The Oxidase Reaction of Flavin-Dependent Thymidylate Synthase (FDTS) Reveals the Nature of Inhibitor Binding

Abstract

Escalating antimicrobial resistance presents an urgent need for novel antibiotics, particularly against Mycobacterium tuberculosis (M. tb), which claims 1.5 million lives annually. A promising avenue for antibiotic development is inhibiting DNA biosynthesis by targeting Thymidylate Synthase enzymes (TS) that produce 2′-deoxythymidine-5′-monophosphate (dTMP), an essential building block of DNA. Two forms of TS enzymes are known, and M. tb has been shown to rely on a unique flavin-dependent thymidylate synthase (FDTS), which is distinct from the human TSase enzyme. This work explores the mechanism and binding modes of inhibitors of the MtbFDTS enzyme by utilizing a non-productive oxidase reaction (O₂ reduction to H₂O₂) catalyzed by the flavin cofactor. We discovered that inhibitors of MtbFDTS that bind competitively at the nucleotide-binding site potentiate the oxidase reaction. Conversely, inhibitors of the enzyme that bind at the folate’s binding site abate the oxidase reaction. We exploit this contrasting kinetic behavior to show that naphthoquinones inhibit FDTS by binding competitively at the nucleotide’s active site, activating the enzyme’s oxidase reaction, and consequently producing substantial peroxide as a byproduct. We confirm our predicted inhibitor binding modes through direct measurement of folinic acid binding to the ternary enzyme-flavin-naphthoquinone complex, providing a revised kinetic mechanism for the FDTS-catalyzed reaction. Our findings show naphthoquinones inhibit the native FDTS reaction and divert the enzyme’s activity to produce reduced oxygen species, which sheds light on their antimycobacterial activity and will be critical for future inhibitor development and high-throughput screening (HTS) methods.

Graphical Abstract

Tuberculosis (TB) is caused by the bacteria Mycobacterium tuberculosis (M. tb) and remains a persistent and pervasive infectious disease.^1,2 Given the emergence of extensive and multiple drug-resistant strains, new targets that interrupt basic cellular processes need to be identified to develop novel antibiotic drugs. One attractive strategy for drug development is via the inhibition of DNA biosynthesis by targeting the enzymes in the pathway that produce 2′-deoxythymidine-5′-monophosphate (dTMP), which is essential for cell replication and survival.³ There are no current antibiotic drugs that target dTMP biosynthesis, which presents a promising approach for treating drug-resistant strains of M. tb and other serious bacterial infections (e.g., C. difficile, H. pylori, and all Rickettsia).⁴

Thymidylate synthase enzymes catalyze the de novo biosynthesis of dTMP through reductive methylation at the C5 position of the substrate, dUMP (2′-deoxyuridine-5′-monophosphate) (Fig. 1).⁵ In human cells, dTMP is produced by the classical thymidylate synthase (TSase) encoded by the thyA gene (TYMS gene in humans), which has been well-characterized and is an enzyme targeted by chemotherapeutic drugs such as 5-fluorouracil (5-FU) via its active metabolite 5-fluoro-deoxyuridylate (5F-dUMP).⁵ In nearly 30% of bacterial genomes, however, dTMP is biosynthesized by flavin-dependent thymidylate synthase (FDTS) enzymes, that are encoded by the thyX gene.⁶ While both classical TSase and bacterial FDTS share the common reaction of converting dUMP to dTMP, they exhibit notable disparities in protein sequence, structure, cofactors, catalytic mechanisms, and inhibition properties (Fig. 1a and b).^7–9 It has been shown that M. tb contains both thyX and thyA genes; however, only thyX is essential for survival, giving precedence to MtbFDTS being a possible antimycobacterial target.¹⁰

Figure 1: Thymidylate biosynthesis and inhibition.

Molecular mechanisms for a) classical TSase; and b) FDTS. c) An active site view from the crystal structure of TmFDTS in complex with FAD, dUMP, and folinic acid (PDB code: 4GTA). The substrate dUMP (magenta) is positioned by the key residues Arg90 and Arg174. The folinic acid (cyan) forms an aromatic stacking with His53 and the FAD (yellow). d) Known inhibitors for FDTS.

Classical TSase is a structural dimer that uses dUMP and N5, N10-methylene-5,6,7,8-tetrahydrofolate (CH₂H₄folate) as substrates, producing dTMP and dihydrofolate (H₂folate) (Fig. 1a).⁵ In contrast, FDTS assembles as a homotetramer, has a solvent-exposed active site at the interface of three subunits, and dUMP, CH₂H₄folate, and nicotinamide adenosine dinucleotide phosphate (NADPH) as substrates, with flavin adenine dinucleotide (FAD) serving as the catalytic cofactor (Fig. 1b).¹¹ FDTS catalyzes FAD reduction with NADPH, generating reduced flavin (FADH₂), which transfers the methylene moiety from CH₂H₄folate to dUMP via a Schiff-base intermediate and subsequently reduces a putative exocyclic enone-intermediate to form dTMP (Fig. 1b).⁸ Mechanistic studies of FDTS have proposed a preliminary kinetic and chemical mechanism that is substantially different from that of the classical TSase.^7,8,12,13 Due to the distinct differences between human TSase and bacterial FDTS enzymes, selective inhibition appears feasible, yet requires more detailed mechanistic and biochemical studies.⁴

FDTS inhibitor discovery is still in its early stages and there are only a few reports of selective inhibitors.^14–18 The known inhibitors for FDTS fall into four main classes: nucleotide analogs, folate analogs, naphthoquinones, and thiazolidines (Fig. 1d). The nucleotide analogs, consisting of various 5-substituted uridylates (1), presumably mimic the substrate/product and display moderate potency (IC₅₀ values in the low μM) and reasonable selectivity for FDTS enzymes over classical TSase.¹⁵ The folate inhibitors, raltitrexed and folinic acid (2), have been crystallized with TmFDTS,¹¹ and the benzo[b][1,4]oxazin-3(4H)-one analog (3)¹⁹ also mimics the folate; however, selectivity is challenging since these analogs may also inhibit Classical TSase.¹¹ The naphthoquinones (NQ) (4)^16,20,21 and thiazolidine (5)¹⁸ analogs were identified by high-throughput screens and do not structurally resemble the substrate or product. The mechanism of the naphthoquinone inhibitors has been investigated by computational methods that predict that these molecules bind the dUMP’s active site; however, this has not been confirmed experimentally.¹³ Since there is a lack of crystal structures of FDTS bound to its inhibitors, understanding the mechanism of action and nature of inhibition is crucial to improving potency and selectivity.

The FDTS reaction requires the cofactor (FAD) and two substrates (NADPH and CH₂CH₄folate) in addition to the nucleotide substrate, dUMP, and detailed kinetic characterization is still lacking. Several studies on FDTS use assays that follow NADPH consumption (a decrease in the UV-visible spectra at 340 nm) to evaluate reaction kinetics and screen for inhibitors. Previously, we have shown that TmFDTS has a nonproductive oxidase reaction where the reduced FADH₂ reacts with molecular oxygen (Fig. 1b); a catalytic property that has yet to be characterized for the MtbFDTS enzyme.²² Importantly, since the oxidase reaction of FDTS competes with dTMP formation in the presence of molecular oxygen, screening assays that only monitor NADPH consumption may not accurately follow the desired catalytic reaction. Unless FDTS assays are performed anaerobically²² or in the presence of sufficiently saturating CH₂CH₄folate,²³ the oxidase reaction will consume the reducing substrate, NADPH, to form peroxide, and not the product, dTMP (Fig. 1b).

To test for the presence of an oxidase reaction for the MtbFDTS enzyme, we performed a series of kinetic studies in the absence of CH₂CH₄folate, only allowing the enzyme to turn over peroxide as a product. By following NADPH consumption, we have previously shown that the TmFDTS enzyme’s oxidase reaction increases with substrate, dUMP, in a concentration-dependent manner.²² This result suggests that dUMP and NADPH bind sequentially to the FDTS enzyme, and reduction of the FAD cofactor is facilitated in the dUMP-FAD-NADPH-FDTS complex. To test this hypothesis, we performed oxidase reaction assays with the MtbFDTS enzyme in the presence of substrate (dUMP) and product (dTMP) and monitored NADPH consumption as a function of time. We detected notable NADPH oxidation by the FAD-MtbFDTS enzyme in aerobic reaction mixtures, confirming the presence of a significant oxidase side-reaction. We further observed an increase in rate in the presence of the substrate, dUMP, as previously reported for TmFDTS (Fig. 2a).²² Additionally, we also observe that the oxidase reaction is also activated by the product, dTMP, of the MtbFDTS reaction (Fig. 2a). To our knowledge, this is the first observation of another ligand (i.e., dTMP) besides the substrate, dUMP, activating the oxidase reaction suggesting this kinetic property may be a general feature of ligand binding to the nucleotide’s active site.

Figure 2: Effect of dUMP and folate on the MtbFDTS oxidase reaction.

a) Relative NADPH oxidation rates measured without ligand (control) and in the presence of folinic acid, methotrexate, dUMP, or dTMP (>10× K_ACT; up to 250 μM final concentration). Concentration dependence of oxidase reaction after titration of b) dUMP and c) folinic acid into a reaction containing MtbFDTS and 200 μM NADPH. Data were fit to the single-site activity curve equation (see Materials and Methods) to obtain the activation constant (K_ACT = 17 ± 4 μM) for dUMP and to an inhibition model (IC₅₀ = 29 ± 6 μM) for folinic acid. All assays were performed in triplicate.

To further understand how ligand binding impacts the oxidase reaction of MtbFDTS, we assessed the impact of folates on the oxidase reaction. Based on the crystal structure of the TmFDTS enzyme (Fig. 1c) and our previous experiments on single turnover FAD reduction, we expected that folates bind to the same active site location as NADPH.¹¹ Indeed, we discovered that both folinic acid and the antifolate, methotrexate, display inhibition of the oxidase reaction of MtbFDTS in a concentration-dependent manner (Fig. 2a and 2c). This observation suggests a mechanism of inhibition for folate analogs that directly competes with NADPH binding, thereby impeding reduction of the FAD and the oxidase reaction of MtbFDTS. Importantly, these results show that ligands binding to the folate’s active site will inhibit the oxidase reaction of the enzyme, which contrasts the behavior of ligands that activate the oxidase activity and bind to the nucleotide’s active site (Fig. 2).

The contrasting kinetic behavior of nucleotide activation and folate inhibition observed for the oxidase reaction of MtbFDTS led us to investigate the naphthoquinone inhibitors to better understand the nature of binding and inhibition. The naphthoquinones were discovered by high-throughput screens, and computational studies have proposed that these molecules bind in the dUMP’s active site; however, this has yet to be experimentally confirmed.¹³ We performed a range of MtbFDTS oxidase reactions using 5-hydroxy-2,4-naphthoquinone (plumbagin), a known antimycobacterial agent and FDTS inhibitor,²¹ and a related biologically active compound, 2-hydroxy-1,4-naphthoquinone (lawsone). We monitored the rate of NADPH consumption in the presence of these naphthoquinones and remarkably, we found a substantial increase in the rate of NADPH consumption (5–10-fold) in both reactions containing plumbagin and lawsone (Fig. 3a). We performed careful controls without the enzyme (see Supporting Information, Fig. S1) to demonstrate that the MtbFDTS was catalyzing the reaction of NADPH and not a bimolecular reaction in solution with the naphthoquinone. Notably, the naphthoquinones activate the oxidase reaction several times greater than the substrate, dUMP (Fig. 2a). These results suggest that the naphthoquinones bind to the same active site location as the nucleotide substrate, dUMP.

Figure 3: Naphthoquinone molecules activate MtbFDTS oxidase activity.

a) Relative oxidase rates measured without ligand (control) and in the presence of folinic acid, dUMP, plumbagin, or lawsone (>10× K_ACT; up to 250 μM final concentration). Rate (s⁻¹) of oxidase reaction following NADPH consumption after titration of b) Plumbagin and c) Lawsone into an enzyme reaction of MtbFDTS and 200 μM NADPH. Data were fit to the single-site activity curve (see Materials and Methods) to determine activation constants (K_ACT = 24 ± 6 μM for plumbagin; K_ACT = 9 ± 4 μM for lawsone) d) Hydrogen peroxide formation was determined by an Amplex red/HPR assay²³ from MtbFDTS reactions conducted with NADPH (black) or with NADPH and 100 μM plumbagin (magenta). Inset shows the initial velocity for the reactions (c.a. 10-fold rate increase). All assays were performed in triplicate.

To assess the impact of naphthoquinones on the MtbFDTS-catalyzed oxidase reaction we monitored the hydrogen peroxide (H₂O₂) being produced from the enzyme in the presence of these inhibitors (Fig. 3d). We quantified the peroxide formed as a function of time in the oxidase reactions using an Amplex red - Horse Radish Peroxidase assay²³ and observed substantially more peroxide formation (c.a. 8–12 fold increase relative to the control experiment) in the presence of the naphthoquinone. This observed rate increase for H₂O₂ formation (Fig. 3d) is similar to the observed rate increase in NADPH consumption in the presence of plumbagin (Fig. 3b). Since quinone moieties may also react with H₂O₂, the measured rates may not accurately reflect actual peroxide formed. Due to these limitations, we are not able to conclude from our results whether naphthoquinones activate the reduction of the FAD by NADPH, or if they increase the reaction of the reduced FADH₂ with molecular oxygen. Nevertheless, the naphthoquinones inhibit dTMP synthesis, yet activate FDTS’s oxidase reaction, which increases peroxide production, and could have a deleterious impact on the host cell.

The activation of the oxidase reaction by naphthoquinones displays a concentration dependence and saturation kinetics consistent with the formation of the naphthoquinone-FAD-NADPH-FDTS enzyme complex. Fitting the concentration dependency to a saturation curve provides an activation constant (termed K_ACT) reporting the half-maximum concentration of naphthoquinone needed to fully activate the MtbFDTS oxidase reaction.²² To gain insight into the relative potency of the naphthoquinone inhibitors, we compare the activation constants found for plumbagin and lawsone with those of the substrate dUMP (Fig. 3b and 3c, and Fig. 2b). We observed that the naphthoquinone molecules activate the MtbFDTS enzyme at lower concentrations than dUMP. To compare these observed kinetic constants with ligand binding, we measured the affinity of dUMP and 5-FdUMP to the FAD-MtbFDTS complex by spectrophotometric titration (Fig. 4a).²⁰ Our results agree with previous reports for the MtbFDTS, showing low μM binding for dUMP and less affinity for 5-FdUMP.²⁰ Notably, our data reports an apparent K_d value for dUMP that is significantly lower (1.1 ± 0.5 μM) than the observed activation constant, K_ACT (Fig. 4a and Fig. 2b) for the oxidase reaction. The observed binding constant for folinic acid to MtbFDTS is also c.a. 10-times lower than the measured IC50 of the oxidase reaction (Fig. 4b and Fig. 2c). Unfortunately, spectrophotometric titrations of the naphthoquinones were complicated by convoluted spectra, which prevented reliable monitoring of FAD absorbance changes at 454 nm during ligand titration. While the activation constants observed are likely an underestimate of binding affinity (due to enzyme concentrations used and kinetic complexity of the FDTS reaction), these values allow for an estimate of the potential for these inhibitors to compete against dUMP binding.

Figure 4. Binding properties of MtbFDTS.

Spectrophotometric titrations of a) 5-FdUMP (black) or dUMP (green) into a solution of FAD-FDTS; and b) folinic acid into a solution of FAD-MtbFDTS (red), FAD-MtbFDTS-Plumbagin (blue), and FAD-MtbFDTS-dUMP (green). Absorbance differences ΔA₄₅₄ for FAD upon ligand binding are monitored as a decrease in 454nm (see inset).²⁴ All assays were performed in triplicate. c) A proposed kinetic and inhibition mechanism for the MtbFDTS enzyme.

We next performed a series of binding experiments to test for the existence of the naphthoquinone-FAD-MtbFDTS ternary enzyme complex predicted by our kinetic experiments on the oxidase reaction. Based on the TmFDTS crystal structure in Fig. 1c of the dUMP-FAD-CH₂H₄folate-TmFDTS¹¹ enzyme complex, we predict a synergistic binding between the folate and the dUMP due to extensive π-stacking interactions. In this direction, we measured the binding affinities for folinic acid by spectrophotometric titration into the FAD-MtbFDTS enzyme or the dUMP-FAD-MtbFDTS ternary complex achieved by saturating the enzyme with dUMP. We observe a K_d of 1.3 ± 0.2 μM for folinic acid binding to the FAD–MtbFDTS complex (Fig. 4b), however, under saturating concentration of dUMP we observed a much higher affinity for the dUMP-FAD-MtbFDTS ternary complex (Fig. 4b). When a titration of folinic acid was performed in the presence of saturating naphthoquinone inhibitor, a similar increase in binding affinity for the folinic acid was observed for the naphthoquinone-FAD-FDTS complex (Fig. 4b). These results strongly support our kinetic predictions that dUMP, and naphthoquinone inhibitors share the same binding site.

The collective kinetics of the oxidase reaction and binding behavior of the MtbFDTS enzyme reveal important information regarding the kinetic mechanism of the MtbFDTS enzyme.²³ Previous detailed Lineweaver-Burk (double-reciprocal) analysis of the MtbFDTS enzyme suggested a sequential mechanism of binding between dUMP and CH₂H₄folate, and sequential binding of NADPH and CH₂H₄folate.¹² Our results support the sequential binding between dUMP and CH₂H₄folate, as indicated by the contrasting kinetic behavior during the oxidase reaction, and by the increased binding affinity of folinic acid to the dUMP-FAD-MtbFDTS complex (Fig. 4b). On the other hand, the inhibition of oxidase activity by folates for the MtbFDTS enzyme, and the previous reports of folate inhibition of the single-turnover reduction of FAD within the TmFDTS enzyme¹¹ strongly support a ping-pong mechanism between NADPH and CH₂H₄folate (Fig. 4c). This prediction is also consistent with the crystal structure of TmFDTS (Fig. 1c), which indicates the dUMP-FAD-CH₂H₄folate-FDTS complex has no other binding sites available for NADPH within the reactive complex. Taken together, our results demonstrate that naphthoquinone moieties inhibit the enzyme by competing for the nucleotide binding site and that folates bind tightly to the naphthoquinone-FAD-FDTS enzyme complex, providing a mechanism of action for naphthoquinone inhibitors (Fig. 4c).

Our results provide a tool to identify the mechanism of inhibition and predict the location of inhibitor binding for the MtbFDTS enzyme. The activation of the oxidase reaction by naphthoquinones suggests a mechanism where these compounds inhibit MtbFDTS through competitively binding to the nucleotide’s active site. The observed increased oxidase reaction leads to substantial peroxide formation (Fig. 3d) and may explain the known antimycobacterial activity of various naphthoquinone compounds. This work illuminates how the NADPH oxidation assay should be monitored when screening for FDTS inhibitors, since molecules that compete for the same active site as dUMP, and are expected to inhibit dTMP synthesis, may activate the oxidase side reaction and increase NADPH consumption. Our kinetics observations may provide a means to evaluate potency (e.g., measurement of activation constant, K_act), which will be critical for establishing a Structure-Activity Relationship (SAR) of current and yet-to-be-discovered FDTS inhibitors. The revised kinetic and inhibition mechanism (Fig. 4c) may guide development of compounds designed to inhibit thymidylate biosynthesis yet exploit the oxidase reaction of FDTS to produce increased peroxide.

Supplementary Material

The Supporting Information is available free of charge at https://pubs.acs.org. Detailed Materials and Methods for experimental procedures and additional supporting data and control experiments are contained in Figure S1. (PDF)

ABBREVIATIONS

M. tb

Mycobacterium tuberculosis

FDTS

Flavin-dependent thymidylate synthase

CH₂H₄folate,

N⁵, N¹⁰-Methylene-5,6,7,8-tetrahydrofolate

dTMP

2′-Deoxythymidine-5′-monophosphate

dUMP

2′-Deoxyuridine-5′-monophosphate

FAD

Flavin adenine dinucleotide

NADPH

Nicotinamide adenine dinucleotide phosphate

NQ

naphthoquinone