The benzimidazole based drugs show good activity against T. gondii but poor activity against its proposed enoyl reductase enzyme target

The enoyl acyl-carrier protein reductase (ENR) enzyme of the apicomplexan parasite family has been intensely studied for antiparasitic drug design for over a decade, with the most potent inhibitors targeting the NAD + bound form of the enzyme. However, the higher affinity for the NADH co-factor over NAD + and its availability in the natural environment makes the NADH complex form of ENR an attractive target. Herein, we have examined a benzimidazole family of inhibitors which target the NADH form of Francisella ENR, but despite good efficacy against T. gondii, the IC 50 for T. gondii ENR is poor, with no inhibitory activity at 1 μ M. Moreover similar benzimidazole scaffolds are potent against fungi which lack the ENR enzyme and as such we believe that there maybe significant off target effects for this family of inhibitors.

eukaryotic and prokaryotic FAS pathways. 1 The Eukaryotic system, termed FAS I, uses one large polypeptide complex, but Prokaryotes use a series of discrete monofunctional enzymes termed the FAS II pathway to achieve the same goal 2 . Enoyl acyl-carrier protein reductase (ENR), which carries out the final stage of FAS II synthesis, has been the main focus of drug development programs in this area. This has resulted in a range of potent ENR inhibitors being developed such as isoniazid, the diazaborine family and the common antibacterial triclosan. 3,4,5,6 The potency and easy synthesis of triclosan has resulted in it being found in a range of common household items such as toothpastes, mouthwashes and chopping boards. Common to all of these potent ENR inhibitors is that they bind to the NAD + complexed form of ENR, despite the NADH co-factor having a higher affinity than NAD + for ENR and likely being equivalently prevalent in nature. 7 Moreover, the NADH/NAD + ratio has been shown to have a significant effect on the potency of isoniazid, an ENR inhibitor in M. tuberculosis. 8 There are very few examples of inhibitors which have been designed against the NADH form of ENR. However, a benzimidazole family of inhibitors designed as such has modest activity against Francisella tularensis (F. tularensis) displaying an IC50 value of 300nM. 9,10 The co-crystal structure of a benzimidazole/NADH/ ENR complex shows the inhibitor makes no π-stacking interactions with the bound cofactor, a feature common to triclosan and its derivatives, but instead makes a hydrogen bond to the conserved catalytic Tyr and NADH cofactor. 10 The potency of the benzimidazole family has been shown for F. tularensis, E. coli, B. anthracis, Y. Pestis, S. aureus and MRSA, but to date no antiparasitic testing has been carried out. 9 Despite its eukaryotic nature, Toxoplasma gondii (T. gondii) uses a typical FAS II pathway making it susceptible to inhibitors of ENR. 11 This pathway is located in the apicoplast organelle and not surprisingly apicomplexan parasites that also retain an apicoplast such as Plasmodium (species) and Eimeria (species) possess this pathway and are known to be susceptible to ENR targeted drugs. 11,12,13 Although FAS II and ENR play an important role in liver-stage development of P. falciparum it is not essential for the survival of blood stages where triclosan would therefore appear to have an off-target effect 14,15 The development of a novel set of ENR/NADH inhibitors against F. tularensis has provided a new potential therapeutic avenue for the development of T. gondii inhibitors. Here we demonstrate the benzimidazole family of compounds shows no inhibition of T. gondii ENR (TgENR) at 1μM. However, the same inhibitors show promising activity with a MIC50 value of between 1 and 10μM against two different strains of T. gondii parasites cultured in vitro. The ability of these compounds to curtail T. gondii growth, but not affect ENR activity, suggests that they have an off-target effect. Consistent with this idea, a structurally similar compound Chlormidazole is active against fungi which have a type I and not a type II fatty acid biosynthesis pathway and lack an ENR homologue, indicating an alternative primary target in fungi, most likely 14 alpha methylase. Although T. gondii lacks this enzyme, data mining of the PubChem compound library has shown a number of similar scaffolds with different targets which may explain some of the off-target affects evident against T. gondii.
Johnson et al reported that the 3,4-dichloro substituted benzimidazole 1 was potent against F. tularensis ENR (FtENR) complexed with NADH 9 . Hit compound 1 and two derivatives 2 and 3 were synthesized from commercially available 5,6-dimethylbenzimidazole and substituted benzyl bromides using NaH and KI in DMF, in moderate to good yields (scheme 1). HPLC determined purity to be > 95% for each compound.
To investigate if the benzimidazole compounds had a similar binding mode in TgENR as described for FtENR, co-crystallisation experiments were conducted. 10 In the first instance crystals were grown in the presence of 1.6mM inhibitor, which was more than sufficient for the FtENR/NADH/Benzimidazole crystal structure. However, this was insufficient to produce a TgENR/NADH/Benzimidazole crystal complex. Instead, 3.2mM of inhibitor was used and crystals were grown in the Morpheus crystal screen from Molecular Dimensions. Several different crystallisation conditions yielded diffracting crystals from which data was collected to identify any bound inhibitor. The highest diffracting crystals grew in condition G6 (0.1M Na-Formate, NH 4 -Acetate, Na 3 -Citrate, NaK-Tartrate, Na-Oxamate, 1M Sodium HEPES, MOPS pH7.5, 30% v/v P500MME_P20K). The crystals were flash frozen and data were collected on beamline I02 at the Diamond SRS. Full data collection and processing statistics are found in table 1. The coordinates have been deposited within the protein data bank, accession number 4O1M. Iterative cycles of model building and refinement were carried out to 2.0 Å resolution in COOT and REFMAC5 with PDB_REDO optimizing the refinement procedure, with resulting R fact and R free values of 0.17 and 0.20, respectively. 21,22,23 The resulting, refined map showed clear and continuous electron density for the bound NADH cofactor. However, there was only a small region of strong positive F obs -F calc density within the proposed binding site ( Figure 1A). Moreover, the position of this density is not consistent with the proposed binding site for the benzimide inhibitor ( Figure 1B). The inhibitor could not be refined either in a similar position to that seen in FtENR or in an alternative conformation. The features seen within the F obs -F calc map within the ENR binding site are consistent with the density associated with water and bound crystallization additives ( Figure 1C). Despite data for several crystallization conditions being collected and processed no evidence for benzimidazole binding could be seen, with all structures showing features consistent with those seen in Figure 1.
Since the inability to co-crystalise TgENR with the benzimidazole inhibitors is not in of itself evidence that the family are poor inhibitors, enzyme and cellular assays were used to investigate their potency. The enzyme assays showed no inhibitory activity even at 1μM concentration for compounds 1-3 which showed that the benzimidazole compounds are very poor inhibitors of TgENR.
This discrepancy between binding of benzimidazole to TgENR and FtENR was investigated by comparing the two binding sites. A number of sequence changes are observed including significant changes in the loop region Ser 242 -Asp 249 (TgENR numbering). A further change which may affect binding is the substitution of Met/Ala at the base of the binding pocket ( Figure 1D). The Methionine residue in FtENR makes packing interactions with the bound inhibitor and appears to be important for stabilising the inhibitor ( Figure 1D). The replacement with Ala in TgENR reduces the ability of this residue to co-ordinate benzimidazole binding. This is an important aspect of binding, as unlike with other ENR inhibitors the benzimidazole compound is not coordinated by forming stacking interactions with the bound cofactor. 10 A further change is the presence of Phe242 (TgENR numbering) which creates steric hindrance within the binding site, however we have previously shown that this residue is capable of flipping out of the binding site, to remove the steric hindrance. 24 These subtle changes will likely further reduce the potency of the benzimidazole family of inhibitors which have been shown to have only modest activity >100,000nm-300nm against FtENR. 10 In the absence of a crystal structure we conducted docking simulations using AutoDock 4.2 of compound 1 to see if it could bind in a mode similar to that observed in FtENR. 25 The predicted binding mode of 1 within TgENR was similar to that of FtENR with no significant steric clashes when Phe242 is flipped out of the active site. However, it is clear from the enzyme assay data that this is not a potent interaction with 1 μM and 3.2mM being insufficient for inhibition and co-crystallisation, respectively.
In conjunction with TgENR enzyme assays, parasite replication assays were also conducted using the RH strain (type 1) and Pru strain (type 2) T. gondii stably transfected with the yellow fluorescence gene (RH-YFP). This allows for parasite viability to be directly recorded from the relative fluorescence intensity. The fluorescence intensities were measured after 72 hours using a Synergy H4 Hybrid Multi-Mode Microplate Reader with excitation/emission wavelength at 510/540nm. Compounds 1, 2 and 3 were effective against type 1 parasites with MIC 50 values of 2.5μM, 4 μM and >10 μM, respectively (Figure 2A). Type 2 T. gondii were also inhibited with compounds 1 and 3 at 10μM ((p<0.005)( Figure  2B)). The difference in activity of compound 2 maybe due to subtle differences in the two types of strains used. This activity is far in excess of what was expected based on the effects of the same compounds on TgENR enzyme, suggesting that they have an additional offtarget effect against T. gondii.
The promising parasite replication assay data shows that these compounds may be a promising lead for novel T. gondii inhibitors. To investigate their potential further, compounds 1, 2 and 3 were tested for cellular toxicity on both fibroblast and prostate cancer cells ( Figure 2C, D). Importantly, none of the compounds showed significant toxicity, in either of the two different assays even at a concentration of 10μM against mammalian cells ( Figure 2C, D). The low toxicity paired with the good MIC50 values make this family of inhibitors a promising avenue for inhibitor design, however the poor IC50 values makes TgENR an unlikely target. To further develop the benzamidazole family we looked for possible targets other than ENR in the PubChem database.
The PubChem search showed that similar small compound scaffolds have been developed which target Sterol 14 alpha-demethylase, Galanin receptor 3, Pteridine Reductase 1, transient receptor potential cation channel C4 (TRPC4) and aldosterone synthase ( Table 2). The crystal structure for the Trypanosoma brucei Pteridine Reductase (3GN2) shows that the additional amine group of the bound inhibitor (4) is essential in the formation of two hydrogen bonds with Asp 161 which is absent in the benzamidizole compound. 26 This would make it an unlikely target within Toxoplasma. The Galanin receptor and TRPC4 have been shown to be targets of similar compounds 5 and 6 (table 1). However, these two proteins are not present within Toxoplasma, making them unlikely targets. A further target is sterol 14 alpha-demethylase, which is inhibited by the antifungal drug Chlormidazole (6) (table 2). Sterol 14 alpha-demethylase is found in a range of fungi and bacteria such as Mycobacteria species as well as the Leishmania and Trypanosoma parasites. However, it is absent from apicomplexan parasites with BLAST searches revealing only poor matches to putative cytochrome P450 enzymes in T. gondii. Moreover, T. gondii, as well as other apicomplexan parasites lack a cholesterol biosynthesis pathway, instead relying on scavenging mechanisms. 27 Further studies have found that different sterol 14 alpha-demthylase targeting drugs were more effective against a different cytochrome P450 CYP121 in Mycobacteria. 28 An additional sterol 14 alpha-demthylase inhibitor itraconazole is effective against T. gondii, however, an additional off target has not been found and it has been hypothesized that there may be several general non-specific targets in T. gondii which account for the cellular activity. 29 A similar scaffold has been shown to display activity against steroid 11-beta-hydroxylase and aldosterone synthase (7,8). Taken together this would suggest that despite the benzimidazole inhibitor family offering potential as a novel inhibitor against F. tularensis ENR it has a likely off target effect which accounts for the good MIC50 value established against the parasite in vitro but poor IC50 value observed against Toxoplasma ENR.
Further studies will be required to determine the exact target of the benzimidazole family of inhibitors and establish a clear structure-activity relationship (SAR). However, they present the potential for novel T. gondii inhibitors given their good anti parasitic activity and low toxicity against both prostate and HFF cells. (A) Refined TgENR/NADH crystal structure with the 2F obs -F calc density map (blue) and 1F obs -1F calc density map (green) contoured at 1.6σ and 4.0σ, respectively. (B) The same view as in (A) but with the FtENR/NADH/benzimide complex superimposed in red to show the predicted position of the benzimide inhibitor. (C) Representative density of the refined TgENR/NADH structure with waters and glycerol modeled into the inhibitor binding site with the map contoured at 1.5σ. For AC, nitrogen, oxygen, chlorine, phosphorous and carbon are colored blue, red, green, orange and yellow (TgENR)/red(FtENR), respectively. (D) Overlay of FtENR (blue sidechains) and TgENR (green sidechains) inhibitor binding sites with bound inhibitor (yellow), NADH cofactor (green) shown in stick format. The position of the critical Tyr residue which forms a hydrogen bond to the bound inhibitor is shown along with those residues which are significant different between species.

Figure 2. Efficacy and toxicity of compound 1-3 against T. gondii
(A) Growth of RH-YFP in human Foreskin fibroblasts (HFF) measured as fluorescence intensity. HFF were infected with differing numbers of RH-YFP tachyzoites as indicated, with uninfected control HFF also shown. HFF infected with 3200 RH-YFP tachyzoites treated with pyrimethamine/sulfadiazine (p/s) or 0.1% DMSO are additional positive and negative controls, respectively (left panel). The effect of compound 1-3 against Toxoplasma tachyzoites was determined by measurement of RH-YFP fluorescence as shown in the right hand panel. HFF were infected with 3200 RH-YFP and treated with compounds 1-3 at the concentration indicated. Toxicity was measured using WST stain. (B) The effect of compounds 1-3 on Toxoplasma growth was determined by infecting confluent fibroblasts with YFP expressing Pru strain T. gondii tachyzoites and measuring YFP florescence at 72 hours. Compounds 1 and 3 effectively reduced parasite growth however at 10μM (p<0.005). (C) Compounds 1-3 were incubated with confluent luciferase expressing PC3-Luc cells to determine their toxicity against prostate cancer cells. 1% Triton X was included as a positive control for cytotoxicity. After 96 hours luciferin (150μg/ml) was added and cytotoxicity was determined by a reduction in bioluminescent activity from the cells. Neither compounds 1-3, nor the vehicle control resulted in a reduction of bioluminescent activity from the PC3-Luc cells. (D) The viability of host HFF cells was assessed by WST-1 staining after 72 hours of incubation of indicated compounds at 10μM concentration. Effect of various concentrations of DMSO present in the HFF culture medium served as a standard toxicity curve.