Optimization of a Noncanonical Anti-infective: Interrogation of the Target Binding Pocket for a Small-Molecule Inhibitor of Escherichia coli Polysaccharide Capsule Expression

We previously identified a small-molecule inhibitor of capsule biogenesis (designated DU011) and identified its target as MprA, a MarR family transcriptional repressor of multidrug efflux pumps. Unlike other proposed MprA ligands, such as salicylate and 2,4-dinitrophenol (DNP), DU011 does not alter Escherichia coli antibiotic resistance and has significantly enhanced inhibition of capsule expression. We hypothesized that the potency and the unique action of DU011 are due to novel interactions with the MprA binding pocket and the conformation assumed by MprA upon binding DU011 relative to other ligands.

ABSTRACT

We previously identified a small-molecule inhibitor of capsule biogenesis (designated DU011) and identified its target as MprA, a MarR family transcriptional repressor of multidrug efflux pumps. Unlike other proposed MprA ligands, such as salicylate and 2,4-dinitrophenol (DNP), DU011 does not alter Escherichia coli antibiotic resistance and has significantly enhanced inhibition of capsule expression. We hypothesized that the potency and the unique action of DU011 are due to novel interactions with the MprA binding pocket and the conformation assumed by MprA upon binding DU011 relative to other ligands. To understand the dynamics of MprA-DU011 interaction, we performed hydrogen-deuterium exchange mass spectrometry (HDX-MS); this suggested that four peptide regions undergo conformational changes upon binding DU011. We conducted isothermal calorimetric titration (ITC) to quantitatively characterize MprA binding to DU011 and canonical ligands and observed a distinct two-site binding isotherm associated with the binding reaction of MprA to DU011; however, salicylate and DNP showed a one-site binding isotherm with lower affinity. To elucidate the binding pocket(s) of MprA, we selected single point mutants of MprA that included mutated residues predicted to be within the putative binding pocket (Q51A, F58A, and E65D) as well as on or near the DNA-binding domain (L81A, S83T, and T86A). Our ITC studies suggest that two of the tested MprA mutants had lower affinity for DU011: Q51A and F58A. In addition to elucidating the MprA binding pocket for DU011, we studied the binding of these mutants to salicylate and DNP to reveal the binding pockets of these canonical ligands.

INTRODUCTION

The rapid increase in antibiotic resistance has rendered many common antibiotics ineffective for clinical use. In particular, the rising resistance among Gram-negative bacteria is concerning, particularly for organisms such as Escherichia coli, a major cause of infections, such as urinary tract infections, meningitis, osteomyelitis, and bloodstream infection, in all age groups (1–4). Despite the urgent need for new anti-infectives, the pace of the development of newer therapeutics lags behind the growing need. Antibiotics that have been released in the market recently are largely modified versions of older drugs and do not target novel bacterial factors (5).

The polysaccharide capsules of E. coli are established virulence factors and, thus, make attractive targets for antimicrobial intervention. Indeed, the loss of capsule results in reduced bacterial burden and mortality in animal models of urinary tract infection and sepsis, respectively (6–10). We have previously described a novel coregulation of the polysaccharide capsule and MprA, a transcriptional repressor of the multidrug efflux pump EmrAB (9). The deletion of mprA attenuates E. coli in several sepsis models (9, 11). MprA belongs to the well-known family of MarR proteins, many members of which are known to be master regulators of genes that play significant roles in vital bacterial functions (12–14). Previous studies have shown that 2,4-dinitrophenol (DNP) and salicylate bind to MprA at micromolar concentrations, resulting in the induction of the emrAB genes (13, 15).

More recently, DU011, a drug-like small molecule, was demonstrated to target MprA and inhibit capsule expression in E. coli at submicromolar concentrations (9, 10), showing far greater potency than canonical ligands without significant upregulation of emrAB. Furthermore, in vitro treatment of E. coli with DU011 did not increase resistance against four classes of clinically relevant antibiotics at concentrations in excess of capsule inhibition concentrations (9). In vivo DU011 treatment of mice abrogated bloodstream infections with uropathogenic E. coli (9).

The mechanism through which DU011 interacts with MprA is unknown and will inform efforts to target MprA and other MarR family members that coregulate antibiotic resistance and virulence factors. In this study, we sought to understand the binding mechanism of DU011 to MprA. Hence, we employed a combination of biochemical, biophysical, in silico, and cellular approaches to delineate the ligand-binding pocket of MprA and the conformational changes in MprA upon binding DU011 and other ligands. These studies revealed a novel binding mode of DU011 to MprA.

RESULTS AND DISCUSSION

Characterization of MprA conformational dynamics in the presence of DU011 by HDX-MS.

Hydrogen-deuterium exchange mass spectrometry (HDX-MS) has been used to study the dynamics of protein structures and the conformational changes that occur in those proteins upon ligand binding (16). Amino acid residues on the surface of a protein are solvent exposed. Therefore, these surface-exposed residues preferentially exchange hydrogen for deuterium (a hydrogen isotope) when incubated in deuterated water. The state of deuteration is monitored over time by mass spectrometry and, thus, relates conformational changes over time, providing insight into the structural dynamics of the protein. This unliganded (apo) model of the protein serves as the starting point to monitor any conformational changes that occur upon ligand binding. Conformational changes induced by ligand binding are expected to result in the differential exposure of amino acid residues to the deuterated solvent.

To investigate the conformational changes in MprA with DU011, a time course of 0, 5, and 10 min was conducted with MprA exposed to deuterated water with and without 50 μM DU011. After quenching the reaction with non-heavy water, pepsin cleavage and mass spectrometry were performed to identify the differential deuterium incorporation into MprA. To map the differentially incorporated deuterium in the presence of DU011, which is indicative of ligand-induced conformational change, we first constructed a homology model of E. coli MprA, as its crystal structure remains undetermined. Four regions (regions I to IV) were shown to incorporate deuterium differentially (Fig. 1). The pattern of deuteration in the presence of DU011 was not different in the time frame measured, suggesting a rapid conformational change upon DU011 exposure not captured within the limits of the HDX-MS.

FIG 1.

MprA amino acids that were found to have differential uptake of the deuterium isotope, as identified by HDX-MS, are shown in a homology model of MprA. Region I (blue), region II (red), region III (yellow), and region IV (green) are highlighted in their respective colors on the model and correspond to regions highlighted in the amino acid sequence of MprA.

MprA, like other MarR proteins, is a winged helix-turn-helix (HTH) protein that exists as a dimer (15, 17). The amino- and carboxy-terminal helices interact to form the dimerization interface (Fig. 1). Each subunit has one DNA-binding domain, which usually consists of an HTH motif. The winged-HTH DNA-binding domain of one protomer does not come into contact with that of the other subunit. The distance between these domains and, hence, the ability to bind cognate DNA is determined by the dimerization interface, which typically is modified by ligand-protein interaction (17). The predicted ligand-binding domain in the MarR family lies near the DNA binding domain and may play a role in conformational changes in the DNA binding helix (18).

As noted, there are four regions of notable differential exchange. Amino acid residues within region I are located at the N terminus and form the dimerization interface. After DU011 exposure, the conformational changes in region I likely follow a secondary change related to DU011-MprA binding elsewhere. In contrast, amino acid residues in regions II, III, and IV are adjacent to the DNA binding HTH domain. These regions contain known ligand binding in other MarR family members (18, 19) and plausible regions of DU011 binding.

Evaluation of region I to IV MprA mutants in E. coli capsule expression.

We next sought to identify the roles of key residues in the regions identified by HDX-MS in the gene-regulatory function of MprA. Plasmid-borne, site-directed mutants of mprA were transformed into the UTI89ΔmprA host strain and tested for complementation by determining the restoration of bacterial encapsulation, as measured by bacterial susceptibility to K1 phage virus (20). The K1 phage virus requires an intact polysaccharide capsule to enter and lyse the cell. In the absence of a capsule, as observed for the UTI89ΔmprA noncomplemented host strain, the cells are resistant to K1 phage lysis. We expected functional mutants of MprA to restore the capsule and, thus, become K1 phage sensitive. We anticipated that MprA mutants with ligand-binding defects, unlike wild-type MprA, would resist DU011 abrogation of capsule expression and K1 phage susceptibility. MprA mutants, therefore, showed three distinct phenotypes: those that complemented the UTI89ΔmprA strain and restored capsule but were resistant to DU011, those that complemented the UTI89ΔmprA strain but were not resistant to DU011, and those that did not complement the UTI89ΔmprA strain and, therefore, could not be assessed for resistance (see Table S1 in the supplemental material). A schematic diagram of the K1 phage assay and the results for representative MprA mutants in the three phenotypic categories are shown in Fig. 2.

FIG 2.

Schematic diagram of the K1 phage assay and representative results of complementation of UTI89ΔmprA strain with MprA mutants to identify functional MprA mutants that were also resistant to DU011. K1 phage polysaccharide-dependent cell lysis was used to determine the presence or absence of the capsule. Complementation with wild-type MprA results in restoration of the capsule and subsequent cell lysis; however, DU011 results in unencapsulation and subsequent resistance to K1 phage cell lysis. MprA mutants shown belong to three phenotypes: (i) those that complement the UTI89ΔmprA strain (sensitive to K1 phage lysis) but are resistant to DU011 treatment (remain sensitive to K1 phage), (ii) those that complement the UTI89ΔmprA strain but are sensitive to DU011 (resistant to K1 phage lysis), and (iii) those that do not complement the UTI89ΔmprA strain (resistant to K1 phage lysis). C, complement; R, DU011 resistant; NR, not DU011 resistant; NC, not capsule complementing.

Many region I mutations retained function (R15, A16, S17, R18, H19, E20, D21, Y24, Q25, I27, and C33) (Table S1). These mutations resulted in complete or partial resistance to DU011. These amino acid residues may alter the DU011 binding site allosterically without affecting DNA binding. Residues E26A and L32A were also functional proteins but were not resistant to DU011. Mutations in residues L29, T30, and R31 were associated with nonfunctional protein and were hypothesized to play a more significant role in the dimerization interface or alter the DNA binding domain even in the absence of ligand binding.

Within regions II and III, mutations in residues A50, Q51, F58, E65, and S66 resulted in functional protein but showed resistance to DU011. Amino acids in region IV are predicted to be close to or within the DNA binding domain. Consistent with this prediction, mutations in some of these residues resulted in nonfunctional proteins. However, mutations in A80, G82, and T86 resulted in functional but DU011-resistant proteins.

Quantitative characterization of MprA binding to DU011, 2,4-dinitrophenol, and salicylate.

DU011 inhibits encapsulation at submicromolar levels (9). To determine the binding affinity and stoichiometry of DU011 binding to MprA, we conducted isothermal titration calorimetry (ITC) assays. We also compared MprA-DU011 binding to the binding of two other well-known MprA ligands, DNP and salicylate. Using an alternative method, DNP bound MprA with a binding affinity (K_d) of ∼15 μM (11). Salicylate is a known ligand of MarR family members, including MarR from Escherichia coli, but acts at high micromolar concentrations (17, 21). Comparing the chemical structures of the three ligands (Fig. 3A), DU011 contains two additional ring structures with attached moieties compared to the single-ring structures of salicylate and DNP (Fig. 3A). Further, DU011 has a volume of 488.1 ų, while the volumes of DNP and salicylate are only 150.79 ų and 134.04 ų, respectively, a greater than 3-fold difference (22). Thus, we would expect distinct differences between the binding modes of MprA to DU011 and the binding of MprA to DNP and salicylate.

FIG 3.

MprA binding to DU011, salicylate, and DNP. (A) Chemical structures of DU011, 2,4-dinitrophenol (DNP), and salicylate. (B) Isothermal titration calorimetry (ITC) thermogram and resulting binding isotherm showing unique two-site binding pattern between MprA and DU011. (C) ITC thermogram and binding isotherm for interaction between MprA and DNP. (D) ITC thermogram and binding isotherm for interaction between MprA and salicylate.

DU011 binds in a manner distinct from those of DNP and salicylate. A two-site binding isotherm was observed with the binding reaction of MprA to DU011 (Fig. 3B). We observed tight binding of DU011 to both sites, with associated K_ds of ∼30 nM and <1 nM (Table 1). The K_d associated with the second value is only an estimate and likely lower, as this is below the limit of detection of the instrument. All experiments, even with various titration conditions, yielded the same two-site binding pattern with only minimal variations in K_d, providing strong evidence that this two-site binding pattern is not an experimental artifact but rather a biochemically and likely biologically relevant phenomenon. The concentration of MprA in our ITC experiments was monomeric; therefore, the measured stoichiometry (N) indicates two molecules of DU011 bind per MprA dimer in two nonequivalent sites.

TABLE 1.

Dissociation constants and stoichiometry associated with the binding reactions between various forms of MprA and ligands DU011, DNP, and salicylate^a

Ligand and parameter	Value for:
	MprA WT	Q51A	F58A	Q51A/F58A
DU011
Kd (1) (nM)	30 ± 8	62 ± 17	119 ± 40	114 ± 26
N (1)	0.307 ± 0.064	0.476 ± 0.042	0.403 ± 0.029	0.731 ± 0.082
Kd (2) (nM)	≤1	2 ± 1	3	3 ± 0.5
N (2)	0.567 ± 0.086	0.511 ± 0.069	0.429 ± 0.065	0.343 ± 0.024
DNP
Kd (μM)	21.4 ± 7.7	14.0 ± 1.2	21.1 ± 1.2	164 ± 7
N	1.13 ± 0.19	0.682 ± 0.059	0.822 ± 0.065	0.976 ± 0.091
Salicylate
Kd (μM)	55.3 ± 2.9	72.6 ± 2.5	329 ± 42	464 ± 78
N	1.06 ± 0.13	1.02 ± 0.08	0.985 ± 0.015	1.0

^aThe values reported are averages from three separate experiments.

To determine the uniqueness of this two-site binding between DU011 and MprA, we also determined the binding affinity of MprA for its ligands, DNP and salicylate. In our ITC experiments, we observed one-site binding isotherms for DNP (Fig. 3C) and salicylate (Fig. 3D) that were distinctly different from the two-site binding isotherm for DU011 (Fig. 3B). The stoichiometry suggests that one DNP molecule and one salicylate molecule bind to each subunit (protomer) of the MprA dimer in equivalent positions (Table 1). The K_d values associated with the binding of DNP and salicylate are ∼22 μM and ∼55 μM, respectively. The K_d measured for the binding of MprA to DNP is consistent with that previously determined, which utilized an alternative method (15). Thus, these data demonstrate that DU011 is a novel type of MprA ligand. The unique binding phenomenon that we observe between MprA and DU011 may be directly related to the unique in vivo functionality of DU011 inhibiting capsule production.

Potential ligand binding pockets of MprA.

To elucidate the DU011-binding pocket(s) of MprA, we initially selected five different single point mutants of MprA that showed differential deuterium labeling by HDX-MS (Fig. 1) and complemented the UTI89ΔmprA strain by restoring the capsule in vitro, as measured using K1 phage assays (Fig. 2). We selected L81A as a control mutation that did not complement the UTI89ΔmprA strain and remained unencapsulated. The selected mutants included residues predicted to be within the putative binding pocket of most MarR family members (18) (Q51A, F58A, and E65D) as well as three on or near the DNA-binding domain (L81A, S83T, and T86A) (Fig. 4). The T86A mutant is on the DNA recognition helix and was selected as a control to elucidate if DU011 binds MprA allosterically or interferes with its DNA-binding capabilities directly. Our ITC studies indicate that only two of the six MprA mutants had increased dissociation constants for DU011: Q51A and F58A (Table S2). In addition to MprA with single Q51A and F58A mutations, a double mutant was constructed. Our data show that the combined Q51A and F58A mutations result in a relatively mild diminution, with 2- to 4-fold changes in K_d for both DU011 binding sites (Table 1). Indeed, the K_d of the Q51A/F58A double mutant is the same as that of the F58A protein alone (Table 1). These data suggest that these residues are involved in binding DU011 to both sites, but these residues are not the residues necessary for high-affinity binding to either DU011 binding site.

FIG 4.

Overlay of the cartoon diagrams of monomeric MprA homology model with the dimeric template (PDB entry 3ZMD) used for generating the homology model. The homology model is shown in green and the template is shown in cyan; the winged helix-turn-helix motif (wHTH) is highlighted in orange. The box denotes the location of specific MprA residues mutated to study the role of these residues in ligand binding. These residues are shown as sticks and are highlighted red in the closeup of the putative ligand-binding pocket and wHTH.

To better understand the role of Q51 and F58 in ligand binding by MprA, we conducted ITC experiments characterizing the binding of the Q51A and F58A point mutants as well as the double mutant, Q51A/F58A, to DNP and salicylate. The data reveal that mutation of Q51 or F58 to an alanine alone does not affect DNP binding; however, mutation of both residues results in a nearly 8-fold increase in K_d (Table 1). Modification of Q51 also appears to have little effect on salicylate binding; however, mutation of F58 results in a 6-fold increase in K_d. These data suggest that salicylate occupies, at least partially, a binding site similar to at least one of the DU011-binding sites, and that F58, in particular, is important for the binding of DU011 and salicylate. These data, however, do not preclude some importance of residue Q51 in binding at least to DNP, as it has been well established that the binding abilities of many multiligand binding transcription factors are typically not dependent upon a single residue for high-affinity ligand binding and recognition. One prominent example is the multidrug binding transcription repressor, QacR, in which mutations in the expected multidrug binding pocket should have abrogated binding affinity but did not (23–25). Hence, the current finding that the Q51A/F58A double mutant results in an 8-fold loss of binding affinity for DNP simply underscores the idea that multiligand binding proteins typically utilize a series of miniature binding pockets within a larger binding pocket area to accommodate chemically and structurally dissimilar compounds (25), and changes to several residues are necessary to affect binding.

To explore further the possibility that DU011, DNP, and salicylate occupy similar or overlapping binding pockets, we performed ligand competition experiments. Based on the experimentally determined K_d values for each ligand, MprA was incubated with saturating concentrations of one of our three ligands of interest overnight at 4°C. The MprA-ligand complex then was loaded into the cell of the ITC instrument, and one of the other two ligands was titrated into this complex. The binding or lack of binding of the titrated ligand, as measured by the loss or gain of heat or no change in heat, indicates the ability of a compound to compete with the bound ligand (Table 2). The data reveal that DU011 is still able to bind both of its MprA sites even after DNP or salicylate has been added to saturable amounts, i.e., greater than 10-fold above their binding affinities (Table 2 and Fig. S1). However, the enthalpic change associated with DU011 binding is reduced; this difference in the enthalpic change associated with DU011 binding could be the result of a change in the protonation or deprotonation of MprA when it is bound to DNP or salicylate. Conversely, little to no binding is observed when DNP or salicylate is titrated into an MprA-ligand complex (Table 2 and Fig. S2 and S3). Given the very high affinity of DU011 for MprA compared to that of DNP or salicylate, we expected that DU011 would be capable of displacing DNP and salicylate if this compound occupied a similar or overlapping binding site that these two molecules occupy. Conversely, DNP and salicylate were not expected to displace DU011 because they bind far more weakly (at least less than 1,000-fold). This is consistent with the experimental data. Comparing the binding of DNP and salicylate to MprA, we observe that DNP has a slightly higher affinity for MprA than salicylate (∼2.7-fold) (Table 1). Thus, we observe some weak binding of DNP to the MprA-salicylate complex (Fig. S2B). However, the lack of saturation of DNP indicates that it is not able to fully displace salicylate, supporting the notion that they occupy overlapping but not necessarily identical sites.

TABLE 2.

Binding observations from ligand competition experiments^a

Cell	Syringe
	DU011	DNP	Salicylate
DU011		No binding	No binding
DNP	Two-site binding		No binding
Salicylate	Two-site binding	Weak, nonspecific binding

^aExperiments were performed in triplicate.

Model for the MprA-DU011 complex.

The binding mode of DU011 to two nonequivalent MprA sites is unique. To our knowledge, a similar ligand binding mechanism has not been observed between other MarR family members and their ligands. Furthermore, it is intriguing that the second DU011-binding site appears to be associated with a much higher binding affinity than the first. One possible explanation is that ligand-free MprA has only one accessible binding site for DU011. Upon binding of the first DU011 molecule, MprA undergoes a conformational change that allows the second site to become accessible for DU011 binding. Thus, DU011 binding could be cooperative and allosteric, as the second binding site might not overlap the initial site. Our mutational studies suggest that at least one DU011-binding site is near the putative ligand-binding pocket observed in several MarR family members and that this pocket can be occupied by other MprA ligands, including DNP and salicylate. Unfortunately, the location of the second DU011-binding site remains unclear. We hypothesize that DU011 initially binds the putative binding pocket on one subunit of the MprA dimer, leading to a conformational change at the dimer interface to reveal a second site on the same subunit (Fig. 5). However, it is also possible that upon initial DU011 binding to the first subunit, MprA shifts the dimer interface to reveal a second nonequivalent binding site on the second subunit. This is in contrast to the binding of DNP and salicylate, which bind a single site on each subunit at equivalent positions. Neither DNP nor salicylate binds the two MprA subunits cooperatively. We hypothesize that DNP and salicylate occupy a site, most likely an overlapping site, similar to the first DU011-binding site (Fig. 5). Structures of MprA bound to DU011, salicylate, and DNP will be necessary to elucidate their binding mechanisms fully, but unfortunately, crystals capable of yielding quality data have yet to be obtained. Structural data will provide further information about the characteristics of ligand-protein interaction between a drug-like small molecule and an important bacterial virulence target. Indeed, these data will be used to inform future rational drug design.

FIG 5.

Hypothesized MprA ligand-binding sites. The overlay of the MprA homology model (green) with its template (cyan) show the wHTH motif in orange and residues Q51 and F58 as red sticks. The hypothesized binding sites of DU011, DNP, and salicylate are shown in colored circles matching the colors of the displayed chemical structures of the molecules. The two nonequivalent binding sites of DU011 are shown in red and blue; the precise location of the second site (shown in blue) is not known. The equivalent binding sites of DNP and salicylate on each subunit are shown in purple.

MATERIALS AND METHODS

Bacterial strains, plasmids, phage, and growth conditions.

All E. coli strains, plasmids, and phages used in the present study are listed in Table 3 (9, 26–28). Bacteria were grown in Luria-Bertani medium (LB) with shaking at 250 rpm at 37°C. Phage lysates were prepared from 50-ml cultures of E. coli strain UTI89 (for K1F phage) or MG1655 (for T7 phage) and stored at 4°C over drops of chloroform, as described previously (29).

TABLE 3.

E. coli strains, plasmids, and phages used in the present study

Strain, plasmid, or phage	Description	Reference or source
Bacteria
UTI89ΔmprA	mprA deletion mutant of K1 E. coli cystitis isolate	9
DH5α/pASKA-mprA	Protein expression vector with pASKA containing mprA	9
Plasmids
pASKA-mprA	mprA deletion mutant complemented with pASKA containing mprA	28
pMCSG7-mprA	mprA expression vector with N-terminal hexahistidine tag and TEV cleavage site	This study
Phage
K1F phage (K1Fφ)	K1 capsule-specific phage virus	27

HDX-MS.

HDX-MS was performed by the Proteomics and Metabolomics core at the Duke Center for Genomics and Computational Biology. The 6× His-tagged MprA was purified using HisPur cobalt resin (Thermo Fisher Scientific). MprA was incubated with and without DU011 in D₂O for 1, 5, or 10 min. After each time point, the reaction was quenched at pH 2.5 and the protein subjected to peptide cleavage. Mass spectrometry was used to measure the deuterium incorporation in peptide fragments.

Site-directed mutagenesis and identification of functional proteins.

Amino acid residues of interest identified from the HDX-MS were individually targeted for site-directed mutagenesis using the QuikChange Lightning site-directed mutagenesis kit (Agilent) according to the manufacturer’s protocol. Mutations were confirmed by Sanger sequencing by GENEWIZ, Inc. Plasmids with confirmed mutations were transformed into the UTI89ΔmprA strain by electroporation. The K1 phage assay was used to determine the presence or absence of the polysaccharide capsule, as previously described (20). The K1 phage is a lytic phage that is dependent on the polysaccharide capsule for the entry and lysis of the bacterial cell (27). Since the UTI89ΔmprA strain is unencapsulated, the K1 phage cannot lyse the bacteria, resulting in their continued growth in LB. When complemented with a plasmid encoding a functional MprA protein, the capsule is restored, allowing for the entry and lysis of the bacteria.

Homology model generation.

The Phyre2 web portal for three-dimensional protein structure prediction was used to generate a monomeric model of MprA from its sequence (30). The one-to-one threading function generated the model of MprA using MarR from Streptomyces coelicolor (PDB entry 3ZMD) as the template. This template was selected because it is one of the few solved structures of a MarR family member without chain breaks and with at least 19.2% sequence identity to MprA. From this template, 81% of MprA residues were modeled with 99% confidence.

MprA expression and purification.

The pMCSG7-mprA expression vector (Table 3) was transformed into C41(DE3) cells (Lucigen). Cultures (1.5 liters) were grown in Luria-Bertani broth and 50 μg/ml ampicillin until the optical density at 600 nm (OD₆₀₀) reached 0.5. MprA expression was induced with 0.5 mM IPTG (isopropyl-β-d-thiogalactopyranoside) for 3 h at 37°C. Cells were then centrifuged and resuspended in buffer A [50 mM Tris, pH 7.5, 250 mM NaCl, 10% glycerol, and 1 mM Tris (2-carboxyethyl)phosphine hydrochloride (TCEP) as a reducing agent] with protease inhibitor cocktail and 1 mg DNase I. The cells were lysed by sonication (Fisher Scientific), and clarified lysate was loaded onto a Ni²⁺-nitrilotriacetic acid (Ni-NTA) column. After washing the column with buffer A plus 25 mM imidazole, MprA was eluted in buffer A with 100 to 300 mM imidazole. SDS-PAGE was used to determine the purity of the eluted fractions; fractions containing pure MprA were combined and buffer exchanged to reduce the imidazole concentration to <10 mM. The protein was incubated with 1 mg of tobacco etch virus (TEV) protease overnight at 4°C. Cleaved MprA was separated on the Ni-NTA column and then loaded onto a Superdex 75 column. SDS-PAGE revealed dimeric fractions to be ≥95% pure.

ITC assays.

Purified MprA protein and ligands of interest were prepared in buffer containing 50 mM Tris, pH 7.5, 150 mM NaCl, 10% glycerol, 1 mM TCEP, and 3.6 to 7.5% dimethyl sulfoxide. Titrations with wild-type or point-mutated MprA in the cell and ligand in the syringe were performed using a VP-ITC microcalorimeter (Malvern Panalytical). The concentrations of protein and DU011 were 30 μM and 300 μM, respectively, for the ITC experiments conducted to study the binding of MprA (wild type and point mutated) to DU011. In our ITC experiments with the other two ligands, the concentrations of protein and salicylate were 70 μM and 2 mM, respectively; the concentrations of protein and 2,4-dinitrophenol (DNP) were 74 μM and 1.39 mM, respectively. Experiments were conducted at 25°C with a stirring speed of 307 rpm. Appropriate blank data subtraction and data analysis were executed using the program ORIGIN 5.0.

To ensure the validity of the unique two-site binding pattern observed for MprA-DU011 binding, we included several experimental variations and controls. We performed replicates of our initial experiment using a construct of MprA that included a TEV cleavage site to remove the N-terminal hexahistidine tag used for affinity chromatography to eliminate the possibility of this tag interacting with DU011. Furthermore, we performed experiments at three different temperatures, including 15, 25, and 30°C, and varied injection volume and spacing ranging from 3 to 5 min. Lastly, replicate experiments were performed by multiple individuals to ensure reproducibility. All of these experimental conditions yielded the same two-site binding pattern, with only minimal variations in K_d. Thus, we compared the binding data collected at 25°C as described for MprA-DU011 binding to the binding data collected at this same temperature for MprA-salicylate and MprA-DNP binding.