Allosteric inhibitors of Mycobacterium tuberculosis tryptophan synthase

Abstract

Global dispersion of multidrug resistant bacteria is very common and evolution of antibiotic‐resistance is occurring at an alarming rate, presenting a formidable challenge for humanity. The development of new therapeuthics with novel molecular targets is urgently needed. Current drugs primarily affect protein, nucleic acid, and cell wall synthesis. Metabolic pathways, including those involved in amino acid biosynthesis, have recently sparked interest in the drug discovery community as potential reservoirs of such novel targets. Tryptophan biosynthesis, utilized by bacteria but absent in humans, represents one of the currently studied processes with a therapeutic focus. It has been shown that tryptophan synthase (TrpAB) is required for survival of Mycobacterium tuberculosis in macrophages and for evading host defense, and therefore is a promising drug target. Here we present crystal structures of TrpAB with two allosteric inhibitors of M. tuberculosis tryptophan synthase that belong to sulfolane and indole‐5‐sulfonamide chemical scaffolds. We compare our results with previously reported structural and biochemical studies of another, azetidine‐containing M. tuberculosis tryptophan synthase inhibitor. This work shows how structurally distinct ligands can occupy the same allosteric site and make specific interactions. It also highlights the potential benefit of targeting more variable allosteric sites of important metabolic enzymes.

1. INTRODUCTION

Mycobacterium tuberculosis (Mtb), the etiological agent of the tuberculosis (TB) disease, is the deadliest pathogen worldwide. The World Health Organization projected that globally, in 2017, ~10 million people developed TB, which resulted in the death of ~1.3 million among HIV‐negative people and an additional 300,000 among HIV‐positive people.1 TB targets men, women and children predominantly in poor countries as only 6% of all cases were reported in Europe and the Americas. It is estimated that 1.7 billion of the world's population have a latent TB infection and are at risk of developing active TB disease. The existing treatment for uncomplicated TB is 6–9 months long and involves administering rifampicin (RIF), the most effective first‐line drug against TB, in combination with isoniazid (INH), pyrazinamide and ethambutol. However, resistance to first‐line agents, namely RIF and INH is becoming a major issue. In 2017 there were 558,000 cases reported of RIF‐resistant TB (RR‐TB), and of these, 458,000 were multi‐drug resistant TB (resistant to both INH and RIF). Cases of extensively drug resistant TB (XDR‐TB), or multiple drug resistant TB (MDR‐TB) that is also resistant to fluroquinolones and at least one second‐line injectable, are also on the rise. Discovery of new therapeutic measures, especially those that involve new drug targets or those with novel mechanism of action, are critical to subvert existing clinical drug resistance, and hold the potential to shorten TB treatment duration in humans.2

One promising avenue lies in the pathway for L‐Trp biosynthesis. Studies of survival of M. tuberculosis in macrophage and mouse infection models showed that anthranilate synthase component I, TrpE,3 as well as functional Mtb tryptophan synthase (MtTrpAB)4 are essential under conditions when an adaptive immune response triggers expression of host indoleamine 2,3‐dioxygenase (IDO‐1)—an enzyme responsible for L‐Trp breakdown—or possibly even before this defense mechanism is executed.4, 5 The essentiality of tryptophan synthase was also established in the M. marinum zebrafish embryo model and M. tuberculosis acute mouse model (C57BL/6J mice).4, 6 Moreover, L‐Trp biosynthetic pathways have been shown to be important for survival of other bacteria.7, 8, 9 It is now evident that for some obligate and opportunistic pathogens the availability of L‐Trp, either supplied by the environment or synthesized in cellulo, is a prerequisite for survival and sustained infection. Some species rely on external L‐Trp sources and maintain either no or only limited functionality of the tryptophan operon, while others preserve the complete pathway for de novo L‐Trp biosynthesis.7, 8

In the light of these discoveries, the L‐Trp biosynthetic pathway, absent in animals and humans, has become an attractive drug target in bacterial diseases, even though the involved enzymes are only essential under certain conditions – that is, when exogenous L‐Trp is limited. Tryptophan synthase in particular has emerged as an important drug target for the treatment of TB. The TrpAB bifunctional enzyme catalyzes the final two steps of tryptophan biosynthesis in bacteria, fungi and plants and uses pyridoxal 5′‐phosphate10 (PLP) as a cofactor.11, 12, 13, 14, 15, 16 It is composed of two protein chains, α17 and β18 and forms a linear heterotetrameric αββα complex. Enzyme minimal functional unit19 contains two active sites connected via 25 Å long channel.12 Structurally, TrpA adopts a canonical (β/α)₈‐barrel fold (TIM barrel) with several additional secondary structure elements, whereas TrpB consists of two three‐layer (αβα) sandwich domains.20

The active site of TrpA is located at the top of the central β‐barrel, with two acidic residues involved in catalysis. Another structural element, loop αL6, serves as a lid closing over the binding pocket. TrpA converts indole‐3‐glycerol phosphate (IGP) into glyceraldehyde‐3‐phosphate (G3P) and indole. Indole then travels across the α/β interface to the β active site. TrpB catalyzes PLP‐dependent β‐replacement reaction in which indole displaces the hydroxyl group of L‐Ser to produce L‐Trp. The TrpB active site is located in a cleft and carries the covalently attached PLP cofactor. The N‐terminal domain encompasses the communication domain (COMM) that plays a key role in coordinating activity of the two active sites.21 The multistep reaction mechanism involves enzyme‐cofactor and substrate covalent adducts.

The enzyme is allosterically regulated by alternating the α‐ and β‐subunits between open (low activity) and closed (high activity) conformations.22 In open conformations, active sites are freely accessible to substrates, and in closed states, sites are solvent inaccessible, whereas the tunnel connecting the α and β sites is open. This switching prevents the escape of the intermediate, indole, produced by the α subunit. This allows the substrate channeling from the α to β active site and triggers ligand‐mediated allosteric signaling.23 This mechanism was first proposed based on the structures of complexes with α‐aminoacrylate and quinonoid moieties together with F6 and F9 inhibitors. These structures show the α and β subunits in closed conformations with blocked access into the α and β sites from solution.24, 25 The interplay between open and closed conformations is critically important as it synchronizes the α‐ and β‐catalytic cycles. The regulation ensures that resources are efficiently utilized to produce L‐Trp—a scarce and most energetically expensive amino acid to bio‐synthesize.26 The chemical transformations are highly controlled by allosteric effects and other factors, for instance the binding of monovalent cations to TrpB is linked to substrate channeling.27, 28 Formation of the α‐aminoacrylate Schiff base intermediate (E_AA) from L‐Ser and PLP in TrpB triggers movement of the TrpB COMM domain towards a closed state (β^C), which subsequently activates TrpA by closure of the αL6 loop (α^C). In a reciprocal process, IGP substrate binding to TrpA promotes the α^C state, which in turn, activates TrpB (β^C). The two protein chains convert back to open states when the L‐Trp external aldimine, E_A,ex2, is produced. Through the years of study, the tryptophan synthase has become a prototype system to explore idiosyncrasies of allosteric regulation and substrate channeling.14, 23, 24, 25, 27, 29, 30

Tryptophan synthase was studied extensively biochemically, structurally and computationally for the past 60 years, but most research focused on a model system from Salmonella typhimurium and orthologs from Escherichia coli and Pyrococcus furiosus. Only recently the field has expanded to include TrpAB from other organisms.4, 19, 31 In M. tuberculosis TrpAB research the major breakthrough involved developing recombinant coexpression system in E. coli, purification of fully active αββα complex and obtaining crystals that diffracted to high resolution and permitted detailed visualization of enzyme inhibitor interactions and insight into the catalytic mechanism.4 In general, the mechanism of L‐Trp synthesis is preserved in all known tryptophan synthases (TrpAB) and majority of features are conserved (the α site, the β site, the tunnel connecting active sites and allosteric regulation). At the same time, some differences were reported between TrpAB enzymes suggesting that they can be exploited to design species‐specific inhibitors.31 This was specifically proposed for the allosteric sites. We have previously reported an inhibitor BRD4592 that selectively targets MtTrpAB and binds to a new allosteric site with high affinity.4 Another group discovered and partly characterized compounds belonging to two distinct chemical classes. The latter study, however, was lacking in detailed structural information due to low resolution of the data (4 Å), limiting its value in structure‐guided drug discovery. Here we present crystal structures of MtTrpAB with two inhibitors from that research, termed GSK1 and GSK2, at 2.4 Å resolution. As BRD4592, these compounds are strong inhibitors.4, 6 The MICs for GSK1 and GSK2 were determined in M. tuberculosis H37Rv to be 0.76 and 1.1 μM, respectively,6 and for BRD4592 in various M. tuberculosis strains they are in the range of 1.6–3 μM.4 This work will help to advance medicinal chemistry efforts.

2. RESULTS

2.1. Crystallization of MtTrpAB inhibitor complexes

M. tuberculosis TrpA (α subunit) and TrpB (β subunit) were coexpressed from individual vectors in E. coli and recombinant proteins were purified and crystallized as described previously.4 Crystals of complexes were produced by soaking TrpAB with two previously published in vivo and in vitro inhibitors of MtTrpAB (GSK1 and GSK2).6 GSK1 (3R,4R)‐4‐[4‐(2‐chlorophenyl)piperazin‐1‐yl]‐1,1‐dioxothiolan‐3‐ol) and GSK2 (1‐[2‐fluorobenzoyl]‐N‐methyl‐2,3‐dihydro‐1H‐indole‐5‐sulfonamide) were discovered by a collaborative team from University of Birmingham and GlaxoSmithKline (GSK) through an in vitro phenotypic screen that aimed to identify new compounds with anti‐tubercular activity.6 GSK1 (3R,4R and 3S,4S racemate) and GSK2 are commercially available, but their interaction with TrpAB has not been characterized structurally, partly due to limited resolution of the diffraction data (available only for GSK2 complex).

2.2. Structures of complexes with allosteric inhibitors

The orthorhombic crystal form (P2₁2₁2₁) of MtTrpAB was used in our experiments to determine structures of complexes with GSK1 and GSK2. This crystal contains two complete αββα heterotetramers in the asymmetric unit and therefore it has four independent copies of αβ minimal catalytic unit. The structures were refined to 2.4 Å resolution. The electron density for ligands is of good quality and it allows reliable interpretation of compounds conformation and contacts in the binding pocket. For GSK1 we modeled 3R,4R stereoisomer that shows the best agreement with the data (lowest R and R_free) and is consistent with the in vivo and in vitro data reported previously.6 In both structures, the enzyme is in α‐aminoacrylate (E_AA) form, with α open and β closed conformation (α^Oβ^C) and has PLP covalently attached, as reported previously.4 The ligands geometry and environment are very similar in all four copies in the crystal. Here we describe binding of ligands to chains C (α) and D (β) as these show the lowest temperature factors. Both GSK1 and GSK2 bind to the same binding pocket, as described for azetidine derivative BRD4592.4 Consistent with an allosteric mechanism of inhibition, the ligands bind outside of the α and β active sites in a cavity located at the α/β interface (Figures 1, 2, 3). The ligand binding pocket intersects the hydrophobic tunnel through which indole presumably travels from the α‐ to the β‐catalytic site.

Figure 1.

Overall structure of α^Oβ^C conformation of tryptophan synthase αββα heterotetramer in surface representation. The α subunits are green and purple and β subunits are beige and salmon. The atoms of GSK2 molecules are shown as spheres and E _AA moieties are shown as ball‐and‐sticks

Figure 2.

Binding mode of allosteric inhibitors of MtTrpAB: (a) GSK1 interactions (3R,4R stereoisomer); (b) GSK2 interactions; (c) interactions of GSK2 N‐methyl moiety with oxygen atoms3; (d) electron density and structural formula for GSK1; (e) electron density and structural formula for GSK2 and (f) binding of BRD4592 (2R, 3S, 4R stereoisomer) and its structural formula. 2mFo − DFc electron density map for GSK1 and GSK2 is contoured at 1σ level. Hydrogen bonds are shown in cyan, interactions of N‐methyl in gray. Water molecules are represented by red spheres

Figure 3.

Comparison of the binding mode for GSK1 (beige), GSK2 (salmon) and BRD5492 (gray) ligands. Top figure shows location of the ligands in αβ heterodimer and bottom insert shows different orientations of ligands in the binding pocket and adjustment of protein side chains upon ligand binding

This allosteric site, extensively characterized for BRD4592, is now being shown to accept three different chemical scaffolds [i.e., BRD4592,4 GSK1, GSK26 (this work)]. The interactions of GSK1 and GSK2 are predominantly through hydrophobic interactions between aromatic rings and β subunit aromatic side chains, including βF188 and βF202, as well as β subunit hydrophobic residues, including βP208, βI184, and βL34 (Figure 3). The binding of GSK1 and GSK2 is very similar to that reported earlier for BRD4592.4 Although of distinct scaffolds, GSK1 and GSK2 have generally similar chemical organization with a larger hydrophobic portion and a smaller hydrophilic “head.” In GSK1 it is sulfolane and in GSK2 it is sulfonamide moiety, respectively. Similar to the BRD4592 azetidine ring, these hydrophilic moieties make direct contacts with protein and provide binding specificity. These interactions, however, are different for each compound. GSK1 makes three direct hydrogen bonds to main chain amides of αG66, βH294 and βG295 utilizing two out of three oxygen atoms of the sulfolane ring moiety (Figure 2a). In addition, there are two hydrogen bonds to water molecules which are coordinated by side chains of αD136 and βN185, as well as main chain atoms. Therefore, this compound bridges α and β subunits through direct and water mediated bonds. Similarly, GSK2 accepts two direct hydrogen bonds from main chain amides of αG66 and αM67 and donates hydrogen bond to carboxylate of αD136 (Figure 2b). All these interactions are with the α subunit. Interestingly, for GSK2, the methyl group of the sulfonamide is inserted into solvent region between α and β subunits. In the crystal structure there are six ordered water molecules forming a well‐ordered network and coordinated by side chains of αD136 (interacting with inhibitor), αY62, βT308 and several main chain atoms (Figure 2b). This ordering effect is rather striking and suggests that GSK2 contributes significantly to stabilization of the α/β interface, although it binds directly to the α subunit only. In comparison, for BRD4592 there are direct and water‐mediated hydrogen bonds that connect the secondary amine and the hydroxyl groups of the azetidine ring with both subunits αD64, αG66, and βH294 (Figure 2f). Therefore, the contacts made by the GSK1 and GSK2 interactions with TrpAB have some similarities to those made by BRD4592. But only direct interaction to αG66 is being shared among all three ligands.

3. DISCUSSION

Allosteric regulation is widely distributed in biology and plays a critical role in many cellular processes. Function of many enzymes evolved by tuning and synchronization of the sequence and structural transitions necessary for chemical transformations. Ligand binding at specific allosteric sites leads to conformational changes that can significantly alter enzyme activity. Such controls have been described for many important metabolic enzymes and they participate in modulation of metabolite level by providing chemical feedback to catabolic and anabolic networks about the state of the cell. The mechanisms of allosteric controls in proteins have been extensively investigated both biochemically and structurally. Many metabolic ligands serve as allosteric regulators linking various cellular processes and coordinate production of metabolites in response to outside signals. The list of allosteric systems includes the transport of metabolites across cell membranes, signal transduction, protein synthesis by the ribosome, transcriptional regulation, gene expression and many others. 32 It has been recognized that targeting allosteric sites for drug discovery serves as an attractive alternative to enzyme active sites since they are less conserved and therefore can offer higher selectivity. As such, allosteric sites may have less metabolic interference and off‐target side effects and can be better optimized for pharmacokinetic properties. They provide new opportunities for novel drug candidates.

Here, we compared the interaction modes of GSK1 and GSK2 with the BRD4592 allosteric inhibitor bound to the α open and β closed α‐aminoacrylate (α^Oβ^C) state of the MtTrpAB enzyme. Each of the compounds belongs to a different chemical scaffold (sulfolane, indole‐5‐sulfonamide, and azetidine, respectively) yet they share similar features. GSK1 and BRD4592 have sulfolane and azetidine rings providing strong hydrogen bonding potential and GSK2 has a sulfonamide functionality attached to an indoline that can make hydrogen bonds with the amino acid residues as well.

Both GSK1 and GSK2 inhibitors bind to the same site of the αβ heterodimer, as described previously for the BRD4592 (PDB id: 5TCI).4 The ligand‐binding pocket intersects the hydrophobic tunnel that is distinct from the reported for StTrpAB intra‐channel binding site of the F6 inhibitor observed at very high ligand concentrations.24 The TrpAB tunnel shows some sequence variation but the hydrophobic moieties of all three compounds can be well accommodated in the pocket. Binding of the three ligands studied here appears to be driven primarily by hydrophobic interactions between aromatic rings and β subunit aromatic side chains, including βF188, βF202, βF211, βY200, βW191, and βH294 as well as β subunit hydrophobic residues including βP208, βI184, and βL34 (Figure 3b). There are few direct and water‐mediated hydrogen bonds connecting the hydrophilic head of the GSK1 sulfonate (Figure 2a) and GSK2 sulfonamide (Figure 2b) to the TrpAB protein. Analogous interactions were also observed for the BRD4592 azetidine ring (Figure 2f). Generally, the inhibitors bind in a similar fashion, but interactions require some rearrangements of both, the ligand and protein. First, conformational changes in the form of small rotational adjustments are needed for each ligand to fit the pocket (Figure 3b). Second, the protein residues also move to accommodate the three different ligands (Figure 3b). For example, F188 and F202 slightly reposition their side chains. Despite these parallels, the specific interactions, especially those involving hydrogen bonds, are different for each inhibitor, even though the same protein region is engaged. Here, the only common biding motif is the αG66 amide. There are also relatively few water molecules preserved in the binding site for all three structures.

GSK1 and BRD4592 make direct hydrogen bonds with both α and β subunits. GSK2 binds directly via a H‐bond only to the α subunit. The binding of GSK2 is puzzling as it causes the ordering of solvent on the interface between subunits indicating the sulfonamide methyl group may be the key moiety for making hydrogen bonds with water molecules through the N‐methyl group33, 34 (Figure 2c). Four water molecules, carbonyl oxygen atom of αAsp64 and carboxylate oxygen atom of αAsp136 are at the C…O distance 3.4–3.9 Å predicted and observed in the crystal structures to form (NZ)CH…O contacts equivalent of weak hydrogen bonds. Interactions of methyl groups with ordered solvent may be considered in the design of protein inhibitors, because they can provide additional favorable interactions.

All amino acid mutations that were identified in Mtb to provide resistance to these inhibitors (αP65, αG66, αD136, αY108 and βI184, βN185, βF188, βP208, βF293) are clustered along the inhibitors binding site. These data support a model where their respective side chains would sterically interfere with inhibitor binding or would alter the binding pocket such that ligand binding would be disrupted (Figure 4). Some of these mutations are predicted to impact binding of hydrophobic part of the ligands and some ‐ the hydrophilic moieties. We have shown this previously by solving the crystal structure of MtTrpAB with the αG66V BRD4592 resistance mutation.4 The isopropyl side chain of αV66 sterically interferes with inhibitor binding, making BRD4592 specific for those TrpABs from mycobacterial species that have glycine in position 66 of the α subunit. Given that GSK inhibitors bind to that same pocket that is created only in the absence of side chain in position 66, these compounds should share BRD4592 species selectivity. All the resistant variants have been obtained in vivo and they are adequately active to support L‐Trp synthesis. A few mutants were tested both in vivo and in vitro. Some of these variants occur naturally in TrpABs from other species, for example, G66V is in TrpAB from S. aureus and G66V E. coli and S. thyphimurium. TrpB‐F188L mutant activity in vitro in the presence or absence of inhibitor was comparable, but it is much weaker than WT.6 The TrpAB G66V mutation showed very similar activity to WT4 and it is a naturally occurring variant of TrpAB.

Figure 4.

Resistance mutants mapped on the allosteric ligands binding pocket. The resistance mutants include residues in TrpA (green) (P65Q, G66V, Y108C, D136V shown in green) and TrpB (beige) (I184S, N185S, F188L, P208L, F293C shown in beige). BRD4592 (dot surface) is marking location of ligand binding pocket and α‐aminoacrylate (in orange) is marking location of the β active site

The mechanism of inhibition for GSK1 and GSK2 has not been studied in detail; however our investigation of BRD4592 indicated a complex network of effects. The molecule shifts equilibrium towards β^C conformation, which makes the TrpAB complex more stable but at the same time it reduces enzyme flexibility and freezes communication between subunits that is needed to complete the catalytic cycle. It also prevents indole migration between the active sites. These two elements effectively block both α and β half‐reactions. Additional, time‐dependent component, promotes L‐Trp product inhibition due to increased L‐Trp affinity in the presence of BRD4592. Given that GSK1 and GSK2 bind to the same site as BRD4592, they are likely to act in a similar fashion. This is partly confirmed by the size exclusion chromatography data which indicate increases TrpAB complex formation in the presence of the compounds.6 The fact that three distinct, independently identified chemical scaffolds bind to the same allosteric site and make specific interactions, provides promising chemistry for the developments of new therapeutics against TB. Our structures validate the inhibitor discovery approach and provide the chemical basis for the development of novel allosteric inhibitors targeting TrpAB.

4. MATERIALS AND METHODS

4.1. MtTrpAB gene cloning, protein expression and purification

The gene cloning, protein expression and purification were performed as in detail reported previously.4 Briefly, for the TrpAB purification, frozen cells were thawed, sonicated and spun down for 1 h at 30,000 × g at 4°C. The protein complex was purified using a Flex‐Column containing 5 ml Ni²⁺ Sepharose (GE Healthcare Life Sciences) preequilibrated with buffer containing 50 mM HEPES:NaOH, pH 7.5; 150 mM KCl; 20 mM imidazole, pH 8.0; 5% glycerol; 1 mM PLP; 1 mM L‐Serine and 1 mM TCEP. The column was connected to a Van‐Man vacuum manifold (Promega). After elution the TrpAB complex was concentrated to about 2 ml and loaded on a Superdex 200 16/70 size exclusion column (GE Healthcare Life Sciences) equilibrated with the same buffer. Successful crystallization required the presence of excess of the TrpA subunit. Fractions containing the protein complex were collected, and purification buffer was replaced with crystallization buffer (20 mM HEPES:NaOH, pH 7.5, 150 mM KCl, 1 mM PLP, 1 mM L‐Ser, 1 mM TCEP) using Amicon 50‐kDa‐cutoff concentrator (Millipore). The TrpAB complex was concentrated to ~38 mg/ml.

4.2. MtTrpAB crystallization and ligand soaking

Crystallization experiments also were conducted as described previously,4 with slight modifications. INDEX, PEG/Ion (Hampton Research), MCSG1 (Anatrace) and PEGsII Suite (Qiagen) screens were used for protein crystallization. The Peg/Ion F6 condition, which had produced the best crystals previously4 was optimized using Additive Screen (Hampton Research). The sitting‐drop vapor‐diffusion method was used with the help of the Mosquito liquid dispenser (TTP LabTech) in 96‐well CrystalQuick plates (Greiner Bio‐One). Crystallizations were performed with the protein‐to‐buffer ratio of 2:1; 1:1 and 1:2. After setting up, the crystallization plates were wrapped in aluminum foil and kept at 16°C.

Diffraction‐quality crystals of MtTrpAB suitable for ligand soaking appeared after 2 weeks in PEG/Ion F6 with additive C1 (8% v/v tacsimate, pH 8.0, 20% w/v PEG3350, 0.1 M sodium malonate pH 7.0) and PEG/Ion E8 (0.2 M sodium malonate pH 7.0; 20% w/v PEG 3350) conditions. To prepare TrpAB complexes, GSK1 (3R,4R/3S,4S racemate) and GSK2 inhibitors were dissolved in DMSO to 0.25 and 0.5 M, respectively (note that GSK1 was not fully dissolved). Crystals were soaked for 2–4 min in the crystallization solution containing 6–12 mM of inhibitor and cryoprotected with 17% (v/v) ethylene glycol. In order to get TrpAB complex in the α^Oβ^C conformation, after initial soaking the crystals were transferred to a droplet with the same composition and 25 mM l‐serine. All crystallization plates were kept in the dark.

5. DATA COLLECTION, STRUCTURE DETERMINATION AND REFINEMENT

Prior to data collection at 100 K, all cryoprotected crystals were flash‐cooled in liquid nitrogen. The x‐ray diffraction experiments were carried out at the Structural Biology Center 19‐ID beamline at the Advanced Photon Source, Argonne National Laboratory. The diffraction images were recorded on the PILATUS3 6 M detector. The data sets were processed with the HKL3000 suite.35 Intensities were converted to structure factor amplitudes in the Ctruncate program36, 37 from the CCP4 package.38 The data collection and processing statistics are given in Table 1.

Table 1.

Data processing and refinement statistics

Structure	MtTrpAB‐GSK1	MtTrpAB‐GSK2
Data processing
Wavelength (Å)	0.9792	0.9792
Resolution range (Å)a	30–2.40 (2.44–2.40)	30–2.40 (2.44–2.40)
Space group	P212121	P212121
Unit cell (Å)	a = 134.92, b = 160.04, c = 165.15	a = 135.11, b = 159.23, c = 164.97
Unique reflections (merged)	137,941 (6,867)	138,216 (6,845)
Multiplicity	24.9 (20.7)	12.7 (10.7)
Completeness (%)	99.8 (99.9)	99.7 (99.7)
Mean I/sigma(I)	13.60 (1.80)	9.93 (1.72)
Wilson B‐factor (Å2)	31.5	27.7
R‐mergeb	0.345 (2.178)	0.308 (1.649)
CC1/2c	(0.697)	(0.661)
Refinement
Resolution range (Å)	29.89–2.41	29.88–2.40
Reflections work/test	135,152/2,677	135,430/2,691
Rwork/Rfree d	0.158/0.192 (0.244/0.293)	0.153/0.195 (0.222/0.250)
Number of non‐hydrogen atoms	21,114	21,418
Macromolecules	19,514	19,491
Ligands/solvent	366/1,234	732/1,195
Protein residues	2,602	2,609
RMSD(bonds) (Å)	0.003	0.004
RMSD(angles) (°)	0.60	0.64
Ramachandran favorede (%)	97.87	97.87
Ramachandran allowed (%)	1.98	1.90
Ramachandran outliers (%)e	0.16	0.23
Rotamer outliers (%)e	1.03	0.62
Clashscore	2.57	2.85
Average B‐factor (Å2)	37.48	31.07
Macromolecules	37.03	30.02
Ligands	49.77	50.12
Solvent	40.86	36.50
Number of TLS groups	8	28
PDB ID	6USA	6U6C

^aValues in parentheses correspond to the highest resolution shell.

^bRmerge = Σ_hΣ_j|Ihj–<Ih>|/Σ_hΣ_j Ihj, where I_hj is the intensity of observation j of reflection h.

^cAs defined by Karplus and Diederichs.41

^d R = Σ_h||Fo|–|Fc||/Σ_h|Fo| for all reflections, where Fo and Fc are observed and calculated structure factors, respectively. Rfree is calculated analogously for the test reflections, randomly selected and excluded from the refinement. As defined by Molprobity.42

^eThe outliers on the Ramachandran plots for GSK1 are G239 of molecules A, C, E, and G and for GSK2 are N‐terminal residues (E8 in molecule D is poorly defined), A6 and P8 in molecule B and P8 in molecule H and also include G239 of molecules A, C, E.

The α^Oβ^C MtTrpAB structure (PDB code 5TCG) was refined against new datasets to yield α^Oβ^C‐GSK1 and α^Oβ^C‐GSK2 models. The structures were refined by manual corrections in Coot39 and crystallographic refinement in Phenix.40 The refinement statistics are given in Table 1. The atomic coordinates and structure factors have been deposited in the Protein Data Bank under accession codes 6USA (MtTrpAB‐GSK1) and 6U6C (MtTrpAB‐GSK2).

Michalska K, Chang C, Maltseva NI, et al. Allosteric inhibitors of Mycobacterium tuberculosis tryptophan synthase. Protein Science. 2020;29:779–788. 10.1002/pro.3825