Functional clues from the crystal structure of an orphan periplasmic ligand‐binding protein from Treponema pallidum

Abstract

The spirochete Treponema pallidum is the causative agent of syphilis, a sexually transmitted infection of major global importance. Other closely related subspecies of Treponema also are the etiological agents of the endemic treponematoses, such as yaws, pinta, and bejel. The inability of T. pallidum and its close relatives to be cultured in vitro has prompted efforts to characterize T. pallidum's proteins structurally and biophysically, particularly those potentially relevant to treponemal membrane biology, with the goal of possibly revealing the functions of those proteins. This report describes the structure of the treponemal protein Tp0737; this polypeptide has a fold characteristic of a class of periplasmic ligand‐binding proteins associated with ABC‐type transporters. Although no ligand for the protein was observed in electron‐density maps, and thus the nature of the native ligand remains obscure, the structural data described herein provide a foundation for further efforts to elucidate the ligand and thus the function of this protein in T. pallidum.

Short abstract

PDB Code(s): 5U2P

Abbreviations

ABC

ATP‐binding cassette

AUC

analytical ultracentrifugation

DSF

differential scanning fluorimetry

LBP

ligand‐binding protein

rTp0737

a recombinant version of gene product tp0737

PCR

polymerase chain reaction

PIPE

polymerase‐incomplete primer extension

SV

sedimentation velocity

Introduction

Despite the efficacy and widespread availability of chemotherapeutic strategies, the sexually transmitted human disease syphilis remains a significant threat to human health on a global scale.1, 2, 3 Worldwide rates and prevalences of the disease have failed to decline,1 and alarming increases in case rates, even in developed countries, have been observed.4, 5, 6, 7 Furthermore, the emergence of macrolide resistance in Treponema pallidum, the etiological agent, has been alarming.8 Finally, decades of intensive efforts have failed to yield a syphilis vaccine.

Treponema pallidum subsp. pallidum is an obligate parasite of humans. This microorganism's lifestyle and its minimized genome (containing only 1041 genes9) necessitate that it acquires the vast majority of its nutrients from its human host. In many cases, nutrient procurement is achieved via ABC‐type transporters.3 In T. pallidum, these systems typically initially secure target small‐molecule compounds through their binding to membrane‐tethered ligand‐binding proteins (LBPs) that protrude into the periplasm.3 These lipoproteins ostensibly facilitate the translocation of the nutrient to the organism's cytoplasm via the action of transmembrane permeases, whose actions are driven by cytoplasmic ATPases.10 Many such systems have been elucidated in T. pallidum, including transporters for transition metal ions,11, 12 L‐methionine,13 D‐glucose,14, 15 purine nucleosides,16 and riboflavin.17

The gene product of tp0737 is an unusual protein in the context of the biology of T. pallidum. The protein bears sequence homology to sugar‐binding proteins of ABC‐type transporters, and is annotated as such in databases.9 However, the protein does not harbor the N‐terminal sequence signatures that are ordinarily necessary for the post‐translational modification of lipoproteins. Rather, an N‐terminal sequence that putatively targets the protein to the periplasm (i.e. to signal peptidase I) is present.18 Whereas this may suggest that Tp0737 can exist free in the treponemal periplasm, it is possible that the signal sequence is not cleaved, thus serving as a “signal anchor” that tethers the protein to the cytoplasmic membrane, with its soluble portion harbored in the periplasm.19, 20, 21 Moreover, the genetic environment of tp0737 suggests that the mRNA derived therefrom is monocistronic. This organization is unusual insofar as T. pallidum commonly transcribes the LBP, permease, and ATPase(s) of ABC transporters on the same mRNA.16, 17, 22

These unique features and the unknown identity of the protein's native ligand prompted us to study the structure of Tp0737 recombinantly expressed without its N‐terminal signal sequence (rTp0737). The crystal structure was determined at a resolution of 1.76 Å. The structure reveals a two‐lobed protein with a deep cleft between the lobes; as predicted, it resembles a LBP for an ABC‐type transporter. Unfortunately, no electron density for a potential ligand could be observed. Although a series of biophysical studies also did not elucidate the identity of any ligand, the rTp0737 structure sets the stage for future efforts in ligand discovery.

Results and Discussion

Structure description

The crystal structure of rTp0737, a recombinant fragment of Tp0737, was determined and refined at a resolution of 1.76 Å (Table 1, Fig. 1). All residues are visible in the electron‐density maps except for 28‐39, 182‐184, 394‐396, and 430‐434 (N.B.: although only 408 residues are present in the protein construct, the numbering scheme reflects the primary structure of the immature, unprocessed protein). Overall, rTp0737 comprises two domains that are connected via three crossover points (Fig. 1). In this report, the two domains are termed the “N domain” (residues 40‐142 and 315‐360) and the “C domain” (residues 146‐312 and 364‐429), as the N‐ and C‐termini are located in those respective domains. The N and C domains have similar structures insofar as they both are composed of a central (mostly parallel) β sheet that is flanked on both sides by helices and regions with no regular secondary structures. The central β‐sheet in the N domain encompasses five β‐strands; these are all parallel except one [10; Fig. 1(A)]. The analogous β‐sheet in the C domain is more complex, as it contains three parallel strands and one antiparallel strand. A secondary, minor sheet is formed by the packing of strands 9 and 11 against one of the strands of the main sheet; both sheets in the C domain share strand 5. There are 14 α‐helices in the structure, and DSSP23 identifies an inordinate number of 3₁₀ helices: 12. There is no evidence that the large number of 3₁₀ helices has significant functional consequences for the protein. The three crossover regions (residues 143‐145, 313‐314, and 361‐363) do not have regular secondary structures and appear to form a hinge between the two domains.

Table 1.

Diffraction and Structure Solution Statistics

Data Set	rTp0737	rTp0737‐SeMet
PDB accession no.	5U2P
Data collection
Wavelength (Å)	0.97918	0.97929
Temperature (K)	100	100
Space group	P3121	P3121
Unit cell dimensions (Å)
a	55.925	55.887
b	55.925	55.887
c	237.632	237.426
α = β (°)	90	90
γ (°)	120	120
Resolution (Å)	37.43–1.76 (1.79–1.76)a	48.4–1.95 (1.98–1.95)
Completeness (%)	97.4 (92.4)	98.8 (79.8)
Multiplicity	3.5 (2.5)	7.1 (3.9)
Unique reflections	43,052 (1,994)	32,407 (1,274)
Rr.i.m. b	0.052 (0.471)	0.061 (0.311)
<I>/σI	19.6 (2.9)	12.7 (2.6)
Wilson B (Å2)	22.8	18.8
Phasing
Sites found	N/A	4
Overall figure of merit	N/A	0.250
Automatically built residues	N/A	368
Refinement
Resolution (Å)	37.54–1.76
No. nonsolvent atoms	3,235
No. water atoms	232
No. ions	3
Average B‐factors
Overall (Å2)	36.64
Protein (Å2)	36.53
Solvent (Å2)	38.20
Ion (Å2)	34.86
R‐values
R work	0.182
R free	0.222
Ramachandran Statistics
Outliers (%)	0.0
Most favored region (%)	98.7
r.m.s. deviations
Bonds (Å)	0.005
Angles (°)	0.90

^aNumbers in the parentheses are reported for the highest‐resolution shell of reflections.

^b R _merge ( $R_{\text{merge}}={\sum }_{hkl}{\sum }_i|I_{h,i}-〈I_h〉|/{\sum }_{hkl}{\sum }_i{I}_{h,i}$ where the outer sum (hkl) is over the unique reflections and the inner sum (i) is over the set of independent observations of each unique reflection) was multiplied by the factor (N/(N − 1))^1/2 to estimate R_r.i.m.

Figure 1.

The structure of rTp0737. (A) Schematic of the overall topology of the protein. Helices are depicted as green rectangles; lettered rectangles are α‐helices, while those with lowercase roman numerals are 3₁₀ helices. β‐strands are represented by purple, numbered arrows. The relative sizes of these features are approximately to scale. The N and C domains are labeled. (B) Ribbons representation of rTp0737. Secondary structural features are colored as in part (A), with regions of irregular secondary structure colored light blue.

Between the domains is a large, solvent‐filled cleft. There is no electron density for a specifically bound ligand in this cleft nor anywhere else in the structure. Besides bound water molecules, the only other chemical entities bound to the proteins are a bromide anion, two chloride anions, and ethylene glycol, all of which were present in the protein‐storage buffer and/or the cryoprotection buffer.

Comparisons to other structures

A hidden‐Markov‐model‐based strategy24 was used to search the Protein Data Bank for sequences similar to that of rTp0737A. The closest sequence match (16% identity) was to the GacH acarbose/maltose‐binding protein from Streptomyces glaucescens (accession number 3K0125). However, that protein had a significantly different conformation than that of rTp0737, so a secondary‐structure‐matching26 search of the Protein Data Bank was initiated, revealing several X‐ray crystal structures with significant structural homology to that of rTp0737. The first match (accession number 3OO9), with an r.m.s.d. of 2.9 Å over 316 aligned C_α atoms is AcbH from Actinoplanes sp. SE50/110.27 The latter protein is the LBP of a putative ABC transporter for β‐D‐galactose. Such proteins bind ligands that are available extracytoplasmically and deliver them to permeases that traverse the cytoplasmic membrane, where the action of an intracellular ATPase drives the import of the ligand into the cytoplasm, usually against a concentration gradient.28 The second match (accession number 3QUF; r.m.s.d. of 2.8 Å over 302 C_α atoms) was to a protein from Bifidobacterium longum supsp. infantis. Although there is no publication describing the structure and function of this protein, it is presumed to be the LBP of an ABC transporter. Other, weaker matches (i.e. with either less coverage or higher r.m.s.d.'s) include other ABC‐type LBPs, such as those with putative specificities for raffinose (accession number 2HFB) and maltotriose (accession number 2GHB). Despite these evident homologies, we observed no bound ligand in the rTp0737 electron‐density maps.

As intimated from the structural matches enumerated above, rTp0737 is topologically and structurally related to polypeptides that serve as the LBPs for the ATP‐dependent transport of nutrients into bacterial cytoplasms. Specifically, rTp0737's topology conforms to that of “Type II” LBPs in an early nomenclature system29 and to that of “Cluster D” in a more recent classification scheme.30 Indeed, the structural organization of rTp0737 is most closely related to proteins in “Cluster D‐I”, which contains carbohydrate‐binding proteins with specificities for maltose and glucose. Like rTp0737, Cluster D‐I proteins have a five‐stranded β‐sheet in the N domain in the order 2‐1‐3‐n‐4, whereas the C domain has the order 1‐2‐n‐3, where n is respectively the first strand after a crossover from one domain to the other. Further, these proteins have an “extra subdomain” in their respective C domains. In rTp0737, this subdomain comprises α‐helices G, H, and N, along with 3₁₀ helices v, vi, and xii.

Significantly, the closest structural matches to rTp0737 are LBPs without any bound ligand. Unliganded LBPs tend to exist in solution in an open‐cleft form, with the two domains not closely apposed.15, 31, 32, 33 When ligand binds, the two domains move closer to one another, effectively closing the cleft and, in many cases, shielding the ligand from solvent.16, 34, 35, 36 Thus, the open form of the rTp0737 observed in the crystal structure is consistent with its lack of bound ligand.

To glean possible hints regarding the cognate ligand of Tp0737, we examined the putative binding pocket (Table 2; Fig. 2). We find that no single class of amino acids predominates in the cleft. Further, we compared the surface features of the cleft to a structurally homologous protein, BxlE of Streptomyces thermoviolaceus OPC‐520 (PDB accession number 3VXC; there is no attendant publication). BxlE binds to a disaccharide called xylobiose, and the binding site is marked on the left side of Figure 2. There are substantial areas of hydrophobicity in the BxlE xylobiose‐binding site, where aromatic residues stack on the surfaces of the linked xylose monomers. The analogous areas of Tp0737 are largely devoid of such features, corresponding to the paucity of aromatic residues in the putative ligand‐binding cleft (Table 2). Moreover, this area on rTp0737 is flatter than the corresponding area in BxlE, and is rich in hydrogen‐bond donors and acceptors. The overall arrangement of arrangement of amino‐acid side chains in the cleft does not immediately suggest the identities of potential ligands beyond the fact that they would likely have to be largely polar.

Table 2.

Residues Whose Side Chains Line the Cleft of rTp0737

Polar	Charged	Hydrophobic
H50	R47	P55
S52	K49	P102
S100	D77	L138
S103	D78	V145
H105	D282	V259
N142	K286	I263
S146	E287	Y274
N148	E313	L279
S191	D385	V280
T193		A283
S195		L352
T199		V362
S277		P364
H278		A383
T309		A386
Q315		A388
H349
Q391

Figure 2.

Surface features of rTp0737 and a related LBP. The two LBPs, BxlE (left) and rTp0737 (right) have been split into their respective N and C domains, and these have been individually rotated such that the ligand‐binding surfaces are facing forward; in effect, they have been “opened” along their hinge regions to present the ligand‐binding surfaces. Surfaces contributed by carbon atoms are colored gray, while those representing nitrogens are blue, oxygens are red, and sulfurs are yellow. Importantly, these surfaces are not the result of an electrostatic calculation. Boxes on BxlE represent the xylobiose‐binding site.

Ligand screening of rTp0737

The obvious structural similarities between LBPs and rTp0737 prompted screening experiments to be undertaken for the binding of small molecules to the purified protein. Differential scanning fluorimetry (DSF) was used for this purpose.37 Briefly, the method involves the measurement of the fluorescence of a dye that is included with the protein/ligand solution. The fluorescence intensity of the dye is enhanced when it binds to exposed, hydrophobic regions of the unfolded protein, allowing the observer to estimate an apparent melting temperature, $T_{\mathrm{m}}^{\text{app}}$, for the protein in the presence of a temperature gradient. Thus, the protein/ligand/dye solution is subjected to increasing temperature, and the fluorescence of the solution is monitored. Specific binding of a ligand should shift the $T_{\mathrm{m}}^{\text{app}}$ to a higher value.

DSF was employed in a series of 96‐well plates, each of which contained 95 ligands and a negative control (water; see Materials and Methods). In the negative‐control wells, rTp0737 exhibited a $T_{\mathrm{m}}^{\text{app}}$ of approximately 48°C (Fig. 3). Overall, 950 ligands/conditions were screened, but none caused a positive excursion from the negative‐control $T_{\mathrm{m}}^{\text{app}}$ greater than 0.5°C. Thus, 0.5°C probably represents the assay's intrinsic noise. By contrast, in a similar DSF assay performed on a different LBP protein (Tp0684), a compound that bound to Tp0684 with a K _D of approximately 30 μM resulted in a +3.5°C change in $T_{\mathrm{m}}^{\text{app}}$.15 Given the lack of a response above the noise level and the results of the earlier assay, we concluded that no candidate ligands for rTp0737 were identified. The tested ligands included small molecules that Cluster D‐I members are known or suspected to bind, including D‐glucose, D‐galactose, maltose, maltotriose, raffinose, and xylobiose. Furthermore, not only carbohydrates were tested; amino acids, phosphorus sources, sulfur sources, cofactors, peptides, and osmolytes were among the tested compounds.

Figure 3.

A typical DSF profile for rTp0737. The minimum in this plot was taken as $T_{\mathrm{m}}^{\text{app}}$: 48°C.

This negative result could have one of several causes. Most likely, the true ligand (if any) of rTp0737 was not among the tested compounds. However, there are other possibilities. For instance, buffering agents are known to interfere with ligand binding in some LBPs.15, 22, 34 To examine whether the phosphate buffer of the assay was preventing ligand binding, the entire DSF screen was performed again in the presence of Tris as the buffering agent. Still, no significant shifts in $T_{\mathrm{m}}^{\text{app}}$could be discerned.

Another potential cause of false negatives in the DSF assay was the solution behavior of the protein. Specifically, rTp0737 could form oligomers or nonspecific aggregates. The interprotein contacts in these assemblies could, in turn, occlude ligand access to the cleft or disfavor the domain motion thought to be necessary for ligand binding.

To address the possibility of oligomers and aggregates, the hydrodynamic properties of rTp0737 were probed using the sedimentation velocity (SV) mode of analytical ultracentrifugation (AUC). At a concentration of 1 mg/mL (roughly twice that employed in the DSF assay), the protein was predominantly monomeric (Fig. 4). More precisely, 95% of the mass present in this rTp0737 preparation could be accounted for by the monomeric form of the protein (3.1 S). A small amount (3%) appears to be either a higher‐mass contaminant or a dimer of rTp0737 (4.2 S). Therefore, it is very unlikely that oligomerization or aggregation was affecting the protein's ability to bind ligands in the DSF assay.

Figure 4.

The c(s) distribution for 1 mg/mL rTp0737. The dominant peak corresponds to the monomeric form of the protein, as computed using the Svedberg equation.

Conclusions and implications

The structure of rTp0737 (Fig. 1) conforms closely to other protein structures that serve as LBPs for ABC transporters. Hence, it may be expected that Tp0737 fills this role as well. However, a screen of a large panel of likely nutrients that might be imported by such a transporter revealed no putative ligands for the protein.

That T. pallidum cannot be continuously cultivated in vitro underscores the incomplete understanding of the spirochete's nutritional requirements. It is therefore possible that the true ligand of Tp0737 may not have been included in the chemical panel of known or likely nutrients used in the DSF experiments detailed above. Expanding the chemical space of the screen could reveal the protein's native ligand. In this regard, it is instructive to examine the chemical nature of the amino‐acid residues that line rTp0737's cleft (Fig. 2, Table 2). These residues are a mixture of polar, charged, and hydrophobic, with only one aromatic residue present (Y274). The aromatic portion of this residue is mostly buried, with only the side chain's hydroxyl group presented to the cleft. The lack of aromatic amino acids in the cleft contrasts with most carbohydrate‐binding LBPs, which have a mix of polar and aromatic residues to bind the sugar moiety(ies).38, 39 Again, the evidence favors an atypical ligand.

The bilobed fold of the LBPs is also used for other purposes by microorganisms, e.g. as chemotaxis receptors40 and transcription factors.41 Although there is no evidence for either of these functions for Tp0737, these examples underscore the functional versatility of the fold. Further investigations along these lines are complicated by the aforementioned intractability of culturing T. pallidum in vitro. Another fact that hampers future analysis of this protein is that homologues are not found in genetically tractable spirochetes such as T. denticola and Borrelia burgdorferi.

Despite the difficulties and enigmas presented above, the crystal structure of rTp0737 offers some opportunities. The amino‐acid side chains present in the cleft (Table 2) are informative in this regard. The ligand probably has a mixed chemical character, with a number of polar groups poised to interact with the polar side chains arrayed in Tp0737's cleft. Future investigations of the ligand's identity will proceed taking heed of these structural clues.

Materials and methods

Cloning, overexpression, and protein preparation

To produce a recombinant derivative of Tp0737 (rTp0737) in Escherichia coli, the DNA fragment encoding amino acid residues 28‐436 (cloned without the residues comprising the N‐terminal predicted transmembrane helix (residues 1‐26) plus Cys27) of Tp0737 was polymerase chain reaction (PCR) amplified from T. pallidum genomic DNA by the polymerase incomplete primer extension (PIPE) cloning method using ends‐specific primers (PIPE insert). The expression vector, pSpeedET (DNASU, AZ), which encodes an N‐terminal TEV‐protease cleavable expression and purification hexa‐histidine tag (MGSDKIHHHHHHENLYFQG), was PCR amplified with PIPE‐vector primers. The PIPE‐insert and PIPE‐vector were mixed to anneal the amplified DNA fragments together.42 E. coli HK100 competent cells were transformed with the mixtures (PIPE‐vector and insert) and selected for kanamycin resistance on LB agar plates. Cloning junctions/fragments were verified by DNA sequencing. A verified plasmid in HK100 cell was then used for soluble protein expression. Briefly, E. coli HK100 cells were grown at 37°C in LB medium containing 40 µg/mL of kanamycin until the cell density reached an A ₆₀₀ of 0.5. The culture was then induced for 3 h at 37°C with 0.2% (w/v) L‐arabinose. The procedures for expression and purification of the recombinant proteins were essentially as previously described.43

For the production of selenomethionine labeled protein, tp0737 was recloned into a pProEx HTb vector (Invitrogen) and methionine mutations (L139M and L336M) were introduced into the plasmid using the QuikChange Multi site‐directed mutagenesis kit (Agilent Technologies). The plasmid was then transformed into a methionine auxotroph E. coli B834 (DE3). The recombinant protein was overproduced and purified as described previously.44

Protein concentrations were determined in buffer A (20 mM Hepes, 0.1 M NaCl, pH 7.5, 2 mM n‐Octyl‐β‐D‐glucopyranoside) using spectrophotometry. Extinction coefficients were calculated using the ProtParam tool of ExPASy (www.expasy.org).45

Crystallization and cryoprotection

Crystals of rTp0737 were obtained by mixing 3 μL of Tp0737 (∼24 mg/mL in buffer A) with 3 μL of crystallization buffer (0.2 M NaBr, 0.1 M Bis‐Tris‐Propane, pH 7.5, 20% (w/v) PEG 3,350) and incubating them over 0.4 mL of the reservoir (containing crystallization buffer) for 7 days. The crystals were transferred to the stabilization buffer (0.2 M NaBr, 0.1 M Bis‐Tris‐Propane, pH 7.5, 20% (w/v) PEG 3,350, 20% (v/v) ethylene glycol). After about 1 min in this solution, the crystals were flash‐cooled in liquid nitrogen. Toward solving the phase problem, crystals of a selenomethionyl derivative of Tp0737 (SeTp0737) were grown by mixing 3 μL of SeTp0737 (∼23 mg/mL in buffer A) with 3 μL of crystallization buffer (0.1 M Tris, pH 8.0, 0.2 M NaCl, 20% (w/v) PEG 6,000) and incubating the drops over 0.4 mL of reservoir solution containing the crystallization buffer for 10 days. The crystals were transferred to the stabilization buffer (0.1 M Tris, pH 8.0, 0.2 M NaCl, 20% (w/v) PEG 6,000, 25% (v/v) ethylene glycol) and cryoprotected as above.

Data collection, processing, and structure determination

Native crystals of rTp0737 diffracted X‐rays to a d_min spacing of 1.76 Å and had the symmetry of space group P3₁21 using radiation of wavelength 0.97918 Å at beamline 19‐ID at the Structural Biology Center of Argonne National Laboratory (Table 1). The data were reduced and scaled using HKL2000.46 Data from the mutant, selenomethionyl derivative of rTp0737 were acquired at the same facility at a wavelength of 0.97929 Å (Table 1). The single‐wavelength anomalous diffraction (SAD) method was used to calculate phases using HKL‐3000.47 In brief, the HKL‐3000 user interface was utilized to reduce, integrate, and scale the data, to find Se sites (four sites were found) using SHELXD,48 to refine the sites and calculate initial phases using MLPHARE,49 to employ density modification using DM50 (no noncrystallographic symmetry is present), and to perform automated model building using ARP/wARP.51 After these steps, the model covered 94% of the input sequence. This model was subsequently rigid‐body refined against the native data (the space groups of the two crystals were sufficiently isomorphous to support this strategy) in PHENIX.52 The model was completed using iterative rounds of model building and adjustment in Coot53 and refinement in PHENIX, including individual B‐factor, TLS, and positional refinement. Riding hydrogen atoms were included in the protein model throughout refinement. The final model comprises 384 amino‐acid residues, 177 water molecules, 2 chloride ions, 1 bromide ion, and a single molecule of ethylene glycol. The model geometry and related statistics are detailed in Table 1. Residues 182‐184 and 394‐396 (given the uncertain knowledge of the protein's processing in T. pallidum, the numbering for the unprocessed protein is used throughout this paper) were not evident in electron‐density maps. The structure has been deposited in the Protein Data Bank with the accession number 5U2P.

Analytical ultracentrifugation

Sedimentation velocity (SV) AUC was conducted in a Beckman‐Coulter Model XL‐I ultracentrifuge. After assembly of the centrifugation cell, which consisted of a dual‐sector, 1.2‐cm path‐length Epon centerpiece positioned between two sapphire windows enclosed in an aluminum housing, 400 μL of sample buffer was introduced into the reference sector, and 400 μL of rTp0737 at an approximate concentration of 1 mg/mL was dispensed into the sample sector. The cell was inserted into an An50‐Ti rotor, which was then placed into the centrifuge. The rotor was temperature‐equilibrated at 20°C for 2.5 h under vacuum, and then accelerated to a speed of 50,000 rpm. Interference optics were used to monitor the concentration profiles. The c(s) methodology was used to analyze the data54, 55 with both radially independent and time‐independent noise being calculated and accounted for in the analysis.56 Partial‐specific volume, solution viscosity, and solution density were calculated using SEDNTERP.57 The molar mass associated with a given c(s) peak was determined using equations presented by Schuck and coworkers.58 GUSSI59 was used to render the c(s) figure.

Differential Scanning Fluorimetry

The thermal stability of rTp0737 in the absence and presence of small molecules was performed in a 96‐well PCR‐plate (Bio‐Rad Laboratories, Inc., Hercules, CA) using a BioRad CFX96 real‐time PCR instrument coupled to a C1000 thermal cycler. Compounds from Biolog's Phenotype MicroAray (PM) supplied in 96‐well microplates (BIOLOG, Inc.) were dissolved in 50 μL of sterile water to obtain a final concentration of around 10–20 mM. Screenings were performed with plates PM1, PM2A, PM3B, PM4A, PM5, PM6, PM7, PM8, and PM9. Each plate contains 95 compounds and a blank (no ligand) control. Complete plate contents are detailed on the BIOLOG website (www.biolog.com).

The assay mixture (20 μL) contained 10 μM purified protein in a buffer containing 10 mM phosphate, 100 mM NaCl, pH 7.4. This mixture also contained SYPRO Orange that had been diluted 1000‐fold from the purchased stock concentration (Thermo Fisher Scientific, Inc., Waltham, MA). Two microliters of the resuspended Biolog compounds were added to each well. Samples were heated from 4°C to 95°C. The fluorescence values were recorded as a function of temperatures, and the negative first derivative values of these curves (−dF/dT) were calculated. The minima of these curves were taken at the apparent melting temperature (T _{m , app}).

Conflicts of Interest

The authors claim no conflicts of interest.