;Drewe & Dunn, 1985, 1986Houben & Dunn, 1990;Lim et al., 1991). Tryptophan synthase has become a prototype system to study the peculiarities of allostery and substrate channeling (Hilario et al., 2016;Niks et al., 2013;Rhee et al., 1996;Rowlett et al., 1998;Spyrakis et al., 2006). TrpA is also one of the model proteins that have been used to investigate protein-folding mechanisms (Wu & Matthews, 2002;Bilsel et al., 1999;Yang et al., 2007;Vadrevu et al., 2008;Wu et al., 2007;Michalska et al., 2015). The sparsity of biochemical/structural investigations of other orthologs possibly stems from challenges in obtaining high-quality TrpAB samples and also from interest being focused on very detailed mechanistic aspects rather than on species-specific variations. Importantly, though, as shown by our recent study of M. tuberculosis TrpAB (MtTrpAB; Wellington et al., 2017), these so-far ignored differences, especially within the nonconserved tunnel lining, may have profound consequences for the discovery and design of new allosteric inhibitors. Therefore, to fill the important phylogenetic gaps in our understanding of TrpABs and to uncover potential unique features of other orthologs to facilitate future drug-discovery efforts, we biochemically characterized three TrpABs from Gram-positive and Gram-negative pathogens: Legionella pneumophila Philadelphia, F. tularensis and S. pneumoniae (LpPhTrpAB, FtTrpAB and SpTrpAB, respectively). In addition to kinetic properties and inhibitor-binding capabilities, we also provide high-resolution structural information gathered using X-ray crystallography for the FtTrpAB and SpTrpAB complexes and for two subunits: LpPhTrpA and that from L. pneumophila Paris (LpPaTrpA).

TrpAB gene cloning
The gene cloning was performed as reported previously (Kim et al., 2011). Briefly, F. tularensis Schu 4, L. pneumophila Philadelphia, L. pneumophila Paris and S. pneumoniae TIGR4 genomic DNAs were used as templates for PCR of the genes coding for the TrpA and TrpB subunits of tryptophan synthase. Vector-compatible primers for the amplification of the DNA fragments coding for the subunits were designed using an online tool (https://bioinformatics.anl.gov/targets/ public_tools.aspx; Yoon et al., 2002). The TrpA subunit peptides that were cloned were as follows: 1-269 for FtTrpA, 1-272 for LpPhTrpA and LpPaTrpA, and 1-258 for SpTrpA. The TrpB subunit peptides that were cloned were as follows: 1-396 for FtTrpB, 13-396 for LpPhTrpB and 4-407 for SpTrpB. Purified PCR products were treated with T4 DNA polymerase in the presence of dCTP (Eschenfeldt et al., 2010) according to the vendor's specification (New England Biolabs, Ipswich, Massachusetts, USA). The protruded DNA fragment for each of the TrpA subunits was mixed with T4 DNA polymerasetreated vector pMCSG68 (PSI:Biology-Materials Repository) to allow ligation-independent cloning (Aslanidis & Jong, 1990;Eschenfeldt et al., 2009). Similarly, the protruded DNA fragment for each of the TrpB subunits was mixed with T4 DNA polymerase-treated vector pRSF with kanamycin resistance, which had an identical ligand-independent cloning site to pMCSG68. Both subunits from each genomic DNA were individually transformed into E. coli BL21-Gold (DE3) cells and grown in the presence of the corresponding antibiotic. A single colony of each transformant was picked, grown and induced with isopropyl -d-1-thiogalactopyranoside (IPTG). The cell lysate was analyzed to confirm a protein of the correct molecular weight. The solubility of the TrpA subunit was analyzed via small-scale Ni 2+ -affinity purification and overnight TEV protease cleavage. Once the DNA sequences of the TrpA and TrpB subunits had been verified, both subunit plasmids from each genomic DNA were co-transformed into E. coli BL21-Gold (DE3) cells in LB medium containing ampicillin (150 mg ml À1 ) and kanamycin (25 mg ml À1 ). Cotransformed colonies were analyzed using Ni 2+ -affinity purification, and overnight TEV protease cleavage was performed to verify that the complex was soluble and stable.

Expression of TrpAB and purification for crystallization
To express SpTrpAB and FtTrpAB, starter cultures were grown overnight at 37 C and 200 rev min À1 in LB medium with ampicillin (100 mg ml À1 ) and kanamycin (30 mg ml À1 ) research papers 652 Karolina Michalska et al. Tryptophan synthases from bacterial pathogens supplemented with 40 mM K 2 HPO 4 . The following morning, LB-PO 4 -glucose (2 g per litre) medium with antibiotics was inoculated with the overnight cultures. After reaching an OD 600 of 1.0 at 37 C, the SpTrpAB cultures were transferred to 4 C and, after 1 h, to 18 C. After a subsequent 15 min incubation, the cultures were induced with 0.5 mM IPTG and incubated at 18 C overnight to produce the native protein.
SeMet-labeled FtTrpAB and native SpTrpAB were purified using the procedure described previously (Kim et al., 2004). The harvested cells were thawed and 1 mg ml À1 lysozyme was added. This mixture was kept on ice for 20 min with gentle shaking and was then sonicated. The lysate was clarified by centrifugation at 36 000g for 1 h and filtered through a 0.45 mm membrane. The clarified lysate was applied onto a 5 ml nickel HisTrap HP column (GE Healthcare Life Sciences) and the His 6 -tagged protein was released with elution buffer (500 mM NaCl, 5% glycerol, 50 mM HEPES pH 8.0, 250 mM imidazole, 10 mM -mercaptoethanol). This was followed by a bufferexchange step using a customized desalting column (Sephadex G-25 Fine XK 26/20, GE Healthcare Life Sciences) equilibrated with buffer consisting of 20 mM Tris-HCl pH 7.5, 500 mM NaCl, 2 mM DTT. All of these steps were performed using an Ä KTAxpress system (GE Healthcare Life Sciences). The fusion tag was removed by treatment with recombinant His 7 -tagged Tobacco etch virus (TEV) protease. Nickelaffinity chromatography was used to remove the His 6 tag, uncut protein and His 7 -tagged TEV protease (Blommel & Fox, 2007). The SpTrpAB ortholog was subjected to an extra purification step via size-exclusion chromatography on a Superdex 200 HiLoad 26/60 column (GE Healthcare Life Sciences) in crystallization buffer (200 mM NaCl, 20 mM HEPES pH 8.0, 2 mM DTT). The FtTrpAB protein was dialyzed against crystallization buffer consisting of 250 mM NaCl, 20 mM HEPES pH 8.0, 2 mM dithiothreitol (DTT) and the proteins were then concentrated to 68 mg ml À1 (FtTrpAB) and 33.6 mg ml À1 (SpTrpAB) using an Amicon Ultra centrifugal filter device with a 10 000 molecular-weight cutoff (Millipore, Billerica, Massachusetts, USA), flash-cooled in liquid nitrogen and stored at À80 C.
The TrpAB protein concentration was determined spectrophotometrically by measuring the absorbance at 280 nm on a NanoDrop ND-1000 spectrophotometer (Thermo Scientific) against buffer containing an equimolar concentration of PLP. The concentration was calculated using extinction coefficients of 34185 and 39435 M À1 cm À1 , respectively, computed from the amino-acid sequence.

Expression of TrpA and purification for crystallization
An LB medium starter culture was supplemented with 40 mM K 2 HPO 4 and ampicillin (150 mg ml À1 ) for LpPhTrpA and LpPaTrpA, grown and shaken overnight at 37 C and 200 rev min À1 . The starter cultures were used to inoculate 1 l of enriched M9 medium for large-scale SeMet-labeled protein production, which was carried out as described above. From each litre of cell culture, 8 g of cell pellet containing SeMetlabeled LpPhTrpA or LpPaTrpA protein was obtained and was consequently resuspended in lysis buffer and stored at À80 C.
SeMet-labeled LpPhTrpA and LpPaTrpA were purified in the same manner as SeMet-labeled FtTrpAB. However, instead of dialyzing these proteins against crystallization buffer, they were buffer-exchanged using an Amicon Ultra centrifugal filter device with a 10 000 molecular-weight cutoff (Millipore, Billerica, Massachetts, USA) with 250 mM NaCl, 20 mM HEPES pH 8.0, 2 mM DTT, flash-cooled in liquid nitrogen and stored at À80 C. Protein concentrations were also determined with a NanoDrop ND-1000 using extinction coefficients of 24870 and 23505 M À1 cm À1 , respectively, computed from the amino-acid sequence.

Expression and purification for enzymatic assays
For each ortholog, a starter culture was grown overnight at 37 C and 200 rev min À1 in LB medium with ampicillin (100 mg ml À1 ) and kanamycin (30 mg ml À1 ) and supplemented with 40 mM K 2 HPO 4 . The following morning, 4 l LB-PO 4glucose (2 g per litre) medium with antibiotics was inoculated with 30 ml of the overnight culture and was grown at 37 C and 200 rev min À1 . After reaching an OD 600 of 1.0 the cultures were transferred to 4 C to cool, and after 1 h the temperature was increased to 18 C. After 15 min, protein expression was induced with 0.5 mM IPTG. The cells were incubated at 18 C overnight. The harvested cells containing TrpAB were resuspended in lysis buffer [500 mM NaCl, 5%(w/v) glycerol, 50 mM HEPES pH 8.0, 20 mM imidazole, 10 mM -mercaptoethanol, protease inhibitor (one tablet per 50 ml of extract), 1 mM PLP] and stored at À80 C. All three native proteins were purified using the procedure described above for FtTrpAB. The samples were concentrated to 40 mg ml À1 (LpPhTrpAB), 40 mg ml À1 (SpTrpAB) and 140 mg ml À1 (FtTrpAB), flash-cooled in liquid nitrogen in 35 ml droplets and subsequently used in enzymatic assays. MtTrpAB was purified as described previously (Wellington et al., 2017).

Crystallization
The FtTrpAB and SpTrpAB proteins were crystallized using sitting-drop vapor diffusion at 16 and 24 C, respectively, in a CrystalQuick 96-well round-bottom plate (Greiner Bio-One North America, Monroe, North Carolina, USA). A 400 nl droplet of the protein (35 or 34 mg ml À1 ) with 1 mM PLP and 1 mM l-Ser (FtTrpAB) or 0.5 mM PLP (SpTrpAB) was mixed with a 200 nl droplet and 400 nl crystallization reagent and allowed to equilibrate against 135 ml crystallization reagent. The nanopipetting was performed using a Mosquito nanolitre liquid-handling system (TTP Labtech, Cambridge, Massa-chusetts, USA). The plates were then incubated within a RoboIncubator automated plate-storage system (Rigaku). Automated crystal visualization (Minstrel III, Rigaku) was utilized to locate several crystals. The best crystals of SeMet-labeled FtTrpAB were obtained from 0.2 M calcium acetate, 0.1 M imidazole-HCl pH 8.0, 10%(w/v) PEG 8000. The SpTrpAB crystals grew from 0.2 M ammonium acetate, 0.1 M Tris-HCl pH 8.5, 25% PEG 3350.
LpPhTrpA (at 25 mg ml À1 ) and LpPaTrpA (at 62.5 mg ml À1 ) were screened in the same manner, but without the addition of extra ligands, using a droplet consisting of 400 nl protein solution and 400 nl crystallization reagent that was allowed to equilibrate over 135 ml of the respective reservoir condition. The proteins were screened against the MCSG 1-4 screens (Microlytic) and the Index screen (Hampton Research) at 16 C. The best crystals of SeMet-labeled LpPhTrpA were obtained from 0.01 M sodium citrate, 33%(w/v) PEG 6000. The SeMet-labeled LpPaTrpA crystals grew from 0.2 M sodium chloride, 0.1 M bis-Tris pH 6.5, 25%(w/v) PEG 3350.

Data collection
The crystals were cryoprotected in their respective mother liquors supplemented with 10% (SpTrpAB, LpPhTrpA and LpPaTrpA) or 25% (FtTrpAB) glycerol and were subsequently flash-cooled in liquid nitrogen. X-ray diffraction data were collected on the Structural Biology Center 19-ID beamline at the Advanced Photon Source, Argonne National Laboratory. The images were recorded on an ADSC Q315r detector. The data sets were processed with the HKL-3000 suite (Minor et al., 2006). Intensities were converted to structure-factor amplitudes in the CTRUNCATE program (French & Wilson, 1978;Padilla & Yeates, 2003) from the CCP4 package (Winn et al., 2011). The data-collection and processing statistics are given in Table 1.

Structure solution and refinement
The SpTrpAB structure was solved by molecular replacement in Phaser (McCoy, 2007) Karplus & Diederichs (2012). § R = P hkl jF obs j À jF calc j = P hkl jF obs j for all reflections, where F obs and F calc are the observed and calculated structure factors, respectively. R free is calculated analogously for the test reflections, which wre randomly selected and excluded from the refinement. } As defined by MolProbity (Chen et al., 2010).
Genomics of Infectious Diseases, unpublished work). The initial model was autobuilt in PHENIX (Adams et al., 2013) and was further improved by manual correction in Coot (Emsley & Cowtan, 2004) and crystallographic refinement in PHENIX (Afonine et al., 2012). The FtTrpAB, LpPhTrpA and LpPaTrpA structures were solved by the SAD method using selenium absorption peak data in SHARP (Vonrhein et al., 2007) or HKL-3000 (for LpTrpA; Minor et al., 2006) and were autobuilt in Buccaneer (Cowtan, 2006). The final model was obtained using alternating manual rebuilding in Coot and maximum-likelihood refinement in PHENIX (Afonine et al., 2012). The refinement statistics are given in Table 1.

Preparation of material for kinetic assays
Prior to kinetic and/or biophysical characterization, MtTrpAB was dialyzed for 2-4 h in TrpAB buffer (20 mM HEPES pH 8.0, 100 mM KCl, 1 mM TCEP, 40 mM PLP) to remove glycerol. After dialysis for 2-4 h, the buffer was exchanged with fresh buffer and dialysis continued overnight. The three other orthologs, however, were stored in 20 mM HEPES pH 8.0, 200 mM NaCl, 2 mM DTT buffer containing no glycerol after purification and did not require dialysis before use.
The compounds F9, F6 and IPP were custom-synthesized by GVK Bio (Cambridge, Massachusetts, USA). The MtTrpAB inhibitor BRD4592 was synthesized internally at the Broad Institute as described previously (Wellington et al., 2017).

Measurement of enzyme kinetics by UV absorption
Enzyme kinetics for each ortholog were determined over 30 min under saturating substrate conditions (200 mM indole and 60 mM l-Ser) in 1 ml TrpAB buffer. An Agilent Technologies Cary 400 Series UV-Vis spectrophotometer set to 290 nm was used for UV absorption measurements. A baseline reading with no enzyme was established, after which enzyme was added every 2 min to give a final concentration range from 50 nM to 2.4 mM. Product progress curves were determined at appropriate enzyme concentrations over a 10 min period in which product generation was linear to determine the K m and k cat parameters. A value of Á" = 1890 M À1 cm À1 was used for the indole to l-Trp conversion. In all cases, these enzymes were studied at room temperature (22 C). These experiments were performed on triplicate test occasions with triplicate replicates in each case.

LC-MS assay
For the liquid chromatography-mass spectrometry (LC-MS) assay, all reagents were prepared in a 96-well plate with a final reaction volume of 50 ml. Compound IC 50 reactions were run at substrate K m conditions (10 mM indole, 20 mM l-serine). Compound concentrations ranged from 0 to 200 mM. 10Â K m substrate solutions were prepared, with 5 ml additions of both indole and serine solutions to the wells. The final concentrations of each protein were as follows: 100 nM SpTrpAB, 5 nM FtTrpAB, 600 nM LpPhTrpAB and 100 nM MtTrpAB prepared in TrpAB buffer.
Standard curves for l-Trp and indole were included with each mass-spectrometry experiment for quantification purposes only. An l-Ser standard curve was also included as a biological check for each ortholog. Final l-Ser standard curve concentrations included 48, 24, 12, 6, 3, 1.5, 0.75 and 0 mM at saturating (500 mM) indole (5Â solution at 2.5 mM indole with 10 ml additions). After all compound, substrate and standard curve solutions had been prepared, 30 ml of a 1.67Â protein solution was added to each well to start the reaction.
After mixing and allowing 10 min incubation at room temperature, the reactions were quenched using 150 ml 0.1% formic acid in methanol followed by storage at 4 C for at least 2 h. The sample plates were then centrifuged for 15 min at 3900 rev min À1 ($3061g) and an aliquot of the supernatant was diluted 1:10 with water. 3.75 ml of this final solution was injected and analyzed. l-Trp and indole were detected by UPLC-MS (Waters, Milford, Massachusetts, USA). Compounds were quantified by selected ion recording (SIR) on an SQ mass spectrometer by negative electrospray ionization. The SIR method was set for l-Trp at 203.4 m/z and for indole at 116.3 m/z. Mobile phase A consisted of 0.1% ammonium hydroxide in water, while mobile phase B consisted of 0.1% ammonium hydroxide in acetonitrile. The gradient ran from 2% to 95% mobile phase B over 2.65 min at 0.9 ml min À1 . An Acquity BEH C18, 1.7 mm, 2.1 Â 50 mm column was used with the column temperature maintained at 65 C.

Data analysis
Kinetic experiments were run in triplicate and the reported values represent the average of at least three independent experiments. K m , k cat and IC 50 data were plotted using GraphPad Prism 7.0 and Origin 8.0.

Protein preparation
The recombinant tryptophan synthases from the pathogenic bacteria F. tularensis, S. pneumoniae and L. pneumophila Philadelphia have been produced for detailed characterization and comparison with the previously studied enzymes from S. typhimurium, E. coli and M. tuberculosis (Wellington et al., 2017). The level of pairwise sequence identity between the TrpBs from these organisms ranges from 51% to 59%, with the exception of the FtTrpB/StTrpB pair, which show 81% conserved residues. The TrpAs are more variable, with only 25-33% sequence identity for most pairs and 50% for the FtTrpA/StTrpA pair ( LpPhTrpAB) or TrpB (SpTrpAB) subunits were equipped with an N-terminal His 6 tag, which was subsequently removed by treatment with TEV protease. The resulting proteins carry an additional three N-terminal residues SNA on the tagged subunit. In addition to TrpABs, TrpAs from the L. pneumophila strains Paris and Philadelphia (LpPaTrpA and LpPhTrpA, respectively; 99% identical) have been produced for crystallographic studies, also with a removable N-terminal His 6 tag. FtTrpAB and LpPhTrpA were produced as SeMetlabeled derivatives, while all other proteins were expressed in the native form. The purified proteins were at least 90% pure as judged by PAGE.

Structure determination
The SpTrpAB protein was crystallized in space group P2 1 with the entire heterotetramer present in the asymmetric unit (Fig. 1, Table 1). The structure, which was determined at 2.45 Å resolution, was solved by molecular replacement. In chains A and C, corresponding to TrpA (amino-acid residues 1-258), residues 1, 180-189 and 182-187, respectively, were not modeled owing to a lack of interpretable electron density. Similarly, in TrpB (amino-acid residues 4-407) the N-terminal SNA sequence and the C-terminal end (residues 403-407) are not present in the respective chains B and D. The other ortholog, FtTrpAB, crystallized in space group C222 1 and the asymmetric unit contains only one module. This structure was solved by experimental SAD phasing and was refined to 2.80 Å resolution. In FtTrpAB, TrpA (chain A; residues 1-269) lacks the N-terminal SNA sequence and residues 183-191, while in TrpB (chain B; residues 1-396) the C-terminal residue is not present. For L. pneumophila only the TrpA subunit could be crystallized. The LpPaTrpA and LpPhTrpA structures were determined by experimental SAD phasing at 1.91 and 2.02 Å resolution, respectively. The LpPhTrpA protein crystallized in the orthorhombic space group P2 1 2 1 2 1 . The asymmetric unit contains one molecule of TrpA and the model lacks the N-terminal SNA residues, residues 57-59, residues 180-186 and the C-terminal residue 272. LpPaTrpA also crystallized in space group P2 1 2 1 2 1 with one chain in the asymmetric unit. The N-terminal SN residues and residues 180-187 and 270-273 are missing from the final model.

Kinetic characterization
Simultaneously with structural characterization, we performed kinetic analyses of the three new orthologs (FtTrpAB, SpTrpAB and LpPhTrpAB) and compared them with the MtTrpAB reference. A UV-based assay was used to measure the production of l-Trp from indole and l-Ser. Firstly, the enzyme concentration versus catalytic rate relationship was determined to identify the linear rate dependencies. Both the SpTrpAB and FtTrpAB enzymes displayed specific activities that were comparable to (SpTrpAB, 1.4 M l-Trp s À1 M À1 enzyme) or higher (FtTrpAB, 26 M l-Trp s À1 M À1 enzyme) than that of MtTrpAB (2.0 M l-Trp s À1 M À1 enzyme), with the rate being linearly dependent on enzyme concentration over the entire tested range. The LpPhTrpAB enzyme, however, was less active than the MtTrpAB enzyme, displaying a biphasic dependency with both components appearing to be linear. The specific activity at low enzyme concentrations (50-800 nM) was much lower (0.17 M l-Trp s À1 M À1 enzyme), while the higher concentration range (1000-2400 nM) displayed an improved but still significantly lower specific activity (0.38 M l-Trp s À1 M À1 enzyme) (Fig. 4). The source of this higher order effect is not obvious, but could be explained by the equilibrium between subunits and dimers and tetramers, with higher protein concentrations favoring the more active oligomeric state. We have observed such an equilibrium for the MtTrpAB enzyme (Wellington et al., 2017). The specific activity order is as follows: FtTrpAB >> MtTrpAB, SpTrpAB >> LpPhTrpAB.
These data were used to set the appropriate enzyme concentrations (5 nM FtTrpAB, 100 nM MtTrpAB, 100 nM SpTrpAB and 600 nM LpPhTrpAB), resulting in linear l-Trp production progress curves over a 10 min reaction period, to determine the apparent K m and k cat parameters using the LC-MS assay. The apparent K m values are similar across all of the species for both substrates tested (indole and l-Ser). The k cat values were reproducible across experiment replicates and substrates, suggesting that saturation was achieved for the independent substrate in each case. The absolute k cat values were consistent with the specific activities described above, following the activity order FtTrpAB >> MtTrpAB, SpTrpAB >> LpPhTrpAB (Fig. 5).

Inhibition studies
In addition, the three TrpAB orthologs were profiled against the reported commercially available inhibitors  Table 2 Primary structure identity and structural similarity between orthologous TrpA and TrpB.
The first number corresponds to the percentage sequence identity (calculated in EMBOSS Needle; Rice et al., 2000), followed by r.m.s.d. (in Å ) for C -atom superposition for the number of pairs given in parentheses (calculated in CCP4; Winn et al., 2011, Krissinel & Henrick, 2004 and IPP (indolepropanol phosphate; CID identifier 3713), as well as the recently discovered MtTrpAB inhibitor BRD4592 (CID identifier 54650477; Wellington et al., 2017) (Fig. 6). The LC-MS-based assays examined inhibition of the reaction with indole and l-Ser as substrates. F9 was found to be a potent inhibitor (IC 50 = 114 nM) of FtTrpAB under substrate K m conditions (10 mM indole, 20 mM l-Ser), while only slightly inhibiting LpPhTrpAB. Interestingly, F9 appears to be an activator of SpTrpAB (Fig. 6)    observed for the FtTrpAB enzyme, with IC 50 = 1.46 mM for F6 and IC 50 = 0.08 mM for IPP. A different profile was seen when using the MtTrpAB inhibitor BRD4592. All three orthologs are slightly inhibited; however, a measureable IC 50 was only obtained for the SpTrpAB ortholog (IC 50 = 21 mM) (Fig. 6).

Structural comparison with other TrpAB orthologs
We have determined the structures of the FtTrpAB and SpTrpAB heterotetramers and of the subunits LpPaTrpA and LpPhTrpA. The overall structures of the complexes, along with the subunits, are essentially identical to those of the orthologs characterized previously, with the heterotetramer representing the complete functional unit (Fig. 1). Despite the rather low sequence identity of the TrpAs, the three polypeptides superpose with r.m.s.d.s of 1.4-1.9 Å amongst themselves and with the orthologs MtTrpA or StTrpA (Table 2, Fig. 7). The enzyme from F. tularensis, which is the most closely related to StTrpAB, shows even better agreement, with an r.m.s.d. of 0.8 Å for corresponding StTrpA C atoms. A similar pattern is observed for the TrpBs, which overlap with r.m.s.d.s of 0.7-1.0 Å .
As expected in the absence of any TrpA ligand, the subunit adopts an open conformation with a disordered loop L6, regardless of whether the subunit is complexed with TrpB or alone. In isolated LpPhTrpA parts of loop L2 could not be modeled, indicating its high flexibility. The TrpA binding pocket and these critical loops are generally well conserved in terms of composition, including the catalytic residues, one of which is provided by loop L2. One important feature, although only noted at the sequence level owing to disorder, is the lack of conservation in the N-terminal region of loop L6. In the Salmonella enzyme this section carries Arg179, which has been shown to provide loop stabilization via hydrogen bonds between the guanidinium group and the main-chain atoms (Schneider et al., 1998). With the exception of FtTrpA, this residue is replaced by much smaller and in some cases hydrophobic residues, Ile in SpTrpA, Leu in LpTrpA and Thr in MtTrpA, and cannot form interactions equivalent to those of Arg179. It has previously been shown that an Arg179Leu mutation reduces the affinity of the substrate IGP for StTrpA and slows the TrpAB reaction (Brzović et al., 1993). It is not clear that this is a valid assumption for the other orthologs; however, MtTrpAB indeed has a higher K m for IGP than StTrpA. In addition, it is also consistent with the relative rank order of specific activities observed across this panel of TrpAB orthologs, although only in the context of the reaction.
Within the ordered fragments of the TrpA pocket, some sequence variability is observed at the positions of Pro129Sp (the equivalent residues are Pro135Mt, Ala130Ft, Ala129St and Val129Lp), Met100Sp (Met100Lp and Met106Mt but Leu101Ft and Leu100St) and Tyr23Sp (replaced by Phe in FtTrpA, LpTrpA and StTrpA). Notably, though, despite the good superposition of the main-chain atoms throughout most of the subunit, the side chains adopt slightly different conformations (Fig. 7). The most pronounced discrepancy is observed for Phe212Sp, a residue that T-stacks against the aromatic ring of indole in the ligand-bound StTrpA state (Weyand & Schlichting, 1999). The position of this residue is affected by the mobile L6 loop in the substratebound closed state that reinforces the proper placement of the Phe side chain with respect to the substrate moiety. Without such constraints, in SpTrpA, as well as in LpTrpA, it points somewhat outside of the binding pocket towards the helical layer of the protein. In FtTrpA it is oriented more towards the cavity, but its position is still halfway from the state achieved in the substrate-bound complex (Fig. 7). Interestingly, this residue is replaced by Leu218 in the MtTrpA ortholog, where it also swings outside the binding pocket. The catalytic Glu52Sp and its equivalents in other orthologs also display some conformational diversity; in some cases, such as FtTrpA or StTrpA, it points towards the protein core, while in others (SpTrpA and MtTrpA) it faces the binding pocket. There are no apparent structural differences between TrpA in the TrpAB complex versus TrpA alone. The only exception is a slight movement of loop L2 towards the active site of TrpA in the heterodimer unit.
In our FtTrpAB and StTrpAB structures the subunits exist in the open conformation, or more precisely in the expanded open conformation eO reported previously for several StTrpAB structures [PDB entries 2j9z , 1qoq (Weyand & Schlichting, 1999) and 1kfb (Kulik et al., 2002)], the P. furiosus ortholog [PDB entries 5e0k (Buller et al., 2015) and 1wdw (Lee et al., 2005)] and MtTrpAB (PDB entry 5tcf; Wellington et al., 2017), suggesting that this state may be more common than previously indicated. The active site carries a PLP moiety covalently attached to Lys91Sp (Lys86Ft, Lys101Mt). The active site is very conserved both in terms of sequence and the conformation of the PLP cofactor and side chains, with a few exceptions. FtTrpB and SpTrpB share an Ala with StTrpB (Ala84, Ala89 and Ala85, respectively), but MtTrpB has an equivalent Ser99 that makes a direct hydrogen bond to PLP. This interaction is missing in the other three orthologs. Thr87 is present in SpTrpB (and Thr97 in MtTrpB), which is replaced by glycine in FtTrpB and StTrpB. There is no obvious role for this substitution. Two important catalytic residues, threonine (Thr114Sp, Thr109Ft, Thr124Mt and Thr110St) and aspartic acid (Asp310Sp, Asp304Ft, Asp319Mt and Asp305St), show a very different conformational behavior in the open state of -subunit orthologs. The threonine, which is involved in coordination of the substrate/product carboxylate, shows nearly the same conformation in all four orthologs, while the conformations of the aspartic acid, which is involved in interaction with the amino group of the reagents, are very different. Larger conformational diversity is also observed for Gln118, a residue that is conserved in all four enzymes. However, only in MtTrpB does this residue form a direct hydrogen bond to O3 of the PLP cofactor. The side chains of a few other residues (Gln89, Ser234 and Lys381 in FtTrpB) also show somewhat different conformations, but these are much less pronounced. The phosphate group of PLP is Comparison of TrpAB orthologs. (a) Superposition of SpTrpB (yellow, TrpA, chain C; coral/cyan, TrpB, chain D) with FtTrpB (blue), MtTrpB (purple; chains A and B; PDB entry 5tcf; Wellington et al., 2017) and StTrpB (gray; PDB entry 1bks; Rhee et al., 1996). PLP from SpTrpAB is shown in a sphere representation. TrpA is shown to indicate the mutual orientation of the subunits. (b) Superposition of TrpA extracted from the TrpAB heterodimers. (c) Stereoview of the TrpA active-site superposition of SpTrpA (yellow), FtTrpA (blue) and StTrpA in complex with IPP (gray; PDB entry 1qop; Weyand & Schlichting, 1999). anchored by interaction with the N-terminal dipole of helix H9, direct hydrogen bonds to several main-chain amino groups (helix H9 and a short loop between S7 and H9) and three conserved side chains (His85, Ser234 and Asn235 in FtTrpB and His90, Ser240 and Asn241 in SpTrpB). These small changes in sequence and conformational propensity may explain the differences in substrate affinities and reaction rates.
The structures of the FtTrpAB and SpTrpAB heterotetramers provide a new set of high-quality models and enable comparison of the intermolecular tunnel connecting the TrpA and TrpB catalytic pockets. In contrast to the active sites, the composition of the tunnel, which is mostly encompassed by TrpB, varies between the orthologs (Fig. 8), although generally SpTrpAB shares some features with MtTrpAB while FtTrpAB is similar to StTrpAB. This is consistent with the relative specific activities and the conservation of local primary sequence. The cross-comparisons indicate a number of differences. For example, one side of the SpTrpB tunnel contains Tyr311, His285 and the neighboring Leu284, with the tyrosine rotated towards the active site of TrpB, where it could potentially interfere with the reaction. The opposite side contributes Val174, Leu178 and Leu192.
In FtTrpB all of the former residues are replaced by phenylalanines (Phe305, Phe279 and Phe278, respectively), while the leucines are conserved and Val174Sp is replaced by Cys169Ft. A similar scenario is present in StTrpB (Phe306 and Phe280), with the exception of Tyr279St, which substitutes for Phe278Ft. In MtTrpB the equivalent residues are Tyr320, His294 and Phe293, resembling the SpTrpB composition, but in this case the tyrosine ring points in a different direction, making a hydrogen bond to His294. Such an arrangement would be more constrained in SpTrpB owing to the proximity of Leu196, a residue that is substituted by a much smaller Ala in the other enzymes. MtTrpB also contains phenylalanines (Phe188 and Phe202) instead of the leucines that are conserved in the three other TrpBs, and Ile184Mt takes the place of Val174Sp. Previous data for the StTrpB ortholog showed that large side chains, such as Phe or Trp, in this position hamper indole channeling (Anderson et al., 1995;Schlichting et al., 1994;Weyand & Schlichting, 2000). Therefore, it appears that these variations in the residues composing the tunnel may have a direct impact on the rate of indole transfer and influence the kinetic activities of these enzymes. This may represent a fine-tuning of the enzyme activity without directly involving the residues in the catalytic sites.  Generally, the tunnel displays some level of flexibility and can adapt to enable indole translocation or to specifically bind certain inhibitors. For instance, we showed previously that in MtTrpAB Phe188 changes conformation to accommodate BRD4592 (Wellington et al., 2017) both in the open and closed states of the subunit, while in StTrpAB Phe280 and Tyr279 swing away to provide space for the F6 molecule (Hilario et al., 2016) in the open state (Fig. 8). The latter work also proposed that the indole moiety enters TrpB in the vicinity of Leu21St (conserved as Leu24Sp, Leu34Mt and Leu20Ft), Leu174St and Phe280St, which need to move to open up a farther segment of the channel that is lined with residues that do not present major obvious obstacles. In principle, an analogous mechanism can be envisioned for the very similar enzyme from F. tularensis. In the other two orthologs alternative mechanisms are most likely to exist. In the SpTrpB/MtTrpB structures, in which Phe280St is replaced by a histidine, this residue adopts a conformation that is compatible with an open channel both in the O (SpTrpB/ MtTrpB) and C (MtTrpB) states. Moreover, in MtTrpB such an architecture is stabilized by a hydrogen bond to Tyr320Mt (in O and C ) and another to Asn185Mt (in Mt C ), suggesting that it represents the most common conformational state. An analogous interaction with asparagine might be created in SpTrpB upon subunit closure, while His-Tyr bonding would require the concomitant movement of Tyr311Sp and Leu196Sp. This coordinated movement is potentially a necessary step for the COMM-domain shift and TrpB closure, as otherwise Leu170Sp would clash with Tyr311Sp. On the other hand, the mycobacterial enzyme may need to undergo a different adjustment on the opposite side of the tunnel. Here, there are two bulkier phenylalanine residues, Phe188 and Phe202. In both cases these residues appear to be mobile, as in some structures of MtTrpAB Phe202 exists in double conformations while Phe188 has been shown to rotate in the complex with the BRD4592 inhibitor. However, for Phe188 in this alternative state the access from subunit is blocked; thus, it is possible that the ligand-free conformation of Phe188 corresponds to the open-tunnel state with only a minor adjustment required.

Allosteric contacts
Previous investigations of allosteric communication between the TrpAB subunits recognized a number of key interactions at theinterface that transmit activation signals. One of them is the main-chain-main-chain hydrogen bond between Ser178 and Gly181 in StTrpAB (Spyrakis et al., 2006;Schneider et al., 1998). The former residue is preserved in FtTrpB; however, the other two orthologs contain valine. On the other hand, the glycine residue (Gly181Sp, Gly182Ft and Gly187Mt) belongs to the highly conserved GVTG motif of the L6 loop. In the S. typhimurium TrpA C state the conserved threonine residue from this motif, Thr183, binds through its hydroxyl group to the carboxylate of the catalytic Asp60 (Asp61Sp, Asp63Ft and Asp68Mt), in addition to the main-chain-main-chain inter-action with the L2 loop. Deletions or point mutations within the L6 loop, such as Thr183Ala in StTrpA, dramatically reduce the -subunit activity (Yang & Miles, 1992). Similar modifications in the L2 loop, including changes to Pro57St (Pro60Sp, Pro58Ft and Pro65Mt) and Asp56St (Asp59Sp, Asp57Ft and Asp64Mt) reduce TrpA activity, although significant effects only occur in the context of the TrpAB complex, i.e. not when the subunit alone is assayed (Ogasahara et al., 1992;Rowlett et al., 1998).
In the available O and C states of the mycobacterial enzyme, the side chain of Asp64Mt (the main chain of Ser63 in eO ) interacts with Lys181 from the COMM domain, while the carbonyl group of Asp68 binds to Arg189 in some of the subunits, as seen before in the StTrpA ortholog (Weyand & Schlichting, 1999). In the SpTrpAB eO state there is also a hydrogen bond between the Ser58 carbonyl group and Lys171, but Arg179 is too distant to interact with the catalytic aspartate. None of these contacts is observed in the reported FtTrpAB structure, either owing to disorder or to longer distances between the relevant atoms.
Overall, the available data suggest that the geometry and contacts established by loops L6 and L2 have a pronounced effect on the enzyme activity. Transition from O to C triggers the closure of L6, which, together with the L2 and H6 elements, activates the catalytic aspartate residue. Changes in these elements or in their neighborhood possibly lock L6 into a low-activity open state (Spyrakis et al., 2006), thus preventing the proper positioning of the catalytic aspartic acid. Simultaneously with the -subunit malfunction, destabilization of the L2-H6 interactions in mutants reduces the -subunit activity (Ogasahara et al., 1992), with the detrimental effect partly alleviated by cation binding. Monovalent cations have been shown to stabilize the StTrpAB enzyme, with large cations (Cs + and NH 4 + ) exhibiting the most pronounced effect (Rowlett et al., 1998). These effects might result from the chain of interactions linking L2 to H6 and further, via the monovalent cation-binding site (MVC), to the active site of the subunit. The MVC is established by a set of residues localized in the proximity of the channel and the active site of TrpB, which interact with the cation through four main-chain carbonyl moieties (in S. typhimurium and M. tuberculosis) and a threonine side chain (only in M. tuberculosis owing to the presence of Pro in the equivalent position in StTrpB). While no monovalent cations have been modeled in the current structures, by analogy to the data collected from the MtTrpAB and StTrpAB systems the MVC must be created by Tyr311Sp, Gly313Sp, Ala273Sp, Gly237Sp and Thr275Sp in SpTrpAB and Phe305Ft, Ser307Ft, Gly267Ft, Gly231Ft in FtTrpAB, with Pro269Ft replacing the threonine residue. Depending on the size of the cation, either all residues equivalent to those in StTrpAB and MtTrpAB would be involved in cation binding, or only a subset, where the unfilled valencies in the coordination sphere may be completed by water molecules. As mentioned above, the MVC is indirectly connected to the H6 element of the COMM domain and to TrpA via either a histidine (His285Sp and His294Mt) or a phenylalanine (Phe279Ft, Phe280St), research papers IUCrJ (2019). 6, 649-664 switching between hydrophobic Phe-Phe contacts (FtTrpB and StTrpB) and the well defined His-Tyr hydrogen bond seen in MtTrpB and most likely to be present in the activated form of SpTrpB. It is not clear how this different organization of the MVC and its interactions with other structural elements affect the sensitivity of the protein to different cations or how the signal transduction is affected.

Enzymatic properties
In the -elimination reaction of TrpAB, with a k cat of between 1.7 and 78.6 min À1 , all of the investigated enzymes appear to be poorer catalysts of the indole-to-tryptophan conversion than the previously studied MtTrpAB (k cat = 197 min À1 ; Wellington et al., 2017 1 ), EcTrpAB (348 min À1 ; Lane & Kirschner, 1983) and StTrpAB (288 min À1 ; Raboni et al., 2007), at least under the given experimental conditions: at room temperature (20-22 C) at pH 7.6-8.0 in the presence of potassium ion. Similarly, the K m for serine is at least 35 times higher for the SpTrpAB and FtTrpAB enzymes (18.3-43.2 mM) than for those previously characterized (0.37, 4.4 and 0.58 mM for EcTrpAB, MtTrpAB and StTrpAB, respectively). Interestingly, however, the K m for indole is at least approximately three times lower for all of the currently tested orthologs than those reported for MtTrpAB and StTrpAB and is comparable to that of EcTrpAB.

Inhibition
Several inhibitors have been designed to study the mechanistic details of TrpAB. A number of them are competitive indole-3-glycerol phosphate analogs that bind to subunit , such as IPP and similar indole-3-alkyl 1-phosphates (Kirschner et al., 1975), indole-3-acetyl amino acids (Marabotti et al., 2000) or aryl compounds linked via an amide/ sulfonamide/thioether/thiourea to a phosphoalkyl moiety Sachpatzidis et al., 1999). The IC 50 parameters for these inhibitors against TrpAB have not been determined, with the exception of thioether-linked substrate analogs (Sachpatzidis et al., 1999), which showed nanomolar values for the reaction of StTrpAB. In addition to competitive inhibition of the reaction, some of the -binders, for example indole-3-acetyl-amino acids, IPP and F9, exert allosteric effects on subunit (Marabotti et al., 2000;. The more promiscuous ligand F6 has been found to bind not only to the active site of TrpA but also to the intersubunit tunnel, close to the active site (Hilario et al., 2016). The influence of competitive inhibitors of TrpA on the TrpB reaction has been linked to their ability to remodel the site, with the higher degree of ordered TrpA structure triggering more pronounced changes in TrpB .
Here, we have tested the commercially available compounds IPP, F6 and F9 against the reaction. Notably, we observed potent inhibition only for FtTrpAB, which is the most similar to the prototypical StTrpAB of all the tested enzymes. It therefore seems that the allosteric effect influencing the activity of TrpB is sensitive to local sequence variations and structural features, and consequently might be unique to a subset of orthologs. Alternatively, it is also possible that the lack of TrpB susceptibility originates directly from the poor affinity of these inhibitors for TrpA, but we have not investigated such a scenario biochemically. From a structural perspective, the TrpA active sites are similar enough to at least bind to the very close substrate mimetic IPP, suggesting that the former argument for the lack of inhibition is more likely. Another explanation of these differences involves long-distance effects within and between subunits. The activation of SpTrpAB by the -binders is unexpected and surprising. However, allosteric sites serve modulatory purposes and a single binding pocket may exert activatory or inhibitory roles. It is therefore possible that the binding of the same ligands to various TrpAB orthologs may result in opposite kinetic effects because of small sequence variations.
In agreement with our previous work demonstrating that BRD4592 inhibition is limited to orthologs containing a glycine residue in the L2 loop of TrpAB, such as in the case of the MtTrpAB enzyme, no significant effect was observed for all of the tested synthases. The weak inhibition of SpTrpAB, which carries the smallest side chain among the tested enzymes (Val61 in place of Gly66 in MtTrpAB, Leu59 in FtTrpAB and Met58 in LpPhTrpAB), supports the previous conclusion that any substitution in the loop would drastically reduce the size of the BRD4592 binding pocket, limiting the inhibitor affinity.

Conclusions
Tryptophan synthases have been shown to be conditionally essential enzymes in a number of important human pathogens, but the enzymes of the family have remained unexplored beyond a limited number of representatives. To broaden our perspective on TrpABs, we have purified and characterized three enzymes from L. pneumophila, F. tularensis and S. pneumoniae to uncover the potential unique features of TrpABs and to support future drug-discovery efforts. X-ray crystallography and biochemical studies show a remarkable structural conservation of the architecture and the catalytic and allosteric sites of the enzyme, suggesting preservation of the catalytic mechanism and regulation. At the same time, these enzymes display local sequence and structural differences in the catalytic, allosteric and metal-binding sites. These enzymes also exhibit differences in kinetic properties and their response to inhibitors, yet they display some correlations between biochemical properties and sequence/structural conservation. Notably, not all enzymes were inhibited by the tested compounds. In fact, for the S. pneumoniae ortholog the reaction was more efficient in the presence of -binders. Some of the differences can be explained structurally; however, others may result from the altered conditions in which these enzymes operate in cellulo. Nevertheless, understanding these dissimilarities may provide a basis for the design of new species- specific tryptophan synthase inhibitors against both the and active sites as well as the allosteric sites, which show higher conformational and sequence variability. Recognition that the targeting of unique allosteric sites may have species-specific effects may be important for the treatment of coexisting infections.