Synthesis and biological evaluation of thiazole derivatives as LbSOD inhibitors

Abstract Leishmaniasis is considered as one of the major neglected tropical diseases due to its magnitude and wide geographic distribution. Leishmania braziliensis, responsible for cutaneous leishmaniasis, is the most prevalent species in Brazil. Superoxide dismutase (SOD) belongs to the antioxidant pathway of the parasites and human host. Despite the differences between SOD of Leishmania braziliensis and human make this enzyme a promising target for drug development efforts. No medicinal chemistry effort has been made to identify LbSOD inhibitors. Herein, we show that thermal shift assays (TSA) and fluorescent protein-labeled assays (FPLA) can be employed as primary and secondary screens to achieve this goal. Moreover, we show that thiazole derivatives bind to LbSOD with micromolar affinity.


Introduction
Tropical neglected diseases (NTDs) are among the leading causes of mortality in tropical countries around the world 1,2 . Leishmaniasis is endemic in 98 countries and South America countries accounting for 90% of worldwide cases of cutaneous leishmaniasis (CL), which is caused mainly by Leishmania braziliensis [3][4][5] .
The treatment of patients with CL relies on pentavalent antimonial or, as a second choice, pentamidine or amphotericin B, which requires parenteral administration 6 . However, these drugs show either low efficacy or limited safety profile, such as a high incidence of nausea, lethargy, urticaria, hepatotoxicity, and cardiotoxicity 7,8 . In addition, resistance to the available drugs is an increasing limitation to the treatment of patients with leishmaniasis 2,9,10 . Therefore, there is an urgent need to identify novel drugs to fight leishmaniasis.
One of the first steps to achieve this goal is to identify metabolic differences between the parasite and the human host that might be targeted for selective modulation 11 . All aerobic organisms employ reactive oxygen species (ROS) for intercellular signaling and synthesis of important biological substances 12,13 . ROS are also a part of the human innate immune response, once they are involved in the recruitment of inflammatory cells to sites of inflammation 14 and/or infection 12 . For that reason, the protozoan parasites have developed an efficient protection against ROS, whose first step is controlled by superoxide dismutase (SOD, E.C.1.15.1.1), an enzyme responsible for the dismutation of superoxide into hydrogen peroxide and molecular oxygen 5,11 . Accordingly, FeSOD activity has proven essential for Leishmania survival in the host 15 .
The superoxide dismutase from L. braziliensis presents iron as a metal prosthetic group (FeSOD) whereas the human counterpart employs Cu/Zn (CuZnSOD) 16,17 . This difference along their low-sequence similarity suggests that it is possible to inhibit the FeSOD from trypanosomatids selectively [18][19][20][21] . ROMERO et al. (2017) have shown that phthalazine derivatives kill promastigotes and amastigotes forms of Leishmania spp, by FeSOD inhibition. Besides, MORENO-VIGURI et al. (2016) have reported that arylaminoketone derivatives inhibit FeSOD from trypanosomatids but have little effect over human CuZnSOD.
These results suggest FeSOD from L. braziliensis is a promising target for drug design efforts. However, no medicinal chemistry effort to target this protein has been carried out before. Aiming at overcoming this knowledge-gap, this work describes how thermal shift assays (TSA) and fluorescent protein-labeled assays (FPLA) can be employed as primary and secondary HTS-friendly alternatives to indirect-kinetic assays that measure SOD catalytic activity. TSA relies on environment-sensitive fluorophores that reversibly binds to exposed hydrophobic regions when protein unfolds due to temperature increase [22][23][24] . Whilst FPLA requires covalent bond formation between the protein and the probe 25 . Finally, we show that this screen-counter-screen strategy led to the identification of a micromolar inhibitor of L. braziliensis FeSOD.

Chemistry
Chemical reagents and solvents were purchased from Sigma-Aldrich (St. Louis, MO, USA) with analytical grade purity. Melting points were determined in open capillaries on a Buchi apparatus and are uncorrected. Thin-layer chromatography (TLC) was carried out on aluminium-supported silica gel plates (Merck 60 F254 ) with visualization by UV light 254 nm in the appropriated system for each compound. The Infrared spectra (1% KBr, cm À1 pellets) were recorded on a Bruker IFS66 spectrophotometer (Billerica, MA, USA), the wave numbers were given in cm À1 and are uncorrected. The 1 H NMR and 13 C NMR spectra were recorded on a VARIAN VNMRS 400-MR (Palo Alto, CA, USA), 400 MHz for 1 H and 75.4 MHz for 13 C. The 1 H spectra were recorded in DMSO-d 6 while 13 C spectra were recorded in CDCl 3 and DMSO using tetramethylsilane (TMS) as the internal standard. The peak and abbreviations were used to indicate multiplicity: s (singlet), d (doublet), dd (double doublet), ddd (double doublet, doublet), t (triplet), and m (multiplet). The chemical shifts were reported in d units and the coupling constants (J) were reported in Hertz. Mass spectra were recorded on a Varian MAT 711 spectrometer (Palo Alto, CA, USA) 70 eV electron impact.

Synthesis of 2-aminothiazole derivatives (Ju-436 and Ju-533)
The 2-aminothiazole derivatives were obtained from the mixture of equimolar amounts 0.175 g (2.3 Â 10 À3 mol) of thiourea dissolved in 15 ml of methanol followed by the addition of 0.513 g (2.3 Â 10 À3 mol) of substituted acetophenone halide dissolved in 20 ml methanol, to form a white suspension. The reaction was left for 15 min at room temperature under magnetic stirring. After that, the mixture was dissolved in 0.6 ml of HCl (dropwise). The reaction was heated to 90 C in an oil bath, the pH adjusted between 4 and 5. The reaction time was 2 h. The reaction was monitored by thin layer chromatography. The product was ice-cooled and filtered. The compound was purified by crystallization from methanol.  General procedure for the synthesis of imines (Ju-445-Ju-555) The 4-(4-bromophenyl)-thiazol-2-imine-arylidene-substituted and 4-(3 or 4-nitrophenyl)-thiazol-2-imine-arylidene-substituted derivatives were synthesized from the mixture of equimolar amounts 2.3 Â 1.0 À3 mol of 4-(4-bromo, 3-nitro or 4-nitrophenyl)-2-aminothiazole and 2.3 Â 1.0 À3 mol of substituted benzaldehydes. The substituted benzaldehydes were dissolved in 15 ml of MeOH under stirring at room temperature. After that, 0.08 ml of HCl (conc.) dissolved in MeOH was added dropwise. After 5 min. under cold stirring, the starting material (4-bromo, 4-nitro or 3-nitrophenyl)-2-aminothiazole previously dissolved in 15 ml of MeOH was added dropwise. The pH was then adjusted between 4 and 5 and the reaction was warmed to 75 C and refluxed in an oil bath for 2-24 h. The progress of the reaction was monitored by TLC. At the end, the solvent was reduced to one-half volume and ice-cooled. The precipitate was filtered off and recrystallized from the appropriate solvent.

Biology
LbSOD heterologous expression and chromatographic purification E. coli BL21 (DE3) cell carrying the pETM11-LbSOD plasmid were cultured at 37 C/180 rpm in Luria Bertani medium supplemented with 30 lg/mL kanamycin until OD600 reached 0.6-0.9. At this moment, 1 mM (final concentration) isopropyl-b-D-thiogalactoside (IPTG) was added to the culture and the temperature was reduced to 20 C. After 16 h, the cells were harvested by centrifugation (8000 rpm -SN93-10, 4 C, 30 min) and resuspended in lysis buffer (PBS 50 mM, NaCl 100 mM pH 7) supplemented with 1 mM phenylmethanesulphonylfluoride (PMSF). Next, cells were incubated with Lysozyme 0.5 mg/mL, for 30 min and then disrupted by sonication (10 Â 15 s bursts with 30 s intervals between each burst, 9 Watts). These steps were carried out in ice-bath. The soluble fraction was clarified by centrifugation (15000 rpm, 20 min, 4 C) and then loaded onto a HisTrap HP column (GE Healthcare), pre-equilibrated with PBS 50 mM, NaCl 100 mM, 20 mM Imidazole pH 7.0.
The column was washed with 20 column volumes of PBS 50 mM, 100 mM NaCl, 20 mM Imidazole pH 7.0 and then steps of increasing concentration of imidazole (50-500 mM) were used. Purification was monitored by UV absorbance measurement at 280 nm and the level of protein purity was confirmed by sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE) 12%. The protein fraction was dialyzed in PBS 50 mM, NaCl 100 mM pH 7 with Amicon Ultra centrifugal filters (10KDa MWCO, Millipore), 4000 rpm at 4 C.
The enzyme was subjected to TEV (Tobacco Etch Virus) protease digestion (1 mg per 20 mg LbSOD), 4 C overnight. After the proteolysis step, LbSOD was loaded again onto His-Trap column to separate cleaved and uncleaved His-tag LbSOD from TEV protease. The column was pre-equilibrated with PBS 50 mM, NaCl 100 mM pH 7 and then the elution of cleaved LbSOD was performed with 10 column volumes of this same buffer. The purification fractions were analyzed by UV measurement at 280 nm and gel electrophoresis (SDS-PAGE 12%) to confirm purification of the protein.

Thermal shift assays (TSA)
The assays were carried out in the RT-PCR Applied Biosystems 7500 (Applied Biosystems, Foster City, CA, USA), in triplicate, using 96-well PCR plate (PCR plates 96 well BioRad V R , Hercules, CA, USA), sealed with transparent capping strips (Flatcap strips BioRad V R , Hercules, CA, USA). The plates were centrifuged for 2 min, 2000 rpm, 25 C. After the plates were heated from 25 to 85 C in increments of 1 C per minute and fluorescence signal was monitored with SYPRO Orange V R dye, with 492 (excitation) and 610 nm (emission) wavelengths.
The melting temperature (Tm) values were calculated by nonlinear fitting of the melting curves to a Boltzmann sigmoidal function, using GraphPad Prism version 5.0 for Windows (GraphPad V R Software, San Diego, CA, USA, www.graphpad.com). Comparisons between Tm of different conditions (DTm) were considered as significantly different when p < .05, according to the Kruskal-Wallis test followed by Dunn's post-test for multiple comparisons.

TSA optimization
Conditions were optimized to LbSOD: protein concentration, buffer, and DMSO concentration. The protein concentration was evaluated (1-5 mM) in the presence of SYPRO Orange V R (Thermofisher, Waltham, MA, USA) (1:100 dilution) in ultrapure water qsp 20 mL. Then, the buffers (sodium citrate, sodium phosphate, tris-HCl, and glycine) and pHs (4-9) were evaluated at a final concentration of 50 mM for each buffer, SYPRO Orange V R (1: 100 dilution) and LbSOD. Finally, the influence of DMSO (2.5, 5, 10% v/v) on the thermal stability of the protein was investigated.

Screening of thiazoles derivatives by TSA
The effect of the thiazoles derivatives over the melting temperature of LbSOD was evaluated at a single concentration (50 lM final concentration) and an equivalent volume of DMSO was employed as a control (DTm ¼ 0.0). Briefly, LbSOD (5 lM) was diluted in 50 mM sodium phosphate buffer (pH 7.0) supplemented with 100 mM NaCl and SYPRO Orange V R (1:100 dilution). The 96 well plates were heated from 25 to 85 C in increments of 1 C per minute. The results were analyzed as described in the previous section.

Fluorescent protein-labeled assays
LbSOD (2 mg/mL) was incubated with fluorescein-5-isothiocyanate (FITC) dye (10 mg/mL) for 2 h, at 25 C, with continuous stirring. Then, the solution was loaded on a Hi-Trap HP desalting column (GE Healthcare), previously equilibrated with 50 mM sodium phosphate buffer (pH 8) and 1.5 column volumes of the buffer were injected. The absorbance of the collected fractions was monitored at 280 nm and 490 nm, so correct to the contribution of dye to Absorbance280nm could be calculated according to Equation (1): Aprotein stands for the protein absorbance after correction; A280 is the absorbance at 280 nm of the purified sample; Amax: maximal dye absorption at 280 nm or correction factor (CF) corresponding to 0.3 to fluorescein.
Next, LbSOD was diluted to 5 mM and the influence of putative inhibitors (50 mM) over its fluorescence was measured at 25 C for 10 min on an Applied Biosystems 7500 RT-PCR machine. The experiments were carried out in triplicate using 96-well PCR plates (PCR plates 96 well BioRad V R ). Fluorescence data was recorded on the Applied Biosystems 7500 Software v2.0 and then analyzed on GraphPad Prism V5.0 software (La Jolla, CA, USA).

Isothermal titration calorimetry
ITC experiments were carried out in a Microcal VP-ITC Model, at 298 K. All the solutions were thoroughly degassed prior to the titrations to avoid the formation of bubbles during the experiment. The reference cell contained distilled water. The sample cell was filled with LbSOD (20 mM) diluted in sodium phosphate buffer 0.05 M pH 7 with DMSO (5%v/v). The injection syringe was loaded with ligand 400 mM (previously dissolved in the same solution as the protein). Control experiments were carried out with sodium phosphate buffer 0.05 M pH 7 and DMSO (5%v/v) in the sample cell to determine the dilution heat. Injections were started after baseline stability had been achieved. The ligands were titrated in the sample cell through 21 injections of 13 mL and the first injection of 10 mL. The solutions of the sample cell were stirred during the experiment at 200 rpm to ensure mixing.
The heat variation, following each injection of the ligand in the sample cell, were employed to build the titration isotherms, as available in Origin package supplied with the calorimeter.
The thermodynamic parameters DG (Gibbs free energy), DS (entropy), and DH (enthalpy) were calculated using the Equation (2): Where R is the universal gas constant, T is the temperature (Kelvin).

Results and discussion
Previous studies have shown that 1,3-thiazole derivatives have leishmanicidal activity 26 and that phthalimide-thiazole derivatives have higher affinity to L. infantum FeSOD than human CuZnSOD 27 . These results suggest the thiazole ring is a suitable scaffold upon which novel SOD inhibitors might be developed. Hence, we designed a series of 2,4 substituted thiazole derivatives as potential LbSOD inhibitors

Synthesis of thiazole derivatives
The starting material 2-aminothiazole-4-substituted Ju-436 and Ju-533 were prepared, as described previously, 28 with minor modifications. The condensation of substituted benzaldehydes in position 2 was carried out to obtain the respective Schiff bases, as outlined in Scheme 1. The spectroscopic data showed the formation of both starting material and Schiff bases, with characteristic 1 HNMR and 13 CNMR absorption of CH ¼ N between 4.92-4.04 and 171.76-168.25 ppm, respectively. The stereochemistry of CH ¼ N bond is assumed to be E, as reported in literature for thiazole Schiff bases derivatives 29 .

Expression and purification LbSOD
E. coli Bl21 (DE3) cells were transformed with plasmid LbSOD-pET-M11 and LbSOD expression was performed with IPTG 1 mM at 20 C for 16 h.
The LIC vector employed in this work, pETM11, was constructed with 6xhistidine tag, which allows the expression of the His-tag fused LbSOD and its purification by affinity chromatography. The purification was based on the interaction of histidine residues with nickel ions immobilized on the chromatographic column (Ni-sepharose) and LbSOD was eluted in a higher concentration of imidazole (500 mM) (Figure 1(A); (B)).
Once the His-tag can alter kinetics and structural features of LbSOD, the biological assays were carried out after the tag was removed by proteolytic cleavage with TEV protease (Tobacco etch virus) (Figure 1(C)). The overall yield of the purification steps is 20 mg of LbSOD per liter of culture medium.
Once the pure LbSOD was obtained, the next step consisted the identification of potential inhibitors ( Table 1). Considering that thiazole derivatives interfere with the kinetic assay of LbSOD due to the absorption of light at the same wavelength of the assay (320 nm) (data not shown), the thermal shift assays (TSA) and fluorescent protein-labeled assays (FPLA) were employed to evaluate the potential LbSOD inhibitors.

Thermal shift assays (TSA)
In general, TSA is used to identify the conditions and molecules that influence protein stability 23 . This assay relies on the interaction of a fluorophore with hydrophobic regions of the protein, which are exposed as a result of thermal denaturation 30 . The change in fluorescence is a function of the temperature increase, and, consequently, protein unfolding. Thus, it is possible to calculate the mean thermal transition point (Tm: melting temperature) 31 and thus to observe the effect of different buffers, additives, and/or ligands on a protein 32,33 . The increase in protein stability is related to an increase in the conformational homogeneity of the protein sample. Therefore, the identification of ideal conditions, such as the buffer and pH used in the assay, is of great relevance to assay standardization 34,35 Accordingly, optimal conditions of protein stability were probed by thermal shift assays, varying conditions of concentration de LbSOD, buffer, pH, and organic solvent (DMSO), as described next.

Biological assay standardization
In TSA, a linear increase of the detected signal is observed as a function of the protein concentration: increase the number of sites available for interaction with the dye and consequently, the displayed signal 36 . However, high concentrations of protein may lead to aggregation and errors in the interpretation of the results 30 . The TSA curves at different LbSOD concentrations (Figure 2(A)), depict the expected signal-to-noise improvement as the concentration increases and at 5 mM, it exhibits a well-defined onset and final transition that was considered suitable for the next standardization steps.
Comparison of LbSOD stability at different buffers and pHs suggests that it is stable between pHs 4.0 and 8.0, as DTm values do not exhibit significant statistical differences (p > .05) in this range (Figure 2(B)). Either below (i.e. pH 2.0) or above this range (i.e. pH 8.5 -9) LbSOD stability decreases. MEIER et al. (1995) showed that Propionibacterium shermanha Fe-SOD activity decreases at pH <5 (cytochrome C indirect assay; xanthine/xanthine oxidase system) 37 . Similarly, the activity of FeSOD from Plasmodium vickey also decreases at pH <6.0 16,38 . Considering the reported data and the results described above, pH 7.0 was chosen for subsequent assays.
The organic solvent used to solubilize the inhibitors may affect the stability of the enzyme by influencing the formation of intermolecular interactions, such as hydrogen bonds and hydrophobic interactions that stabilize the three-dimensional structure of the enzyme 39 . Then, the effect of DMSO on the thermal stability of LbSOD was also investigated. The assessment of different DMSO concentrations suggests that up to 5% (v/v) of DMSO does not significantly affect the thermal stability of LbSOD (p > .05) (Figure 2(C)) and, for that reason, this is the concentration of DMSO used in the assays.

Screening of thiazole derivatives
HTS methods afford false-positive compounds due to a number of reasons 40 , but it is a consensus that data from single-concentration    experiments is a ubiquitous shortcoming of this strategy 41 . The fact that several SOD assays rely on indirect methods that are subject to interference by oxi-redoxi conditions makes this concern an ordinary problem for LbSOD hit identification campaigns. In order to increase the number of true-positives, it is common practice to employ a counter-screening method, whose end-point is different from the first screening assay. Accordingly, the first screening method described in this work relies on fluorescein-5-isothiocyanate (FITC) covalent binding to the target protein 42 . Once LbSOD was labeled with FITC, any change in the protein conformation would lead to a change in the fluorescence. In case thiazole derivatives cause a conformational shift that exposes FITC-labeled residues, an increased fluorescence signal would be observed. On the other hand, if ligand binding hides the FITC-labeled residues, a decreased signal is expected. Then, the conformational modification due to thiazole derivatives binding to LbSOD was evaluated at a single concentration. This approach suggests that Ju-450 and Ju-480 are the most promising hits (Figure 3). When protein labeling occurs far from the LbSOD active site or there is minor-induced fit upon ligand binding (e.g. low entropic cost), the change in the fluorescence signal might be low and promising compounds would be discarded. An even more troublesome scenario would be a covalent modification of a residue that is crucial for binding. Considering all these shortcoming of protein covalent labeling, we resorted to thermal shift assays as a secondary counter-screening approach. Although this strategy also depends on a fluorescent label (Sypro-orange V R ), covalent binding is not an issue anymore.
Both Ju-450 and Ju-480 have a p-nitrophenyl moiety at the 4position of the thiazole ring and increase the fluorescence signal in FLPA. On the other hand, Ju-514 has a p-bromophenyl moiety at the equivalent position and reduces the fluorescence signal in the FLPA. Given the overall similarity between these compounds, we decided to include Ju-514 in subsequent assays to make sure no true ligand was excluded from our study ( Figure 4).
Although all evaluated compounds exhibited dose-response profile that are typical of true ligands, Ju-450 and Ju-514 have low affinity to LbSOD (Kd¼ 4.7 ± 0.01 mM and Kd¼ 6.1 ± 0.8 mM, respectively). Hence, structure-activity relationships based on those compounds would be of little help to design potent LbSOD inhibitors. Instead, we decided to focus our efforts on Ju-480, that shows low-micromolar affinity to the macromolecular target (Kd¼ 4.8 ± 12.5 lM) and employ a label-free method to study its thermodynamic signature 43 .

Isothermal titration calorimetry
Once ITC measures the heat released or absorbed upon the ligandmacromolecule interaction, it is possible to directly determine thermodynamic parameters DH and DS 44 . Ju-480 has favorable enthalpic (DH¼ À4.22 ± 1.6970 kcal/mol) and entropic (DS ¼ 0.0037 ± 0.0006 kcal/ mol) contributions to affinity ( Figure 5). The enthalpic contributions are related to the intermolecular interactions, such as hydrogen bonds, Van der Waals and dipole-dipole interactions, etc. whereas the entropic contributions are, in general, due to binding-site desolvation and/or induced fit upon ligand binding 5 . Therefore, our results suggest that Ju-480 affinity might be improved either by the addition of substituents that interact with hydrophobic patches in LbSOD surface, as was observed in the optimization of HIV-1 protease inhibitors, 45 or by structure rigidification, 46 which might reduce the entropic penalty for Ju-480 to adopt the bioactive conformation, as entropic contributions are easier to optimize than enthalpic ones. 43 .

Conclusion
Although TSA is commonly employed for screening putative ligands of promising targets, the results presented in this work highlight that its use alone might lead to the selection of falsepositive LbSOD ligands. On the other hand, the combined use of TSA and FPLA assays lead to the identification of Ju-480, a thiazole derivative that has a low-micromolar affinity to L. braziliensis Superoxide dismutase. Label-free methods confirm that Ju-480 binds to LbSOD and suggests that structure rigidification might increase the ligand's affinity to its macromolecular target.