Solvent-Assisted Slow Conversion of a Dithiazole Derivative Produces a Competitive Inhibitor of Peptide Deformylase

SUMMARY

Due to its potential as an antibiotic target, E. coli peptide deformylase (PDF_Ec) serves as a model enzyme system for inhibitor design. While investigating the structure-functional and inhibitory features of this enzyme, we unexpectedly discovered that 2-amino-5-mercapto-1,3,4-thiadiazole (AMT) served as a slow-binding inhibitor of PDF_Ec when the above compound was dissolved only in dimethylformamide (DMF), but not in any other solvent, and allowed to age. The time dependent inhibitory potency of the DMF-dissolved AMT was correlated with the broadening of the inhibitor's 295 nm spectral band toward the visible region, concomitant with the increase in the mass of the parent compound by about 2-fold. These data led to the suggestion that DMF facilitated the slow dimerization of AMT (via the formation of a disulfide bond), and that the dimeric form of AMT served as an inhibitor for PDF_Ec. The latter is not caused by the simple oxidation of sulfhydryl groups by oxidizing agents such as H₂O₂. Newly synthesized dimeric/dithiolated form of AMT (“bis-AMT”) exhibited similar spectral and inhibitory features as given by the parent compound when incubated with DMF. The computergraphic modeling data revealed that bis-AMT could be reliably accommodated within the active site pocket of PDF_Ec, and the above enzyme-ligand interaction involves coordination with the enzyme resident Ni²⁺ cofactor. The mechanism of the DMF-assisted activation of AMT (generating bis-AMT), the overall microscopic pathway for the slow-binding inhibition of PDF_Ec by bis-AMT, and the potential of bis-AMT to serve as a new class of antibiotic agent are presented.

INTRODUCTION

It has been well-established that in prokaryotes as well as in certain organelles of the eukaryotic system, the initiation of protein synthesis occurs via incorporation of a formylated methionine residue (^FMet) at the N-terminus of the nascent polypeptide chain [1–3]. For many proteins in the prokaryotic system, the ^FMet residue must be removed to elicit the desired physiological functions via a co-translational process referred to as the N-terminal methionine excision (NME) pathway [4–6]. The NME pathway involves two enzymes: peptide deformylase (PDF) and methionine aminopeptidase (MetAP), and these enzymes sequentially remove the N-terminal formyl group and the Met residue, respectively [7]. Since ^FMet is not utilized in the synthetic pathway of cytosolic proteins in eukaryotes, the absence of PDF in the physiological milieu of higher organisms has been widely justified. However, in recent years, eukaryotic homologues of PDF have been discovered in the chloroplasts and mitochondria of plants and animals including humans [3, 8–11]. Nevertheless, the physiological role of the human PDF homolog (which is translated in the cytoplasm and exported to the mitochondrion) is questionable in the light of the fact that proteins synthesized within mitochondria have been shown to retain their N-formyl groups [12, 13]. This feature has led to the suggestion that this enzyme is no more than a vestigial remnant of its ancestral counterpart with no or very little physiological function in normal human tissues [14].

Since the discovery of PDF in E. coli cells [15], it has been realized that the genomes of all bacterial species harbor at least one putative PDF gene [16, 17], and the expression of the enzyme has been shown to maintain the viability of many pathogenic bacteria [18–21]. This coupled with the fact that PDF is non-essential in normal human cells (but may be essential in malignant human cells) [11], is the basis for which the bacterial enzyme has been considered as a prototypical target for designing novel antibiotic agents [22]. In the development of PDF inhibitors as therapeutic agents, most research groups have used actinonin, a natural product inhibitor of the enzyme [23], as a template for the design of inhibitory pseudopeptides. However, due to poor pharmacokinetic profiles, actinonin [24] and many of its derivative compounds [25] do not serve as potent antibiotics under in vivo conditions. This is not surprising [26] since several environmental variables of the physiological milieu alter the efficacies of inhibitors for their putative target sites [27]. Irrespectively, a few actinonin analogues have been recently found to exhibit promising results in model animal systems, and they are currently under different phases of clinical trials [22].

In light of the structural-functional studies, it has been deduced that actinonin harbors a hydroxamate group that directly interacts with the active-site resident metal ion as well as a methionine-like sidechain at the P₁’ position that binds at the S₁’ subsite of the enzyme [28] (A diagram depicting the PDF_Ec subsites in relation to the enzyme’s substrate is depicted in Figure 6 in the Discussion [17, 29–32]). These structural moieties play an important role in the binding of the inhibitor by PDF, which has been shown to exhibit a binding affinity in the nanomolar range for various PDF isozymes [23, 33, 34]. Kinetic studies of the PDF-actinonin interaction led to the suggestion that the actinonin mediated inhibition of PDF occurs via the slow-binding mode, and the overall inhibition pathway conforms to the two-step binding mechanism [33]. As elaborated subsequently, the second step drives the overall equilibrium to yield the stable (isomerized) form of the PDF-actinonin complex, resulting in greatly enhancing the binding affinity of the enzyme-inhibitor complex and eliciting a potent inhibitory profile.

Figure 6.

Representations of bis-AMT and ^FMet-Leu-p-nitroanilide in complex with the active site of PDF_Ec. Panel A depicts the stereo representation of bis-AMT docked to the active site of PDF_Ec. Docking of bis-AMT to the structure of the Ni²⁺-substituted form of PDF_Ec (PDB ID: 1G2A) [28] was performed on a Microsoft Windows^®-based molecular modeling workstation using Autodock 4.0 [46, 47]. The figure was produced using the UCSF Chimera modeling package [49] and shows the best-docked (lowest energy) structure of bis-AMT (van der Waals surface shown in grid format) as well as active site residues of PDF_Ec which were located within 3.5 Å of the inhibitor. The purple lines indicate coordination bonds whereas the cyan lines indicate putative hydrogen bonds. Panels B and C depict the schematic representation of the putative “subsites” and relevant hydrogen bonding interactions within the active site pocket of PDF_Ec with either ^FMet-Leu-p-nitroanilide or bis-AMT bound (Panels B and C, respectively) [17, 29–32]. In both panels, the bound ligand is shown in black, with hydrogen bonds given by thin dashed lines and coordination bonds given by wide dashed lines.

In screening enzyme inhibitors, it is customary to dissolve putative compounds in DMSO or DMF without considering whether or not such solvents would have any effect on the structural integrity of the parent compounds. Although the above solvents are fairly non-reactive (particularly on the time scale of the screening process), they can react with parent compounds and modify them under prolonged storage condition. For example, DMF has been known to react with small molecules such as carboxylic acid esters [35] as well as phosphorus oxychloride, which is commonly exploited in the Vilsmeier-Haack reaction [36]. In addition, the use of DMF as a solvent has been shown to facilitate the formation of alkanesulfonothioates from alkanesulfinyl chlorides [37]. As will be elaborated in the following sections, we unexpectedly discovered that the solution of 2-amino-5-mercapto-1,3,4-thiadiazole (AMT) in DMF (but not in any other solvent) served as a slow-binding inhibitor of PDF_Ec upon aging. This communication elaborates on the plausible mechanism by which DMF activates AMT, and the resultant (dithiolated “bis-AMT” derivative) serves as a slow binding inhibitor of PDF_Ec.

EXPERIMENTAL PROCEDURES

MATERIALS

The commercially available (high purity grade) preparation of 2-amino-5-mercapto-1,3,4-thiadiazole (AMT) was purchased from TCI America, Portland, Oregon. All other reagents were of analytical reagent grade.

METHODS

Protein purification

The full-length, Ni²⁺-substituted form of E. coli peptide deformylase (PDF_Ec) was purified as described previously [38, 39], and the enzyme was judged to be homogeneous via SDS-PAGE analysis. The protein concentration of the purified PDF_Ec was determined via the Bradford method [40] using BSA as a standard and a correction factor of 0.56 as described by Rajagopalan et al [41].

Activity measurements

The activity of the recombinant PDF_Ec was measured via the Aeromonas aminopeptidase-coupled reaction using formyl-methionine-leucine-p-nitroanilide (^FMet-Leu-p-nitroanilide) as substrate [42]. The inhibition studies were performed on either a SpectraMax^® Plus (Molecular Devices) absorption microplate reader or a Lambda 3B (Perkin-Elmer) spectrophotometer. Initial screening of AMT in the various solvents was performed using 50 mM potassium phosphate buffer, pH 7.0, containing 10 – 100 nM PDF_Ec, 0.1 mg/mL BSA, 0.5 units coupling enzyme and 5 – 10% solvent (v/v). The enzyme reaction was initiated by addition of saturating concentrations of ^FMet-Leu-p-nitroanilide (200 – 300 µM) and the absorption was monitored for 10 min at 405 nm. The enzymatic assays for the slow-binding inhibition of PDF_Ec by DMF-activated AMT and freshly synthesized bis-AMT were performed in the above buffer using 250 – 500 µM substrate, 1 – 2.5 units of Aeromonas aminopeptidase, 0.1 mg/mL BSA and 7.5 – 12.5% DMF or DMSO (v/v). During these experiments, the concentration of inhibitor (20 – 1800 µM) was maintained to be much higher than the concentration of the enzyme (3.5 – 6 nM). The reactions were performed in duplicate and were monitored for 90 min at 405 nm after initiation by addition of PDF_Ec. Analysis of the resulting activity traces was according to the method of Sculley et al. [43]:

$${\text{Abs}}_{405}={\mathrm{v}}_1\times \mathrm{t}+[({\mathrm{v}}_0-{\mathrm{v}}_1)/{\mathrm{k}}_{\text{obs}}]\times (1-{\mathrm{e}}^{-{\mathrm{k}}_{\text{obs}}\times \mathrm{t}})+\text{offset}$$ Eq. 1

where v₀ and v₁ are the initial and final steady-state rates, respectively, and k_obs is the observed first-order rate constant for the transition from v₀ to v₁ in the presence of the inhibitor.

To determine the mechanism by which bis-AMT served as a slow-binding inhibitor, plots of k_obs as a function of inhibitor concentration were generated. Based on the hyperbolic dependence of k_obs on the inhibitor concentration, a two step binding mechanism conceived [43–44].

The experimental data were analyzed by the following equations:

$${\mathrm{k}}_{\text{obs}}={\mathrm{k}}_4+{\mathrm{k}}_3\times [\mathrm{I}]/({\mathrm{K}}_{\mathrm{i}}+[\mathrm{I}])$$ Eq. 2

$${\mathrm{k}}_{\text{obs}}={\mathrm{k}}_3\times \{{{\mathrm{K}}_{\mathrm{i}}}^{*}/({\mathrm{K}}_{\mathrm{i}}-{{\mathrm{K}}_{\mathrm{i}}}^{*})+[\mathrm{I}]/({\mathrm{K}}_{\mathrm{i}}+[\mathrm{I}])\}$$ Eq. 3

As per the above models, the binding of bis-AMT to PDF first results in the formation of PDF-bis-AMT collision complex (with a dissociation constant of K_i) followed by a slow isomerization process with forward and reverse rate constants being k₃ and k₄. Note that Eq. 3 was produced via rearrangement of Eq. 2 using the following relationship which describes the resultant effective inhibition constant: K_i^* = K_i × k₃/(k₃ + k₄).

To directly determine the magnitude of K_i^* for the DMF-activated AMT(bis-AMT), PDF_Ec (140 nM) was incubated overnight at 4 °C in the presence of 0.1 mg/mL BSA, 25% DMF (v/v) and increasing concentrations of the inhibitor (0 – 30 µM). Two independent incubations were performed for each inhibitor concentration. The following day, assays of PDF_Ec activity were performed as described above in the presence of the inhibitor at concentrations identical to that maintained in the overnight incubation. The reactions were started by the addition of aliquots of PDF_Ec from the overnight incubated mixture and the reaction progress was monitored for 60 min at 405 nm in the microplate reader.

To test for competitive inhibition of PDF_Ec by the DMF-activated AMT, the steady-state velocity of the PDF_Ec-catalyzed reaction was determined in the presence of 100, 250 or 500 µM inhibitor and increasing concentrations (25 – 300 µM) of substrate. The reaction mixtures were as described above and the reaction progress was monitored continuously for 1 hour at 405 nm after addition of PDF_Ec and the k_obs obtained via Eq. 1. Plots of the dependence of k_obs on substrate concentration were analyzed by Eq. 4 as described by Tian and Tsou [44].

$${\mathrm{k}}_{\text{obs}}={\mathrm{k}}_0/(1+[\mathrm{S}]/{\mathrm{K}}_{\mathrm{m}})$$ Eq. 4

where k₀ is the intrinsic rate constant for the inhibition of the enzyme in the absence of substrate and the K_m is the apparent binding affinity of substrate for the enzyme. Additionally, inhibition data were analyzed via double-reciprocal plots using increasing concentrations of substrate (10 – 100 µM) and the DMF-activated AMT (0 – 100 µM). The reaction was started by the addition of 22 nM PDF_Ec and the initial rate (v) was calculated from the initial 60 seconds of the reaction. The resulting data were plotted as 1 / v versus 1 / [S] and analyzed by simultaneous linear regression analysis (Origin^® 7.0 software) according to the competitive binding equation:

$$1/\mathrm{v}=1/{\mathrm{V}}_{\text{max}}+1/{[\mathrm{S}]}^{*}({{\mathrm{K}}_{\mathrm{m}}}^{*}(1+[\mathrm{I}]/{\mathrm{K}}_{\mathrm{i}}))/{\mathrm{V}}_{\text{max}}$$ Eq. 5

where K_m is the Michaelis constant, V_max is the maximal steady-state rate, and the K_i is the apparent inhibition constant.

The time dependent activation of AMT as a PDF_Ec inhibitor was ascertained by dissolving the compound in 100% DMF at a concentration of 20 mM at time zero (performed in duplicate). The stocks, along with a buffer blank and a previously activated 20 mM stock of AMT in DMF, were then diluted to 500 µM in 45 mM phosphate buffer, pH 7.0 containing 10% glycerol (v/v), 0.1 mg/mL BSA, 300 µM substrate, 1 unit coupling enzyme and a final concentration of 5% DMSO or DMF (v/v). The reactions were started by addition of PDF_Ec and were monitored at 405 nm for 1 hour. This procedure was repeated at various time intervals over the course of one week.

UV/Vis spectroscopy

All UV/Vis absorption spectra were measured on a Molecular Devices SpectraMax^® Plus microplate reader using either a quartz cuvette (1 cm path length) or a polystyrene microplate (path length approximately 0.3 cm). To compare the spectra of inactive AMT and bis-AMT, 200 µL of each (250 µM in 100% DMSO) was prepared and their spectra were obtained after subtraction of a solvent background.

To determine the time dependent changes in the absorption of AMT, the absorption spectrum of the inhibitor (543 µM) was measured in the microplate reader in 45 mM phosphate buffer, pH 7.0 containing 10% glycerol (final DMF concentration = 5%). The absorption values at 350 nm were plotted as a function of time or the corresponding k_obs value obtained at the same time point.

ESI-MS analysis

To determine the time dependent structural changes of active form of AMT, AMT samples were dissolved in 100% DMF (final concentration = 20 mM) at various time points over the course of two weeks. The samples were then diluted in dd-H₂O (0.1% HOAc) and immediately loaded into Agilent^® MSD Trap SL mass spectrometer with a flow rate of 0.3 mL/hr using a syringe pump (KDS 100, DK Scientific Inc.). MS settings for both full MS (m/z 50 to m/z 600) and MS² scans were: capillary voltage, −4500 V; nebulizer press, 15 psi; dry gas follow rate, 5 L/min; and dry temp 325 °C.

Synthesis of bis-AMT

To directly synthesize the disulfide form of AMT (i.e., bis-AMT), the synthetic protocol employed was that described by Mathes and Beber [45]. Briefly, AMT (500 mg) was placed in a glass test tube to which was added 1.67 mL ddH₂O while stirring to form a thick suspension. At this point, 180 mg NaOH was added and the mixture was stirred for 60 min at 4 °C. A solution of 539 mg of ammonium persulfate in 1.67 mL ddH₂O was prepared and added to the reaction in 200 µL increments. The resultant mixture was stirred for 30 min at 4 °C during which a bright yellow precipitate was formed. The precipitate was separated by filtration, washed by approximately 500 mL ddH₂O and dried under vacuum for 2 hours. The dried powder that remained was recrystallized using a mixed-solvent system of 40% ddH₂O in DMSO (v/v). The recrystallized powder was washed with ddH₂O and air dried prior to use.

Molecular Modeling

Docking of bis-AMT to the X-ray crystal structure of PDF_Ec (PDB ID: 1G2A) [28] was performed on a Microsoft Windows^®-based molecular modeling workstation utilizing Autodock 4.0 [46, 47]. The parameters for the active site Ni²⁺ were set as follows: r = 1.170 Å, q = +2.0 and van der Waals well depth of 0.100 kcal mol⁻¹ [48]. A grid of 50 × 50 × 50 points was employed with a spacing of 0.375 Å centered on the active site of the enzyme. Using the Lamarckian genetic algorithm (LGA) as the search engine, 30 independent runs with a maximum number of 27,000 LGA operations were performed on a population size of 150 individuals. The maximum number of energy evaluations was set to 2,500,000 with a step size of 1.0 Å. All other parameters were left as the Autodock default values. The structure of bis-AMT was prepared and energy minimized in ChemBioDraw^® 11 using the MM2 force field prior to docking. Structural depictions of the docked complexes were produced using the UCSF Chimera modeling package [49] from the Resource for Biocomputing, Visualization, and Informatics (University of California, San Francisco; supported by NIH P41 RR-01081).

RESULTS

Discovery of 2-amino-5-mercapto-1,3,4-thiadiazole as a slow-binding inhibitor of E. coli PDF

While undertaking kinetic studies with E. coli peptide deformylase (PDF_Ec), we accidently discovered that 2-amino-5-mercapto-1,3,4-thiadiazole (AMT; Figure 1A) serves as an inhibitor of the enzyme when the latter was dissolved in DMF and the resultant solution was stored for several days. No inhibitory effect of AMT was observed when this solution was freshly prepared in DMF or when the compound was dissolved in any other solvent and allowed to age at 25 °C for any length of time (Table 1). This unexpected observation led to the suggestion that DMF somehow caused slow “activation” of AMT, and the activated product served as an inhibitor of PDF_Ec. To ascertain the mechanistic basis of the DMF-assisted activation of AMT, we determined the time courses of the PDF_Ec-catalyzed reaction as a function of the incubation time of AMT in DMF (Figure 2, Panel A). Note that when AMT was incubated in DMF only for 30 min (open circles), the initial rate of the enzyme catalyzed reaction was barely lower than that of the control (solid circles). On the other hand, when the above preparation was allowed to age for 7 days, the initial rate of the enzyme catalysis decreased by about 50%. To our further interest, the overall reaction progress curve exhibited a downward curvature, as is apparent by the open triangles in Figure 2A. This has been in marked contrast to the linear increase in the reaction progress curve when AMT was dissolved in other solvents (Table 1) or briefly incubated in DMF. The downward curvature in the steady-state reaction profile was suggestive of the fact that the DMF-modified AMT served as a slow-binding inhibitor of the enzyme, a feature similar to that observed for the binding of actinonin (one of the highly potent inhibitors of the enzyme) [33] as well as other inhibitors of PDF [50–56].

Figure 1.

Chemical structures of 2-amino-5-mercapto-1,3,4-thiadiazole (AMT) and bis-AMT in Panels A and B, respectively.

Table 1.

Inhibition of PDF_Ec by AMT when dissolved in selected solvents^a

Solvent	Initial rate (mOD min−1)	% of control
Control (no inhibitor or solvent)	16.9	100
50 mM potassium phosphate, pH 7.0	17.5	103
Ethanol	16.9	100
Acetonitrile	16.3	96
Dimethylsulfoxide	15.9	94
Dimethylformamide (immediately)	17.0	101
Dimethylformamide (1 week post)	7.9	47

^aAll inhibitors were tested at a concentration of 500 µM after more than 2 days incubation in their respective solvent unless stated otherwise. Note that no change was observed at any time of AMT incubation in any solvent except DMF.

Figure 2.

Steady-state and transient kinetic analysis of the inhibition of PDF_Ec activity by AMT in the presence of fixed and varied concentrations of substrate. Panel A depicts the inhibition of PDF_Ec activity by 500 µM AMT as a function of increasing incubation time of the inhibitor in DMF. The reaction progress curves of the PDF_Ec-catalyzed reaction are shown in the absence (solid circles) and presence of 500 µM AMT incubated in DMF for 30 min (open circles) and 7 days (open triangles). The solid lines represent the linear fit of the initial portion of the reaction progress curves, whereas the dashed line represents the best fit of the data by Eq. 1. Panel B shows the dependence of k_obs (open circles) on the concentration of DMF-activated AMT and represent the average k_obs value from two independent assays. The dashed line represents the best fit of the data by Eq. 2, which provided values of 6.4 × 10⁻⁴ and 1.3 × 10⁻⁴ sec⁻¹ for k₃ and k₄, respectively, and a K_i value of 190 µM. The solid line represents the best fit of the data by Eq. 3 (with the value for K_i^* fixed at 3.7 µM) and provided values of 7.3 × 10⁻⁴ and 2.2 × 10⁻⁵ sec⁻¹ for k₃ and k₄, respectively, as well as a K_i value of 129 µM. The inset of panel B shows the reaction progress curves for the PDF_Ec-catalyzed reaction in the presence of 150 µM (open circles), 300 µM (solid squares), 600 µM (open squares) and 1200 µM (solid triangles) inhibitor with the solid lines depicting the best fit of the data by Eq. 1. Panel C shows the dependence of k_obs for the PDF_Ec-catalyzed reaction in the presence of 250 µM DMF-activated AMT as a function of ^FMet-Leu-p-nitroanilide concentration. The solid line indicates the best fit of the data by Eq. 4 which provided a value of 0.0027 sec⁻¹ for k₀ and a K_m of 83 µM. The double-reciprocal plot for the PDF_Ec-catalyzed reaction is given in the inset of panel C at concentrations of 0, 30, 50 and 100 µM AMT (open squares, solid squares, open triangles and solid circles, respectively). The solid lines of the inset represent the simultaneous fit of all the datasets by Eq. 5, which provided values of V_max and K_m of 0.65 µM sec⁻¹ and 54 µM, respectively, as well as a K_i of 170 µM.

Microscopic pathway for the slow-binding inhibition of PDF_Ec

We investigated the effect of AMT (prepared in DMF and stored for an extended period of time) on the steady-state kinetic profile (monitored at 405 nm) for the PDF_Ec-catalyzed reaction. These experiments were performed under the pseudo first-order condition for both substrate and AMT. As shown in the inset of Figure 2, Panel B, as the concentration of AMT increases the initial rate of the enzyme catalyzed reaction decreases, and the overall reaction profile exhibits rapidly attaining downward curvature. These observations led to the suggestion that the DMF-activated AMT served both as a conventional (i.e., the rate of inhibition occurred on the same time scale as the catalytic event) as well as the slow-binding inhibitor of the enzyme [43].

The latter feature is clearly apparent by the fact that the increasing concentration of the DMF-activated AMT enhances the rate of transition from the first (initial) steady-state to the second (final) steady-state. The main panel of Figure 2 (Panel B) shows the dependence of k_obs on the AMT concentration. Note the hyperbolic dependence of k_obs on AMT concentration, which supports the two-step (as opposed to one step) model of slow-binding inhibition (see Discussion). The data of Figure 2, Panel B was analyzed by the model of Eq. 2 (dotted line), with k₃ and k₄ being equal to 6.4 and 1.3 × 10⁻⁴ sec⁻¹, respectively, and K_i equal to 190 µM. It should be emphasized that since the magnitude of k₄ is obtained from the extrapolation of the hyperbolic curve of the figure to zero concentration of the inhibitor, it is not very reliable. To circumvent this, recourse was made to analyze the data of Figure 2, Panel B by Eq. 3 (solid line) by fixing the magnitude of K_i^* as being equal to 3.7 µM. This latter value was obtained by directly measuring the steady-state velocity of the PDF_Ec-catalyzed reaction at equilibrium (achieved after overnight incubation of different concentrations of AMT with DMF) as function of the activated concentration of AMT (data not shown). The analysis of the data provided values of k₃ and K_i of 7.3 × 10⁻⁴ sec⁻¹ and 129 µM, respectively, from which k₄ could be derived (via the relationship K_i^* = K_i × k₃/(k₃ + k₄)) to be 2.2 × 10⁻⁵ sec⁻¹.

To ascertain the effect of substrate (^FMet-Leu-p-nitroanilide) on the PDF_Ec-catalyzed inhibition by AMT [57], we examined the effect of increasing concentrations of substrate on k_obs for the AMT dependent inhibition of the enzyme (Figure 2, Panel C). Note that as the concentration of substrate increases, k_obs decreases in a hyperbolic manner, consistent with the competitive mode of binding of the inhibitor with respect to the substrate. The analysis of the data by Eq. 4 yielded the magnitudes of k₀ and K_m as being equal to 0.0027 sec⁻¹ and 83 µM, respectively. Moreover, the inset of Figure 2, Panel C shows the double-reciprocal plot for the initial velocity (v₀) of the PDF_Ec-catalyzed reaction in the presence of increasing concentrations of DMF-activated AMT at changing fixed concentrations of the substrate. The figure reveals intersecting linear-regression lines at the Y-axis, which is indicative of a competitive mode of inhibition, and global analysis of the data by Eq. 5 yielded the K_m, V_max, and K_i values to be 52 µM, 0.65 µM sec⁻¹, 170 µM, respectively. .

Electronic spectral changes of AMT upon incubation in DMF

We noted that the freshly prepared solution of AMT in DMF was initially colorless, but its color changed to pale yellow upon aging. This observation was suggestive of the fact that the electronic structure of AMT underwent slow changes in the DMF solution. We purported to investigate whether the time dependent changes in the electronic structure of the DMF-dissolved AMT were related to its increased inhibitory potency for PDF_Ec.

Figure 3, Panel A shows the absorption spectrum of AMT as a function of different time of its incubation in DMF. Note that at zero time of incubation, the absorption spectrum of AMT was characterized by a peak at 295 nm. As the incubation time increased, the absorption peak at 295 nm started to decrease in intensity concomitant with the red shift of the absorption profile to 350 nm region. Panel B of Figure 3 depicts the increase in the absorption of the inhibitor as function of incubation time in DMF. Evidently, AMT underwent slow structural modification (see below) in DMF solution. Moreover, to our utmost interest, the increase in absorption of AMT at 350 nm was mirrored by the enhancement in inhibitory potency (given by k_obs) of the compound toward PDF_Ec as a function of the aging time of the compound in DMF solution (Figure 3, Panel C). The similar time dependent profiles supported the premise that the increase in the inhibitory potency of AMT and the changes in its electronic structure were due to the formation of the modified form of the compound (see below). This is further evident by the fact that the observed rate constant for the slow inhibition (k_obs) of the enzyme was apparently linearly correlated with the absorption of AMT at its corresponding time point (Figure 3, Panel D). Clearly, the DMF-modified form of AMT, but not its native (unmodified) form, served as the slow-binding inhibitor of PDF_Ec.

Figure 3.

Time dependent changes in the UV/Vis absorption spectrum and k_obs of AMT as a function of incubation time in DMF. Panel A shows the spectrum of AMT in 45 mM phosphate buffer, pH 7.0 containing 10% glycerol (final DMF concentration = 5%) after different times of incubation in DMF, with the arrows indicating the time dependent changes in the spectrum. Panel B depicts the dependence of the absorption at 350 nm as a function of incubation time in DMF and represents the average of two independent measurements. Panel C shows is the dependence of the k_obs value for the inhibition of PDF_Ec by 500 µM AMT (calculated from the average of two independent measurements at corresponding time-points) as a function of incubation time in DMF. Panel D depicts the correlation between the absorption at 350 nm and the k_obs at corresponding time-points.

Nature of the structural modification of AMT upon incubation in DMF

To ascertain as to the nature of structural modification of AMT upon incubation in DMF, we subjected the above solution to ESI-MS analysis. Figure 4, Panel A shows the full MS spectrum of AMT after incubation in DMF for 14 days. The minor peaks of m/z 74 and m/z 147 appear to be the protonated form of DMF (M_DMF + H⁺) and the protonated DMF dimer (2M_DMF + H⁺), while the peaks of m/z 96 and m/z 169 corresponded to cationized DMF (M_DMF + Na⁺) and cationized DMF dimer (2M_DMF + Na⁺), respectively. On the other hand, the major peak of m/z 265 corresponded to the dimeric form of AMT with an added proton (264 + 1 = 265). This was confirmed by the observation of fragmentation ion of m/z 134 (M_AMT + 1) in MS² analysis of m/z 265, as depicted in the inset of Figure 4, Panel A.

Figure 4.

Mass spectral analysis of AMT upon incubation in DMF. Panel A shows the mass spectrum of AMT (in units of ion current) after 15 days incubation in DMF, with the mass to charge ratio (m/z) indicated by the numbers located above each peak. The inset of Panel A indicates the fragmentation spectrum of the m/z 265 species determined via MS² analysis. Panel B shows the intensity of the m/z 265 peak as a function of incubation time of AMT in DMF.

Based on the initial ESI-MS based assessment of the dimerization of AMT upon incubation in DMF (m/z = 265), we conjectured that the intensity of the above peak would increase as a function of the incubation time. To substantiate or refute the above possibility, we performed the ESI-MS experiment with AMT samples incubated for different times in DMF, and determined the intensity of the m/z 265 peak as a function of the incubation time. As shown in Figure 4, Panel A, there is no m/z 265 peak when AMT was freshly dissolved in DMF (i.e., at zero incubation time). However, as the incubation time increased, the intensity of the above peak started to increase in a linear fashion (Figure 4, Panel B). This trend has been similar to the increase in absorption of AMT solution in DMF at 350 nm (Figure 3, Panel B) coupled with the increase in the magnitude of k_obs for the slow-binding inhibition of the enzyme (Figure 3, Panel C). Hence, the AMT was slowly converted to a dimeric form (bis-AMT) upon incubation with DMF, which showed an increase in absorbance at 350 nm, and was responsible for the inhibition of the enzyme.

We further considered the possibility whether or not the DMF assisted formation of bis-AMT was due to simple oxidation of sulfhydryl group by oxidizing agent present in the organic solvent. To probe this, we incubated AMT with 0.06% H₂O₂ in buffer and measured the inhibitory potency of this reaction mixture for the enzyme (data not shown). As no inhibition of PDF_Ec was observed by the reaction mixture containing H₂O₂, it appeared evident that H₂O₂ did not promote dimerization/dithiolation of AMT to produce bis-AMT and elicit its inhibitory feature. This coupled with the fact that no other organic solvent (i.e., other than DMF; Table 1) promoted the formation of bis-AMT, we surmise that the latter species is not produced by contamination of some unforeseen oxidizing agent (e.g., peroxides) in the DMF solution.

Confirmation of bis-AMT being the inhibitory species for PDF_Ec

In pursuit of unraveling the molecular nature of the putative dimeric form of AMT which was responsible for inhibiting PDF_Ec, we surmised that DMF might promote the formation of the disulfide bond between the two AMT moieties to generate bis-AMT. The latter compound (shown in Panel B of Figure 1) would exhibit a mass of 264 daltons, which is in excellent agreement with the observed m/z ratio of 265 (M_bis-AMT + 1) obtained from the ESI-MS analysis of the DMF-activated inhibitor (Figure 4, Panel A).

To confirm that the dithiolated bis-AMT was the inhibitory species, we synthesized the latter compound as described by Mathes and Beber [45]. A visual inspection of the crystalline form of the freshly synthesized bis-AMT vis a vis its monomeric counterpart (i.e., parent AMT) revealed that the former was bright yellow in color as compared to the pale yellow color of the parent compound. This observation was reminiscent of the color change of AMT upon incubation with DMF (see Figure 3).

Figure 5, Panel A shows the comparative absorption spectra of AMT and newly synthesized bis-AMT (each at a concentration of 250 µM in DMSO). The solid and dashed lines are the spectra of AMT and bis-AMT, respectively. Although the absorption spectra of both these compounds are different in DMSO as compared to that observed when the activated compound was diluted in aqueous solution (containing 5% DMF; Figure 3, Panel A), it is evident that the spectral band of bis-AMT is considerably broad around the 400 nm region and is slightly red-shifted as compared to its monomeric counterpart. These spectral features are qualitatively similar to those observed during the course of the DMF-assisted transformation of AMT to bis-AMT (Figure 3, Panel A).

Figure 5.

UV/Vis absorption spectrum and inhibition properties of bis-AMT. Panel A shows the overlay of the monomeric and disulfide forms of AMT at 250 µM in DMSO (solid and dashed lines, respectively). Panel B depicts the transient-kinetic analysis of the slow-binding inhibition of PDF_Ec by bis-AMT, where the open circles depict the dependence of k_obs on the concentration of inhibitor (as determined by the best fit of the activity traces by Eq. 1). The solid line represents the best fit of the data by Eq. 2, which provided values of 8.0 × 10⁻⁴ and 3.6 × 10⁻⁶ sec⁻¹ for k₃ and k₄, respectively, and a K_i value of 490 µM.

Finally, we evaluated the efficiency of newly synthesized bis-AMT in inhibiting PDF_Ec. This was achieved by measuring the activity of PDF_Ec in the presence of increasing concentrations of bis-AMT. Note that the stock of bis-AMT was prepared in DMSO, which did not activate AMT to serve as the PDF_Ec inhibitor. The steady-state reaction profiles of PDF_Ec in the presence of increasing concentrations of bis-AMT were similar to those observed for the DMF-activated AMT (data not shown), suggesting that bis-AMT rather than its parent compound (AMT) served as the slow-binding inhibitor of PDF_Ec. The k_obs values derived from such plots (by analyzing the raw data by Eq. 1) exhibited a hyperbolic profile (Figure 5, Panel B), which is similar to the data of Figure 2, Panel B. The analysis of the data yielded the magnitudes of k₃ and k₄ being equal to 8.0 × 10⁻⁴ and 3.6 × 10⁻⁶ sec⁻¹, respectively, and the K_i value of 490 µM. Using these values, a K_i^* of 2.2 µM was obtained via the relationship K_i^* = K_i × k₃/(k₃ + k₄), which is in excellent agreement with the K_i^* of 3.7 µM determined directly for the DMF-activated AMT at equilibrium.

DISCUSSION

The experimental data presented in the preceding sections reveals, for the first time (to the best of our knowledge), that 2-amino-5-mercapto-1,3,4,-thiadiazole (AMT) is able to serve as a competitive, slow-binding inhibitor of E. coli peptide deformylase (PDF_Ec) when it is “activated” by DMF as a solvent. The DMF-mediated activation involves the dimerization of two AMT molecules via the dithiol linkage forming bis-AMT, which serves as a slow-binding inhibitor of the enzyme. This conclusion has been corroborated by the fact that directly synthesized bis-AMT exhibits similar spectroscopic and inhibitory features as those obtained with the DMF-incubated parent compound (AMT).

Since the activation/dimerization of AMT in DMF was extremely slow, it appeared likely that some intermediary species would predominate during the course of the overall process. However, this expectation was not supported by the ESI-MS profiles (Figure 4). This coupled with the fact that the activated form of AMT is the dimer of the parent compound leads to the suggestion that DMF, at the best, functions as a slow catalyst. A tentative sequence of steps in involved in the overall dimerization reaction is in Scheme 1. With literature precedents [32–34], we believe DMF promotes initial formylation of AMT, which serves as a site for bridging the two AMT moieties and juxtaposing their thiol groups. These thiol groups are slowly oxidized to produce the dithiolated form of AMT (bis-AMT) with concomitant loss of formic acid. We are currently in the process of probing the existence of putative intermediates as well as the products, and we will report these findings subsequently.

Scheme 1.

Proposed mechanism for the DMF-mediated activation of AMT to bis-AMT.

Our experimental data clearly reveal that bis-AMT serves as a slow-binding inhibitor of PDF_Ec and that the overall microscopic pathway conforms to the two (instead of one) step binding process. Whereas the latter case is essentially a simple one-step binding mechanism (with a kinetically slow rate of binding defined by the magnitudes of the microscopic rate constants, k₁ and k₂), the two-step mechanism represents the condition where the slow-binding process is biphasic in nature and consists of a relatively rapid collision phase (E + I ⇌ EI) followed by the kinetically slow isomerization phase (EI ⇌ EI^*) [43]. Note that given the magnitudes of K_i and K_i^* as being equal to 129 and 3.7 µM, respectively, it is significant that the slow-binding step enhances the apparent binding affinity of AMT for the enzyme by about 30 fold.

It should be pointed out that the two-step binding of inhibitors to various PDF isozymes is fairly common. In fact, the most prominent inhibitor of PDF, actinonin, has been reported to bind via a two-step mechanism which has been characterized for the S. aureus enzyme [33]. The binding of actinonin to Ni²⁺-substituted S. aureus PDF (PDF_Sa) features microscopic rate constants for the EI to EI^* conversion, k₃ and k₄, of 0.033 and ≤ 9.8 × 10⁻⁶ sec⁻¹, respectively [33]. When compared to the values obtained herein for bis-AMT (8.0 × 10⁻⁴ and 3.6 × 10⁻⁶ sec⁻¹ for k₃ and k₄, respectively) it is apparent that whereas the value of k₄ is of the same order of magnitude, the value of k₃ is 40 fold higher for the PDF_Sa-actinonin interaction. This explains the slower onset of full inhibition (viz. isomerization phase) in the case of bis-AMT, which features a t_1/2 of approximately 14 minutes at saturating concentrations of the inhibitor. As a result of the relatively small value of k₃ for the binding of bis-AMT to PDF_Ec, the isomerization phase does not pull the overall equilibrium from the collision complex (E + I ⇌ EI) to the isomerized complex (EI ⇌ EI^*). Van Aller et al. have reported that the difference in binding energy (ΔΔG°) between the two phases is equal to −4.5 kcal mol⁻¹ at 25 °C [33]. Using a standard state of 1.0 M, a ΔΔG° of −3.2 kcal mol⁻¹ was obtained for bis-AMT upon isomerization via the relationship, ΔΔG° = RT ln K_i^*/K_i. It should be noted that although the reported ΔΔG° for actinonin is 1.3 kcal mol⁻¹ lower than that for bis-AMT, in the latter case, the isomerization phase is responsible for more than 40% of the total −7.5 kcal mol⁻¹ of binding energy (based on an average K_i^* of 3.0 µM) as opposed to approximately 30% for actinonin. This demonstrates that the inhibition of PDF_Ec by bis-AMT is strongly dependent upon the isomerization phase for the overall energy of the binding interaction.

The origin of the two-step slow-binding mechanism for enzyme-inhibitor interactions is generally attributed to an initial binding event followed by a conformational change in the structure of the enzyme-inhibitor complex which gives rise to the isomerization phase [58–60]. This is equivalent to the induced-fit model of protein-ligand binding, with the distinction that the adjustment of the protein conformation takes place secondary to the initial binding event. However, it has been indicated in the case of actinonin that structural changes which are responsible for the conversion from EI to EI^* are not evident upon comparison between the bound and free forms of PDF [32, 33]. This observation, coupled with the recent delineation of at least two coexisting conformational states of PDF_Ec (both of which bind actinonin) [34] prompts the question as to whether the slow-binding inhibition of PDF by compounds such as bis-AMT is truly associated with a major conformational change of the enzyme-inhibitor complex or subtle (albeit slow) adjustment in the protein-ligand structure. Revelation of these features must await further structural-functional investigations of the PDF_Ec in the presence of bis-AMT.

In agreement with all other PDF inhibitors characterized [25], the data of Figure 2 demonstrates that the inhibition of PDF_Ec by bis-AMT is competitive in nature. As the majority of PDF inhibitors are peptides or pseudo-peptides that mimic the natural ^FMet-peptide substrates, much of the binding energy for these inhibitors is proposed to arise due to enzyme-inhibitor contacts that are similar to those of the native substrates [25]. However, as bis-AMT is structurally dissimilar from most other PDF inhibitors, it was unclear as to how it served as a competitive inhibitor of the enzyme. To elucidate this feature, we performed molecular modeling studies for docking bis-AMT to the active site of PDF_Ec (PDB ID: 1G2A) [28] via Autodock 4 (see Experimental section). Panel A of Figure 6 depicts the stereo view of the best-docked structure of bis-AMT at the active site of the enzyme. A casual perusal of the docked structure of Figure 6, Panel A reveals several interesting features of the enzyme-inhibitor complex. The most notable structural feature is the orientation of bis-AMT with respect to the active site Ni²⁺. The metal ion is coordinated to the 2-amino group as well as the adjacent ring nitrogen of the thiadiazole ring at distances of 2.55 and 2.81 Å, respectively. These distances are slightly longer than those observed for the hydroxamate of actinonin complex, viz., 2.20 and 2.30 Å, respectively [28]. This is not surprising since bis-AMT is a weaker inhibitor for PDF_Ec than actinonin [23, 34]. In addition to the coordination bonds between the active site resident metal ion and bis-AMT, several potential hydrogen bonds (depicted in cyan in Figure 6) were observed between the inhibitor and active site residues which are in close proximity to the active site of the compound. Specifically, putative hydrogen bonds were indicated between the metal-coordinating 2-amino group of the inhibitor and the side-chain oxygens of Glu133 and Gln50, as well as between the 2-amino group of the distal (solvent exposed) thiadiazole ring and the backbone oxygen of Glu87. The latter hydrogen bonding interaction is important for the bis-AMT interaction, as this represents the only hydrogen bond specific to the dimeric form of the inhibitor versus the AMT monomer. Moreover, the hydrogen bonds involving Glu133 and Gln50 are of particular interest, as these residues are highly conserved amongst PDF isozymes due to their essential function in catalysis [17]. Both residues have been postulated to stabilize the ^FMet-peptide substrates of PDF through the formation of intermolecular hydrogen bonds (see Figure 6, Panel B), with Glu133 also functioning as a proton shuttle [61]. In this regard, it may be concluded that bis-AMT is able to make similar contacts with the enzyme site as those made by the substrate upon binding (Figure 6, Panel C).

It should be further emphasized that although bis-AMT is observed to make van der Waals contacts with both the S₂’ and S₃’ subsites of PDF_Ec [30–32], notably absent in the structure of Figure 6 are contacts between the inhibitor and the S₁’ subsite of the enzyme. (For clarity, refer to the schematic diagram representing the residues that comprise the various subsites of PDF_Ec provided in Figure 6, Panels B and C.) The latter subsite is largely hydrophobic in nature and represents the pocket where the methionine sidechain of the substrate resides upon binding to the enzyme. As PDF is highly specific for N-formyl methionine substrates, the S₁’-methionine interaction has been postulated to be important in substrate recognition and binding [17, 29–32], and consequently the majority of PDF inhibitors exploit this interaction through the use of methionine-like sidechains at their P₁’ positions [25]. Since bis-AMT does not feature such a structural moiety, it is not surprising that only limited contact between the inhibitor and the S₁’ subsite is present in the docked structure of bis-AMT (Figure 6, Panels A and C). Hence, it is likely that the lower affinity of PDF_Ec for bis-AMT versus other inhibitors is a consequence of the lack of interaction between the inhibitor and the enzyme at this position. It is therefore possible that the binding affinity of bis-AMT for the enzyme site may be enhanced significantly through the modification of the bis-AMT structure such that the inhibitor features a hydrophobic moiety in a position where it may interact with the S₁’ subsite of PDF_Ec.

ABBREVIATIONS

PDF

peptide deformylase

Ec

E. coli

AMT

2-amino-5-mercapto-1,3,4-thiadiazole

DMF

N,N’-dimethylformamide

NME

N-terminal methionine excision

^FMet

formyl-methionine

DMSO

dimethylsulfoxide

BSA

bovine serum albumin

HOAc

acetic acid

ESI

electrospray ionization

Sa

S. aureus