Screening,Identification,and Characterization of Mechanistically Diverse Inhibitors of the Mycobacterium Tuberculosis Enzyme,Pantothenate Kinase (CoaA)

Abstract

The authors describe the discovery of anti-mycobacterial compounds through identifying mechanistically diverse inhibitors of the essential Mycobacterium tuberculosis (Mtb) enzyme, pantothenate kinase (CoaA). Target-driven drug discovery technologies often work with purified enzymes, and inhibitors thus discovered may not optimally inhibit the form of the target enzyme predominant in the bacterial cell or may not be available at the desired concentration. Therefore, in addition to addressing entry or efflux issues, inhibitors with diverse mechanisms of inhibition (MoI) could be prioritized before hit-to-lead optimization. The authors describe a high-throughput assay based on protein thermal melting to screen large numbers of compounds for hits with diverse MoI. Following high-throughput screening for Mtb CoaA enzyme inhibitors, a concentration-dependent increase in protein thermal stability was used to identify true binders, and the degree of enhancement or reduction in thermal stability in the presence of substrate was used to classify inhibitors as competitive or non/uncompetitive. The thermal shift–based MoI assay could be adapted to screen hundreds of compounds in a single experiment as compared to traditional biochemical approaches for MoI determination. This MoI was confirmed through mechanistic studies that estimated K_ie and K_ies for representative compounds and through nuclear magnetic resonance–based ligand displacement assays.

Keywords

Tm shift MoI CoaA tuberculosis

Introduction

The World Health Organization estimated in 2010 that ~11.1 million people across the globe are infected with Mycobacterium tuberculosis in a given year, with an associated mortality of 1.3 million.¹ Furthermore, 4% of all cases harbored resistance to at least one frontline drug in the current anti–tuberculosis (TB) regimen.² This, coupled with the extended duration of treatment, makes it imperative to discover new drugs to combat tuberculosis.

Cofactor synthesis pathways are rich sources of potential drug targets, as a single cofactor is used by many proteins in a network. Coenzyme A (CoA), one such cofactor, is an essential and ubiquitous part of the metabolism of many organisms. A recent review by Spry et al.³ estimates that 9% of 3500 identified activities in the database BRENDA use CoA or a derivative as a cofactor.

In most bacteria, CoA is synthesized from pantothenic acid in five steps.³ The first committed step is the phosphorylation of pantothenate by the enzyme pantothenate kinase (PanK or CoaA), with adenosine triphosphate (ATP) serving as the phosphate donor. The PanK reaction is also the rate-limiting step in CoA biosynthesis and, in many bacteria and eukaryotes, feedback regulated by the product of the pathway, CoA, and its thioesters.⁴ Three distinct types of PanK have been characterized thus far. The type I PanK, exemplified by the Escherichia coli CoaA protein, was originally thought to represent the majority of PanK enzymes from prokaryotic sources.^5,6 Type II PanKs predominantly occur in eukaryotic organisms.^7,8 Recently, a third type of PanK was identified in bacteria (encoded by the gene coaX, to be distinguished from coaA). Type III PanKs have a much wider phylogenetic distribution than the better known type I PanKs and are present in 12 of the 13 major bacterial groups, as well as in many pathogenic bacteria.⁹

The Mtb genome has a single copy of the gene encoding the type I PanK, coaA. CoaA has been shown to be essential in M. tuberculosis by transposon mutagenesis¹⁰ and gene knockout experiments.¹¹ The Mtb genome also contains a type III PanK, coaX, but this gene has recently been shown to be nonessential.¹¹ Furthermore, there is low sequence homology between prokaryotic (type I) and eukaryotic (type II) pantothenate kinases.¹² Hence, an opportunity exists for the discovery of Mtb CoaA inhibitors that are specific to bacteria and have minimal interaction with the human enzyme.

The discovery of compounds with exquisite inhibitory effect against purified enzymes does not always translate into antibacterial activity. In addition to permeability and efflux issues, the target enzyme could also exist in different forms inside the cell, with potency optimization happening on the incorrect enzyme form. Kinetic analysis of the E. coli type I PanK indicates that substrates bind in an ordered fashion, with ATP binding first and pantothenate binding to the ATP–PanK complex.⁶ In-house nuclear magnetic resonance (NMR) experiments studying the order of addition of substrates using a nonhydrolysable analog of ATP (ATPγS) and pantothenate corroborate this observation for Mtb CoaA, arguing that Mtb CoaA could exist as free and ATP-bound forms inside the cell. We therefore explored the possibility of screening for mechanistically diverse inhibitors that could bind either the free enzyme (biochemically competitive with ATP), the ATP–CoaA complex (uncompetitive with ATP), or both (noncompetitive inhibitors).

Information on inhibitor mechanism is routinely obtained by detailed biochemical analysis, but this technique is limited in the numbers of compounds that can be characterized. Obtaining mechanistic information for the large numbers of hits produced by high-throughput screening (HTS) necessitates a technique with adequate throughput. In this article, we explore the modulation of inhibitor-enhanced protein thermal stability by the presence of enzyme substrates as a tool to screen for desired inhibitor mechanisms without compromising throughput. We further demonstrate the validity of this method by characterizing the modes of inhibition (MoI) of inhibitors thus discovered, using a variety of techniques.

Materials and Methods

Protein expression and purification

The gene coding Mtb pantothenate kinase (coaA) was cloned, expressed, and purified in E. coli as reported previously.¹³ Briefly, coaA was cloned in a pET15b vector under a T7 promoter and expressed as a Histidine₆-tagged protein in BL21(DE3) cells (Novagen, Gibbstown, NJ) at 37 °C. Protein was purified using metal affinity chromatography (Ni-NTA; Qiagen, Valenica, CA), followed by gel permeation chromatography (GPC) in buffer containing 50 mM Tris-HCl (pH 7.5), 300 mM NaCl, 10% glycerol, 2 mM dithiothreitol (DTT), and 0.1 mM EDTA. Protein aliquots were stored at −70 °C. Only once-thawed aliquots were used for each experiment.

Thermal stability measurements

Thermal stability measurements were recorded in 96-well PCR plates covered with optical film (Bio-Rad, Hercules, CA), using a real-time PCR thermal cycler (initially Mx3000 from Stratagene [La Jolla, CA], subsequently iQ5 from Bio-Rad). Protein solution at a concentration of 0.15 to 0.2 mg/mL in buffer containing 50 mM HEPES-Na (pH 7.5) and 100 mM NaCl in a 50-µL volume was mixed with compound at appropriate concentrations or with equivalent amounts of DMSO. Final DMSO concentration was maintained at 4% (v/v). Protein–DMSO or protein–compound mixes also contained 0.5 mM or 1 mM ATP as required.

The environmentally sensitive dye, SYPRO Orange (Invitrogen, Carlsbad, CA), was used to monitor protein unfolding at a final arbitrary concentration of “6.25X” (the dye is sold as a DMSO stock marked “5000X”). Thermal unfolding was monitored at 0.5 °C intervals from 25 °C to 85 °C. Thermal melting profiles were analyzed using a six-parametric equation (1) in GraphPad Prism version 3.0 (GraphPad Software, La Jolla, CA):

Y = \frac{e^{D} * (Au + Bu * (X+273)) + (An + Bn * (X + 273))}{1 + e^{D}},

(1)

where

D = - Δ H* \frac{1 - \frac{(X + 273)}{(Tm + 273)}}{8.314 * (X + 273)}

where Tm is the midpoint of the unfolding transition, ΔH is the enthalpy change, Au is the maximum value of fluorescence observed, An is the minimum value of fluorescence observed, Bu is slope of the unfolded baseline, and Bn is the slope of the folded baseline.

Enzyme assay for Mtb CoaA characterization

Mtb CoaA activity was measured using a pyruvate kinase–lactate dehydrogenase (PK-LDH) coupled assay. A previously reported assay¹⁴ was modified and adapted to the Mtb enzyme in the 96-well format. This assay has also been optimized in the 384-well format for the HTS campaign (unpublished data). All reactions were performed at 25 °C. Kinetic parameters for pantothenate and ATP using Mtb CoaA were determined in a 100-µL reaction mixture containing 50 mM PIPES–NaOH (pH 7.0), 25 mM KCl, 20 mM MgCl₂, 0.1 mM EDTA, 2 mM DTT, 0.002% Brij, 0.5 mM phosphoenol pyruvate (PEP), 0.24 mM NADH, 10 U/mL PK-LDH, and 20 nM Mtb CoaA (0.75 µg/mL). The reaction was started with the addition of Mtb CoaA to the reaction mix. Change in absorbance at 340 nm (NADH consumption) was monitored for 60 min in a SpectraMax (Molecular Devices, Sunnyvale, CA) spectrophotometer. K_m for ATP was determined by varying the concentration of ATP from 20 µM to 1 mM while maintaining D-pantothenate at a saturating concentration of 3 mM. K_m for D-pantothenate was determined by varying its concentration from 50 µM to 3 mM while maintaining ATP at a saturating concentration of 1 mM.

Inhibitor screening and kinetic parameters

Routine screening and IC₅₀ estimations for the Mtb CoaA inhibitors were carried out by keeping both substrate concentrations at their respective K_m under the buffer conditions mentioned above. For selected inhibitors, competition with ATP was checked by measuring IC₅₀ at a higher ATP concentration (50 × K_m = 6 mM). Compounds listed in Table 1 were further characterized for their kinetic MoI with respect to the ATP substrate. The inhibitory effect was monitored using an appropriate range of compound concentrations in the presence of varying amounts of ATP (1.5 mM to 11.7 µM) and a constant amount of D-pantothenate (3 mM).

Table 1.

Mtb CoaA Inhibitors Representative Compounds from Each Mechanistically Distinct Class Are Listed along with Their Inhibitory Properties

Initial rates in all the reactions were estimated using SoftMaxPro data analysis software (Molecular Devices). Inhibition constants were determined by fitting the data to four different equations, equations (2) to (5), for the competitive, pure noncompetitive, mixed noncompetitive, and uncompetitive modes of inhibition where K_ie and K_ies indicate the dissociation constants for free enzyme–inhibitor complex and substrate bound enzyme–inhibitor complex, respectively.

Competitive inhibition:

v = \frac{V_{max} * [ATP]}{10^{\log Km} (1 + \frac{[I]}{10^{\log K_{ie}}}) + [ATP]}

(2)

Pure noncompetitive:

v = \frac{V_{max} * [ATP]}{10^{\log Km} (1 + \frac{[I]}{10^{\log K_{ie}}}) + [ATP] (1 + \frac{[I]}{10^{\log K_{ie}}})}

(3)

Mixed noncompetitive:

v = \frac{V_{max} * [ATP]}{10^{\log Km} (1 + \frac{[I]}{10^{\log K_{ie}}}) + [ATP] (1 + \frac{[I]}{10^{\log K_{ies}}})}

(4)

Uncompetitive:

v = \frac{V_{max} * [ATP]}{10^{\log Km} + [ATP] (1 + \frac{[I]}{10^{\log K_{ies}}})}

(5)

GraFit 5.0.11 was used for nonlinear regression analysis. Eadie-Scatchard plots (v/[ATP] vs v) using the initial velocity data were used to discriminate between various modes of inhibition.

WaterLOGSY

NMR spectra were recorded at 298K on a Bruker AvanceII 500-MHz spectrometer equipped with a 5-mm triple resonance inverse probe. Reference spectra were recorded for all samples prior to each 1D WaterLOGSY experiment. Pulse program was from Dalvit et al.^15,16 The first water-selective pulse was 6 ms long, whereas the two water-selective 180° square pulses of the double spin echo scheme were 3 ms long. A weak rectangular pulsed field gradient was applied for the duration of the mixing time (1.8 s), which was followed by a gradient recovery time of 2 ms before the detection pulse. The experimental sweep width was 12 379 Hz, with an acquisition time of 1.3 s. Data were multiplied with an exponential function (line broadening 2 Hz) and then processed.

Results

Using changes in protein thermal stability to determine mode of inhibition

The thermal stability of an enzyme can be enhanced by the binding of a ligand, and the extent of stabilization is guided by ligand binding affinity and the number of binding sites. This enhancement in stability can be measured as a positive shift in the midpoint of the melting curve (Tm) in the presence of ligand. The magnitude of observed thermal shifts can be modulated by the addition of a second ligand. If the first ligand were an enzyme inhibitor and the second ligand a substrate, three scenarios are possible ( Figure 1 ; assumption: inhibitor and substrate do not have similar contributions to protein thermal stability on binding. Ligands bind preferentially to the native form of the protein.).

Figure 1.

Deriving mode of inhibition (MoI) using changes in the magnitude of thermal melting points upon binding of inhibitor in the presence and absence of substrate. Tm_EI, Tm observed on inhibitor binding to protein in the absence of substrate; Tm_E, Tm of protein alone; Tm_{EI_S}, Tm observed on inhibitor binding in presence of substrate; Tm_{E_S}, Tm of protein in presence of substrate; ΔTm, shift in Tm on inhibitor binding.

Scenario 1: Inhibitor with higher affinity is partially displaced by a substrate molecule competing to bind at the same site as inhibitor (competitive inhibitor). In such a scenario, the magnitude of an inhibitor-induced thermal shift in the presence of substrate is lowered as compared to the shift observed in its absence.

Scenario 2: Inhibitor preferentially binds to the substrate-bound form of the enzyme (uncompetitive inhibitor). In such a scenario, the magnitude of shift in the presence of substrate is higher as compared to the shift observed in its absence. For purely uncompetitive inhibitors, no Tm shift may be observed in the absence of substrate.

Scenario 3: Inhibitor binds equally to the free and substrate-bound forms of the protein (noncompetitive inhibitor). In such cases, the magnitude of shift in the presence of substrate is either higher as compared to the shift observed in its absence, or little difference is observed.

Effect of Mtb CoaA inhibitors on protein thermal stability

We conducted two HTS campaigns for inhibitors against the Mtb CoaA enzyme. The first screen (campaign 1) challenged a diverse library of ~70 000 compounds, whereas the second screen (campaign 2) challenged a larger library of ~1 000 000 compounds. Nonpromiscuous inhibitors (IC₅₀s independent of protein and detergent concentration in the enzyme assay and Hill slope ~1) were used for MoI studies described below.

Campaign 1 hits

Campaign 1 chemistry focused on triazole and quinolone scaffolds (see Table 1 for representative molecules). Molecules from these two chemical classes increased the thermal stability of Mtb CoaA ( Fig. 2a ), suggesting specific binding without enzyme denaturation. The increased thermal stability also correlated broadly with inhibitor potency (IC₅₀).

Figure 2.

Modulation of inhibitor-induced thermal shift in the presence of adenosine triphosphate (ATP). Compounds were assayed at 25 µM and at three ATP concentrations: 0 (•), 0.5 (□), and 1 (×) mM. (a) CoaA inhibitors from campaign 1 were competitive with ATP in the thermal melting–based mode of inhibition (MoI) assay wherein the Tm shift (ΔTm) decreased with increasing ATP concentration. (b) Inhibitors selected in campaign 2 displayed all three modes of inhibition in the presence of ATP. Compounds 32–34 and 36–40 were competitive with ATP, whereas compounds 21–30 were uncompetitive or noncompetitive with ATP.

In the presence of ATP, this stability enhancement was reduced, as evidenced by a decreased Tm shift when compared to the “no-ATP” condition. Figure 2a shows the progressive decrease in Tm shift in the presence of increasing concentrations of ATP. For example, the triazole 8 shows a Tm shift of 3.8 °C in the absence of ATP, but this value decreases to ~1 °C in the presence of 0.5 or 1 mM ATP. The quinolone 13 shows a Tm shift of 3.3 °C in the absence of ATP, a reduced shift of 1.3 °C in the presence of 0.5 mM ATP, and a further reduced shift of 0.4 °C in the presence of 1 mM ATP. Similar trends were observed for all compounds tested, and this was interpreted as reduction in inhibitor binding in the presence of ATP (i.e., inhibitor competes with ATP for binding to Mtb CoaA; scenario 1 in the discussion above).

Campaign 2 hits

Campaign 2 involved the screening of a larger, more diverse library and resulted in a greater number of potential chemistry start points. To identify hits with diverse mechanisms of inhibition, the high-throughput thermal stability-based MoI assay was used to classify inhibitors as competitive or un/noncompetitive with respect to ATP.

Figure 2b shows the behavior of a representative set of compounds from campaign 2 in the presence and absence of ATP. Unlike molecules from campaign 1, all inhibitors tested did not bind and enhance the thermal stability of CoaA in the absence of ATP. In the presence of ATP, CoaA inhibitors derived from campaign 2 displayed a variety of responses.

Set 1: The binding of compounds such as 36–40 decreased in the presence of ATP. These compounds could be competitive with ATP as discussed in scenario 1 above.

Set 2: Compounds such as 21–28 that had previously not shown binding now bound CoaA with Tm shift values increasing in the presence of ATP. These compounds could be uncompetitive with ATP as discussed in scenario 2 above.

Set 3: Compounds such as 31 showed increased binding in the presence of 1 mM ATP. Yet other compounds such as 35 that bound CoaA were unaffected by the presence of ATP (Tm shift values of <0.5 °C were not considered significant). Both these compounds could be noncompetitive with ATP as discussed in scenario 3 above.

Chemically desirable hits that were also not competitive with ATP in the thermal shift–based MoI assay were prioritized for chemistry expansion. Specifically, biaryl acetic acid and quinoline carboxamide scaffolds (represented in Fig. 2b by compounds 23, 27, 29, 35 and compounds 22, 24, respectively) were picked as start points in a hit-to-lead campaign.

To demonstrate that the thermal shift–based MoI assay reliably picks compounds with desirable MoI, inhibition kinetics was studied in greater detail for selected compounds synthesized during the structure–activity relationship (SAR) exploration of the biaryl acetic acid and quinoline carboxamide series ( Table 1 ). An uncompetitive inhibitor (thiazole class) and a competitive inhibitor (triazole class) were also included in the study for comparison.

Confirming mode of inhibition in enzyme kinetic assays

Mode of Mtb CoaA inhibition was elucidated using the PK-LDH coupled assay adapted to the 96-well plate format. Estimates of K_cat, K_m for ATP, and D-pantothenate were confirmed in four replicates during characterization of the Mtb enzyme and were reconfirmed during eight independent studies on mode of inhibition for the enzyme inhibitors using varying concentrations of ATP. Estimated values of K_cat, K_m for ATP, and D-pantothenate were 5 ± 2 s^–1, 122 ± 25 µM, and 395 ± 9 µM, respectively.

Enzyme kinetics data obtained in the presence of eight inhibitors ( Table 1 ) were modeled using equations (1) to (4). Inhibition constants, K_ie (indicative of the affinity for the free form of enzyme) and K_ies (indicative of the affinity for the substrate-bound form of enzyme) ( Table 1 ), were used along with Eadie-Scatchard plot analysis ( Fig. 3 ) to elucidate the MoI. The plot of v/[S] versus v for compound 20 showed lines intersecting just below the x-axis ( Fig. 3a ), indicating competitive mode with respect to ATP as a major mode of inhibition. This was supported by the significantly lower value of K_ie (0.02 µM) as compared to that of K_ies (1.16 µM). Competition with ATP was initially indicated by a decreased Tm shift in the presence of 25 µM of compound 20 and 1 mM ATP (0.5 °C), as compared to the Tm shift in the absence of ATP (1.2 °C) ( Fig. 2a , Table 1 ). It was also indicated by the loss of inhibition in an enzyme assay in presence of excess ATP. The IC₅₀ of compound 20 increased from 0.02 to 0.7 µM when ATP concentration was increased 50-fold.

Figure 3.

Eadie-Scatchard plots for Mtb CoaA inhibitors with various modes of inhibition. Plots are shown for three compounds from Table 1 and exemplify the three different modes of inhibition (MoIs) observed with respect to adenosine triphosphate (ATP) substrate. (a) The triazole 20 (competitive). Inhibitor concentrations used were 0 (o), 0.013 (Δ), 0.039 (■), 0.116 (□), and 0.347 (•) µM, respectively. (b) The thiazole 25 (uncompetitive). Inhibitor concentrations used were 0 (Δ), 1.56 (■), 3.125 (□), 6.25 (•), and 12.5 (o) µM, respectively. (c) The quinolone carboxamide 41 (mixed noncompetitive). Inhibitor concentrations used were 0 (Δ), 1.56 (■), 3.13 (□), 6.25 (•), and 12.5 (o) µM, respectively.

Similarly, the thiazole 25 was found to be uncompetitive with ATP, as evidenced by the intersection of v/[S] versus v plots on the y-axis ( Fig. 3b ) and the significantly higher value of K_ie (1.1 × 10¹⁸µM) as compared to K_ies (6.4 µM). Higher affinity for the ATP-bound form of the enzyme was also evident in an enzyme assay where the compound’s inhibitory effect was marginally improved by increasing ATP concentration from 1 × K_m to 50 × K_m. In the Tm shift analysis, addition of 25 µM of 25 to Mtb CoaA resulted in almost no Tm shift. However, binding of 25 to protein in the presence of 1 mM ATP resulted in a shift of 4 °C ( Fig. 2b , Table 1 ).

Kinetic data for the quinoline carboxamide 41 fitted best to the equation describing mixed noncompetitive mode of inhibition. K_ie and K_ies estimates for this compound were 16.2 and 5.8 µM, respectively ( Table 1 ), and the plots of v/[S] versus v were neither perfectly parallel to each other nor intersecting on the y-axis ( Fig. 3c ). Compound 41 showed a small Tm shift of 0.8 °C in the absence of ATP but a higher Tm shift of 1.6 °C in its presence ( Table 1 ). This is in accordance with a mixed mode of inhibition with higher affinity in the presence of ATP. The K_ies values for the other two compounds from the quinoline carboxamide class (42, 43) were twofold lower than that of K_ie, thus indicating a mixed mode of inhibition. Tm shifts were observed only in the presence of ATP for these two compounds, with compound 42 showing a Tm shift of 3.3 °C and compound 43 resulting in a shift of 5 °C in the presence of 0.5 mM ATP.

K_ie/K_ies estimates for the representatives of the biaryl acetic acid class (compounds 44, 45, 46) were very similar, again indicating a mixed noncompetitive mode of inhibition. As seen for compound 41, and in keeping with the observed kinetic behavior of these compounds, significant Tm shifts were seen in both the presence and absence of ATP ( Table 1 ).

The kinetic MoI data obtained for the above inhibitors thus confirmed the MoI indicated by the high-throughput thermal melting–based assay.

Mapping of inhibitor binding site by 1D NMR

A robust campaign to improve inhibitor potency is greatly aided by knowledge of the inhibitor binding pocket. In the absence of co-crystal structures, competition with substrate or an exemplary inhibitor often provides part of this information. During a large part of the two chemistry campaigns, we did not have co-crystal structures of Mtb CoaA with any inhibitor. Also, the substrate binding pocket of CoaA is an extended zone that accommodates both ATP and D-pantothenate or the feedback inhibitor, coenzyme A.^13,17,18 To map the binding sites of the various inhibitors in this extended pocket, a series of displacement experiments was planned. CoA or the inhibitors discovered from screening were sequentially used as probes to check if inhibitors bound Mtb CoaA in the extended pocket occupied by CoA and if the various inhibitor classes occupied the same binding site.

Figure 4 shows the binding to Mtb CoaA of a representative quinoline carboxamide, 41, by the ligand-observed NMR technique, WaterLOGSY.^15,16 Transfer of bulk water magnetization to small molecules resulted in negative ligand signals, which changed in sign on binding to the protein. This is seen in Figure 4a , b for 41, showing that the inhibitor bound the CoaA protein. WaterLOGSY experiments recorded in the presence of the known CoaA inhibitor, coenzyme A, resulted in reduced inversion of the quinoline carboxamide signals, demonstrating that coenzyme A could displace 41 ( Fig. 4c ). Similarly, the biaryl acetic acid, compound 44, could displace compound 41 ( Fig. 4d ), as could an ATP-competitive triazole (data not shown). However, this reduced or complete inversion was not observed when ATPγS was used instead of coenzyme A or compound 44 ( Fig. 4e ). The spectra in the presence or absence of ATPγS ( Fig. 4b , e ) appeared very similar, thus indicating that ATP derivatives could not displace 41. This is in accordance with the kinetic analysis for compound 41. Data shown in Table 1 and Figure 3c indicate that this compound shows mixed noncompetitive behavior with respect to the substrate ATP.

Figure 4.

WaterLOGSY experiments showing (a) free 41 and (b) binding of 300 µM 41 to 5 µM Mtb CoaA. (c) 1 mM of the inhibitor, coenzyme A, displaced 300 µM of the quinolone carboxamide 41, as shown by the reduced inversion of compound signals. (d) 60 µM of the biaryl acetic acid 44 displaced 300 µM of 41 and also showed some binding to Mtb CoaA. Note extra positive peaks observed in this spectrum. (e) 300 µM of 41 was not displaced by 1 mM ATPγS.

WaterLOGSY data obtained using the uncompetitive thiazole compound 25 are shown in Figure 5 . Binding of 25 to Mtb CoaA was indicated by the inversion of negative signals associated with free ligand ( Fig. 5b ). Binding of 25 to CoaA was much weaker than that of compound 41, as evidenced by the poor signal-to-noise (S/N) ratio seen in the NMR spectrum. Addition of coenzyme A results in the abrogation of signal inversion due to displacement of the thiazole ( Fig. 5c ). Again, Figure 5d shows that 25 was not displaced by ATPγS, as expected from its uncompetitive behavior.

Figure 5.

WaterLOGSY experiments showing (a) free thiazole 25 and (b) very weak binding of 300 µM of 25 (thiazole) to 5 µM Mtb CoaA. (c) Coenzyme A at 1 mM concentration displaced the inhibitor, as shown by the return to negative sign of 25 signals. (d) This compound was not displaced by 1 mM ATPγS.

The above NMR data were consistent with the observation that the quinoline carboxamide and thiazole classes of inhibitors were not competitive with ATP but interestingly also suggested that their binding site(s) overlapped with that of known ATP competitors such as coenzyme A and a triazole class of inhibitors. Crystal structures of Mtb CoaA in complex with triazole and biaryl acetic acid inhibitors were obtained as this article was being prepared, and the inhibitors were observed to occupy overlapping positions with coenzyme A and each other, as suggested by the NMR displacement experiments (T. Alwyn Jones, personal communication, 2008).

Discussion

Mtb CoaA is an essential bacterial enzyme that has been evaluated in this study as a target for the discovery of anti-TB drugs. Target-driven drug discovery is greatly aided by enzyme targets such as Mtb CoaA that are amenable to a range of biochemical and biophysical studies. Although this frequently leads to the discovery of well-characterized, potent inhibitors, such inhibitors may fail to demonstrate antibacterial activity due to a variety of reasons, such as poor permeability into or efficient efflux out of the bacterial cell. The enzyme may also exist in the cell in forms significantly different from that used to screen for inhibitors. In a target-based, enzyme potency–driven drug discovery campaign, it is not easy to predict permeability or efflux issues during screening or potency optimization phases. It is also difficult to accurately predict the biologically relevant form of the protein. One way to mitigate these risks could be to select chemical start points with maximum structural and mechanistic diversity.

In the specific case of Mtb CoaA as a target, ATP-competitive inhibitors may require very high potency to inhibit the enzyme inside the cell as the intracellular ATP concentration (~1 mM; unpublished data)^19,20 is in ~10-fold excess of its K_m for ATP. As per the Cheng-Prusoff relationship²¹ between IC₅₀ and K_i, a competitive inhibitor may be required at a much higher concentration (>10×) to achieve the same degree of inhibition as obtained by a noncompetitive or uncompetitive inhibitor with equivalent K_i values. Similar arguments have been put forth by Duggleby²² to overcome the phenomenon of “metabolic resistance,” where increasing substrate accumulation on enzyme inhibition by a competitive inhibitor ultimately reverses the effect of the inhibitor. In such cases, inhibitors that are uncompetitive or noncompetitive with accumulating substrate are more effective in the cellular context. Therefore, determination of MoI immediately after an HTS screen could be useful in either prioritizing compounds with desirable MoI or ensuring the inclusion of compounds with mechanistic diversity, thus increasing chances of success of obtaining antibacterial activity.

Thermal melting of proteins has been extensively used to explore protein function^23,24 and search for optimal buffer conditions for purification²⁵ or crystallography,²⁶ as well as in drug discovery to screen for small-molecule ligands that bind the drug target under consideration.^27-29 Changes in the thermal stability of a protein with varying concentrations of ligand have also been used to determine dissociation constants.^30,31 For ligands with similar binding thermodynamics, the magnitude of this change at a single ligand concentration can be used to rank molecules in order of affinity for target.³²

The above examples all stress the ease with which the measurement of protein thermal stability can be adapted to high-throughput formats. In addition to screening for protein ligands, as done by Pantoliano et al.,²⁷ early binding confirmation can be obtained for the large numbers of enzyme inhibitors picked as hits at the end of HTS campaigns by monitoring increases in protein thermal stability. In this article, we also demonstrate that comparing changes in protein thermal stability in the presence of inhibitor with such changes in the presence of both inhibitor and substrate is a quick, effective, and high-throughput tool to classify HTS hits according to their modes of inhibition (i.e., as either competitive or non/uncompetitive with the substrate). In some cases, such as quinoline carboxamides 42 and 43, Tm shifts were observed only in the presence of ATP, even though the kinetic assay indicated a mixed noncompetitive mode of inhibition. This indicates the Tm shift analysis can be used to broadly classify compounds into mixed or un/noncompetitive and competitive MoI. We were able to screen 30 compounds at three substrate/second ligand concentrations or 80 compounds at a single substrate concentration in a 96-well plate (other wells used for controls) in the thermal melting–based MoI assay, as opposed to a single compound analyzed for its MoI using kinetic analysis in a similar plate. Preliminary indications of MoI thus observed were confirmed by detailed kinetic analysis of selected inhibitors, as well as by displacement NMR experiments.

All four classes of Mtb CoaA inhibitors discussed in this study could be optimized to nanomolar potencies regardless of MoI (IC₅₀ ≤50 nM when both substrates of Mtb CoaA were maintained at K_m). To ensure the various mechanistic classes retained their indicated MoIs, thermal stability behaviors of compounds were tracked during the hit-to-lead campaign ( Fig. 2a —competitive inhibitors; Suppl. Fig. S1—mixed/noncompetitive inhibitors). Inhibitors that demonstrated anti-mycobacterial activity were from chemical series shown to not be competitive with ATP. Specifically, some biaryl acetic acid derivatives with IC₅₀s <1 µM had Mtb minimum inhibitory concentrations ranging between 4 and 16 µg/mL. Many of these derivatives had an increased Tm shift in the presence of ATP and hence behaved like the other members of this class discussed above (Suppl. Fig. S1).

Thus, mechanistic information on scaffolds at the onset of chemistry optimization can enhance the chances of identifying potent leads with enzyme inhibition and cellular activity. Such mechanistic information can be efficiently and reliably obtained for large numbers of potential scaffolds using the protein thermal stability-based MoI assay.

Footnotes

Acknowledgements

We thank Rajendra Rane for optimizing the Mtb CoaA assay used in inhibitor screening, Pavithra Vishwanath and T. Raga Deepthi for estimating the intracellular concentration of ATP in mycobacterial cells, and Sunita DeSousa for scientific discussions.

Supplementary material for this article is available on the Journal of Biomolecular Screening Web site at .

References

World Health Organization. Tuberculosis fact sheet n104. 2010. http://www.who.int/mediacentre/factsheets/fs104/en/

World Health Organization. Multidrug and extensively drug-resistant TB (M/XDR-TB): 2010 global report on surveillance and response. 2010. http://www.who.int/tb/publications/2010/978924599191/en/index.html

Spry

Kirk

Saliba

K. J.

Coenzyme A Biosynthesis: An Antimicrobial Drug Target. FEMS Microbiol. Rev. 2008, 32, 56–106.

Vallari

D. S.

Jackowski

Rock

C. O.

Regulation of Pantothenate Kinase by Coenzyme A and Its Thioesters. J. Biol. Chem. 1987, 262, 2468–2471.

Song

W. J.

Jackowski

Cloning, Sequencing, and Expression of the Pantothenate Kinase (CoaA) Gene of

Escherichia coli. J. Bacteriol. 1992, 174, 6411–6417.

Song

W. J.

Jackowski

Kinetics and Regulation of Pantothenate Kinase from

Escherichia coli. J. Biol. Chem. 1994, 269, 27051–27058.

Rock

C. O.

Calder

R. B.

Karim

M. A.

Jackowski

Pantothenate Kinase Regulation of the Intracellular Concentration of Coenzyme A. J. Biol. Chem. 2000, 275, 1377–1383.

Rock

C. O.

Karim

M. A.

Zhang

Y. M.

Jackowski

The Murine Pantothenate kinase (pank1) Gene Encodes Two Differentially Regulated Pantothenate Kinase Isozymes. Gene. 2002, 291, 35–43.

Yang

Eyobo

Brand

L. A.

Martynowski

Tomchick

Strauss

Zhang

Crystal Structure of a Type III Pantothenate Kinase: Insight into the Mechanism of an Essential Coenzyme A Biosynthetic Enzyme Universally Distributed in Bacteria. J. Bacteriol. 2006, 188, 5532–5540.

10.

Sassetti

C. M.

Boyd

D. H.

Rubin

E. J.

Genes Required for Mycobacterial Growth Defined by High Density Mutagenesis. Mol. Microbiol. 2003, 48, 77–84.

11.

Awasthy

Ambady

Bhat

Sheikh

Ravishankar

Subbulakshmi

Mukherjee

Sambandamurthy

Sharma

Essentiality and Functional Analysis of Type I and Type III Pantothenate Kinases of

Mycobacterium tuberculosis. Microbiology 2010, 156, 2691–2701.

12.

Leonardi

Chohnan

Zhang

Y. M.

Virga

K. G.

Lee

R. E.

Rock

C. O.

Jackowski

A Pantothenate Kinase from Staphylococcus aureus Refractory to Feedback Regulation by Coenzyme A. J. Biol. Chem. 2005,280, 3314–3322.

13.

Das

Kumar

Bhor

Surolia

Vijayan

Invariance and Variability in Bacterial PanK: A Study Based on the Crystal Structure of Mycobacterium tuberculosis PanK. Acta Crystallogr. D Biol. Crystallogr. 2006, 62, 628–638.

14.

Brand

L. A.

Strauss

Characterization of a New Pantothenate Kinase Isoform from

Helicobacter pylori. J. Biol. Chem. 2005, 280, 20185–20188.

15.

Dalvit

Pevarello

Tato

Veronesi

Vulpetti

Sundstrom

Identification of Compounds with Binding Affinity to Proteins via Magnetization Transfer from Bulk Water. J. Biomol. NMR. 2000, 18,65–68.

16.

Dalvit

Fogliatto

Stewart

Veronesi

Stockman

WaterLOGSY as a Method for Primary NMR Screening: Practical Aspects and Range of Applicability. J. Biomol. NMR. 2001, 21, 349–359.

17.

Ivey

R. A.

Zhang

Y. M.

Virga

K. G.

Hevener

Lee

R. E.

Rock

C. O.

Jackowski

Park

H. W.

The Structure of the Pantothenate Kinase-ADP-Pantothenate Ternary Complex Reveals the Relationship between the Binding Sites for Substrate, Allosteric Regulator, and Antimetabolites. J. Biol. Chem. 2004, 279, 35622–35629.

18.

Yun

Park

C. G.

Kim

J. Y.

Rock

C. O.

Jackowski

Park

H. W.

Structural Basis for the Feedback Regulation of Escherichia coli Pantothenate Kinase by Coenzyme A. J. Biol. Chem. 2000, 275, 28093–28099.

19.

Mathews

C. K.

Biochemistry of Deoxyribonucleic Acid–Defective Amber Mutants of Bacteriophage T4. 3. Nucleotide Pools. J. Biol. Chem. 1972, 247, 7430–7438.

20.

Buchholz

Takors

Wandrey

Quantification of Intracellular Metabolites in Escherichia coli K12 Using Liquid Chromatographic–Electrospray Ionization Tandem Mass Spectrometric Techniques. Anal. Biochem. 2001, 295, 129–137.

21.

Yung-Chi

Prusoff

W. H.

Relationship between the Inhibition Constant (KI) and the Concentration of Inhibitor Which Causes 50 Per Cent Inhibition (I50) of an Enzymatic Reaction. Biochem. Pharmacol. 1973, 22, 3099–3108.

22.

Duggleby

R. G.

The Application of Metabolic Resistance Theory to the Selection of Preferred Target Enzymes for Therapeutic Drugs. Comput. Biomed. Res. 1988, 21, 579–592.

23.

Carver

T. E.

Bordeau

Cummings

M. D.

Petrella

E. C.

Pucci

M. J.

Zawadzke

L. E.

Dougherty

B. A.

Tredup

J. A.

Bryson

J. W.

Yanchunas

Jr. . Decrypting the Biochemical Function of an Essential Gene from Streptococcus pneumoniae Using ThermoFluor Technology. J. Biol. Chem. 2005, 280, 11704–11712.

24.

Giuliani

S. E.

Frank

A. M.

Collart

F. R.

Functional Assignment of Solute-Binding Proteins of ABC Transporters Using a Fluorescence-Based Thermal Shift Assay. Biochemistry 2008, 47, 13974–13984.

25.

Mezzasalma

T. M.

Kranz

J. K.

Chan

Struble

G. T.

Schalk-Hihi

Deckman

I. C.

Springer

B. A.

Todd

M. J.

Enhancing Recombinant Protein Quality and Yield by Protein Stability Profiling. J. Biomol. Screen. 2007, 12, 418–428.

26.

Vedadi

Niesen

F. H.

Allali-Hassani

Fedorov

O. Y.

Finerty

P. J.

Jr. Wasney

G. A.

Yeung

Arrowsmith

Ball

L. J.

Berglund

. Chemical Screening Methods to Identify Ligands That Promote Protein Stability, Protein Crystallization, and Structure Determination. Proc. Natl. Acad. Sci. U. S. A. 2006, 103, 15835–15840.

27.

Pantoliano

M. W.

Petrella

E. C.

Kwasnoski

J. D.

Lobanov

V. S.

Myslik

Graf

Carver

Asel

Springer

B. A.

Lane

. High-Density Miniaturized Thermal Shift Assays as a General Strategy for Drug Discovery. J. Biomol. Screen. 2001, 6, 429–440.

28.

Cummings

M. D.

Farnum

M. A.

Nelen

M. I.

Universal Screening Methods and Applications of ThermoFluor. J. Biomol. Screen. 2006, 11, 854–863.

29.

Crowther

G. J.

Napuli

A. J.

Thomas

A. P.

Chung

D. J.

Kovzun

K. V.

Leibly

D. J.

Castaneda

L. J.

Bhandari

Damman

C. J.

Hui

. Buffer Optimization of Thermal Melt Assays of Plasmodium Proteins for Detection of Small-Molecule Ligands. J. Biomol. Screen. 2009,14, 700–707.

30.

Matulis

Kranz

J. K.

Salemme

F. R.

Todd

M. J.

Thermodynamic Stability of Carbonic Anhydrase: Measurements of Binding Affinity and Stoichiometry Using ThermoFluor. Biochemistry 2005, 44, 5258–5266.

31.

Maryanoff

B. E.

McComsey

D. F.

Lee

Smith-Swintosky

V. L.

Wang

Minor

L. K.

Todd

M. J.

Carbonic Anhydrase-II Inhibition: What Are the True Enzyme-Inhibitory Properties of the Sulfamide Cognate of Topiramate?

J. Med. Chem. 2008, 51, 2518–2521.

32.

M. C.

Aulabaugh

Jin

Cowling

Bard

Malamas

Ellestad

Evaluation of Fluorescence-Based Thermal Shift Assays for Hit Identification in Drug Discovery. Anal. Biochem. 2004, 332, 153–159.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.16 MB