Rational Design of Benzylidenehydrazinyl-Substituted Thiazole Derivatives as Potent Inhibitors of Human Dihydroorotate Dehydrogenase with in Vivo Anti-arthritic Activity

Human dihydroorotate dehydrogenase (hDHODH) is an attractive therapeutic target for the treatment of rheumatoid arthritis, transplant rejection and other autoimmune diseases. Based on the X-ray structure of hDHODH in complex with lead compound 7, a series of benzylidenehydrazinyl-substituted thiazole derivatives as potent inhibitors of hDHODH were designed and synthesized, of which 19 and 30 were the most potent with IC50 values in the double-digit nanomolar range. Moreover, compound 19 displayed significant anti-arthritic effects and favorable pharmacokinetic profiles in vivo. Further X-ray structure and SAR analyses revealed that the potencies of the designed inhibitors were partly attributable to additional water-mediated hydrogen bond networks formed by an unexpected buried water between hDHODH and the 2-(2-methylenehydrazinyl)thiazole scaffold. This work not only elucidates promising scaffolds targeting hDHODH for the treatment of rheumatoid arthritis, but also demonstrates that the water-mediated hydrogen bond interaction is an important factor in molecular design and optimization.

crystallographic investigations revealed that the water-mediated binding modes were specific to inhibitors harboring benzylidenehydrazinyl-substituted thiazole skeletons. Compared to other published structures of hDHODH inhibitors, the benzylidenehydrazinyl-substituted thiazole analogs exhibited greater flexibility within the binding site due to the two rotatable bonds of the amphipathic methylenehydrazine linker connecting the hydrophilic and hydrophobic portions of the ligands. Water molecules located at the P3 and P4 positions acted as anchors to help maintaining the bioactive conformation of the inhibitors in this series. Therefore, the benzylidenehydrazinyl-substituted thiazole inhibitors developed in this study exhibited relatively high binding affinities against hDHODH. Of these inhibitors, compounds 19 and 30 were the most potent with the IC 50 values in the double-digit nanomolar scale. Moreover, 19 exhibited remarkable anti-arthritic efficacy in vivo and markedly alleviated foot swelling in a dose-dependent manner. Compound 19 also displayed good pharmacokinetic profiles and a half-life of 9.69 h indicating its promise as a scaffold for further development as an anti-arthritic agent. Interrogating the distributions and binding patterns of all the bridging water molecules in the 35 X-ray crystal structures of hDHODH deposited in the Protein Data Bank 29 suggested that the locations and binding modes of the binding-enhancing interfacial water molecules in the ubiquinone-binding site tend to be complementary with the polar moiety of the ligands and should be exploited rationally in the future drug design targeting hDHODH enzyme. In addition to providing novel chemical scaffolds for the treatment of rheumatoid arthritis, the present study also sheds light on the importance of water-mediated hydrogen bond networks for molecular design and optimization.

Results and Discussion
Lead Generation and X-ray Structure Analysis. Compound 7, a benzylidenehydrazinyl-substituted thiazole derivative, was first identified in a hierarchical structure-based virtual screen by our group as a potent hDHODH inhibitors with an IC 50 value of 0.365 μ M 6 . To elucidate the binding mode of 7 and promote the structure-based development of hDHODH inhibitors of this series, compound 7 was co-crystallized with hDHODH, yielding X-ray structures that diffracted to 2.07 Å. The coordinates of the crystal structure have been deposited in the Protein Data Bank with PDB ID code 4LS0. The X-ray structure of 7 and hDHODH reveals that the formyl-substituted phenyl ring moiety of 7 is located at the inner region of the ubiquinone-binding pocket of hDHODH. A hydrogen bond is observed between the oxygen atom of the aldehyde group of 7 and the guanidine group of Arg136 at subsite S2 (Fig. 2). Two buried water molecules (W627 and W660) in this hydrophilic region form a complicated water-bridging hydrogen bond network with the ligand and residues Gln47, Arg136 and Thr360. More specifically, W627 forms hydrogen bonds with the sidechain nitrogen atom of Gln47 and the sidechain oxygen atom of Thr360, while W660 is more versatile and participates in the formation of four hydrogen bonds with the sidechain nitrogen atom of Gln47, the backbone oxygen atom of Thr360, the oxygen atom of the formyl group of 7, and the secondary amine nitrogen atom of the hydrazine group of 7. Another structural water molecule, W661, is located near the thiazole moiety and forms a bridge between the nitrogen atom of the thiazole group, the secondary amine of the hydrazine moiety and the carbonyl oxygen atom of residue Ala55. Thus, W661 not only enhances the intramolecular interactions within 7, but also participates in polar interactions between the ligand and hDHODH within the hydrophobic subsite S1. In addition to the hydrogen bond interactions in the hydrophilic region, the hydrophobic interactions between 7 and hDHODH near the entrance of the ubiquinone-binding site (S1, Fig. 2) are also of crucial importance. Because the 4-phenyl-thiazole moiety extends to the entrance of the pocket and fits the pocket shape well, hydrophobic effects and VDW interactions occur between the 4-phenyl-thiazol moiety and residues Leu42, Met43, Leu46, Leu58, Phe62, Leu68, Phe98 and Met111.
Analysis of the binding mode of the lead compound in the X-ray structure highlighted three water molecules (W627, W660 and W661) embedded in the binding site. Compared with the water molecule W627, which consistently bridges Gln47 and Thr360 of the receptor and the polar groups (like carbonyl, carboxyl and cyano) of the ligands in subsite S2 (Fig. 2), W660 and W661 exhibit unexpected binding modes by engaging the 2-(2-methylenehydrazinyl)thiazole moiety of 7 in close contacts with the receptor. The methylenehydrazine linker that connects the hydrophilic and hydrophobic portions of the inhibitor provides two rotatable bonds and results in a more flexible structure. Therefore, the presences of W660 and W661 act as anchors to maintain the skeleton of 7 in its bioactive conformation, thereby enhancing the binding affinity between 7 and hDHODH.
To identify more potent hDHODH inhibitors and to determine if the three water molecules, particularly W660 and W661, are present in other X-ray structures of hDHODH in complex with the benzylidenehydrazinyl-substituted thiazole analogs, lead optimization was conducted in three phases: i) a substructure-based database search to diversify the scaffold of the lead compound and obtain beneficial information; ii) hydrophilic modification of the benzene ring located at subsite S2 to improve polar contacts and determine if variation of the hydrophilic substitutions will influence the presence of the buried water molecule located within the binding site; and iii) hydrophobic optimization at subsite S1 to strengthen nonpolar contacts and to determine if the appearance of the structural water molecule is affected by stronger hydrophobic interactions.

First-Round Optimization of hDHODH Inhibitors.
To detect what functional groups are more favorable for the binding at subsite S2, the substructure-based database searching approach was applied in the first round against the currently available databases (such as SPECS and MayBridge database) to rapidly obtain the derivatives. Using unsubstituted (E)-2-(2-benzylidenehydrazinyl)-4-phenylthiazole as the query scaffold, a focused library containing several thiazole derivatives was extracted from the SPECS database 30 , and all compounds in this focused library (8)(9)(10)(11)(12)(13)(14) were purchased from the vendor. HPLC analysis data confirmed the purity of these compounds to be ≥95% (Supplementary Table  S1). However, none of the analogs were active (Table 1). Combined with the binding information of 7, we then analyzed the structure-activity relationship of compounds 8-14 and synthesized 15-18. Compounds 15 and 18 exhibited moderate inhibition of hDHODH with increased lipophilicity, resulting in sharply decreased ligand lipophilicity efficiency (LLE) 31,32 . In contrast to the formyl group of 7, compounds 8-18 contained a hydroxyl-substituted or unsubstituted aromatic ring in subsite S2 (Fig. 2), which undermined the hydrogen bond interactions between the compounds and residues Arg136 and Gln47. This led to a sharp decrease in inhibitory activity as well as a disfavored increase in ligand lipophilicity. By comparing the inhibitory activity of the compounds in the first-round, it was indicated from the opposite side that a carbonyl-substituted aromatic ring in R 2 is more preferred than the hydroxyl-substituted or unsubstituted one by subsite S2. In addition, an o-chlorine atom in the benzene group of the 4-phenyl-thiazol moiety was tolerated (18). Accordingly, in the next round of optimization, the accepted o-chlorine atom located in hydrophobic subsite S1 was retained and more polar group was introduced at subsite S2.
Second-Round Optimization of hDHODH Inhibitors. Based on results of the first round, a carboxyl group was introduced into the phenyl to replace the hydroxy group. The resulting compound 19 had an IC 50 value of 0.032 μ M against hDHODH, a 43-fold increase compared to the hydroxy-substituted derivative 18. Coupled with a reduction in lipophilicity of more than one log unit lipophilicity, 19 exhibits a pronounced advantage over 18 in terms of LLE. Obviously, the greatly increased activity of 19 is attributable to the strong salt bridge interaction between the carboxyl group and Arg365. However, due to an inappropriate orientation and steric clashes, compounds with a meta-substitution (20) or para-substitution (21) of the carboxyl group were unable to directly interact with Arg136, thus leading to a loss of activity. Esterification of the ortho-substituted carboxyl group of 19 with methyl alcohol (22) or ethyl alcohol (23) affected the aforementioned salt bridge, resulting in decreases in potency of 101-and 58-fold, respectively, compared with 19. The esterification also increased the lipophilicity of 22 and 23 compared to 19, leading to an obvious decrease in LLE.
Compound 23 was slightly more potent than 22, which may be attributable to the stronger hydrophobic interactions between the ethyl group and the small hydrophobic pocket formed by residues Val134 and Val143 at subsite S4 (Fig. 2). To detect subsite S4, a fluorine atom was introduced (24). Unfortunately, 24 displayed a 7-fold decrease in inhibition activity compared with 19. This decrease is attributable to the effect of the electron-withdrawing nature of the fluorine atom impacting the orientations of the two oxygen atoms of the carboxyl group, which slightly weakens the salt bridge and the water-mediated hydrogen bond interaction network. In an attempt to establish polar contacts with residues Tyr356, His56 and Tyr147, an amino group was introduced at subsite S3 (25). 25 exhibited a sharp decrease in inhibitory activity, which was likely due to steric clashes. In summary, the second round of optimization revealed that the derivative with an ortho carboxyl-substituted group on the benzene ring at subsite S2 was strongly preferred.
Analysis of the X-ray Structure of 19 with hDHODH. Using our established X-ray crystallographic protocols (see Experimental Section), we obtained a 2.20 Å X-ray structure of compound 19 in complex with hDHODH. The coordinates of the complex structure have been deposited in the Protein Data Bank as the entry 4LS1. The binding mode of compound 19 is presented in Fig. 3A,B. A similar water-bridging hydrogen bond network was observed around the benzylidenehydrazinyl-substituted thiazole portion of inhibitor 19. Crystallographic water molecules (W592, W620, and W659) were observed at nearly the same location and function in the same way as those depicted in Fig. 2. W659 mediates the interaction between Gln47, Thr360 and the carboxyl group. W620 is also tetra-coordinated with the ligand and the protein by simultaneously forming two hydrogen bonds with the receptor and another two with compound 19. A 1.1-Å deviation is apparent between W592 and W661 in Fig. 2. Compared to W661, W592 is closer to Ala55 and slightly further from the hydrazine moiety. However, the bridging functions of W592 are unaffected, and W592 is involved in three hydrogen bonds with the nitrogen atom of the thiazole group, the secondary amine of the hydrazine moiety and the carbonyl oxygen atom of residue Ala55. The novel binding modes of W592 and W620, coupled with the typical binding pattern of W659, result in the formation of a complicated hydrogen bond network that contributes to the stability of the bioactive conformation of the flexible scaffold provided by the benzylidenehydrazinyl-substituted thiazole moiety, which enables favorable polar interactions with hDHODH. Moreover, in subsite S2 of the ubiquinone-binding site, the carboxyl of the benzoic acid moiety forms a nearly perfectly oriented salt bridge with the side chain of Arg136. The combination of this favorable salt bridge and the strong water-bridging hydrogen bond network increase the potency of compound 19 compared to compound 18.
Notably, the chlorophenyl group exhibits a dual binding mode at the entrance of the binding tunnel. The chlorophenyl moiety perfectly accommodates the concave surfaces formed by Met43 and Leu46 on the left side and Ala59, Phe62, and Met111 on the right side (Fig. 3B). In addition to the shape complementarity, the o-substitution of a chlorine atom on the benzene group presumably has the same favorable effects on binding affinity as the methyl group by stabilizing the bioactive conformation of compound 19. This conformation more closely matches the conformer observed in the X-ray structure of the hDHODH and 19, thus decreasing the conformational reordering required upon binding [33][34][35] . The amphipathic character, salt bridge, water mediation effect and conformational advantages combine to establish compound 19 as the most potent hDHODH inhibitor among the benzylidenehydrazinyl-substituted thiazole derivatives, with an IC 50 of 32 nM.

Third-Round Optimization of hDHODH Inhibitors. A third round of modification was launched
to further determine if the water-bridging hydrogen bond network would be destroyed when the R 4 group (2-chlorine-Ph) of 19 was replaced by more diverse substituents. When there was no substitution on the phenyl moiety, the hydrophobic interactions were weakened, resulting in a slight decrease in activity (26). When another chlorine atom was introduced at the 5-position of the phenyl moiety, compound 30 (IC 50 = 0.043 μ M) did not exhibit any advantage over 19. Because 30 exhibited increased lipophilicity but similar potency, resulting in a decreased LLE value. Baumgartner et al. 11 proposed a new, predominantly polar subsite S5 (Fig. 2) consisting of the hydroxy group of Tyr38 and the backbone carbonyl of Leu67 in the hydrophobic section. This subsite was probed by attaching a -OCH 3 group to the 3-position of the phenyl moiety (27). Unfortunately, no obvious polar interactions were observed between the proposed subsite S5 and the -OCH 3 moiety (Fig. 3C). However, 27 was equipotent with 19 in terms of LLE, as its reduced lipophilicity compensated for its decreased potency. Efforts to explore this subsite were continued, and 28 was synthesized. The lipophilicity of 28 was reduced by two log units at the expense of a sharp decrease in potency, making 28 less attractive than 19. Therefore, attempts to generate polar contacts with residues Tyr38 and Leu67 at subsite S5 by modifying the 3-position of the phenyl group were unsatisfactory. In addition, the binding mode of 27 in the crystal structure ( Fig. 3C) also suggested that substitutions at the 4-position of the phenyl moiety could be accommodated. The biological activity data for 29 and 32 indicated that small hydrophobic groups such as -CH 3 are favored at the 4-position, whereas larger groups such as phenyl, may collide with Leu67 and Leu68 and result in a loss of activity. Interestingly, when the biphenyl group was converted into a naphthalene group (31), the latter was capable of forming π -π stacking interactions with Phe62. Thus, compound 31 exhibited moderate inhibitory activity. None of the compounds in the third round performed better than 19, but this round revealed that small substituting groups such as -Cl, -CH 3 and -OCH 3 are favored at the entrance of the binding site, whereas substituents at the 4-position of the phenyl moiety are actually size-constrained. More importantly, the crystal structure of hDHODH in complex with 27 was solved (PDB ID: 4LS2), which confirmed the functions of the versatile water molecules observed in the other structures.
Analysis of the X-ray Structure of 27 with hDHODH. The X-ray structure of hDHODH with compound 27 was obtained, and the binding mode of 27 is shown in Fig. 3C,D. Interestingly, there are only two structural water molecules (W501 and W502) in the inner polar site of the tunnel-like binding site. These water molecules form nearly the same water-bridging hydrogen bond network observed in compounds 7 (W627 and W660) and 19 (W659 and W620). W501 mediates the interaction between Gln47, Thr360, and the polar carboxyl group of 27. The versatile tetra-coordinated W502 appears in the same place and mediates interactions in the same manner as W660 and W620, even though some modifications to the benzene motif stretching along the hydrophobic entrance of the binding site are apparent. Nevertheless, an acetate molecule instead of a water molecule (W661 in Fig. 2, W592 in Fig. 3A) is in close proximity to the thiazole moiety of the inhibitor, which came from the crystallization agent. The size of the final 2Fo-Fc electron density map at this site is too big for one water molecule but most suitable for one acetate molecule, and the acetate molecule happened to locate at the entrance of the inhibitor binding pocket and formed hydrogen bonds with compound 27 in the complex structure. In addition to the polar contacts with the nitrogen atom of the thiazole group and the secondary amine of the hydrazine moiety, this polar solvent molecule also forms hydrogen bonds with W502 at a distance of 3.4 Å. This structure indicates that the water molecules positioned at W501 and W502 (Fig. 3C) are stable and complement the binding of the scaffold of the benzylidenehydrazinyl-substituted thiazole inhibitor and hDHODH. However, the water molecule located around the thiazole portion of the ligand is loosely bound in the hDHODH-ligand system and is susceptible to be replaced by a solvent molecule. As shown in Fig. 3B,D, the hydrophobic entrance of the binding pocket is exposed to solvent, and its openness is largely influenced by the rotation of the flexible residue Phe62.
In addition to the water-bridging interactions, the salt bridge between residue Arg136 and the carboxyl moiety of 27 is also a valuable feature for binding affinity. The -OCH 3 group at the 3-position of the phenyl moiety matches the concave surface perfectly, forming close hydrophobic interactions with the residues of Leu42, Met43 and Leu46 in subsite S1. Figure 3C also indicates that little space remains between the 4-position of the phenyl moiety and the receptor at the entrance of the binding site, and only small substitutions can be accommodated, as verified using compounds 29 and 32. This round of X-ray crystallographic exploration confirmed once again that the binding of the benzylidenehydrazinyl-substituted thiazole inhibitor to hDHODH is accompanied by two tightly bound water molecules and a loosely bound water molecule that is exposed to solvent.

Analysis of Structural Water Molecules in the Ubiquinone-Binding Site.
In each of the crystal structures of the three complexes derived in our study, an unexpected water bridging pattern was discovered (Fig. 4A). Figure 4B presents the alignment of all 32 reported protein-ligand complex structures of hDHODH deposited in the Protein Data Bank (Supplementary Information Table S2), with the addition Scientific RepoRts | 5:14836 | DOi: 10.1038/srep14836 of the 3 structures solved in this study. In total, there are 43 interfacial water molecules clustered into four positions in the ubiquinone-binding site of these 35 crystal structures, of which 58.1% are located in the P1 position and 23.3% are located in the P2 position. Of the five water molecules at the P3 position, three were identified in this study. In addition, two water molecules were observed at the P4 position for the first time in this study.
Apparently, more than half of the interfacial water molecules are located at the P1 position, where they perform the general functions of bridging Gln47, Thr360 and polar groups (like carbonyl, carboxyl and cyano) of the ligands. Therefore, the water molecule at the common P1 position is adequately considered in molecular docking or virtual screening work 6 . The appearance of a water molecule at the P2 position is always accompanied by polar groups such as that of A771726. In detail, the water molecule at the P2 position is involved in an intricate hydrogen bond network between residues Gln47, Arg136 and the carbonyl group of the amide moiety in the A771726-like ligand. This binding mode was observed in several X-ray structures, such as 1D3H, 3F1Q, 3FJL, 3G0U, 3G0X and 3U2O. By contrast, unlike the water molecule commonly observed near the P1 or P2 position, the water molecules located at the P3 and P4 positions were rarely observed in other previously published X-ray structures of hDHODH. Interestingly, the co-occurrences of water molecules at the P3 and P4 positions were found together with the scaffold of the benzylidenehydrazinyl-substituted thiazole inhibitors designed in this study. These water molecules cooperate with each other to form a complicated hydrogen bond network that facilitates the binding of the inhibitors in this series to the active site of hDHODH. The water molecule at P3 region was tetra-coordinated with the ligand and the protein through an unexpected bridging pattern in which the water molecule simultaneously forms two hydrogen bonds with the ligand (the benzylidenehydrazinyl portion) and an additional two hydrogen bonds with the protein (the residues Gln47 and Thr360). Moreover, three rounds of verification demonstrated that the water molecule at the P3 position is robust, tightly bound and not shifted by molecular decoration at the hydrophilic or hydrophobic portions of lead compound 7. The water molecule at P3 position appears to be a specialized binding partner that maintains the bioactive conformation of the benzylidenehydrazinyl-substituted thiazole inhibitor, which has a more flexible methylenehydrazine linker connecting the polar and nonpolar portions of the ligand in the binding site compared to other known inhibitors. The water molecule located at the P4 position can also bridge the nitrogen atom of the thiazole group, the secondary amine of the hydrazine moiety, and the carbonyl oxygen atom of residue Ala55. Thus, this water molecule enhances intramolecular interactions in addition to participating in polar interactions between the ligand and hDHODH in the hydrophobic subsite S1. Therefore, the water molecules at the P3 and P4 positions actively contribute to maintaining the bioactive conformations of the benzylidenehydrazinyl-substituted thiazole inhibitors in the ubiquinone-binding site of hDHODH. However, the water molecule at the P4 position is bound more loosely than the water molecule located at the P3 Position, and thus is susceptible to be replaced by the polar acetate molecule (Fig. 4) because the P4 position is exposed to solvent.
Analysis of the distributions and binding modes of the water molecules in the ubiquinone-binding site of hDHODH indicated that the hydrophilic pocket of the active site is water accessible and that the locations and binding patterns of the water molecules complement the polar moiety of the ligands that   likely has an advantage in terms of its terminal half-life (9.69 h). The oral bioavailability of 19 could be improved by administration in its salt formation and their drug-like properties along with pharmacokinetic characteristics will be optimized further in the future work.
In vivo Anti-arthritic Efficacy of Compound 19. Compound 19 and methotrexate were injected intraperitoneally once per day for 28 days into the Wistar rats with collagen-induced arthritis (CIA). The swelling scores of the arthritis and morphological observations of the rats' joint tissues were employed to examine the anti-arthritic effect of compound 19 (Fig. 5). With the onset of arthritis, the swelling scores were an indication of the disease states. The maximum score was 8, which is the sum of the scores from both hind paws of each rat and indicates a severe arthritis state. During the experiment, the control rats had normal eating patterns, shiny hair and continuously increasing body weight. By contrast, the rats in the model group exhibited a rough and dull hair at the beginning of the trial, and a slower increase in body weight relative to the normal group after day 9 (Fig. 5A). Although treatment with compound 19 or methotrexate had no obvious influence on increase in body weight, compound 19 displayed significant dose-dependent anti-arthritic effect (p < 0.05) and obviously alleviated foot swelling (Fig. 5B,C). The arthritis scores of the CIA rats reached their peak and stabilized (Fig. 5B) around 8 after day 18, presenting a severe arthritis state, while the arthritis scores of the compound 19 treated rats decreased and were around 5 (5 mg/kg) and 3 (30 mg/kg) at the day 28, which indicated that compound 19 can display substantial anti-inflammation effects and alleviate foot swelling in a dose-dependent manner (Fig. 5B,C). Besides, histological changes in the joint tissues of the normal, CIA, compound 19 and methotrexate treated rats were investigated by the hematoxylin and eosin (H&E) staining method. In the joint tissues of the CIA rats with serious arthritis (Fig. 5D), pathological features including the infiltration of the inflammatory cells (marked with asterisk), synovial degradation (marked with solid triangles), cartilage erosion in the joint tissues (marked with solid arrows) and inflammatory lesions of some cells (highlighted with an oval) were clearly observed. Besides, due to the serious arthritis, the articular cavity was full of exudate which was larger than the other compound treated groups. Compared to the CIA rats, the previously mentioned pathological changes were evidently alleviated in the joint tissues of the rats treated with compound 19 in both dosage, demonstrating that compound 19 displayed anti-arthritic effects in a dose-related manner. The significant in vivo anti-arthritic effects of 19 not only confirmed that hDHODH is an effective anti-arthritic target but also confirmed that this compound is a promising

Conclusion
In summary, this work presented a series of benzylidenehydrazinyl-substituted thiazole inhibitors to suppress the bioactivity of hDHODH. Compound 19 displayed the lowest IC 50 value among these compounds, with notable anti-arthritic efficacy and acceptable pharmacokinetic profiles in vivo. These results suggest the potential of 19 for application as an immunosuppressive agent targeting hDHODH. More importantly, water-bridged interactions between the ligands and hDHODH were identified in the three X-ray structures solved in this study. Two robust bridging water molecules co-occurred with the ligands to help maintain their bioactive conformations by generating a strong water-mediated hydrogen bond network between the flexible 2-(2-methylenehydrazinyl)thiazole linker and hDHODH. The locations and H-bonding patterns of the water molecules are highly complementary with the polar moieties of the ligands in the ubiquinone-binding site of hDHODH and should be considered in future drug design and optimization workflows targeting hDHODH.

Methods
In vitro Enzyme Assay. The plasmid coding for human DHODH was kindly provided by Prof. Jon Clardy (Harvard Medical School) 7 . hDHODH (Met30-Arg396) was expressed and purified in the same manner as described in our recently published work 38 . We determined the effect of benzylidenehydrazinyl-substituted thiazole derivatives to inhibit hDHODH by using a DCIP assay method 38 . The assay buffer contained 50 mM HEPES pH 8.0, 150 mM KCl, 0.1% Triton X-100, 100 μ M UQ 0 and 120 μ M DCIP. The results were presented as the mean ± standard deviation of three independent experiments (Table 1).
Crystal Structure Determination. Co-crystallization of hDHODH with compound was performed as previously described 38 . X-ray diffraction data were collected at 100 K at the synchrotron beamline BL17U1 at SSRF (Shanghai, China). Statistics of data collection, processing, and refinement were summarized in Table 3. The data were processed with MOSFLM 39 , and scaled using the SCALA program from the CCP4 suite 40 . Structural elucidation and refinement were carried out using the CCP4 suite of programs 40,41 . The PDB entry 1D3G (stripped of ligands and water molecules) was used as a molecular replacement model for phasing of the X-ray data. Iterative model building and refinement were performed using Coot and Refmac5, respectively 42 In vivo Efficacy Study. Adult male Wistar rats (120-140 g) were purchased from Shanghai SLRC Laboratory animal company and housed in standard pathogen-free cages. The environment was climate-controlled (20-25 °C, 50-60% relative humidity) with a 12-hour light and 12-hour dark cycle as described previously 38 . Sterilized food and water were provided naturally. All procedures involving animals conform to the Chinese government guidelines for animal experiments, and were conducted with approval of the Institutional Animal Care and Use Committee of East China University of Science & Technology. Bovine type II collagen (CII, Chondrex, USA) was emulsified in the same manner as that conducted previously to get the CIA model 44 . Specifically, 49 rats were randomly assigned into five groups: normal control (solvent, i.p., n = 6), CIA model (solvent, i.p., n = 11), Methotrexate-treated (0.3 mg/kg, i.p., n = 10), compound 19-treated (5 mg/kg, i.p., n = 11) and compound 19-treated (30 mg/kg, i.p., n = 11) groups. Methotrexate, reported to has a therapeutic effect on rheumatoid arthritis 45 , was used as the positive control in this study. The main workflows of this experiment were consistent with that reported recently 38 . During the 28 days of the experiment, compound 19 (5 mg/kg and 30 mg/kg) and Methotrexate (0.3 mg/kg) were injected intraperitoneally once every day. At the same time, the normal control group and model group were administered the same amount of solution. The solvent of 19 was made up of 5% DMSO, 45% saline and 50% PEG400.
In this study, the clinical-mimic arthritis was observed and the severity was assessed by the same criteria as that applied in reference 65: 0, normal, without any swelling; 1, mild swelling; 2, swelling; 3, significant swelling; 4, severe swelling. The maximum score was 8, which was the sum of the scores from both hind paws of each rat. When arthritic signs appeared, the incidence and severity of the arthritis in rats with different treatments were evaluated via arthritis scores recorded every 2 days under blinded  conditions. Measurement of body weight was taken every 3 days. All Wistar rats were killed after 28 day's trails. Pathological detections against the whole knee joints of the rats were conducted using the same method as described previously 38 . Mean ± standard deviation (mean ± SEM) was applied to present the data in this study. SPSS was used to evaluate the statistical significance of differences among the groups. P ≤ 0.05 was the threshold to be statistically significant 46 .

Chemistry.
A series of benzylidenehydrazinyl-substituted thiazole derivatives were designed and synthesized as potential hDHODH inhibitors. These analogs were primarily prepared via the cyclization of thiosemicarbazone with 2-bromo-1-phenylethanones 47 , and the thiosemicarbazones were obtained from the condensation reaction of commercially available substituted benzaldehydes and thiosemicarbazide 47 . The preparation of 2-bromo-1-phenylethanones began with commercially available acetophenones with diverse substituents 48 . Following this design route, we successfully acquired compounds 15-18 and 20-21 (as shown in Supplementary Scheme S1). When the substituted benzaldehyde was transformed to the o-carboxybenzaldehyde, further intramolecular cyclization between the carboxyl group and the second nitrogen atom in the hydrazine group occurred in addition to the anticipated reaction (as shown in Supplementary Scheme S2). To obtain the target compounds, extensive efforts were made to regulate the hydrolysis of the intramolecular amide bond. Fortuitously, this problem was resolved by referring to an article in which NaOH was used in dry THF 49 . However, several substituted benzaldehydes, such as 24a and 25a in Supplementary Scheme S2, were commercially unavailable. Therefore, a five-step workflow comprising esterification, benzylic bromination, lactonization, benzylic bromination and hydrolysis was designed to obtain their crucial o-carboxybenzaldehyde intermediates (24V and 25V). For this workflow, 2-methyl benzoic acid derivatives were used as the starting products (Supplementary Scheme S3) [50][51][52][53][54][55][56][57] . The preparations of the esterified derivatives (22,23) are shown in Supplementary Scheme S4. For these compounds, the esterification reactions of 2-formylbenzoic acid were firstly performed to obtain the methyl ester and ethyl esters without affecting the aldehyde group 58 . Thiazole hydrazine was obtained via the condensation of thiosemicarbazide and 2-bromo-1-phenylethanones in 1,4-dioxane at room temperature. Finally, the target products, the benzylidenehydrazinyl-substituted thiazole analogs, were obtained via a dehydration reaction of methyl or ethyl 2-formylbenzoate and thiazole hydrazine 59 . In general, all reaction conditions were mild, and no dangerous or toxic materials were used in the reactions. The structures of all products were verified by 1 H NMR, 13

C NMR and HRMS (ESI).
General methods. All chemical reagents and solvents were obtained from commercial sources and used without further purification. Thin-layer chromatography (TLC) was carried out to monitor the process of reactions. Purification of compounds was achieved by column chromatography with silica gel (HaiYang, Qingdao) 200-300 mesh. 1 H NMR and 13 C NMR spectra were recorded on a Bruker AM-400 spectrometer with chemical shifts expressed as ppm (in CDCl 3 or DMSO-d 6 , Me 4 Si as internal standard). Melting points were recorded on a WRS-1B-digital melting point apparatus. The mass spectra were measured at The Institute of Fine Chemistry of ECUST. The MS data were obtained by Agilent 5975C GC-MS instrument and Agilent LC-MS 1200 and 6120 system. Analytical HPLC was performed on a Hewlett-Packard 1100 chromatography system equipped with photodiode array detector and a Zorbax XDB-C18 column (250 mm × 4.6 mm) was used to determine the purity of the products. The mobile phase A was water with 1‰ TFA and mobile phase B was acetonitrile. A gradient of 30-100% B over 20 minutes was run at a flow rate of 1.0 mL/min. Compounds synthesized in our laboratory were generally varied from 95% to 99% in purity, and the biological experiments were employed on compound whose purity is at least 95%.
General procedure for the preparation of compounds 24V and 25V (Supplementary Scheme S3). A solution of the substituted benzoic acid derivative (24I or 25I) (20 mmol) and concentrated sulphuric acid (0.2 mL) in MeOH (50 mL) was refluxed until TLC analysis indicated the total consumption of the starting material. Then, solvent was concentrated and pH was adjusted to 7.0 with 10% aq. NaHCO 3 solution. The mixture was extracted with ethyl ether (3 × 50 mL), and the combined extracts were dried with anhydrous sodium sulphate and concentrated under reduced pressure to give the desired methyl esters 24II or 25II. Then they were both confirmed by GC-MS and used in the next step without further purification.
A mixture of 24III or 25III (10 mmol) and calcium carbonate (60 mmol) in 1,4-dioxane (50 mL) and water (50 mL) was heated at reflux for 5h. The insoluble material was removed by filtration after cooled to rt. The filtrate was concentrated under vacuum and then extracted with CH 2 Cl 2 (3 × 50 mL). The combined extraction was dried over anhydrous sodium sulphate. Removal of solvent under vacuum provided the desired product with 90-95% yield as white solid (24IV or 25IV).
24IV or 25IV (9 mmol), bromosuccinimide (9.9 mmol) and azobisisobutyronitrile (0.4 mmol) were added to 1,2-dichloroethane (25 mL), the mixture was refluxed for 4 h. Then, the reaction mixture was kept in an ice bath for 2 hours and filtered. The solvent was removed under reduced pressure. Water (20 mL) was added and the resulting mixture was refluxed for 5-6 h, then cooled to room temperature and extracted with EtOAc 3 times. The combined organic phase was dried and concentrated under reduced pressure. The residue was purified by column chromatography (PE : EtOAc = 8 : 1, v/v), 23V or 24V with 70-80% yield as white solid was obtained.  General procedure for the preparation of Schiff bases of thiosemicarbazone compounds 15a, 19a-21a, 24a-25a (Supplementary Scheme S1 and S2). A mixture of substituted aldehyde (0.01 mol) and thiosemicarbazide (0.01 mol) in ethanol (40 mL) was refluxed on a water bath for 5 h in the presence of few drops of glacial acetic acid as catalyst. The progress of reaction was monitored by TLC at appropriate time interval. After completion of the reaction, the solution was cooled and filtrated. The solid was washed with water and dried. Finally, the crude product was recrystallized from ethanol to obtain the product with 75-90% yield as white solid.  General procedure for the preparation of compounds 15b-18b, 26b-32b (Supplementary Scheme S1 and S2). A solution of substituted acetophenone (10 mmol) in 30mL of anhydrous ether was cooled in an ice bath. Then 0.1 g of anhydrous aluminum chloride was introduced, and at the same time, bromine (10 mmol) was added gradually through a separatory funnel with continuous stirring. After the bromine was added, the solvent and dissolved hydrogen bromide were removed at once under reduced pressure with a slight current of air. Then, stirred overnight, the progress of reaction was monitored by GC-MS. After completion of reaction, the reaction mixture was extracted with saturated aqueous sodium bicarbonate (60 mL)/ ethyl acetate (60 mL). The organic layer was dried with anhydrous sodium sulfate and concentrated at reduced pressure. At last, the resulting residues were used for next step without purification.
General procedure for the preparation of compounds 15-18, 20-21 (Supplementary Scheme S1). Substituted phenacyl bromide (0.01 mol) was added to a solution of thiosemicarbazone (0.01 mol) in ethanol (20 ml). The mixture was heated to reflux for 3-5 h. The reaction progress was monitored by TLC at appropriate time interval. Cooled in an ice bath, the solid was collected by filtration, suspended in water and neutralized with NaHCO 3 to get the desired product. The crude product was recrystallized from ethanol to obtain the product with 70-88% yield as yellow solid. (15). mp 227.5-229.0 °C. 1  General procedure for the preparation of compounds 19c, 24c-32c (Supplementary Scheme S2). To a solution of thiosemicarbazone (0.01 mol) in ethanol (20 mL), substituted phenacyl bromide (0.01 mol) was added. The mixture was heated to reflux for 3-5 h. The progress of reaction was monitored by TLC at appropriate time interval. After completing the reaction, and cooling the mixture in an ice bath, the solid was collected by filtration, then suspended in water and neutralized with NaHCO 3 to get the desired product. The product was recrystallized from ethanol, with 70-88% yield as white solid.   General procedure for the preparation of compounds 25d (Supplementary Scheme S2). To a solution of SnCl 2 ·H 2 O (5 mmol) in concentrated HCl (10 mL), 25c (1.0 mmol) was added. After stirring for 6 h at 80 °C, water was added and the mixture was poured into 2N aqueous NaOH at 0 °C. The solid obtained by filtration was dissolved in DMF, and removed the undissolved solid by filtration. Then the filtrate was concentrated to give white needles 25d with the yield of 82%.  13  General procedure for the preparation of compounds 19, 24-32 (Supplementary Scheme S2). To a solution of 19c/24c/25d/26c-32c (0.50 mmol) in dry THF (7.0 mL), NaOH (1.75 mmol) was added. After stirring over night at rt, the reaction mixture was acided to pH 4-5, and the solid was collected by filtration. Then the solid obtained by filtration was dissolved in 1N NaOH (30 mL), and the mixture was extracted with DCM (30 mL). The aqueous layer was acided to pH 4-5 with 1N HCl, the formed product was filtered as a green powder with the yield of 30-45%. (19). mp 183.9-185.8 °C. 1 (25). mp 220.9-221.6 °C. 1 151.0, 140.8, 135.1, 133.7, 132.9, 132.6, 132.4, 130.9, 130.4, 129.3, 128.6, 128.6, 128.1, 126.9, 126.5, 124.6,