1-Deoxy-d-xylulose 5-phosphate reductoisomerase as target for anti Toxoplasma gondii agents: crystal structure, biochemical characterization and biological evaluation of inhibitors

Toxoplasma gondii, the causative agent of toxoplasmosis, is an obligate coccidian parasite member of the phylum Apicomplexa [1]. As all apicomplexans, T. gondii possesses a complex and heteroxenous life cycle alternating between sexual stages that occur exclusively in the intestinal epithelium of their definitive hosts (the family Felidae), and asexual stages that can take place virtually in any warm-blooded animal, including humans [2]. Therefore, the pathogenesis of T. gondii is profoundly influenced by the growth rate of its asexual stages [3]. It has reported that, in the United States, ∼11% of the population aged 6 years and older have been infected with T. gondii [4]. In humans, the primary routes of infection are foodborne, caused by the consumption of raw or undercooked meat that contain tissue cysts (bradyzoites), or by ingestion of contaminated vegetables or water containing sporulated oocysts [5,6]. In healthy and immunocompetent individuals, toxoplasmosis typically remains asymptomatic or manifests with flu-like symptoms, since the infection is efficiently controlled by a fully functional immune system [7,8]. On the other hand, the disease poses a significant concern in immunocompromised individuals, often leading to the reactivation of latent infection with severe clinical manifestations, such as chorioretinitis, encephalitis, pneumonitis and sepsis-like symptoms [9]. Moreover, acute T. gondii infections of pregnant women are a relevant risk for the unborn, as maternal-to-fetal transmission of this infection can result in devastating ophthalmic and neurological consequences as well as fatalities for the fetus [10].

Currently, the gold-standard treatment for toxoplasmosis remains the antifolate combination of pyrimethamine and sulfadiazine (PYR-SLZ) [11]. Despite the advancements in target-based drug development in the post-genomic era, human toxoplasmosis lacks sufficient treatment options [12]. Furthermore, all the current available regimens suffer from several limitations and negative aspects that compromise patient compliance and overall effectiveness. These limitations include: a lack of specificity, that can cause severe and potentially fatal side effects; and the inability to effectively target bradyzoites, the dormant cyst-form of the parasite, responsible for the latency of the infection [13]. Therefore, novel, safer and more efficient therapeutic options are urgently needed.

Since its discovery, the apicoplast, a non-photosynthetic plastid organelle in the apicomplexan parasites has emerged has an attractive target for anti-infective drugs due to its absence in mammalian cells [14,15]. This plastid is responsible for essential metabolic pathways for the parasite such as fatty acid, haem, iron sulfur cluster, and synthesis of isoprenoids among others [16]. The 2-C-Methyl-d-erythritol 4-phosphate (MEP) pathway, also called non-mevalonate pathway, for the biosynthesis of isopentenyl pyrophosphate and dimethylallyl pyrophosphate (DMAPP), crucial metabolites for biosynthesis of isoprenoids and essential for the organism, has gathered high interest as a potential drug target (Figure 1). Its enzymes are highly conserved in plastid-bearing organisms and most gram-negative bacteria, and there are no mammalian orthologues, as humans and animals use the mevalonate pathway for the biosynthesis of isoprenoids [17–20].

The 1-Deoxy-d-xylulose 5-phosphate reductoisomerase (DXR, IspC, EC 1.1.1.267), the second and rate-limiting enzyme in the MEP pathway, catalyzes the isomerization and the reduction of 1-Deoxy-d-xylulose 5-phosphate (DXP) into MEP. The catalysis occurs with dependency of a metallic dication (Mg²⁺) and NADPH as cofactors (Figure 2A). DXR enzymes have been extensively investigated and characterized for their key role in the pathway, and in particularly, for the development of novel antibacterial and antimalarial agents [21,22]. Moreover, several structural studies of this enzyme from different organism have been reported with high and medium resolution, such as the crystal structure of DXR from Escherichia coli [23], Mycobacterium tuberculosis [24], Plasmodium falciparum [25], but not from T. gondii.

Fosmidomycin (1) (Figure 2B), a natural product originally isolated from the bacteria Streptomyces lavendulae [26] was identified as a potent and specific inhibitor of the DXR enzymes from E. coli [27] and P. falciparum [28]. Unfortunately, this molecule faces several limitations as a clinical therapeutic related to its pharmacokinetics: short plasma half-life and rapid plasma clearance [29] as well as poor bioavailability due to the ionized nature of the phosphonate group of fosmidomycin at physiological pH [30]. Its polarity affects membrane permeability, except E. coli [31] that actively transports the compound. However, it is not effective in M. tuberculosis [32] and T. gondii [33] probably due to the lack of uptake systems.

Given the significance of DXR inhibitors and with the aim to identify improved inhibitors of TgDXR, in the present study we successfully cloned, expressed, purified and biochemically characterized the recombinant TgDXR. We report the X-ray structure of TgDXR in a tertiary complex with the cofactor NADPH and the inhibitor fosmidomycin, further defined by small-angle X-ray scattering (SAXS) analysis. Moreover, we assessed the in vitro activity of previously described β-thia and β-oxa isosters of reverse hydroxamic acid analogues of fosmidomycin [34–36] against the activity of the recombinant TgDXR enzyme and T. gondii proliferation. These findings could support the future design and development of novel anti-toxoplasma agents.

We set out to functionally and structurally characterize the His₁₀-TgDXR, which contains the catalytic domain. To achieve this, we constructed a plasmid, which contains the catalytic centre consisting of the NADPH binding site as well as the substrate-binding site of the TgDXR protein (amino acid 182–632) (Supplementary Figure S1). After expression, TgDXR was purified to homogeneity as observed by a single symmetric peak on the size exclusion chromatography (SEC) profile and SDS–PAGE analysis (Figure 3). The purified TgDXR was then evaluated for its enzymatic activity through a spectrophotometric assay measuring the oxidation of NADPH over time under various conditions. The optimum of its catalytic capacity was reached with 100 µM of DXP, 100 µM of NADPH, 4 mM of MgCl₂ and a pH of 7.5 at concentration of 100 nM of TgDXR (Figure 3). The kinetic parameters of TgDXR were also evaluated: the K_m value for DXP was determined to be 30.58 ± 6.33 µM, which is comparable with the TgDXR K_m value obtained previously [37] and the reported DXR K_m values from other species, such as PfDXR (67 µM) [38] and MtDXR (47 µM) [39]; the K_m for NADPH was determined as 51.30 ± 11.84 µM (Figure 3). The V_max was shown to be 0.83 mmol/min/mg, comparable to PfDXR (1.04 mmol/min/mg) [38].

TgDXR was crystallised in the presence of the known inhibitor fosmidomycin using the sitting drop method. The 3 µl drops consisted of 1.5 µl protein (7.5 mg/ml in 20 mM TRIS pH 7.5, 150 mM NaCl, 2% glycerol supplemented with 1 mM fosmidomycin, 4 mM MgCl₂ and 4 mM NADPH) mixed with crystallization buffer 1.5 µl 200 mM Na citrate tribasic, 27% PEG smear low, 150 mM HEPES pH 7.8. Crystallization was performed at 12°C and crystals appeared within a few days and grew to their final dimensions within two weeks. Before the crystals were flash-cooled in liquid nitrogen, the crystallization drops were overlaid with mineral oil and the crystals were dragged through it during crystal harvesting. A high-resolution dataset was collected and phased using the model of the DXR protein from E. coli (PDB entry: 1K5H) [40] revealing a dimer in the asymmetric unit.

The structure of TgDXR was solved and refined to 2.5 Å resolution (data-collection and refinement statistics are summarized in Supplementary Table S2). The asymmetric unit contains one homodimer. The r.m.s. deviation between Cα atom positions within in the two subunits when they are superimposed is 0.3 Å using all 347 Cα atoms. Residues 25–472 could be modelled for both molecules from the electron-density map except the loop ranging from amino acid 182–221 and were therefore not modelled in either of the two subunits. First, we describe the overall structure for monomer A.

TgDXR is composed of three domains: an N-terminal NADPH-binding domain, a connective domain and a C-terminal α-helical domain (Figure 4A). These are arranged in a V shape, where the N-terminal and C-terminal domains form the two arms and the central domain lies at the vertex.

The N-terminal domain (residues 25–171) consists of a Rossmann fold, with a β-sheet containing seven parallel β-strands (β1–β7, including a kink at residue 56 in β2), which is flanked by a total of six α-helices (α1–α6). The N-terminal domain is connected via a long loop (residues 172–227) to the catalytic domain. Of this loop, the largest part is not visible in the electron density and appears to be flexible although NADPH and the inhibitor fosmidomycin are present. The catalytic domain (residues 228–385) includes a four-stranded β-sheet ordered β9–β8–β10–β11, where β10 is positioned antiparallel to the other strands. This sheet adheres to the other two domains by virtue of a layer of helices. β8 extents into a flexible loop consisting of a broken helix α7 which then returns into the four-stranded β-sheet via α helix α8.

The C-terminal domain residues (386–472) feature a four-helix bundle. The dimer interface is created by interactions between the catalytic and connecting regions of each subunit. A twisted eight-stranded β-sheet is formed using the four β-strands of each catalytic domain, with the respective β11 strands positioned antiparallel at its central point (Figure 4B). Further antiparallel interactions between the β12 strands of each subunit are found on the concave surface of this larger sheet; interactions at the C-terminal end of the β12 strand links them to the sheet, thus forming an imperfect ten-stranded β-barrel as the core of the dimer interface.

In the electron density, NADPH could be unambiguously identified and modelled. NADPH is bound to the Rossmann fold by interactions typically observed, while the pyrophosphate moiety interacts with the consensus sequence GGGNGA, establishing further interactions with the co-substrate NADPH which is bound at the identical position as found in other DXR proteins [23,25,41,42]. The binding of the adenine and pentose phosphate moieties of NADPH is identical to that observed in the structure of the E. coli DXR, in contrast, the nicotinamide ring of NADPH is ordered in this fosmidomycin complex (Figure 4C).

Within the solvent-shielded cavity that is formed upon closure of the ‘lid’ residues 281–285 the inhibitor backbone lies parallel to the β-indole of Trp283 at a distance of ∼4 Å. The hydroxamic acid moiety of the inhibitor binds to the side-chain of Glu231. The sidechain of Glu231 itself is stabilized and positioned via interaction with Lys299 (Figure 4C). Fosmidomycin further interacts with the sidechains of Asn298 and His280. The sidechain of Glu302 also points towards the fosmidomycin, however the distance is roughly 4 Å indicating a very weak interaction or maybe this interaction is water mediated which cannot be conclusively modelled at this resolution.

The TgDXR protein was successfully crystallised and forms a dimeric conformation in solution. However, parts of the protein were not visible in the electron density, likely due to the flexibility, e.g. parts from the N-terminus and more importantly the flexible loop region from the TgDXR protein. These loop regions are special in this TgDXR variant and are not present in homologous structures. We used SAXS to determine the structure of the TgDXR protein in solution. In the SEC-SAXS elution profile (Figure 5A; and Supplementary Figure S2A), TgDXR elutes in one homogenous peak. Evaluation of the data revealed a dimer in solution with the same shape and orientation as determined by X-ray crystallography, with an R_g value of 3.33 nm and a D_max value of 10.44 nm (Supplementary Table S3). With the Ensemble Optimisation Method (EOM), we modelled the missing amino acids from the loop region and the N-terminus to each protomer, which completed the structure. The corresponding scattering data with the EOM fit (χ²: 1.207) is shown in Figure 5B and the most representative EOM model in Figure 5C. The remodelled loops (37 amino acids each), cover the area between both protomers of TgDXR. Its flexibility, evidenced by the absence of electron density in the crystal structure, led us to analyse the loop's position with EOM, resulting in a 62% occupancy in this conformation.

The loop is located at the back side of the TgDXR protein (Figure 5D) and therefore does not play an immediate role in catalysis.

To identify the binding site essential for the inhibitory activity of fosmidomycin, we obtained point-mutated TgDXR proteins (section ‘Site-directed mutagenesis and expression of TgDXR mutants’) and conducted in vitro TgDXR enzyme inhibition assays (section ‘Enzymatic assays of TgDXR mutants Glu321Ala, His280Ala and Asn298Ala’). We compared the reaction rates of TgDXR wild-type with TgDXR mutants Glu231Ala, His280Ala, and Asn298Ala. Unfortunately, the Glu231Ala mutant could not be expressed and therefore the analysis of this mutation was not possible. Both other mutant proteins (His280Ala and Asn298Ala) displayed significantly decreased activities of only 5–6% activity compared with the wild-type DXR protein (Figure 6), indicating that these amino acids are essential for binding of the substrate DXP. This is in line with the DXR protein from E. coli, where mutations of the same positions and amino acids also resulted in an almost 19-fold reduction of the activity as observed for the wildtype protein [43]. Due to the low activity, the observed inhibition with fosmidomycin was measured but due to the initial low activity a valid IC₅₀ could not be determined for these mutants.

Based on previous studies, demonstrating the effectiveness of reversed hydroxamic acid analogues of fosmidomycin (1) as DXR inhibitors, with either a bivalent sulfur or oxygen atom replacing the β-methylene group of the main chain of the linker [34–36], we tested 5 reverse thia (Figure 7, 2–6) and 4 reverse oxa analogues (Figure 7, 7–10) for their enzymatic activity against the purified recombinant truncated His₁₀-TgDXR and determined their inhibitory concentrations (IC₅₀) and inhibitory constant (K_i) values.

Interestingly, all the investigated compounds possessed TgDXR inhibitory activity. In particular, the reverse thia isosters revealed a marked increase on the inhibitory activity compared with 1 and the cognate reverse oxa compounds (Figure 8, Table 1 and Supplementary Figure S5). Moreover, the 3,4-difluorophenyl- (3) and 3,4-diclorophenyl- (4) substituted thia analogues were the most active inhibitors with comparable submicromolar IC₅₀ and K_i values (Figure 8, Table 1 and Supplementary Figure S5).

To determine if reverse thia and reverse oxa analogues of 1 (Figure 7) could inhibit T. gondii proliferation in vitro, we conducted an evaluation to assess their anti-parasitic activity and their IC₅₀ values against the proliferation of T. gondii (type II ME49 strain) with an ³[H]-uracil incorporation assay in vitro. Interestingly, contrary to 1 that showed no activity accordingly to previous reports [32–37,44]. The TgDXR inhibitors 3, 4, 5, 6, and 7 demonstrated activity against T. gondii growth (Figure 9, Table 2 and Supplementary Figure S6). Moreover, as was first shown in the enzymatic assays, a significant increase in activity for the thia isosters compared with the oxa isoster was observed. Notably, the thia analogue 4 which bears a 3,4-dichlorophenyl moiety in the α-position of the linker demonstrated the most potent inhibitory activity (IC₅₀ = 5.46 µM) (Figure 9, Table 2 and Supplementary Figure S6).

To assess the host cell cytotoxicity of the investigated compounds, MTT assays with Hs27 cells and the TgDXR inhibitors were performed. Similar to 1, all the reverse thia and oxa analogues did not show any detectable host cytotoxicity at the concentration range of 0.19–200 µM (Figure 9, Table 2 and Supplementary Figure S6). Therefore these TgDXR inhibitors appear to have a good therapeutic index.

To elucidate a possible binding mode of compound 4 in TgDXR and to predict which enantiomer of 4 could pose the eutomer, the S- and R-enantiomer of 4 were docked into the X-ray crystal structure of TgDXR from which fosmidomycin was removed. The docking revealed possible binding modes where the R-enantiomer of 4 binds outside the fosmidomycin binding pocket whereas the S-enantiomer of 4 binds into the binding pocket (Figure 10). In its predicted binding mode, the S-enantiomer of 4 binds similarly to fosmidomycin with the phosphonic acid and the hydroxamic acid groups occupying the same positions. The hydroxamic acid group in 4 however is turned compared with fosmidomycin, which might be a result of the change in direction in which this group is attached to the core of 4. This, however, does not allow for interactions of the carbonyl group with Asn298 of TgDXR in contrast with fosmidomycin. Yet, 4 places the 3,4-dichlorphenyl substituent in a previously unoccupied pocket, which increases the number of hydrophobic interactions to TgDXR, which are important for increasing the affinity with the enzyme.

Toxoplasmosis is a widespread disease and current treatment regimens are related to adverse effects and additionally, are not able to eradicate the latent phase of the infection. Thus, the identification of novel drug targets is crucial for the development of novel, potent and safer anti-toxoplasma drugs for improving the health of patients at risk of toxoplasmosis [45]. The MEP pathway has been well established as a promising drug target for novel antimicrobial agents, due to its crucial role on the viability of the microorganism and, additionally, its absence in the human host, that allows the development of selective inhibitors [46].

DXR inhibitors are a class of antimicrobial agents studied intensively in P. falciparum, E. coli and M. tuberculosis [34–36]. From previous SAR studies, it has been shown that the retro-inversion of the hydroxamate moiety of 1 and its N-methylation strongly improve the hydrophobic interaction of Trp212 of EcDXR [47] and that the addition of the α-phenyl substitution of the N-methylated reverse fosmidomycin analogues strongly inhibits the growth of the apicomplexan parasite P. falciparum [48,49].

DXR inhibitors against T. gondii have been already investigated by Cai et al. [37], precisely α-phenyl and α-pyridine substituted fosmidomycin analogues, pyridine containing phosphonate compounds, non-hydroxamate phosphonate inhibitors containing a pyridine and 1-hydroxy-5-phenylpyridin-2-one moiety without a phosphonate group. All the compounds were active against TgDXR, but none of them inhibited the parasite growth, probably due to the permeability barrier posed by parasite plasma membrane(s) of T. gondii that is lacking the fosmidomycin transporter GlpT [33].

In this study, we present the first crystal structure of TgDXR in the tertiary complex with NADPH and fosmidomycin. No magnesium ion was observed in this structure despite the fact that it is essential for activity and was included in the crystallization experiments, which also was the case for the E. coli DXR protein [23]. The overall structure of TgDXR resembles its counterparts in various species [23,25,40,41]. A search for homologous structures revealed that overall the DXR protein are very similar, with an r.m.s.d. of 1.2–1.8 Å for the overall structure (calculated using the EBI fold server). This shows that the overall fold is conserved. Furthermore, also the amino acids involved in DXP substrate binding and conversion are conserved among the different enzymes. Not surprisingly, also the binding site for fosmidomycin at TgDXR consists of the same amino acids as seen in the structure of the EcDXR, indicating that the domain of TgDXR is similar to the EcDXR domain (Figure 4C). The distance of the hydroxymate region of fosmidomycin to NADPH is 4.4 Å, which is too large for conversion (indicated by the yellow line in Figure 4C). With the substrate DXP, the distance to NADPH is reduced, as an additional bond length is strategically positioned to facilitate the occurrence of the reaction. Since the structure is of medium resolution, we could not identify water molecules with certainty, likely however they play a role in catalysis as shown before [23]. Moreover, the parts of the protein that were not visible in the electron density were determined through SAXS, unveiling their peculiarity. With SAXS however only the position of the C-α atoms can be determined, and atomic detail of the side chains is still lacking which would allow a more detailed description.

In addition, we performed a biological investigation on TgDXR, repurposing already investigated reverse α-phenyl-β-oxa and β-thia-substituted analogues of (1) which were previously described as potent inhibitors of P. falciparum, E. coli and M. tuberculosis DXR enzymes [34–36]. We show that all the investigated compounds possess TgDXR inhibitor activity, with the reverse thia analogues presenting a more pronounced activity than the reverse oxa analogues. Moreover, we noticed that inhibitors with a the 3,4-halogen substitution of the phenyl moiety of the reverse thia analogues confers the highest potency against the enzymatic activity of TgDXR (Figure 8, Table 1 and Supplementary Figure S5). In the cell-based assays, we demonstrated that the TgDXR inhibitors 3–7 inhibit T. gondii proliferation in contrast with 1 (Figure 9, Table 2 and Supplementary Figure S6). As shown in previous studies, T. gondii is unaffected by the activity of compound (1) probably due to its lack of the GlpT transporter, the glycerol-3-phosphate (G3P) transporter responsible for the drug uptake of (1) in E. coli [31,33]. T. gondii does not possess a protein with significant sequence similarity to E. coli GlpT. This suggests that the activity of the derivatives of (1) on T. gondii may be due to the use of a different transporter protein or their increased lipophilicity, as indicated by their higher ClogP values compared with (1) in Figure 11, that could facilitate passive transport through the membranes [50].

TgPiT and TgPT2 are plasma membrane transporters found in T. gondii. Notably, TgPT2, which is only present in coccidian parasites, is essential for parasite growth. These proteins, possessing phosphate transport activities [51], could be potential candidates for transporting reverse fosmidomycin derivatives.

Additionally, we observed once again a trend within the thia analogues, showing higher potency compared with the cognate oxa analogues. Previous studies showed that the increased activity of the α-phenyl-β-thia-substituted analogues against EcDXR and MtDXR enzymes is attributed to the interaction of the sulfur atom with the conserved Met298 within the flexible loop of the enzymes [34,48,50]. Furthermore, in this context, the 3,4-dichlorophenyl substituted compound 4 emerged as the most potent one in inhibiting T. gondii proliferation, with an IC₅₀ of 5.46 µM, presumably due the improved lipophilicity provided by the 3,4-dichlorophenyl substitution (Figure 11) that enhances cellular uptake [52].

In summary, in the present study we report the first X-ray structure of TgDXR co-crystallized with fosmidomycin and the cofactor NADPH and its biochemical characterization, as well as the identification of reverse α-phenyl-β-thia- and β-oxa-isosters of fosmidomycin as novel anti-T. gondii agents with in part potent TgDXR inhibitory activity. The ternary co-crystal structure of TgDXR and 1 will serve as starting points for the design and development of improved TgDXR inhibitors. Further efforts are needed for the development of enhanced derivatives and to explore the efficacy in vivo.

The amino acid sequence of the investigated protein TgDXR (NCBI Reference Sequence: XP_018635719.1) was compared with a multiple sequence alignment using Clustal Omega sequence analysis tool (https://www.ebi.ac.uk/Tools/msa/clustalo/) [53] with its default settings with DXR enzymes from other species that have been extensively studied: P. falciparum DXR (PfDXR, NCBI Reference Sequence: AAD03739.1), M. tuberculosis DXR (MtDXR, NCBI Reference Sequence: OHO19719.1) and E. coli DXR (EcDXR, NCBI Reference Sequence: WP_302347400.1) and visualized and the percentage of identity analysed with Jalview software version 2.11.2.7 [54] (Supplementary Figure S1).

The gene of 1-Deoxy-d-xylulose 5-phosphate reductoisomerase of T. gondii (TgDXR) ME49 strain was identified using ToxoDB database (http://ToxoDB.org). The gene of TgDXR (NCBI: NC_031478.1), optimized for expression in E. coli, was obtained from GenScript (Piscataway, Township, NJ, U.S.A.) as a synthetic gene subcloned into the pET-16b plasmid, using NdeI/BamHI restriction sites. The resulting expressed protein carries an N-terminal decahistidine tag (His₁₀). However, the purification of the recombinant His₁₀-tagged TgDXR proved to be challenging. To overcome these issues and to generate a soluble recombinant protein, another expression plasmid was designed. We deleted 181 amino acids (2–182) of the TgDXR sequence representing the bipartite apicoplast targeting peptide [37] in order to generate a soluble truncated His₁₀-tagged TgDXR containing the sequence of the catalytic domain (Supplementary Figure S1).

The generation of the truncated His₁₀-TgDXR commenced with the deletion of 181 amino acids via Site-Directed-Mutagenesis method [55]. For this purpose, the coding sequence of the truncated TgDXR was amplified from the pET-16b plasmid construct with a polymerase chain reaction (PCR) (Phusion® High-Fidelity PCR Kit, New England BioLabs, Frankfurt am Main, Germany, #E0553S) using the primers TgDXR-del181AA_For and TgDXR-del181AA_REV (Supplementary Table S1). Then, the coding sequence was purified through an 1% agarose gel purification [56] and ligated with a KLD- Enzyme Mix (KLD Enzyme Mix, New England BioLabs, Frankfurt am Main, Germany, #M0554S). Subsequently the plasmids were transformed into an E. coli DH5α (DH5α Competent Cells, Thermo Fisher Scientific, Waltham, MA, U.S.A., #EC0112), which then were streaked onto Luria-Bertani (LB) agar-plates containing ampicillin (100 µg/ml) (Ampicillin, Thermo Fisher Chemicals, Thermo Fisher Scientific, Waltham, MA, U.S.A., #J60977.14). To confirm the successful cloning, positive clones were identified through PCR. Additionally, DNA sequencing was performed using T7 forward and reverse primers for further verifications. The TgDXR expression plasmid was extracted with a plasmid miniprep kit (Monarch® Plasmid Miniprep Kit, New England BioLabs, Frankfurt am Main, Germany, #T1010L).

The expression plasmid TgDXR was transformed into the chemically competent E. coli BL21 (DE3) cells (BL21 (DE3) Singles™ Competent Cells — Novagen, Merck KGaA, Darmstadt, Germany, #70235). Then E. coli transformants were streaked onto LB agar-plates containing ampicillin (100 µg/ml) (Ampicillin, Thermo Fisher Chemicals, Thermo Fisher Scientific, Waltham, MA, U.S.A., #J60977.14) and incubated overnight at 37°C.

To express the His-tagged protein, 100 ml overnight pre-cultures were prepared using fresh colonies from LB agar plates or 50% glycerol cryo stock stored at −80°C. The day after, 1 l of freshly prepared and autoclaved LB media (10 g Triptone, 5 g yeast extract and 5 NaCl) was supplemented with 100 μg/ml of ampicillin and inoculated with the pre-cultures to an optical density at 600 nm (OD₆₀₀) of 0.1. The main culture was then incubated with shaking at 37°C and 180 rpm. At an OD₆₀₀ of 0.6, protein expression was induced with the supplementation in the culture of isopropyl-β-d-thiogalactopyranosid (IPTG, Merck KGaA, Darmstadt, Germany, #I6758) to a final concentration of 1 mM. The main culture was further incubated for additional 2 h with shaking at 37°C and 180 rpm. Afterwards, cells were harvested via centrifugation at 5000g, for 15 min at 4°C (rotor SLC-6000, Sorvall, Thermo Fischer, Waltham, MA, U.S.A.), the supernatant discarded. The cell pellet was used subsequently or snap frozen in liquid nitrogen and short-term stored at −20°C.

Truncated His₁₀-TgDXR purification involved two main steps: immobilized metal ion affinity chromatography (IMAC) and SEC.

Cells were thawed at 4°C and suspended with ice-cold lysis buffer A (50 mM NaH₂PO₄, 300 mM NaCl, 20 mM imidazole, pH 8) supplemented with protease inhibitor cocktail (cOmplete Protease Inhibitor Cocktail, Roche, Basel, Switzerland). Subsequently, cell lysis was performed through the application of shear force using a cell disruptor/homogenizer (M-110P Microfluidizer, Microfluidics Inc., Westwood, MA, U.S.A.). This method was employed to effectively break down the cells and release their content. To eliminate cell membranes and other insoluble detritus, ultracentrifugation of the lysate was performed at 100,000g, for 45 min at 4°C (Beckman Optima XE, Beckman Coulter Inc., Brea, CA, U.S.A.). Then, the collected supernatant was purified via Ni-IMAC at 4°C on a Ni²⁺ pre-treated HiTrap IMAC FF 5 ml column (Cytiva Life Science, Marlborough, MA, U.S.A., #17092104) using a protein purification system (Äkta Purifier 10, GE Healthcare, Chicago, IL, U.S.A.). The protein was loaded on the column with a flow rate of 1 ml/min. After binding, the resin was washed with buffer B (50 mM NaH₂PO₄, 150 mM NaCl, 20 mM imidazole, pH 8) to remove the non-specifically bound proteins. Afterwards, the protein was eluted with a gradually increasing concentration of imidazole with buffer C (50 mM NaH₂PO₄, 150 mM NaCl, 300 mM imidazole, pH 8). As determined by absorption at λ 280 nm, the fractions containing proteins were then pooled, concentrated, and subjected to a SEC to further improve its purity. SEC was performed on an Äkta Purifier 10 equipped with a pre-equilibrated Superdex 200 Increase 10/300 GL column (Cytiva Life Science, Marlborough, MA, U.S.A., #28990944) at a flow rate of 0.5 ml/min with buffer D (20 mM Tris–HCl, 150 mM NaCl, 2% Glycerol, pH 7.5). Dimeric protein was collected, concentrated, aliquoted, and snap-frozen for storage at −80°C.

The concentration of the protein was assessed by measuring its absorbance with a microvolume UV-Vis spectrophotometer (NanoDrop™ One, Thermo Fisher Scientific, Waltham, MA, U.S.A.) using protein-specific parameters, including the molecular mass (51,808 Da) and the extinction coefficient of the protein (34,420 M⁻¹ cm⁻¹) obtained with the web-software tool Expasy ProtParam [57].

To biochemically characterize the catalytic ability of the truncated TgDXR for the conversion of DXP into MEP in the presence of the cofactors Mg²⁺ and NADPH, and to determine the optimum conditions for maximum enzyme activity, the enzymatic assays were monitored at 340 nm (maximal absorbance of NADPH) in different conditions.

The enzymatic activity was evaluated in fixed conditions, using 50 mM HEPES buffer containing 50 µg/ml of bovine serum albumin (BSA) containing 100 nM of purified TgDXR protein in dimeric state. In each experimental evaluation, only one parameter was varied: either the substrate (DXP) concentration (ranging from 7.8 to 500 µM), or the cofactor NADPH (ranging from 7.8 to 500 µM), or Mg²⁺ concentration (ranging from 1 to 6 mM) or the pH (ranging from 5.5 to 9) (Figure 3).

For the determination of the kinetic parameters, various concentration of the substrate DXP or NADPH were employed for the determination of the apparent K_m (substrate concentration that yield a half-maximal velocity) and V_max (maximum velocity) for the enzyme. These parameters were calculated by a non-linear regression with the software GraphPad PRISM™ (Version 9.5.1; San Diego, CA, U.S.A.) plotting the initial velocity of the reactions and using the Michaelis–Menten model (Figure 3).

Site-directed mutagenesis of the truncated His₁₀-TgDXR was performed on the pET16-b plasmid (see above) using the Site-Directed-Mutagenesis method in order to generate Glu231Ala, His280Ala, and Asn298Ala mutants. Briefly, the coding sequence of truncated TgDXR was amplified from the pET-16b plasmid using PCR with primers reported in Supplementary Table S2. The amplified sequence was purified via 1% agarose gel and ligated using KLD Enzyme Mix. The plasmids were then transformed into E. coli DH5α cells and streaked onto LB agar plates with ampicillin (100 µg/ml). Positive clones were confirmed by PCR and further verified by DNA sequencing using T7 primers. Finally, the TgDXR expression plasmid was extracted using the Monarch^® Plasmid Miniprep Kit. The mutant TgDXR plasmids were used to transform E. coli BL21 (DE3) competent cells for expression under conditions identical to those used for wild-type expression. Cells were cultured at 37°C in LB medium with shaking until reaching an optical density at 600 nm of 0.6. Protein expression was induced with 1 mM IPTG, followed by an additional 3-h incubation at 37°C. Cells were harvested by centrifugation (5000g for 15 min at 4°C) and stored at −80°C until needed. The histidine-tagged proteins were subsequently purified as described for the wild-type protein (see section ‘Purification of truncated His₁₀-TgDXR’). Structural characterization of the TgDXR catalytic domain.

TgDXR was crystallized by sitting-drop vapor-diffusion at 12°C at a concentration of 7–10 mg/ml with 20 mM TRIS pH 7.5, 150 mM NaCl, 2% glycerol containing 1 mM fosmidomycin, 4 mM MgCl₂ and 4 mM NADPH. Crystals formed after a couple of days, were cryo-protected with 20% ethylene glycol, washed in mineral oil and flash frozen in liquid nitrogen. Diffraction data were collected at 100 K at beamline ID30-B (ESRF, Grenoble, France) using a 0.9793 Å wavelength. Data reduction was performed using XDS [58] and aimless [59] from the CCP4 suite [60]. The structure was solved via molecular replacement with Phaser [61]. The initial model was refined alternating cycles of manual model building in COOT [62,63] and automatic refinement using Phenix [64] version 1.19.2_4158. Data collection and refinement statistics are reported in Supplementary Table S2.

We collected the SEC-SAXS data on beamline BM29 at the ESRF Grenoble [65,66]. The BM29 beamline was equipped with a PILATUS 2M detector (Dectris) at a fixed distance of 2.827 m.

The SEC-SAXS measurement was performed at 10°C with a TgDXR protein concentration of 8.00 mg/ml. The SEC-SAXS run was performed on a Superdex 200 increase 10/300 column (300 µl inject, buffer: 20 mM Tris–HCl, 150 mM NaCl, 40 mM MgCl₂, 2% glycerol, pH 7.5) with a flowrate of 0.5 ml/min. We collected 1500 frames with an exposer time of 2 s/frame and scaled the data to absolute intensity against water.

All used programs for data processing were part of the ATSAS Software package (Version 3.0.5) [67]. Primary data reduction was performed with the programs CHROMIXS [68] and PRIMUS [69]. With the Guinier approximation [70], we determined the forward scattering I(0) and the radius of gyration (R_g). The program GNOM [71] was used to estimate the maximum particle dimension (D_max) with the pair-distribution function p(r). We used the partially solved crystal structure as template in an EOM [72,73] and added the missing amino acids from the loop region and the N-terminus to each protomer.

For the molecular docking the two enantiomers of compound 4 were drawn and converted into 3D using the ChemDraw19 suite. The enantiomers of 4 were subsequently docked into the X-ray crystal structure of DXR from which 1 was removed utilizing a combination of AutoDock as a docking engine and the DrugScore2018 distance-dependent pair-potentials as an objective scoring function [74,75]. During docking, default parameters were used, except for the clustering RMSD cut-off, which was set to 2.0 Å [74–76]. The docking grids were centred on the 1 binding site. Binding modes were considered valid if they were contained in the largest cluster with the most favourable docking energies, which comprised at least 20% of all docking poses.

For figure preparation of the crystal structures of TgDXR enzyme we utilized PyMOL software suite (www.pymol.org) [77].

Fosmidomycin (1) (Fosmidomycin sodium salt, Invitrogen, Thermos Fisher Scientific, Waltham, MA, U.S.A., #FR-31564) was dissolved in Dulbeccós Phosphate Buffered Saline ((1x), Gibco-Thermo Fisher Scientific, Waltham, MA, U.S.A., #14190144). DXR inhibitors (((2-(hydroxy(methyl)amino)-2-oxoethyl)thio)(phenyl)methyl)phosphonic acid (2); ((3,4-difluorophenyl)((2-(hydroxy(methyl)amino)-2-oxoethyl)thio)methyl)phosphonic acid (3), ((3,4-dichlorophenyl)((2-(hydroxy(methyl)amino)-2-oxoethyl)thio)methyl)phosphonic acid (4), ((3,5-difluorophenyl)((2-(hydroxy(methyl)amino)-2-oxoethyl)thio)methyl)phosphonic acid (5), ((3,5-dimethoxyphenyl)((2-(hydroxy(methyl)amino)-2-oxoethyl)thio)methyl)phosphonic acid (6), ((2-(hydroxy(methyl)amino)-2-oxoethoxy)(phenyl)methyl)phosphonic acid (7), ((3,4-difluorophenyl)(2-(hydroxy(methyl)amino)-2-oxoethoxy)methyl)phosphonic acid (8), ((3,4-dichlorophenyl)(2-(hydroxy(methyl)amino)-2-oxoethoxy)methyl)phosphonic acid (9), ((2-(hydroxy(methyl)amino)-2-oxoethoxy)(p-tolyl)methyl)phosphonic acid (10) employed in this study (Figure 7) were prepared using procedures from previously published methods [34–36] and dissolved in DMSO (dimethyl sulfoxide, ≥99%, Thermo Scientific Chemicals, Waltham, MA, U.S.A., #A12380.36). Staurosporine (Merck, Darmstadt, Germany, #S4400) and pyrimethamine (Merck, Darmstadt, Germany, #219864) were dissolved in DMSO. All the compounds were prepared as 10 mM stock solutions and stored at −20°C. Before use, these solutions were thawed and diluted in culture medium to produce the appropriate concentrations (ranging from 0.0003 to 200 µM).

The enzymatic assays were conducted at 30°C in 96 well plates using a total reaction volume of 150 µl containing 100 nM of purified TgDXR protein in dimeric state, 100 µM of NADPH and 4 mM of MgCl₂ as cofactors, 100 µM of DXP as substrate in 50 mM HEPES buffer (pH 7.5) containing 50 µg/ml of BSA. For the screening, DXR inhibitors were tested for their inhibitory activity and their IC₅₀ measurements at concentrations ranging from 100 µM to 3.05 nM, in dilution steps 1:2.

To optimize and to ensure the interaction of NADPH and its enzyme binding pocket, the assay solution was incubated for 10 min at 37°C. Then, the reaction was commenced with the addition of 100 µM of DXP to the complete assay mixture. The reaction was monitored by measuring the absorption at 340 nm every minute for 1 h using a microplate reader (Tecan® 200 Pro, Tecan Group, Männedorf, Switzerland). The initial velocity of the reactions were calculated, the values were then used in the software GraphPad PRISM™ (Version 9.5.1; San Diego, CA, U.S.A.) for determinations of the IC₅₀ values. The inhibitor constant values (K_i) of all inhibitors were determined using the Michaelis–Menten formula K_i = IC₅₀/(1 + [S]/K_m), [S] is the concentration of DXP (100 μM) and K_m was calculated as 30.58 μM.

After purifying the TgDXR mutants Glu321Ala, His280Ala and Asn298Ala, their activities were tested in comparison with the TgDXR wildtype. Enzymatic assays were conducted under the same conditions described in section ‘T. gondii DXR enzyme inhibition assays’.

T. gondii tachyzoites of the ME49 strain (ATCC/LGC Standards GmbH, Wesel, Germany, #50611), were cultured and maintained by repeat passage in monolayers of human foreskin fibroblasts Hs27 (ATCC/LGC Standards GmbH, Wesel, Germany, #CRL-1634) as host cells. Cultures were grown in Iscove's modified Dulbecco's medium (IMDM; Gibco-Thermo Fisher Scientific, Waltham, MA, U.S.A., #12440053) supplemented with 10% heat-inactivated fetal bovine serum (FBS Standard; South America origin, FBS, 2 µm sterile filtered, PAN-Biotech, Aidenbach, Germany, #P30-3306) and 50 mM 2-mercaptoethanol (Gibco-Thermo Fisher Scientific, Waltham, MA, U.S.A., #21985023) at 37°C and 5% CO₂ as previously described [78–80].

With the aim to assess the inhibitory in vitro effect on parasite growth, T. gondii growth assays were performed as described previously [78–80]. Compounds, previously diluted in culture media in appropriate stock solutions, were added to confluent monolayers of Hs27 cells in 96-flat well plates at various concentrations. Then, freshly harvested tachyzoites were added to the cultures at a multiplicity of infection of 1:1 (parasite/host cell ratio). As controls, untreated and uninfected Hs27 cells, 24 h pre-stimulated and T. gondii infected cells with human interferon γ (IFN γ) (300 U/ml) (Merck, Darmstadt, Germany, #I17001) and T. gondii infected cells only were employed. After 48 h incubation, proliferating toxoplasmas were labelled with 0.3 μCi/well of tritiated uracil (³H-U; 5 mCi, Hartmann Analytic, Braunschweig, Germany, #ART1782) [81]. After 28–30 h of incubation, plates were frozen at −20°C overnight. Cells were then thawed and harvested (Basic96 Harvester; Zinsser Analytic, Skatron Instruments, Northridge, CA, U.S.A.), purified with glass fibre filters (Printed Filtermat A 102 mm × 258 mm; PerkinElmer, Waltham, MA, U.S.A.) and dried at 130°C for 15–20 min. Dried filters were then wrapped into transparent plastic covers, dunked with 10 ml scintillation cocktail (Betaplate Scint; PerkinElmer, Waltham, MA, U.S.A., #1205-440), sealed and then loaded into metal cassettes and the incorporation of ³H-uracil into the RNA of proliferating toxoplasmas was measured using a β-counter, a liquid scintillation counter that measure the Cherenkov radiation (Betaplate Liquid Scintillation Counter 1205; LKB-WALLAK, Melbourne, Australia). All data sets were normalized to 100% of the positive control. The dose-response curves of the compounds were fitted by means of the statistics software GraphPad PRISM™ (Version 9.5.1; San Diego, CA, U.S.A.). The minimal concentrations of compounds required for 50% inhibition of the parasite (IC₅₀ values) were determined by non-linear regression analysis. Pyrimethamine [13,82] was used as reference anti-toxoplasma compound.

All supporting data are included within the main article and its supplementary files. We uploaded the SAXS data to the Small Angle Scattering Biological Data Bank (SASBDB) [85], with the accession codes SASDS47. The structure of TgDXR has been deposited in the Protein Databank (PDB) under accession code: 8S65.

The authors declare that there are no competing interests associated with the manuscript.

This work was supported by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) project number 270650915/GRK 2158 (to T.K., S.H.J.S., and K.P.). The Center for Structural Studies (CSS) is funded by the DFG (grant number 417919780 INST 208/740-1 FUGG, INST 208/868-1 FUGG and INST 208/761-1 FUGG to S.H.J.S.).

Klaus Pfeffer: Conceptualization, Resources, Supervision, Funding acquisition, Project administration, Writing — review & editing. Flaminia Mazzone: Conceptualization, Data curation, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing — original draft, Writing — review & editing. Astrid Hoeppner: Resources, Data curation, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing — original draft, Writing — review & editing. Jens Reiners: Resources, Data curation, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing — original draft, Writing — review & editing. Christoph G.W. Gertzen: Software, Formal analysis, Investigation, Visualization, Methodology, Writing — original draft, Writing — review & editing. Violetta Applegate: Data curation, Formal analysis, Validation, Investigation, Methodology, Writing — review & editing. Mona Abdullah Abdullaziz: Resources, Writing — review & editing. Julia Gottstein: Methodology, Writing — review & editing. Daniel Degrandi: Conceptualization, Supervision, Writing — review & editing. Martina Wesemann: Conceptualization, Methodology, Writing — review & editing. Thomas Kurz: Conceptualization, Resources, Supervision, Funding acquisition, Project administration, Writing — review & editing. Sander H.J. Smits: Conceptualization, Resources, Data curation, Formal analysis, Supervision, Funding acquisition, Investigation, Visualization, Writing — original draft, Project administration, Writing — review & editing.

We gratefully acknowledge the DFG (GRK2158) and the Heinrich Heine University (HHU) for their generous support. We thank Karin Buchholz and Ursula Sorg for scientific consultation. Furthermore, we thank Christian Mammen and Giuseppe Andreacchio for experimental assistance and Eymen Hachani for SEC-MALS measurement and analysis. We acknowledge the European Synchrotron Radiation Facility for provision of synchrotron radiation facilities and we would like to thank Petra Pernot for assistance in using beamline BM29. We also thank the beamline staff of the beamline ID30-B at the ESRF for help during remote collection of the data. Initial crystals were also tested at the EMBL beamline at the DESY synchrotron in Hamburg, Germany which we would also like to acknowledge.

BSA

bovine serum albumin

DMSO

dimethyl sulfoxide

DXP

1-Deoxy-d-xylulose 5-phosphate

EOM

Ensemble Optimisation Method

FBS

fetal bovine serum

IMDM

Iscove's Modified Dulbecco's medium

IPTG

isopropyl-β-d-thiogalactopyranosid

LB

Luria-Bertani

MEP

2-C-methyl-d-erythritol 4-phosphate

PCR

polymerase chain reaction

SAXS

small-angle X-ray scattering

SEC

size exclusion chromatography