Biochemical characterization of Mycobacterium tuberculosis dihydroorotate dehydrogenase and identification of a selective inhibitor

Mycobacterium tuberculosis (MTB) is the etiologic agent of tuberculosis (TB), an ancient disease which causes 1.5 million deaths worldwide. Dihydroorotate dehydrogenase (DHODH) is a key enzyme of the MTB de novo pyrimidine biosynthesis pathway, and it is essential for MTB growth in vitro, hence representing a promising drug target. We present: (i) the biochemical characterization of the full‐length MTB DHODH, including the analysis of the kinetic parameters, and (ii) the previously unreleased crystal structure of the protein that allowed us to rationally screen our in‐house chemical library and identify the first selective inhibitor of mycobacterial DHODH. The inhibitor has fluorescence properties, potentially instrumental to in cellulo imaging studies, and exhibits an IC50 value of 43 μm, paving the way to hit‐to‐lead process.


Tuberculosis (TB) is an infectious disease caused by
Mycobacterium tuberculosis (MTB) that nowadays still claims more than 1.5 million deaths worldwide, making MTB the world's top infectious killer. According to the most recent Global Tuberculosis Report (2022), the COVID-19 pandemic made the TB management even more difficult, reversing years of progress against the disease by increasing the number of TB deaths for the first time after a decade [1]. Moreover, the global scenario is becoming more alarming if we consider the spreading of multidrug-resistant (MDR) MTB strains, mostly referred to as rifampicin/isoniazid-resistant TB, as well as extensively drug-resistant (XDR) MTB strains that are resistant to first-and second-line antitubercular drugs [2]. In this context, where also the adherence to the available long-term antitubercular treatments is poor, the development of antitubercular agents is in high demand as well as the identification of innovative molecular targets guided by the genotypic information derived from MTB genome sequencing [3].
Pyrimidine biosynthesis pathway represents a valid source of therapeutic targets since dysfunctions in nucleotide metabolism are detrimental for bacterial survival because vital for proliferation and for bacterial growth in human blood during septicemia [4][5][6][7][8]. While nondividing cells satisfy their need in pyrimidines through a salvage pathway (Fig. 1A), proliferating cells and infecting pathogens fulfill their large demand for nucleotide precursors with the de novo pathway (Fig. 1A) [5], consisting in a multienzymatic cascade that culminates in the production of uridine monophosphate (UMP), the precursor of all pyrimidine nucleotides [9]. In MTB, the genes encoding five of the enzymes involved in the pyrimidine biosynthesis belong to the pyr operon and are essential for the growth of MTB in vitro [9,10]. However, most of the knowledge of the biochemistry of pyrimidine biosynthesis in MTB comes from studies performed on reference bacteria.
The fourth step in the de novo pathway refers to the oxidation of dihydroorotate (DHO) to orotate (ORO), and it is carried out by the rate-limiting flavoenzyme dihydroorotate dehydrogenase (DHODH), encoded by the gene pyrD (Rv2139) that catalyzes the formation of the 5,6-insaturation of the pyrimidine base. The reaction proceeds through a ping-pong mechanism where the flavin mononucleotide (FMN) is firstly reduced into FMNH 2 , and then it is oxidized again to FMN by a third molecule that acts as an electron acceptor.
DHODHs are divided into class 1 and class 2, depending on sequence similarity, subcellular location and substrate preference [11]. DHODHs belonging to class 1 are cytosolic enzymes and are divided into class 1A and class 1B, which differ for the structural and mechanistic features because they use fumarate [12,13] and NAD +  [14] as electron acceptors, respectively. On the contrary, class 2 enzymes are monomeric proteins that exploit the N-terminal hydrophobic region to be anchored to the inner membrane of the mitochondria in eukaryotes [15,16], and in the cytosolic membrane in some prokaryotes [17]. For class 2 proteins, the FMN cofactor is regenerated by transferring electrons to ubiquinone in humans or to menaquinone (MQ) in MTB. Indeed, MQs (2-methyl-3-polyprenyl-1, 4-naphthoquinones) are the predominant lipoquinones of mycobacteria since they lack the pathway for the synthesis of ubiquinone [11,18]. Since MTB is an intracellular pathogen, it could potentially respond to MQ auxotrophy by salvaging the ubiquinone from the human host; consequently, demonstrating the MTB DHODH (MtDHODH) selectivity toward different lipoquinones will add insights to the complexity of MTB life cycle and crosstalk with the host. Prokaryotic DHODHs are considered potential drug targets in conventional pathogen-directed therapies (PDT). The ubiquinone-binding site in human DHODH (hDHODH) is known to be significantly divergent from its parasitic counterpart and, several groups have reported the discovery of pathogen-specific DHODH inhibitors (Helicobacter pylori, Toxoplasma gondii, Plasmodium falciparum, and Schistosoma mansoni) [19]. Notably, phase Ia/Ib clinical studies of the potent and specific Plasmodium falciparum DHODH inhibitor DSM265 have recently been published with promising results [20]. Moreover, combining host-directed therapy (HDT) and PDT strategies represents a potentially effective method for lowering the insurgence of drug-resistant strains and for shortening the duration of the therapies, a common problem of conventional TB treatments.
MTB pyrD is essential for MTB growth in vitro [21], hence representing a promising drug target in a PDT, that can be potentially combined with HDT targeting human hDHODH. Indeed, unlike parasites such as Helicobacter pylori, Plasmodium falciparum, and Schistosoma mansoni, which lack salvage pathways for pyrimidine nucleotides, in MTB the pyrimidines are synthesized both through the de novo biosynthesis and via the salvage pathways under physiological conditions similarly to mammalian cells [22].
To the best of our knowledge, MtDHODH is fully uncharacterized since the literature lacks data on the biochemical and structural characterization of the enzyme. In particular, the only structural model, which has been deposited in the protein data bank (PDB), refers to a truncated version of the protein (PDB: 4XQ6) not reported in a peer-reviewed, indexed scientific journal. The aim of the present work is to fill the gap between the potential druggability of the enzyme and the lack of biochemical data. Accordingly, herein we present the crystallographic structure of MtDHODH coupled with the biochemical investigation of the protein. Moreover, a structure-based inhibitor screening allowed us to identify the first MtDHODH inhibitor that showed also fluorescent properties, potentially instrumental to in cellulo imaging studies, paving the way to the development of a new class of antimycobacterial agents with fluorescent properties, with interesting applicative perspectives.

Proteins expression, purification, and crystallization
The open reading frame of MtDHODH (identifier: RV2139) was subcloned into pET-28a(+) (GenScript, Piscataway, NJ, USA) plasmid using NdeI/XhoI recognized sequences as restriction sites. The plasmidic vector was designed to induce the expression of the recombinant N-terminal His-tagged protein (predicted molecular mass of 40 161 Da) exploiting E. coli BL21(DE3) (Novagen) as host system. E. coli was transformed with the target construct and grown on an agar plate for 16 h at 37°C, and then cells were inoculated in 1 L of sterile 2XYT liquid broth in the presence of Kanamycin (50 lg/mL) as selector. Bacterial culture was grown in a 5 L shaking flask at 5 g/37°C until they reached optical density at 600 nm (OD600) of 0.6. The expression of the recombinant MtDHODH was induced with 0.2 mM isopropyl-1thio-D-galactopyranoside, and cell suspension was further incubated at 5 g/16°C for 15 h. Culture was harvested and centrifugated at 8 000 g/4°C for 10 min 8 g of bacterial pellet were resuspended (1 g: 9 mL) in lysis buffer (LB) [100 mM HEPES pH 7,8, 300 mM NaCl, 10% (v/v) Glycerol, 0.1% (v/v) Triton X-100 and 1 mM ORO] supplemented with complete EDTA-free protease inhibitor cocktail (Merck) and DNAse. All the following steps were performed at 4°C. Cells were lysed with a Sonics Vibra-Cell VC 130 Ultrasonic Homogenizer (Strokes: 8; Pulse: 30 00 ; Stop: 1 0 ; Amplitude: 45). The lysate was centrifuged at 64 000 g/4°C for 50 min (Beckman Coulter Avanti Centrifuge J-26 XP), and the cleared cell lysate was loaded onto a Qiagen Ni-NTA Agarose column pre-equilibrated with LB supplemented with 10 mM imidazole. After washing the resin with 20 CV of the wash buffer (LB supplemented with 50 mM imidazole), the protein was eluted with the elution buffer (LB supplemented with 300 mM imidazole). The fractions containing MtDHODH resulted to be yellow since the presence of FMN prosthetic group directly bound to the protein core. The positive fractions, checked through 12.5% acrylamide SDS/PAGE, were pooled and concentrated using Amicon 15-10 000 MWCO centrifugal concentrator (Merck). Before undergoing crystallization studies, the concentrated protein was loaded onto a Hiload Superdex 200 16/600 (GE Healthcare, Chicago, IL, USA) column, pre-equilibrated with a size exclusion buffer [100 mM HEPES pH 7,8, 300 mM NaCl, 10% (v/v) Glycerol, % (v/v) Triton X-100 and 1 mM ORO]. Size exclusion chromatography was monitored at 350 and 442 nm wavelengths to avoid interference with the absorbance peak of ORO. Protein quantification was performed by Bradford assay in a calibrated system at 595 nm with a Savatec Onda spectrophotometer UV-21. Analytical size exclusion chromatography was performed using a Superdex 200 10/300 GL column (Cytiva), precalibrated with standard proteins (following manufacturer's instructions), and by fluxing the size exclusion buffer [100 mM HEPES pH 7,8, 300 mM NaCl, 10% (v/v) Glycerol 1 mM ORO and 0.25% n-Undecyl-N,N-Dimethylamine-N-Oxide (UDAO) w/v] as the mobile phase. To evaluate the stability of the protein, different aliquots of the MtDHODH preparation [stock buffer: 100 mM HEPES pH 7,8, 300 mM NaCl, 10% (v/v) Glycerol, % (v/v) Triton X-100 and 1 mM ORO] have been stored for up to 1 week at different temperatures, before being analyzed by standard SDS/PAGE and differential scanning fluorimetry in order to compare the different stocks in terms of thermal denaturation in the presence of SYPRO Orange dye.
For crystallization assay, the purified protein was mixed with ORO to give a final concentration of 2 mM and incubated at 4°C over night. 0.5 lL of the complex protein-ORO were mixed with 0.5 lL of reservoir solution from Classics II Suite (QIAGEN, Hilden, Germany) and underwent crystallization trials by means of robot-assisted (Oryx4; Douglas Instruments, Hangerford, UK), sittingdrop-based spare-matrix strategy. After 3 months, a single bar-shaped and fragile crystal grew in G1 drop condition (0.2 M Sodium Chloride, 0.1 M Tris-HCl pH 8.5 and 25% (w/v) PEG 3350).
X-ray data collection, structure determination, and refinement The crystal was cryoprotected with 12.5% (v/v) Glycerol, flash-cooled in liquid nitrogen and sent to the European Synchrotron Radiation Facility (ESRF), France, where they underwent X-ray diffraction experiment on the beamline ID23-2 with a Pixel Pilatus3_2M as detector [23].
The data were indexed using XDS program; then they were integrated and scaled to a resolution of 3.8 A using the Aimless utilities of the CCP4i2 Program Suite version 8.0.008 [24]. The crystal was assigned to the orthorhombic space group P2 1 2 1 2 1 with the cell dimensions a = 46.0 A, b = 85.2 A, c = 180.9 A, containing 2 molecules in the asymmetric unit, with a Matthews coefficient and a solvent percentage of 2.21 A 3 Da À1 and 44.40%, respectively ( Table 1). The structure was determined by molecular replacement with phenix-PHASER [25] using the predicted structure provided from Alphafold database as a search model (AF-P9WHL1-F1-model). Manual model building was performed with Coot program [26], structure refinement was done with PHENIX [25], and pictures were generated with PyMol [27]. Data collection and refinement statistics are listed in Table 1.

PDB deposition
The atomic coordinates and structure factors of MtDHODH have been deposited in the Protein Data Bank (PDB) as 8OFW code.

DHODH enzymatic and inhibition assays
The enzymatic activity of both MtDHODH and hDHODH was monitored using a Tecan Sunrise spectrophotometer in a coupled reaction which involves the reduction of 2, 6dichloroindophenol (DCIP) reagent. The reduction of DCIP is stoichiometrically associated with the enzymatic oxidation of one equivalent of the substrate DHO. The purified recombinant protein was tested at 0.1 lM alone or complex with the inhibitor MEDS322 (tested at increasing concentration from 12.5 to 300 lM). After 20 min of preincubation at 37°C in the mixture of activity [50 mM Tris-HCl pH 8.0, 300 mM NaCl, 5% (v/v) Glycerol, 0,1% (v/v) Triton X-100, 50 lM DCIP, and 100 lM MQ (dissolved in DMSO)] up to a final volume of 100 lL, the reaction was initiated by addition of DHO at increasing concentration from 1.56 to 200 lM and the reduction was monitored at 600 nm for 10 min. For the calculation of the biochemical parameters toward MQ cofactor, the experimental setting was the same described above keeping the concentration of DHO fixed to 100 lM and varying the concentration of MQ from 12 to 400 lM. The curves were analyzed in their maximum drop in the first 5 min of the enzymatic reaction. Biochemical parameters were calculated using GraphPad Prism software.

Synthesis of MEDS322
Analytical grade solvents (diethyl ether, methanol [MeOH], glacial acetic acid) were used without further purification. When needed, solvents were dried on 4 A molecular sieves. Thin layer chromatography (TLC) on silica gel was carried out on 5 9 20 cm plates with 0.25 mm layer thickness to monitor the process of reactions. Anhydrous Na 2 SO 4 was used as a drying agent for the organic phases. Purification of compounds was achieved with flash column chromatography on silica gel (Kieselgel 60, 230-400 mesh ASTM; Merck) using the eluents indicated. Final compounds were assayed in biological experiments. Their purity was measured by HPLC analyses, showing a chromatogram where the main peak (area at least 95% of all detected peaks) was attributable to the final compound. Some HPLC analyses were performed on a UHPLC chromatographic system (Flexar; Perkin Elmer, Waltham, MA, USA). The analytical column was an UHPLC Acquity CSH Fluoro-Phenyl (2.1-100 mm, 1.7 mm particle size; Waters, Milford, MA, USA). Compounds were dissolved in acetonitrile or methanol and injected through a 20 mL loop. The mobile phase consisted of methanol/water with 0.1% trifluoroacetic acid; two gradient profiles of mobile phase were used to assay the purity of each compound. UHPLC analysis was run at flow rates of 0.5 mL/min, and the column effluent was monitored at 215 and 254 nm, referenced against a 360 nm wavelength. All compounds were routinely checked by 1 Hand 13 C-NMR and mass spectrometry. 1 H-and 13 C-NMR spectra were performed on a Br€ uker 300 MHz Ultra shield spectrometer. For coupling patterns, the following abbreviations are used: br = broad, s = singlet, d = doublet, dd = doublet of doublets, t = triplet, q = quartet, and m = multiplet. MS spectra were performed on Finnigan-Mat TSQ-700 (70 eV, direct inlet for chemical ionization [CI]).

Expression and purification of wild-type fulllength MtDHODH
As of today, 143 structures of DHODHs in 15 different organisms are described in Protein Data Bank (PDB); among which, however, MTB is missing. Indeed, MtDHODH lacks a complete structural characterization because the only provided experimental model (PDB code: 4XQ6) is a N-terminal truncated version of the protein (Val31-Arg353) where the DNA coding for Met1-Ala30 residues was not included in the cloning vector.
In initial attempts, SDS/PAGE and western blot assays of the isolated enzyme revealed proteolytic degradation at the level of the N-terminal domain (data not shown). To overcome the issue, the addition of 1 mM ORO [28] and of twofold more of the protease inhibitor cocktail in the lysis buffer was enough to preserve the protein integrity. Moreover, the strongly hydrophobic nature of the outer additional residues made the full-length structure sticky and more prone to aggregation and, consequently, more susceptible to amorphous precipitation. In this regard, to avoid nonfunctional multimeric adducts, different ligands have been tested (i.e., 100 lM MQ, 0.25% w/v n-Undecyl-N,N-Dimethylamine-Oxide and 0.1% of triton X-100 as detergent). The protein eluted in a single welldefined peak at the elution volume consistent with a monomeric assembly in solution (Fig. S2) and it was concentrated up to 9 mg/mL. Therefore, with an optimized expression and purification protocol, we managed to obtain the pure monodisperse protein in a yield that was consistent with crystallization trials. The stability of the recombinant protein has been evaluated over time (1 week) at different temperatures (namely 20, 4 and À80°C). By means of SDS/PAGE analysis and differential scanning fluorimetry (Fig. S3), we demonstrated that the protein does not undergo limited proteolysis or degradation upon time and that the melting temperature of the protein (57.6°C) remains constant when measured on protein samples stored at different temperatures. Stability insights are further supported by protein activity not being altered over time.

Structural analysis of MtDHODH
The best crystal of full-length MtDHODH diffracted at 3. 8 A of resolution, and the structure shows the typical a/b TIM barrel fold common to other DHODHs, with the FMN molecule occupying an opening of the b-barrel core that is surrounded by eight a-helices (H4-H11; Fig. 2A). Unfortunately, the resolution range did not allow to detect the electron density of the substrate ORO. Residues 116-118 of chain A and residues 283-286 and 340-345 of chain B are absent from the model due to the local weak or missing electron density. Instead, a clear and unambiguous electron density map defines the N terminus, which allowed the reliable building of the so-far uncharacterized N terminus of MtDHODH (Fig. S4). This region, stretching from amino acids 1-40 (3-40 in the structure model), caps the FMN cofactor and forms the binding site of the DHODHs inhibitors competitive with respect to the electron carrier molecules ubiquinone and MQ for hDHODH and MtDHODH, respectively, as observed in their crystal structures (PDB codes 2PRM and 8OFW; Fig. 2B). In particular, the prosthetic group is positioned in its binding pocket mainly composed by loops that include residues Ala66, Asn241, Thr242, Ser263, Gly264, Gly292, Gly293, Gly313, Tyr314, and Thr315 establishing polar contacts with FMN ( Fig. 2A). On the top of the b-barrel, beside the FMN there is the active site composed by the consensus sequence of class 2 DHODHs (Ser-Ser-Pro-Asn-Thr), housing the catalytic Ser179 that is oriented toward the FMN on the opposite side with respect to the N-terminal.
Comparative analysis with the human ortholog revealed a high degree of structural conservation between the two proteins (RMSD of C-alpha atomic positions between MtDHODH and hDHODH is 0.93 A). The main discrepancies reside at the Nterminal domain where the hDHODH misses the H3 helix (Ala32 to Leu41), which in the human protein is replaced by a long loop oriented toward the central core of the enzyme narrowing the ubiquinone-binding pocket and contributing to the steric hindrance as a possible element to be considered for selective molecules (see Discussion below). Moreover, hDHODH is characterized by an amino acidic insertion at level of H5 (aa 168-175 as numbered in human protein, PDB: 2PRM; Fig. 2B).

Biochemical characterization of MtDHODH
MtDHODH enzyme catalyzes the conversion of DHO into ORO using FMN as the oxidizing substrate that is further regenerated by the presence of another electron acceptor that, for the mycobacterial enzyme, corresponds to the MQ. Accordingly, the enzyme activity has been measured by means of DCIP as an alternative electron acceptor, as already demonstrated for other DHODHs [29][30][31]  k cat of 401 min À1 (Fig. 3A,B), demonstrating that the DHO represents the rate-limiting factor of the reaction. Moreover, the protein activity was assayed over the range pH 6-10, using a series of constant ionic strength buffers. Fig. S5 shows the dependence of reaction velocity from pH profile, monitored by the DCIP assay. The enzyme shows an activity maximum at pH 9.0 and has no activity below pH 6.0. With the same experimental setting, we evaluated the capability of mycobacterial enzyme to use ubiquinone as oxidizing cofactor in order to validate the possibility of hostpathogen molecular cross talk at level of electron acceptor. The protein did not show enzymatic activity in presence of ubiquinone sustaining a potential selectivity of inhibitor molecules designed to interfere with MtDHODH activity at level of quinone-binding site.
As depicted in Fig. 4A, an accurate structural analysis of quinone-binding sites can provide the molecular details for the development of selective MtDHODH inhibitor. When compared, the host and the pathogen proteins revealed to have some similarities: the hydrophobic feature of the 'lipophilic patch' is predominantly maintained in accordance with the chemical properties of both ubiquinone and MQ molecules, whereas notable changes concern the peculiar substitution from hDHODH to MtDHODH of Gln47 with Phe10, of Met43 with Arg6, of Pro365 with Glu322 and of Thr63 with Leu25, respectively. Among these discrepancies, the Gln47-to-Phe10 substitution could explain the lack of activity of some inhibitors that have been developed for the human ortholog (Fig. 3C). In particular, MEDS433 establishes an ion bridge extending to the side chain of Arg136 and a hydrogen bond with the side chain of Gln47 that is potentially lost in MtDHODH [32]. Moreover, the molecular architecture of the N-terminal region of MtDHODH is built on a three helices assembly, where the position of helix H3 strongly differs from the corresponding loop in hDHODH that is oriented toward the central core of the protein, narrowing the quinonebinding pocket (Fig. 4B). In order to achieve inhibitor selectivity, the modification of the chemical-physical environment at the level of the active site can be exploited to drive selective drug design.
Furthermore, Arg136 of hDHODH, paramount residue for inhibitors binding activities [15,32], is maintained in its identity in the MTB orthologue (Arg106), representing the key for an initial exploration of hDHODH inhibitor libraries.

Identification of a selective MtDHODH inhibitor
In order to identify a selective MtDHODH inhibitor to be considered as potential antimycobacterial agent, we screened our in-house chemical library consisting in more than 1200 compounds based on the hydroxyazoles scaffold and other heterocycles, with the characteristic of being carboxylic acid bioisosteres and able of establishing strong interactions with the positive ionized residues of proteins as arginine. Among them, we also included molecules, derived from the research pipelines aimed at identifying human [32][33][34][35] and Plasmodium falciparum DHODH inhibitors [36], that did not show activity toward their targets. The selection of potential MtDHODH inhibitors from the in-house library has been performed following two basic principles: (i) identification of molecules having a chemical scaffold similar to others that have been reported with antimycobacterial activity; and (ii) presence of the carboxylic acid terminal or bioisosteres, potentially able of establishing strong interactions with the arginine located in the MQ binding pocket that is highly conserved and interacts with most of the well-known DHODH inhibitors. With his approach, we identified 52 molecules that underwent in vitro investigation through enzymatic assay. Among them, compound 1 (hereafter MEDS322) resulted as the most potent molecule of the tested series. It contains the mentioned carboxylic acid and the thiazolidinone ring that is bioisostere of urea derivatives whose antimycobacterial activity was described by Brown and coauthors [37]. MEDS322, whose synthesis is described below and in the Scheme 1, inhibits MtDHODH with an IC 50 = 43 lM (Fig. 5A) and represents the first selective inhibitor of MtDHODH, since it showed no inhibitory activity on the human orthologue (produced as already described in reference [35]) (Fig. S1). Furthermore, MEDS322 showed fluorescence emission at 530 nm following excitation at 350 nm (Fig. 5B), and additional proof of MEDS322 binding was obtained by measuring the emission intensity of the molecule at increasing concentrations of MtDHODH which resulted in an evident increase of the emission intensity, proportional to the protein concentration. As already demonstrated for other proteins having fluorometric properties [38,39], this effect could be attributed to the transition of MEDS322 from the polar environment of the bulk solvent to the quinone pocket that is characterized by a higher hydrophobicity.
The published structures of hDHODH in complex with various inhibitors together with the here reported first crystal structure of the full-length MtDHODH allowed us to rationalize the molecular determinants of MEDS322 selectivity toward MtDHODH.   A representative binding pose of a hDHODH inhibitor and the related molecular interactions are reported by Sainas et al. [32], which showed the compound MEDS433 snugly inserted in the N-terminal ubiquinone-binding site which is constituted by helices H1 and H2 and a structured loop surrounding the inhibitor (amino acids Leu65-Ala71). Such binding pose guided the docking simulations of compound MEDS322 into the MtDHODH experimental structure. This was performed using the HADDOCK server by imposing docking parameters consistent with the molecular interactions of compound MEDS433 to hDHODH [32]. However, in MtDHODH, the structured loop of hDHODH is replaced by a-helix H3 (amino acids Ala32-Leu41) which enlarges the binding site and that we anticipated to drive inhibitor selectivity. Indeed, our docking simulations showed that the a-helix H3 of MtDHODH is conveniently placed to accommodate the MtDHODH-selective inhibitor MEDS322, whereas the structured loop of hDHODH clashes with the chloroaniline moiety of MEDS322, thus hampering its binding to hDHODH and rationalizing its inefficacy on hDHODH inhibition and its selectivity for MtDHODH. Fig. 6 reports the docking of MEDS322 to MtDHODH and the structural alignment with hDHODH in complex with MEDS433 inhibitor. MtDHODH accommodates the MEDS322 ligand through the interaction of its carboxyl group with Arg106, thus anchoring and orienting the whole inhibitor. The docked molecule engages also in key hydrophobic interaction with residues Leu25 and with Leu38 of the structure, hence playing a determinant role in inhibition and selectivity (Fig. 6B). MEDS322, by targeting the quinone-binding site, sets the stage for a rational structure-based potency improvement for the identification of a lead compound to be further evaluated in in vivo tests. The molecule suffers the absence of an additional polar interaction with respect to the one established with Arg106; indeed, MEDS433 interacts also with Gln47 that is absent in MtDHODH. A possible approach to increase the potency of future MEDS322 derivatives could be the functionalization of thiazolidinone ring to establish a polar interaction with Glu322 that in hDHODH is substituted by a proline residue (Fig. 6A).

Synthesis of the MEDS322 compound
For the synthesis of target compound 1 a chemical strategy, which had already been investigated to obtain similar derivatives, was used [40]. The scheme starts from 1, 4-thioxo-2-thiazolidinone (isorhodanine) 2 that was allowed to react with 3-chloroaniline in refluxing methanol in order to obtain compound 3 in good yield and purity, with products precipitating from the reaction mixture and then collected by filtration. This latter was then condensed with methyl 2-(4-formyl-2methoxyphenoxy)acetate involving a Knoevenagel condensation, performed in acetic acid as solvent, obtaining the target compound 4 as Z isomer. Compound 4 was hydrolyzed with 2 M NaOH to obtain the corresponding acid 1.

MEDS322 preliminary ADME profiling
In order to complete the scenario, as the final step in this work, a preliminary ADME profile of MEDS322 was evaluated (Table 2 and Data S1). The compound presents an acceptable profile in terms of plasma stability and protein binding but its solubility falls in the lower limit to be considered acceptable. While this solubility profile did not preclude the in vitro studies on the compound, this property will be the first descriptor to improve during the following hit-to-lead process in order to reach acceptable drug-like properties for in vivo experiments.