Molecular Mechanism of Action of Antimalarial Benzoisothiazolones: Species-Selective Inhibitors of the Plasmodium spp. MEP Pathway enzyme, IspD

The methylerythritol phosphate (MEP) pathway is an essential metabolic pathway found in malaria parasites, but absent in mammals, making it a highly attractive target for the discovery of novel and selective antimalarial therapies. Using high-throughput screening, we have identified 2-phenyl benzo[d]isothiazol-3(2H)-ones as species-selective inhibitors of Plasmodium spp. 2-C-methyl-D-erythritol-4-phosphate cytidyltransferase (IspD), the third catalytic enzyme of the MEP pathway. 2-Phenyl benzo[d]isothiazol-3(2H)-ones display nanomolar inhibitory activity against P. falciparum and P. vivax IspD and prevent the growth of P. falciparum in culture, with EC50 values below 400 nM. In silico modeling, along with enzymatic, genetic and crystallographic studies, have established a mechanism-of-action involving initial non-covalent recognition of inhibitors at the IspD binding site, followed by disulfide bond formation through attack of an active site cysteine residue on the benzo[d]isothiazol-3(2H)-one core. The species-selective inhibitory activity of these small molecules against Plasmodium spp. IspD and cultured parasites suggests they have potential as lead compounds in the pursuit of novel drugs to treat malaria.

chemotherapeutic in combination with clindamycin and piperaquine 14 and has therefore demonstrated the validity and safety of targeting MEP pathway enzymes as an antimalarial strategy 15 .
Here we report the discovery of a new series of antimalarial IspD inhibitors and determination of their mechanism of action. We have employed a combined approach of chemoinformatics and high-throughput enzymatic screening to identify a new, chemically distinct series of IspD enzyme inhibitors that inhibit P. falciparum growth through disruption of isoprenoid biosynthesis in the Plasmodium parasite. Following chemotype identification, we have further developed this series, building structure-activity relationship around the chosen chemical motif and improving drug-like inhibitor properties. Finally, we have employed a combination of molecular modelling studies, site-directed mutagenesis and structural elucidation to determine the molecular mechanism by which these compounds achieve enzyme inhibition. This work has identified a series of compounds that are chemically

Results
Identification and optimization of benzoisothiazolone PfIspD inhibitors. A high-throughput screen (HTS) was developed to identify inhibitors of purified recombinant PfIspD. An assay protocol and chemoinformatics strategy (Methods: Chemoinformatics strategy) suitable for HTS were optimized and validated for screening with the Z' ranging from 0.85 to 0.95 23 . With active compounds defined as displaying an PfIspD IC 50 of < 20 μ M, a total of 208 hit compounds were identified from the HTS study, comprising over 10 structural chemotypes. Significantly, the 2-phenyl benzo[d]isothiazol-3(2H)-one (BITZ) chemotype was repeatedly identified within a number of hit structures, including 7, which exhibited a PfIspD IC 50 value of 450 ± 79 nM (Table 1).
A number of analogs, based around 7, were synthesized to explore the structure-activity relationship around the BITZ chemotype and analyze the nature and tolerance of the IspD active site. The bi-aryl BITZ template was selected to enable structural and chemical modification to the C and D rings adjacent to the inhibitor core (8-13, Table 1) with the aims of varying lipophilicity, improving solubility, and introducing functionality to enhance potency at PfIspD. A representative four-step synthesis of target compounds is shown (Fig. 3) involving an acid-chloride mediated amide coupling of 2-(methylthio) benzoic acid and 3-iodoaniline, generating a (methyl-thio)benzamide, 5a 24 . Oxidation of 5a to the corresponding sulfoxide, 5b 24 , followed by Suzuki-Miyaura cross-coupling reactions 25 , using a range of para-substituted phenyl boronic acids, was employed to generate intermediates 5c-g. The methylsulfinyl benzamides, 5c-g, were cyclized using a thionyl chloride-mediated ring closure 26 to give selected inhibitors 8-12. Inhibitor 13 was generated using an alternative three-step synthesis where a stable and isolatable acid chloride was employed for the generation of the core BITZ motif ( Supplementary Fig. F1).
Several compounds demonstrated low micromolar to nanomolar activity against P. falciparum and P. vivax IspD enzymes (Table 1), highlighting both the efficacy of these compounds and their potential to act as broad spectrum antimalarial agents. Overall, the series 8-13 demonstrated good correlation between PfIspD enzyme inhibition and inhibition of parasite growth (r 2 = 0.91, Fig. 4).
Given the ongoing and widespread issues of drug resistance to P. falciparum, it is critical that existing drug-resistant parasites are not cross-resistant to novel therapies in development. Therefore, we evaluated the antimalarial potency of BITZ compounds against a series of lab-adapted field isolates of P. falciparum, including strains resistant to chloroquine, mefloquine and artemisinin. We find that 8 effectively inhibits the growth of three lines of drug-resistant P. falciparum parasites ( Table 2). These studies demonstrate the promise of BITZ compounds to inhibit parasite growth in Plasmodium strains with acquired resistance against current antimalarial drugs. In optimising BITZ compounds, it was noted that meta-aryl functionalization across the C-ring consistently gave good levels of enzyme inhibition and that this substitution pattern is required to maintain potent PfIspD inhibitory activity. In comparison, para-substitution across the C-ring lowered activity for series analogs, highlighting the importance of active site-inhibitor shape complementarity (Supplementary Table T1). Para-substitution within the D-ring, using electron releasing or withdrawing groups (8-10) was well tolerated,   (Table 1) and half-maximal inhibitory concentrations determined prior to least-squares linear regression analysis (coefficient of determination, r 2 = 0.91; GraphPad Prism). Mean values given with SEM; n ≥ 3. with PfIspD IC 50 values as low as 73 ± 20 nM and parasite EC 50 of 1.1 ± 0.16 μ M, with P. vivax IspD (PvIspD) IC 50 values of 57 ± 13 nM for the para-chloro analog 9. In order to enhance aqueous solubility, methylene-linked polar heterocyclic groups were incorporated, exemplified by 11-13. Inhibitor 11 showed a 100-fold increase in predicted aqueous solubility compared to 9, a reduction in ClogD from 4.98 to 3.80, and displayed nanomolar inhibitory activity against PfIspD, PvIspD, and cultured P. falciparum, demonstrating that inhibitor activity of this series can be maintained whilst enhancing drug-like properties.

BITZ compounds inhibit isoprenoid metabolism in malaria parasites. To establish that
anti-parasitic activity was the result of methylerythritol phosphate (MEP) pathway inhibition, we performed targeted metabolic profiling of MEP pathway intermediates in parasites treated with our BITZ IspD-inhibiting compounds. Early ring-stage cultures of P. falciparum were treated with compound concentrations at ~5 times the IC 50 of 8 for 10 hours and MEP pathway intermediates were quantified by LC-MS/MS analysis as described elsewhere 27 19 , and therefore confirm that BITZ compounds inhibit the MEP pathway at levels known to affect asexual parasite growth. In malaria parasites, supplementation with the isoprenoid precursor, IPP, has been demonstrated to rescue MEP pathway dysfunction. For example, IPP supplementation has been shown to allow parasite growth during treatment with either the DXR inhibitor, fosmidomycin (FSM), or the IspD inhibitor, 1R,3S-MMV008138 20,21,28 . In order to determine whether BITZ inhibition was specific to activity against the MEP pathway, we tested for IPP rescue during treatment with compound 8. No rescue was observed (p = 0.75; Supplementary Table T2). The observed correlation between enzymatic inhibition and parasite growth inhibition (Fig. 4), as well as metabolic analysis (Fig. 5), suggest that growth inhibition under treatment with BITZ compounds involves inhibition of the IspD enzyme. However, this lack of rescue by IPP supplementation suggests that off-target activity also contributes to parasite growth inhibition.

BITZ compounds inhibit IspD enzymes through covalent active site modification. To gain
insight into the mechanism of enzyme inhibition, we developed a PfIspD homology model 19 based on an E. coli IspD (EcIspD) protein structure (PDB ID: 1I52 29 ). Molecular docking of 8 clearly indicates that the inhibitor binds within the CTP substrate-binding pocket, placing the BITZ warhead proximal to the PfIspD active site cysteine-202 (Cys-202) residue. We therefore predicted that the thiol moiety of Cys-202 would likely react with the sulfur heteroatom of BITZ inhibitors, forming a covalent disulfide adduct, while the carbonyl oxygen would be within hydrogen bonding distance of the backbone nitrogen atoms of Gly-203 and Gly-204 (Fig. 6). Formation of the covalent adduct would occlude the active site, preventing natural substrate binding and render the PfIspD enzyme inactive. The active site Cys-202 residue is conserved in Plasmodium spp. IspD enzymes but not in bacterial homologs (Fig. 7a). BITZ compounds are found to be ineffective against EcIspD enzyme activity (Fig. 7d), thus supporting the proposed involvement of and specific requirement for this Cys-202 residue, which is not present in the E. coli enzyme.
Our modelling studies predicted that mutation of Cys-202 to an alanine (Ala) residue would decrease the sensitivity of PfIspD to this BITZ compound class. Indeed, introducing this PfIspD mutation resulted in a 6-fold decrease in sensitivity: [PfIspD-wt IC 50   for inhibitor activity, but may reflect both a loss of covalent bond formation and a change in compound fit within the active site following introduction of this mutation to the inhibitor binding region.
Covalent inhibition is often marked by an increase in the apparent potency of an inhibitor over time, as irreversible associations are gradually made between the inhibitor and enzyme. To evaluate this premise with respect to our proposed mechanism of enzyme inhibition, PfIspD and various concentrations of 8 were pre-incubated prior to measurement of enzyme activity. As expected for covalent inhibition, the potency of 8 increased over time (PfIspD-wt IC 50 = 330 ± 57 nM at five min. and 81 ± 13 nM at 50 min.; p = 0.0007; Fig. 7c) whereas no such substantial or significant change in PfIspD-C202A inhibition was observed over the same time period (PfIspD-C202A IC 50 data of 860 ± 220 nM at five min. and 470 ± 39 nM at 50 min.; p = 0.32). These results highlight the importance of the proposed covalent interactions with Cys-202 during inhibition of PfIspD by BITZ inhibitors.
Once an inhibitor-adduct has formed, increasing substrate concentration should not relieve covalent enzyme inhibition. Consistent with this model, the susceptibility of PfIspD-wt to inhibition by 8, following prolonged incubation, was not substantially affected by CTP substrate concentration (Fig. 8b). In contrast, inhibition by 8  was somewhat relieved under higher CTP concentrations over shorter incubation periods, as is typically observed for reversible competitive inhibitors (Fig. 8a). Together, these observations are consistent with relatively slow inhibitor-enzyme adduct formation within the CTP binding site, as predicted by our molecular docking model. Inhibition of PfIspD-C202A activity was not significantly affected by CTP concentrations over short or long incubations with 8 (Fig. 8c,d), demonstrating that the CTP-site binding does not substantially contribute to inhibition of mutant PfIspD-C202A and suggests an independent mode of inhibition of PfIspD-C202A.
While both P. falciparum and P. vivax IspD homologs possess an active site Cys-202 residue, the corresponding residue in bacterial IspD homologs is alanine (Ala) (Fig. 7a). In contrast to the inhibitory IspD activity shown by BITZ analogues at P. falciparum and P. vivax IspD homologs (Table 1), we found that recombinant EcIspD was highly resistant to inhibition by 8 (Fig. 7d). Therefore, to confirm the critical role of active site Cys residues in inhibition, we generated a variant EcIspD protein, in which the corresponding Ala residue was substituted with a Cys residue by introduction of the A14C mutation. EcIspD-A14C was rendered sensitive to 8, with an IC 50 of 440 ± 48 nM (Fig. 7d). EcIspD-A14C was crystallized in the presence of 5 mM of 8 (Supplementary Table T3), providing supportive structural data. While electron density corresponding to 8 was not observed, the resulting crystal structure revealed formation of a novel intramolecular bond, leading to occlusion of the CTP binding site, thus providing modest additional evidence for a temporary covalent interaction between the inhibitor and the Cys-14 residue (Supplementary Fig. F2).

Discussion
The non-mevalonate (MEP) pathway of isoprenoid biosynthesis is highly desirable as a target for antimalarial drug development, as the parasite-specific enzymes of this pathway do not have mammalian homologs. Since isoprenoid biosynthesis contributes to a number of essential cellular functions, MEP pathway function is required for asexual replication of Plasmodium spp. parasites. Despite the promise of new MEP pathway-targeting agents, few studies have explored the potential of targeting the second-dedicated enzyme of this pathway, 2-C-methyl-D -erythritol-4-phosphate cytidyltransferase (IspD), for the purpose of antimalarial chemotherapy. We have previously established that the ISPD gene of P. falciparum is resistant to genetic disruption and that the PfIspD enzyme is amenable to small molecule inhibition 19 . We now report a new, structurally distinct class of PfIspD 2-phenyl benzo[d]isothiazol-3(2H)-one (BITZ) inhibitors with antimalarial activity, which were identified through a target-based high-throughput study.
We find that BITZ compounds inhibit IspD enzymatic activity of both P. falciparum and P. vivax and that the IspD-inhibitory activity and antimalarial efficacy of these compounds are correlated (Fig. 4). Furthermore, we find that treatment of P. falciparum with growth-inhibitory concentrations of BITZ compounds is sufficient to disrupt de novo production of isoprenoid precursors. These findings strongly indicate that BITZ compounds inhibit parasite growth, at least in part, through inhibition of cellular IspD, as predicted. Importantly, parasite strains resistant to current generation antimalarials do not possess cross-resistance to BITZ compounds.
Our studies staunchly suggest that these novel IspD-inhibiting BITZ compounds have a distinctive mechanism of action that relies on the formation of a covalent enzyme-inhibitor adduct, with several lines of evidence supporting this mechanism of inhibition. First, BITZ inhibitors are competitive with CTP during short incubation periods but enzymatic inhibition cannot be out-competed following prolonged incubation of PfIspD with the inhibitor, suggesting that a covalent bond is likely formed. Importantly, molecular modelling implicates a specific residue, Cys-202, as a likely site of enzyme modification by BITZ compounds. Supporting this model, we find that mutation of Cys-202 to an alanine residue confers BITZ resistance to recombinant PfIspD enzyme (Fig. 7). In addition, we find that the bacterial IspD enzyme, EcIspD, which possesses an alanine at the analogous position to Cys-202, is insensitive to BITZ compounds, but can be rendered sensitive via replacement of the alanine residue with cysteine. The proposed mechanism of enzyme inhibition thus provides an explanation for the marked species-selectivity of BITZ compounds and their inhibitory activity amongst IspD homologs, despite the overall sequence homology across this group of enzymes.
The demonstrated cysteine reactivity of our BITZ compound series suggests that development of this series may be hampered by possible polypharmacology. However, several recent reviews have noted that covalent mechanisms of action are prevalent among successful drugs 30,31 . Indeed, aspirin, also a covalent modifier, is one of the most widely used drug in the world. The polypharmacology of bioactive molecules 32,33 and approved drugs 34 has similarly been studied and is consistent with the concept that safe and effective drugs are invariably active at many targets. One such study calculates that, on average, approved therapeutics are active against approximately seven different molecular targets 35 . Since rescue with the isoprenoid precursor, IPP, is not observed, it is apparent that parasite growth inhibition by our BITZ compound series does not result solely from activity against IspD. Further studies may shed light on additional targets. Certainly continued vigilance will be required as BITZ compounds and their derivatives progress in development, with continued monitoring for potential promiscuous and off-target reactivity that may lead to toxicity. However, it should be noted that, similar BITZ chemotypes have demonstrated successful medicinal chemistry optimisation [36][37][38][39][40][41][42] . Recently, a molecule containing a BITZ moiety has begun Phase II clinical trials as an antiviral agent, providing important evidence that the potential reactivity of the BITZ moiety alone is not incompatible with a safe therapeutic profile 43 .
In conclusion, we have identified novel PfIspD 2-phenyl benzo[d]isothiazol-3(2H)-one inhibitors, initiated from a HTS study. Following discovery of 7, several structural analogs were generated around the identified BITZ motif which display increased potency, both enzymatically and phenotypically, compared to the initial hit. We have shown that the BITZ motif and corresponding analogues act as potent PfIspD inhibitors and offer potential as a broad spectrum antimalarial agents. In addition, we have demonstrated that these compounds also express inhibitory activity against P. falciparum parasite strains which show drug resistance to existing antimalarial therapies. Our studies clearly define the mechanism of inhibition of the BITZ chemotype and corresponding analogues, with selectivity for PfIspD relying on the presence of an active site Cys residue. A mechanism of inhibition has been deciphered whereby a disulfide adduct is formed between PfIspD Cys-202 and the electrophilic sulfur atom of the BITZ inhibitor core. The next stages of hit development will focus on improving the efficacy, solubility, and PK properties of the most potent PfIspD inhibitors. The potential promiscuity of the BITZ warhead and possible binding events at additional protein contacts will also require careful consideration. However, as interest in the covalent modification of drug targets within drug design increases, similar techniques and experiments to those outlined here may be used to probe other identified Cys modifiers which may be capable of enacting this mode of covalent modification. As more researchers study the MEP pathway and screen for inhibitors of its different enzymes, expectations dictate that it will become a recognized target for the treatment of not just malaria, but also tuberculosis and other bacterial pathogens.

Methods
Chemoinformatics strategy. A chemoinformatics strategy was employed to identify possible inhibitors using known IspD substrates and inhibitors, metal binding moieties, and biphosphate isosteres as query molecules for performing similarity and scaffold-hopping database searches 44 . A number of algorithms were applied, including molecular fingerprints 45 , turbo similarity 46 , principal component analysis, Bayesian modelling 47 , and machine learning to select ~10,000 compounds for screening. The selected structures were chosen from a commercial library of ~500,000 compounds (BioFocus DPI) that were predicted to possess favorable absorption, distribution, metabolism, excretion, and toxicity characteristics 48 .
Scientific RepoRts | 6:36777 | DOI: 10.1038/srep36777 Molecular Modelling. A homology model of PfIspD was constructed using the PHYRE online homology modelling program 49 . PfIspD primary sequence Q8I273 was obtained from UNIPROT (http://www.uniprot. org/, accessed 10/12/14). A number of protein alignments and homology models were constructed by PHYRE, and the model with 98.82% confidence was selected, which was based on an E. coli IspD structure (PDB accession code 1I52) 29 . 1I52 is a 1.50-Å resolution crystal structure of E. coli IspD, complexed with CDP-ME in the active site. The structure of the model was validated using the WHATIF web interface 50 . 8 and other BITZ inhibitors were modelled in silico using the homology model described above in order to visualize the interactions between each analogue and the active site. Using GOLD, protons were added and docking was performed with default parameters, except that GoldScore was used and 50 docking poses were obtained for comparison and analysis. Covalent docking was performed using GOLD with the C-alpha atom of Cys-202 as the link atom to BITZ inhibitor 8 51 .
Scientific RepoRts | 6:36777 | DOI: 10.1038/srep36777 IspD assay conditions. Phosphate released by IspD was quantified using the EnzChek Phosphate Assay Kit (Invitrogen, Life Technologies), as previously described 27 . 2-Amino-6-mercapto-7-methylpurine riboside (MESG) and purine nucleoside phosphorylase (PNP) were diluted and stored according to the manufacturer's instructions. Final concentrations of reagents were as follows: 100 mM NaCl, 25 mM Tris pH 7.5, 7.5 mM MgCl 2 , 1 U/ml PNP, 0.1 U/mL yeast inorganic pyrophosphatase (New England Biolabs). Unless otherwise specified, IC 50 assays contained 50 nM enzyme, 400 μ M CTP (Sigma), 200 μ M MEP (Echelon Biosciences), 2% DMSO (vehicle), and 200 μ M MESG. Reactions were performed in 50 μ L final volumes in 96-well clear flat bottom plates. In general, enzymes, buffers, and inhibitors were pre-warmed approximately 15 minutes at 37 °C, and reactions were then initiated by addition of the MEP substrate. Timed inhibitor 8 IC 50 assays (5, 20, 35, and 50 minute incubations) and comparisons between behavior at 100 μ M and 400 μ M CTP were initiated by the addition of the CTP substrate and contained 100 nM enzyme. Assays using E. coli IspD used 7.5 nM EcIspD-wt or 75 nM EcIspD-A14C. Absorbance at 360 nm was measured over time on a BMG POLARStar plate reader, preheated to 37 °C. Nonlinear regression analysis was performed using GraphPad Prism software. Values reflect the mean and standard error from ≥ 3 independent experimental and were compared using the Student's t-test (two-tailed). Slopes of changing absorbance values were converted to (μ M MEP) (μ M enzyme) −1 s −1 using a phosphate standard curve. P. falciparum culture. Unless otherwise specified, P. falciparum parasites were cultured as previously described 6,53 , in a 2% suspension of human erythrocytes in RPMI-1640 medium (Sigma Aldrich) supplemented with 27 mM sodium bicarbonate, 11 mM glucose, 5 mM HEPES, 1 mM sodium pyruvate, 0.37 mM hypoxanthine, 0.01 mM thymidine, 10 μ g ml −1 gentamicin and 0.5% Albumax (Life Technologies). Cultures were maintained in 5% O 2 /5% CO 2 /90% N 2 at 37 °C. Parasite growth was monitored by microscopy of Giemsa-stained parasites. Parasite sensitivity assays. Asynchronous cultures of P. falciparum strain 3D7 were diluted to 1% parasitemia and cultured at indicated concentrations of inhibitor compounds in 100 μ L culture volumes. During rescue experiments, parasite growth medium was supplemented with 200 μ M isopentenyl pyrophosphate (IPP; Echelon Biosciences). After 72 hours, parasite growth was quantified using PicoGreen dye (Invitrogen) to measure DNA content as previously described 27 . PicoGreen fluorescence was measured (485 nm excitation/528 nm emission) in a POLARStar Omega microplate reader (BMG Labtech). IC 50 values were calculated by nonlinear regression analysis using GraphPad Prism software, and reflect the mean and standard error from ≥ 3 independent experiments. Values were compared using the Student's t-test (two-tailed).