Efficient Dimerization Disruption of Leishmania infantum Trypanothione Reductase by Triazole-phenyl-thiazoles

Inhibition of Leishmania infantum trypanothione disulfide reductase (LiTryR) by disruption of its homodimeric interface has proved to be an alternative and unexploited strategy in the search for novel antileishmanial agents. Proof of concept was first obtained by peptides and peptidomimetics. Building on previously reported dimerization disruptors containing an imidazole-phenyl-thiazole scaffold, we now report a new 1,2,3-triazole-based chemotype that yields noncompetitive, slow-binding inhibitors of LiTryR. Several compounds bearing (poly)aromatic substituents dramatically improve the ability to disrupt LiTryR dimerization relative to reference imidazoles. Molecular modeling studies identified an almost unexplored hydrophobic region at the interfacial domain as the putative binding site for these compounds. A subsequent structure-based design led to a symmetrical triazole analogue that displayed even more potent inhibitory activity over LiTryR and enhanced leishmanicidal activity. Remarkably, several of these novel triazole-bearing compounds were able to kill both extracellular and intracellular parasites in cell cultures.


■ INTRODUCTION
Leishmaniasis is an infectious disease caused by intracellular protozoan parasites from more than 20 Leishmania species and transmitted to humans through the bite of infected female phlebotomine sandflies. According to the World Health Organization (WHO), leishmaniasis is one of the major neglected tropical diseases that affects 12 million people in 98 countries, being especially prevalent in tropical and subtropical areas. 1 Visceral leishmaniasis (VL), also known as Kala-azar, is the most severe form of the disease with more than 60 000 human deaths annually. VL is caused by Leishmania donovani and Leishmania infantum parasites worldwide, including Mediterranean countries. 2 No effective vaccine for humans has so far been developed, and leishmaniasis control mainly relies on chemotherapy. 3,4 However, currently available treatments are inadequate because of high costs, toxicity, drug resistance, and the need for parenteral administration. 5,6 Therefore, there is an urgent need to find new effective innovative drugs against this disease.
Herein, we report a target-based approach for the discovery of novel agents against L. infantum parasites targeting a protein that is essential for parasite survival and exclusive for these parasites. 7,8 We have focused on trypanothione (Try) disulfide reductase (TryR) because it is one of the few genetically and chemically validated drug targets, and validation is one important criterion in target assessment. 9−11 Among other highly relevant functions, this enzyme is essential for antioxidant defense in trypanosomatids. The Try/TryR couple in these parasites substitutes for the glutathione/glutathionedisulfide reductase (GR) pair characteristic of most eukaryotic organisms. Accordingly, the absence of TryR in mammalian cells, the significant differences between TryR and human GR (hGR) (the active sites of the two enzymes have opposite net charges and different volumes), and its crucial role for parasite survival make TryR an attractive target for new chemotherapeutics. 12−15 Since this enzyme was identified, many potent and selective competitive polycationic TryR inhibitors (which mimic the positively charged Try in the active site) have been described. However, there are scarce reports of potent TryR-inhibiting compounds with adequate antiparasitic activity. 16,17 This is so because it has been shown that survival of the parasites is only affected when TryR activity is reduced by more than 90%, probably as a consequence of the high intracellular concentrations of its natural substrate (Try) in the cell. 18 This implies that, to be effective against the parasites, competitive inhibitors must have very high binding affinities so as to give rise to inhibition constants in the low nanomolar range. Nonetheless, one of the most potent (submicromolar) competitive inhibitors of Trypanosoma brucei (TbTryR) described to date is almost inactive against L. donovani. 19 This limitation, together with the need for very high affinities imposed by the high intracellular concentrations of both the natural substrate (between 3 and 50 μM under steady-state conditions and up to 150 μM under oxidative stress) and TryR itself (0.5−1.3 μM), make noncompetitive and/or irreversible inhibitors emerge as more attractive alternatives for targeting this enzyme. 20a,b,21 The crystal structure of TryR from L. infantum (LiTryR) was first solved by Baiocco et al. 22 and is very similar to that of other trypanosomatids. 13,14,16,23,24 This enzyme is a twofold symmetrical homodimer that contains two active sites and two reduced nicotinamide adenine dinucleotide phosphate (NADPH)-and flavin adenine dinucleotide (FAD)-binding domains separated by a large and well-characterized interface. 25 Each Try binding site is located in a large solventexposed cavity at the interface between the two monomers. Several X-ray crystal structures of TryR in complex with a variety of competitive reversible inhibitors have shown that the large pocket allows not only several binding modes for smallmolecule inhibitors but also the simultaneous binding of more than one inhibitor molecule. 23 Such unpredictable binding modes, due to the large size of the active site, make the rational design of competitive, high-affinity inhibitors of TryR a very challenging task.
Bearing in mind the homodimeric nature of this enzyme, back in 2013 we devised an alternative inhibition strategy that aims to disrupt the protein−protein interactions responsible for LiTryR homodimerization. 25 Initially, by means of a combination of molecular modeling and site-directed mutagenesis studies, we identified and validated E436an amino acid located at an interfacial α-helix spanning from P435 to M447as a hotspot for dimer stabilization and assessed its importance for the catalytic activity of the enzyme. As a "proof of concept" of this novel approach, we designed and tested a small library of linear peptides that represent rational variations of this α-helix. From these studies, the 13-mer-modified peptide sequence PKIIQSVGIS-Nle-K-Nle (1) emerged as a potent enzyme dimerization disruptor that also behaves as a strong inhibitor (in the submicromolar range) of the oxidoreductase activity of LiTryR. 25 Next, a variety of helixstabilized cyclic peptides 26,27 and α,β 3 -peptide foldamers 28 were successfully explored in our search to find peptidomi-metics with increased proteolytic stability. However, conjugation with cell-penetrating peptides was always required to facilitate the cellular uptake of these peptide-or peptidomimetic-based LiTryR dimerization disruptors and to kill the parasites in cell culture. 28,29 Thus, the design of small molecules with druglike properties better than those of the prototype peptide or the previous peptidomimetics appeared as a desirable goal.
In this regard, we recently disclosed the results of our first steps toward nonpeptide disruptors of LiTryR dimerization with promising leishmanicidal activity by an α-helical mimetic approach. 30 Among the reported proteomimetic scaffolds, 31−33 the pyrrolopyrimidine 34 and the 5-6-5 imidazole-phenylthiazole 35 cores were selected to dictate the spatial orientations of the side chains of three key residues (K2, Q5, and I9) in the linear peptide prototype 1. Imidazole-based compounds 2 and 3 (Figure 1), bearing a naphthalene or a biphenyl polyaromatic R 3 substituent to mimic the hydrophobic isoleucine residue, emerged as potent inhibitors of the LiTryR oxidoreductase activity, while pyrrolopyrimidines were shown to be much less active. Although these small molecules displayed a moderate capacity to disrupt the LiTryR dimer, this effect was much less pronounced than that observed for the previous peptide-based inhibitors. Remarkably, the imidazole-based compounds were cell-permeable and showed significant leishmanicidal activity against both amastigote and promastigote forms of Leishmania parasites. However, these molecules displayed a cytotoxic activity similar to that observed for their peptidic and peptidomimetic predecessors and exhibited low selectivity indexes (SIs).
1,2,3-Triazole-containing compounds have been widely investigated because they are considered privileged scaffolds in different areas of medicinal chemistry, 36−38 including several examples in the antileishmanial field. 39 Interest in this molecular architecture has also been fueled by its expedient synthesis through "click chemistry". 40 In this scenario, we decided to apply the bioisosteric replacement of imidazole by 1,2,3-triazole in our recently described imidazole-phenylthiazoles. 30 We herein report the synthesis and structure− activity relationship (SAR) studies of novel 1,2,3-triazolephenyl-thiazole compounds of general formulae I (Figure 1) to demonstrate the potential of this novel chemotype for LiTryR dimerization disruption and leishmanicidal activity against L. infantum parasites. A whole set of 26 triazole-based compounds modified at four different positions in the new scaffold (R 1 −R 4 Figure 1. From a peptide dimerization disruptor of LiTryR (1) and in-house imidazoles 2 and 3 to the newly designed triazole-phenyl-thiazole scaffold I. substituents) was prepared and evaluated. Five of these molecules display activities that very closely resemble that of the peptide prototype 1 and likewise behave as noncompetitive, slow-binding inhibitors. 41 Molecular modeling studies identified a putative binding site for these compounds at an almost unexplored central interfacial cavity of the enzyme and provided a structural rationale to the observed SAR results that was supported by ensuing rational design, synthesis, and activity of a new symmetrical triazole analogue with the desired properties.

■ RESULTS AND DISCUSSION
Chemistry. The synthesis of the target triazole-phenylthiazole type I compounds is exemplified by the preparation of the first series of 1,2,3-triazole derivatives 12a−c (bearing R 1 − R 3 substituents similar to those present in the predecessor imidazole analogues 2 and 3), as shown in Scheme 1. The synthetic route started from the previously described bromide intermediate 4a 30 bearing an imidazolidinone ethyl R 2 group (used as a Gln mimetic in the previous imidazole derivatives) by treatment with sodium azide in anhydrous dimethyl sulfoxide (DMSO) at 100°C for 72 h to afford the azide intermediate 5a in 85% yield. The addition of molecular sieves with pore openings of 4 Å was required to avoid the partial hydrolysis of the azide group to the undesired amine. The next step in the synthetic route was the 1,3-dipolar cycloaddition of the azide intermediate with terminal alkynes substituted with the appropriate R 1 group (as a Lys mimetic) to generate the 1,2,3-triazole ring. Initial attempts to use the rutheniumcatalyzed azide−alkyne cycloaddition (RuAAC) 40b from 5a and commercially available N-Cbz-protected terminal alkyne 6 in the presence of catalytic amounts (5 or 10%) of a cyclopentadienyl Ru catalyst [Cp*RuCl(PPh 3 ) 2 ] at different temperatures in dry dimethylformamide (DMF) and an inert atmosphere did not produce the expected 1,5-disubstituted 1,2,3-triazole 7. Only unreacted starting materials or complex decomposition mixtures were observed under these conditions. In contrast, the 1,4-disubstituted 1,2,3-triazole regioisomer 8 could be easily obtained in 56% yield by the 1,3-dipolar copper(I)-catalyzed azide−alkyne cycloaddition (CuAAC) 40a of azide 5a and the terminal alkyne 6 using the CuSO 4 /sodium ascorbate catalyst system at room temperature for 8 h in H 2 O/ EtOH (1:1 mixture) as the solvent. Subsequent treatment of nitrile 8 with aqueous (aq) 20% ammonium sulfide at 80°C for 4 h provided the thioamide intermediate 9 in 74% yield. Next, Hantzsch thiazole synthesis from 1 equiv of 9 and 1 equiv of the commercially available α-bromomethylketone 10a or the previously synthesized 10b and 10c 30 35,24, and 35% yields, respectively. Chromatographic purification by Biotage using reverse-phase columns was always required to obtain the final compounds with purities higher than 95% for their biological evaluation. Linear peptide 1 and imidazole-based compounds 2 and 3 were included as reference compounds. b Enzymatic activity >75 indicates that the IC 50 value is higher than 75 μM (maximum assayed). Results are representative of three independent experiments, each performed in triplicate. c Dimer quantitation assay (enzyme-linked immunosorbent assay (ELISA)). 25 d Percentage of inhibition observed at 20 μM as reliable IC 50 values could not be determined in this case. e Percentages of inhibition observed at 20 μM were 73 and 78% for 12b and 12c, respectively. f Im* = imidazolidinone, THQ* = tetrahydroquinoline (TFA salt), DHB* = dihydrobenzofuranyl, DBF* = dibenzofuranyl.
As will be discussed below, the 1,2,3-triazole compounds 12a−c emerged as more potent dimerization disruptors of LiTryR than the reference imidazoles and also behaved as strong inhibitors of the oxidoreductase activity of the enzyme. These results prompted us to develop a novel series of triazole derivatives 12d−s (see structures in Table 1 and Scheme S1) modified at the R 1 −R 3 substituents to perform a SAR study. Regarding modifications at R 1 and R 2 , the protonated aminobutyl group linked to the triazole was replaced by OH, Me, or COOH (12d−g) or by a shorter aminopropyl substituent (12h, 12i), while the R 2 imidazolidinone (attached to the central phenyl ring through an ethyl linker) was removed (12j, 12k). In terms of R 3 modifications, bulky alkyl substituents (12l, 12m) or a phenyl group linked to the thiazole ring through a polymethylene spacer of a different length (12n−p vs 12a) were explored.
Since the presence of a biphenyl or naphthyl substituent at R 3 was highly relevant for potent enzyme inhibition and antileishmanial activity, other polyheteroaromatic rings were also explored at R 3 (12q−s). The synthesis of the novel triazole analogues 12d−s involved (i) a high-yielding nucleophilic aromatic substitution (SN Ar ) of aryl fluorides with the appropriate alcohols (bearing the R 2 substituent), 30 (ii) CuAAC reactions between azide intermediates and terminal alkynes (containing the R 1 substituent), (iii) Hanztsch thiazole synthesis of thioamides with suitable α-bromomethylketones (with the R 3 substituent), and (iv) and a final step of catalytic hydrogenation to yield the deprotected final compounds using similar protocols to those described above. It should be noted that in the catalytic hydrogenation of the protected triazole compound intermediate 11q (see Scheme S1) bearing a quinoline moiety as the R 3 substituent, the 1,2,3,4-tetrahydroquinoline compound 12q (see Scheme S1) was isolated as a ditrifluoroacetate salt due to the partial hydrogenation and salt formation of the quinoline ring nitrogen atom. The synthetic experimental details of the target 12q−s compounds and the noncommercially available key αbromomethyl-ketone intermediates 10m, 10p−s are included in the Supporting Information (see Schemes S1 and S2). Scaffold-truncated simplified compounds (13 and 14, see Table 1), possessing only two aromatic rings and substituents, were also prepared for the SAR studies following similar synthetic protocols (see also the Supporting Information for experimental details; Scheme S3). To further investigate the SAR of the triazole-based compounds, we next focused on the preparation of an additional series of triazole analogues 19a−e (Scheme 2) disubstituted at both positions 4 and 5 of the thiazole ring. Polyaromatic groups directly attached or linked through an ethyl spacer to the 4-position and a phenyl group at the 5-position were selected as R 3 and R 4 substituents, respectively. As shown in Scheme 2, the preparation of these compounds required the previous synthesis of the non-Scheme 2. Synthesis of Triazoles Disubstituted at the Thiazole Ring (19a−e) a a commercially available α-bromobenzylketones 17a−d. Thus, 17a−c bearing a biphenyl or a phenyloxyphenyl group at R 3 directly attached to the carbonyl were synthesized by bromination of the commercially available benzylketones 16a−c with N-bromosuccinimide (NBS) in p-toluensulfonic acid (p-TsOH) in good to excellent yields.
On the other hand, α-bromobenzyl α,β-unsaturated ketone 17d with a polyaromatic group was obtained by regioselective bromination of the most favorable benzylic position with NBS in p-TsOH of the corresponding benzylketones 16d, which was previously synthesized from 1-chloro-3-phenylpropan-2one by treatment with PPh 3 to give the phosphonium salt 15 followed by a Wittig procedure with the corresponding polyaromatic aldehydes in moderate yields. Next, Hantzsch thiazole synthesis from thioamide intermediate 9 with the commercial 17e or synthetic α-bromobenzylketones 17a−d, followed by catalytic hydrogenolysis of the N-Cbz group and simultaneous olefin hydrogenation of 18a−e using 10% Pd/C in the presence of TFA, afforded the desired unprotected disubstituted thiazole derivatives 19a−e as monotrifluoroacetate salts in low to moderate yields. Finally, we also designed (see section Computational Studies below) and prepared (Scheme 3) a new symmetrical triazole analogue (22) by treating 2 equiv of thioamide intermediate 9 with 1 equiv of αbromomethylketone 20 (previously synthesized by bromination with NBS of commercially available 1,3-diacetyl benzene) under Hantzsch reaction conditions, followed by catalytic hydrogenolysis of the NHCbz group using similar procedures to those described above.
Biological and Structural Studies. Enzymatic Assays. L. infantum TryR Inhibition and Dimerization Assays. All of the novel triazole-phenyl-thiazole modified at substituents R 1 −R 3 (12a−s), disubstituted thiazole analogues bearing R 3 and R 4 substituents (19a−e), and truncated simplified compounds (13 and 14) were evaluated for their effect on both the oxidoreductase activity and the dimerization status of LiTryR. 25 The prototype peptide 1 and imidazole-containing compounds 2 and 3 were used as reference molecules. Screening was performed at different concentrations ranging from 0.1 to 75 μM. A concentration of 1 μM TS 2 was used for this initial screening. This low substrate concentration is sufficient for comparison of compound inhibitory activity in the oxidoreductase assay. The best candidates were then characterized further to obtain their dissociation constants. The IC 50 values determined in the activity (IC 50 act ) and dimerization (IC 50 dim ) assays are shown in Table 1. According to the results obtained with the first series of triazole-based compounds 12a−c (Table 1), analogues containing a naphthyl or a biphenyl group at R 3 (12b, 12c) potently inhibited the TryR oxidoreductase activity, showing similar IC 50 act values (in the low micromolar range) to those obtained for the reference imidazole compounds 2 and 3.
Remarkably, 12a−c were much more potent LiTryR dimerization disruptors than their imidazole counterparts. Thus, in contrast to 2 and 3, for which no reliable IC 50 dim could be measured and only ∼30% dimerization inhibition was observed at a concentration of 20 μM, the IC 50 dim values for 12b, 12c (7.1 and 4.8 μM, respectively) were similar to (or even slightly lower than) those obtained for the potent prototype peptide 1. Furthermore, replacement of the naphthyl or biphenyl groups at R 3 by a phenyl (12b, 12c vs 12a) slightly increased the IC 50 act value, while it significantly decreased (in a 5−8-fold range) the ability to disrupt the LiTryR dimer. We took this result as a clear indication that the presence of a polyaromatic substituent at R 3 is relevant to potent LiTryR dimer disruption.
SAR studies on the second series of triazole compounds (12d−s) with variations at the R 1 −R 3 substituents began with modifications at the R 1 position. As shown in Table 1, the presence of a positively charged amino alkyl group at R 1 appears as an absolute requirement for inhibition of both activity and dimerization, as evidenced by the complete loss of effect observed when the amino group was replaced by hydroxyl, methyl, or carboxylic acid groups (12d−g vs 12a, 12c). As regards the length of the R 1 spacer attached to the amino group, shortening the original tetramethylene linker to a trimethylene brought about a ∼2-fold reduction of potency in both assays (12h, 12i vs 12a, 12c). The existence of the imidazolidinone ring as the R 2 substituent on the central phenyl ring also appeared important for optimal activity since its removal (12j, 12k vs 12a, 12c) resulted in compounds twice or thrice less active in both assays. Modifications at the R 3 substituent (located at the 4-position of the thiazole ring) had a pronounced effect on enzymatic activity inhibition and, even more so, on dimerization disruption. Thus, the presence of a (poly)aromatic substituent at R 3 was crucial for enhanced Scheme 3. Synthesis of the Newly Designed Symmetrical Analogue 22 a potency since compounds 12l, 12m, with bulky aliphatic groups, exhibit much less inhibitory activity on the enzyme and no destabilization effect on the LiTryR homodimer in comparison with the (poly)aromatic analogues 12a−c. In comparison, varying the length of the linker connecting the phenyl group to the thiazole (12n−p vs 12a) turned out to be not so crucial because, in general, it did not significantly affect the inhibitory activity.
In view of these results, and even though the phenylpropyl derivative 12p displayed the lowest IC 50 dim value, for the next round of modifications, we decided to focus on shortening the propylene linker at R 3 to an ethylene (12a) or even removing it altogether (12n) because of the better synthetic accessibility of the corresponding bromoketones. On the other hand, replacement of the naphthyl or the biphenyl group at R 3 in the most potent analogues 12b, 12c by polyheteroaromatics such as a dihydrobenzofuranyl substituent (12r) or a larger hydrophobic tricyclic dibenzofuranyl moiety (12s) also yielded very potent compounds but did not further increase the inhibitory activity of the most potent analogues (12b, 12c) in any of the two assays. In contrast, the introduction of a positively charged tetrahydroquinoline ring (12q) annihilated the dimerization disruption capacity of the molecules even though potent inhibition of the redox activity of the enzyme was retained. In contrast, no activity in any of the assays was observed for the simplified truncated scaffold derivatives 13 and 14, in which the appropriately substituted triazole or thiazole rings were removed. This additional evidence further supports the key role played by these two rings (and the R 1 and R 3 substituents) in this type of inhibitor.
Finally, we also evaluated a new series of triazole compounds (19a−e) disubstituted at the 4-and 5-positions of the thiazole ring with R 3 and R 4 aromatic moieties. Some of these molecules combined excellent inhibition in the redox assay with potent LiTryR dimer disruption ability. Interestingly, both activities were significantly improved by a factor of 2−4 for 19a, in which a phenyl group is bonded at both R 3 and R 4 positions, in comparison with the R 3 phenyl monosubstituted analogue 12n. Introduction of a larger biphenyl or a diphenylether group as the R 3 substituent (maintaining the phenyl group at R 4 ) significantly decreased the inhibitory activity of the highly potent diphenylsubstituted analogue, but their dimerization disruption ability was affected only slightly (cf. 19b, 19c vs 19a). Also, the combination of R 3 polyaromatics attached to the thiazole ring through an ethylene linker with an R 4 phenyl group resulted in a significant deleterious effect on the potency of the compounds as inhibitors of the LiTryR oxidoreductase activity, although the potent disrupting effect on dimerization was maintained (cf. 19d, 19e vs 19b, 19c or the corresponding monosubstituted 12c).
In general, the most active compounds in the enzymatic activity assay were also the best dimerization disruptors in the ELISA with the exception of the highly potent tetrahydroquinoline inhibitor 12q, which did not show any dimerization disruption effect, or the thiazole-disubstituted analogues 19b− e, which behave as moderate redox inhibitors but highly potent dimerization disruptors of the enzyme. All in all, our SAR study showed that (i) a positively charged butylamine (R 1 ), (ii) an imidazolidinone (R 2 ) together with a hydrophobic poly-(hetero)aromatic group as R 3 (12b, 12c; 12r, 12s) substituents, or (iii) the combination of a phenyl as R 3 with an additional phenyl as R 4 (19a) are highly relevant to optimal enzyme inhibition and dimerization disruption in this series.
To gain insight into the selectivity of the compounds for LiTryR, representative molecules 12a−c, 12n, 12r, and 19a were studied as inhibitors of human glutathione-disulfide reductase (hGR), the closest related host enzyme. Remarkably, none of these compounds showed any inhibitory activity against this enzyme. As an example, a comparison of the activities of LiTryR and hGR in the presence of compounds 12b and 12c at 50 μM of their respective substrates is shown in Figure S15.
Mechanism of LiTryR Inactivation. As already described for 1, 41 the peptidic precursor of all other compounds, a progressive loss of LiTryR activity is observed during the course of the reaction in the presence of 12b, 12c, 12r, 12s, and 19a ( Figure 2), selected as the most potent inhibitors in these series. In stark contrast, reaction rates in the presence of the classical competitive inhibitor mepacrine are almost constant until 5,5′-dithiobis(2-nitrobenzoic acid) (DTNB) (which allows substrate regeneration) is exhausted. The progress curves of the reactions in the presence of the selected compounds are typical of slow-binding inhibitors. 42a,b Despite the loss of activity observed during the reaction, dependency of the initial velocities (v i ) on 19a and TS 2 concentrations could be fitted to a model in which a rapid equilibrium is reached between the free enzyme and inhibitor (E + I) and the enzyme/inhibitor (EI) complex (Scheme 4). This analysis, based on v i values, was performed both in a DTNB-coupled assay and in a direct NADPH-oxidation assay

Scheme 4. Two-Step Mechanism of Time-Dependent
Inhibition of LiTryR a a The first step is a rapid equilibrium that generates the enzyme− inhibitor (EI) complex. This process is governed by the association rate (k 3 ) and the dissociation rate (k 4 ) constants. The second step consists of slow and reversible inactivation of the enzyme that is governed by the forward isomerization rate constant (k 5 ) and the much smaller reverse rate constant (k 6 ).
( Figure S1 and Table S1). Our data consistently reveal a nonlinear noncompetitive hyperbolic inhibition mechanism.
Peptide 1 has already been described as a pseudoirreversible time-dependent noncompetitive inhibitor that is able to promote a two-step process of enzyme isomerization (Scheme 4). 41 The kinetic data obtained for 19a, the most potent compound, were also analyzed considering a similar mechanism of enzyme inactivation. In this inactivation model, a rapid and reversible primary inhibition event generates an enzyme−inhibitor binary complex (EI complex). The process of generating an inactive form of LiTryR upon 19a binding is governed by k 5 and k 6 first-order rate constants, which describe an isomerization equilibrium between the initial enzyme−inhibitor complex (EI) and a second high-affinity complex (E*I) (Scheme 4). This slow isomerization process gives rise to a characteristic bending of the reaction curve that is defined by the first-order rate constant k obs , whose value describes the conversion from the initial reaction rate (v i ) to the final steady-state velocity (v s ).
Adjustment to eq 1 (see equations in Experimental Section) of the progress curves of the reactions catalyzed by LiTryR allowed us to obtain the k obs values for six 19a concentrations (4.4, 5.9, 7.9, 10.5, 14.1, and 18.7 μM) at six different TS 2 concentrations (3.1, 6.2, 12.5, 25, 50, and 100 μM) (Figures S2−S7 and Table S2). The different k obs values obtained at each substrate concentration can be used to determine the apparent inhibitory constants K i app and K i * app defined as the dissociation constant for the initial EI complex and the constant for the overall dissociation from E*I to E + I, respectively (eq 2). 42a,b Based on the relationship among k obs , v i , and v s (k 6 = k obs × v i /v s ) at all of the trypanothione disulfide (TS 2 ) and 19a concentrations assayed, k 6 varies between 3.9 × 10 −6 and 6.5 × 10 −5 s −1 . Taking into consideration the extremely slow process of recovery of the enzymatic activity after complete inactivation of LiTryR (400 nM) upon incubation with 25 μM 19a for 16 h and subsequent 2500fold dilution ( Figure S13), a value of 3 × 10 −5 s −1 was selected as a conservative upper value for k 6 . Fixing this value for k 6 rendered very good estimations of K i app and K i * app after the fitting process for all of the substrate concentrations assayed ( Figure 3 and Table 2).
The relationships between the different K i app and K i values in slow-binding inhibitors are the same as those for classical linear reversible inhibitors. 42a,b By applying eq 3 to fit the K i app values obtained at the six different TS 2 concentrations assayed ( Figure S8A), we could estimate the values of α (0.8 ± 0.4) Figure 3. Concentration dependence of the observed rate constants of LiTryR (0.8 nM) inactivation by 19a at different TS 2 concentrations. Plot of the k obs values (±standard errors) as a function of 19a concentration at six different TS 2 concentrations (3.1, 6.2, 12.5, 25, 50, and 100 μM) (k obs determinations for every 19a concentration at the different fixed TS 2 concentrations assayed are shown in the Supporting Information). Curves were fitted using eq 2, and the results of the fits are shown in Table 2.
5.0 ± 2.7 1.3 ± 0.6 a K i app and K i * app values at different TS 2 concentrations were obtained by fitting to eq 2 the k obs values for every 19a concentration at the different fixed TS 2 concentrations assayed. Results are the estimated values of the nonlinear regression ± associated standard errors. and K i (7.8 ± 1.7 μM). A similar approach (eq 4 and Figure  S8B) was followed to estimate the values of α* (5.5 ± 2.9) and K i * (0.5 ± 0.1 μM) for the two-step global process of enzyme inhibition. According to these results, formation of the initial EI complex is not affected by the presence of substrate, a behavior that is characteristic of a pure noncompetitive process (α ∼ 1). The isomerization step from EI to E*I strongly enhances the inhibitory activity of 19a, which is characterized by an overall K i * value of 0.5 μM. This isomerization process is, as expected, slightly impaired by the presence of substrate (α = 5.5) because of the generation of the enzyme−substrate− inhibitor complex, which decreases the concentration of the EI complex and thereafter the rate of E*I generation. Despite this small substrate effect, an α value of 5.5 for the overall process is still in the range of α values characteristic of noncompetitive (mixed) inhibitors. The overall inhibitory process is depicted in Scheme S5. Because of the ability of 19a to cause enzyme dissociation, the characteristic isomerization of the enzyme in this mechanism of slow-binding inhibition is expected to be related to homodimer disruption.
Interestingly, compounds 12b and 12q have similar IC 50 values in the reductase assay but only 12b interferes with dimerization. The kinetic analyses reveal that both compounds are time-dependent inhibitors but only inhibition caused by 12b follows a two-step process of enzyme isomerization ( Figure S11 and Table S3). Plots of k obs vs inhibitor concentration for 12q fit to a straight line that is indicative of a single-step binding mechanism (Figure S12, Table S4, and Scheme S4), which agrees with its inability to disrupt the homodimer.
Computational Studies. Despite numerous attempts to form cocrystals with LiTryR or soak crystals of the enzyme with moderately or highly potent triazole-based dimerization disruptors 12a, 12b, and the heteroaromatic 12r, no diffraction-quality crystal containing an enzyme/drug complex could be obtained. Of note, similar problems were encountered in previous cocrystallization efforts with the predecessor imidazoles 2 and 3. 30 Because of these so-far insurmountable hurdles, we had to rely on molecular modeling tools to try and shed light on the atomistic details of the binding process. Docking the most potent triazoles 12b, 12c into the cavity of one LiTryR monomer that lodges the 435 Pro−Met 447 α-helix of the other monomer, while feasible, failed to account for the SAR results discussed above. Indeed, the biologically essential naphthyl or biphenyl R 3 substituents could not be properly accommodated unless the ligand folded onto itself in an unrealistic fashion due to lack of room. Therefore, we sought alternative binding sites in LiTryR, as found in Protein Data Bank (PDB) entry 2JK6 22b using the FTMap web server. 43 This procedure highlighted the existence of a large, hydrophobic, and putatively druggable cavity right at the center of the dimer that is lined by residues from both monomers and located very close to the interfacial helices that were mimicked by the original peptides and peptidomimetics ( Figure 4).
Interestingly, this central cavity at the LiTryR dimer interface is reminiscent of that present in Plasmodium falciparum GR 44 or hGR and is shown to bind some noncompetitive or uncompetitive inhibitors such as safranin, menadione, 45 xanthenes, 46 or N-arylisoalloxazines. 47 Ligand accessibility to this site was assessed by means of the CAVER web server, 48 which identified several tunnels connecting this intermonomer cavity to the bulk solvent ( Figure 5).
We next performed docking studies with the most potent triazole dimerization inhibitors 12b, 12c and the disubstituted thiazole analogue 19a centering the search within the proposed interfacial cavity of the LiTryR homodimer ( Figure 6). The best ligand poses had in common the positioning of the (hetero)aromatic scaffold inside the connection tunnel and the terminal ammonium group of the R 1 substituent at the hydrogen-bonding distance from the carboxylate of Glu466′ The proposed binding modes also revealed the potential of the imidazolidinone group of the R 2 substituent to act as a hydrogen bond acceptor from the hydroxyl group of either Thr65 (d(N 12b ···O Thr65 ) = 2.9 Å), for the naphthyl compound 12b, or Ser433′ (d(N 12c ···O Ser433′ ) = 2.5 Å), for biphenyl compounds 12c and 19b. The unhindered rotation of the central benzene ring of the scaffold seems to play a fundamental role to distinctly orient the R 2 substituent in each complex. Regarding the R 3 hydrophobic substituents at  . Visualization of the potential access routes from the bulk solvent to the interfacial cavity (2.808 Å 3 , 0.77 "druggability") as determined by the CAVER Web server using default parameters. 48 In comparison, each nicotinamide adenine dinucleotide phosphate (NADP) binding site occupies 1.343 Å 3 and its druggability is 0.63. The tunnels are depicted as a continuous semitransparent gray surface. The central vertical α-helices encompass the amino acid residues originally identified as hotspots for dimerization. The FAD prosthetic groups are displayed as stick models for reference purposes only. the thiazole, the bulky naphthyl and biphenyl aromatic groups for compounds 12b, 12c could be accommodated in one of the hydrophobic subpockets of the interfacial cavity that is mainly defined by Pro435′ and Phe367′, in good accord with the previous FTMap results. In addition, for the R 3 and R 4 disubstituted phenyl-thiazole analogue 19a, one of the phenyl groups attached to the thiazole ring was found to point directly toward the second hydrophobic subpocket Leu72−Leu72′ at the bottom of the cavity, another location pinpointed by FTMap. This binding mode nicely accounts for the abovementioned dramatic improvement of inhibitory potency relative to the monosubstituted phenyl-thiazole 12n.
Our molecular dynamics simulations and binding energy decomposition results using 19a as a representative inhibitor point to Glu436 as one of the major contributors to ligand binding (see Figure S16 and Movie S1 in the Supporting Information). We have previously shown that this residue is a hotspot for LiTryR dimerization. 25 Indeed, in all X-ray crystal structures of TryR, this glutamate's carboxylate from one monomer is involved in two short hydrogen bonds with the peptide backbone of Ser464 and Ala465 from the other monomer. Inhibitor binding appears to promote the loss of these interdimer hydrogen-bonding interactions (Figure S17) due to a wedge effect brought about by hydrophobic interactions involving the phenyl rings on the one side of the molecule and strong electrostatic interactions on the other side, which involve not only the free amino group but also the triazole ring. Indeed, the negative electrostatic potential generated by the two unsubstituted ring nitrogens faces the positive dipole of the short 400 MetGly 405 α-helix. Taken together, the theoretical calculations provide a rationale for the observed SAR results. Furthermore, the residue in hGR (PDB id: 1XAN) that is positionally equivalent to Leu72 in LiTryR is Phe78, whose aromatic side chain provides a flat platform for the stacking of ligands such as safranin or xanthene 45,46 but prevents the binding of our novel compounds, hence their remarkable selectivity. The central cavity putatively targeted by 19a and analogues is lined by the residues listed in Table S5, which also shows the positionally equivalent residues in hGR. The overall lack of identity in several crucial regions is in consonance with the observed marked selectivity for TryR.
To further support the feasibility of the proposed binding mode and improve the binding affinity for the putative target site, a new C 2 -symmetric compound 22 was rationally designed with a view to exploiting the C-2 symmetry of this homodimeric enzyme and simultaneously occupy both entrance/exit tunnels ( Figure 5). The putative LiTryR/19a complex provided the starting point for the structure-based design. As shown in Figure 7A, reproducing the binding mode of 19a within one monomer on the other monomer immediately reveals the possibility of merging both single molecular entities into one by making use of a suitable linker bonded to the R 3 substituent. To this end, a 1,3-disubstituted phenyl was selected as an appropriate spacer to yield the C 2symmetric compound 22, which was also docked into the putative binding site ( Figure 7B).
The good agreement found between the intended binding mode for 22 and that found by the automated docking program 49 (Figure 7) is noteworthy. In addition to the hydrophobic contacts between the aromatic rings and the walls of the cavity, the two terminal R 1 butylammonium groups of 22 are presumed to establish simultaneous salt bridges with Glu466/Glu467 and Glu466′/Glu467′, whereas the carbonyl groups of the two imidazolidinone moieties can establish hydrogen bonds with the protonated nitrogens of Lys61/ Lys61′, the side-chain carboxamide of Gln68, and the backbone NH of Ser433/Ser433′.
The activity of the new symmetric compound 22 was evaluated in the two enzymatic assays. Encouragingly, the compound emerged not only as the most effective LiTryR inhibitor of the entire series, with an IC 50 act value of 0.4 ± 0.03 μM, but also as a very potent dimerization disruptor (IC 50 dim = 8.0 ± 0.2 μM). In comparison with the corresponding monomeric triazole ligand 12n (substituted with an R 3 phenyl group), 22 was 28-and 4-fold more potent, respectively, in both assays (see Table 1). Taken together, these experimental results support the proposed binding mode of the monomeric and symmetric triazole-based compounds at this interfacial site.
Inhibitory Activity on Other TryR Enzymes from the Trypanosoma Genus. TryR is an essential enzyme not only for Leishmania but also for Trypanosoma parasites. Given the high degree of identity/similarity among the amino acid sequences of TryRs from these two genera, some degree of activity against so closely related enzymes would not be unexpected for our compounds. To test this hypothesis, the most active members of these series were tested as inhibitors of the oxidoreductase activity of TryRs from T. brucei (TbTryR) and Trypanosoma congolense (TcoTryR), the causative agents of sleep sickness and nagana, respectively. The results shown in Table 3 demonstrate, in fact, very similar activities for 12a−c, 12n, 12r, 19a, and 22 as inhibitors of the three enzymes.
In Vitro Antileishmanial Evaluation and Cytotoxicity. Compound Internalization into the Parasites. Our earlier attempts to develop LiTryR dimerization disruptors as leishmanicidal molecules were hampered by the poor ability of previously synthesized compounds to cross the parasite plasma membrane. The peptidomimetics herein described, endowed with fluorescent properties, no longer suffer from this limitation. Figure 8 shows the intense blue fluorescence emitted by promastigotes incubated for 1 h with compound 12a. Strong fluorescent spots can be observed in the apical Journal of Medicinal Chemistry pubs.acs.org/jmc Article region of some parasites, just opposite the flagellar pocket that is expected to be the main entry region for these compounds. Activity on Extracellular Parasites and Human Cells. Representative examples of triazole-based LiTryR enzyme inhibitors were tested in vitro against L. infantum axenic promastigotes and amastigotes. Cytotoxicity was assessed employing the human liver cancer cell line HepG2. Miltefosine, the R 9 conjugate of the prototype peptide 1, and imidazole-based compounds 2 and 3 were included as reference drugs. According to the data shown in Table 4, among the 18 triazole compounds tested, 9 showed halfmaximal effective concentration (EC 50 ) values below 10 μM against Leishmania promastigotes or amastigotes (12b, 12c, 12i−k, 12n, 19b, 19d, and 19e) that are similar to those observed for the reference compounds.
Of all the compounds bearing a butylamine at R 1 , an imidazolidinone at R 2 , and different R 3 substituents (12a−c), the biphenylethyl analogue 12c displayed better activity against both promastigotes and amastigotes than those containing naphthyl or phenylethyl substituents (12c > 12b > 12a) and also lower cytotoxicity. This results in an improved selectivity index (SI) of 8.8 for promastigotes (SI p ) and 3.7 for amastigotes (SI a ) in comparison with the SIs of the corresponding biphenyl imidazole analogue 3 (SI p = SI a = 2.7). On the other hand, a combination of a shorter propyl spacer at R 1 and the biphenyl group at R 3 gave 12i, the most potent and selective compound of the entire series, with EC 50 values of 5.8 against promastigotes (EC 50p ) and 5.2 μM against amastigotes (EC 50a ) and the highest SIs (SI p = 10.4 and SI a = 11.6). The imidazolidinone moiety at R 2 could be eliminated without compromising the leishmanicidal activity (12j, 12k), but a significant increase in cytotoxicity was detected, which suggests a positive effect of the imidazolidinone on reducing the cytotoxicity. As regards R 3 , the biphenylethyl substituent could be replaced by a phenyl resulting in an equipotent leishmanicidal analogue (12c vs 12n), in contrast to the higher IC 50 values observed for 12n in the enzymatic assays. However, the SI for 12n was low due to an increase in toxicity. The introduction of polyheteroaromatic rings gave rise to a significant reduction of the leishmanicidal activity (12q−s vs 12c) in contrast to the high potency observed for these analogues in the enzymatic assays. Disubstituted thiazole analogues 19a−e also showed high leishmanicidal activities against amastigotes with EC 50 values in the 8.4−22.6 μM range but, with the exception of 19c, low SIs (SI < 2.5) due to higher toxicities. Remarkably, the symmetrical triazole analogue 22 turned out to be one of the most potent and selective agents against L. infantum amastigotes with an EC 50a value of 4.9 μM and a SI a of 6.2.
It should be noted that a direct correlation between in vitro TryR inhibitory activity and in cellulo effects is not to be expected due to differences in cellular permeability or the possibility that this enzyme is not the only target for these compounds. Therefore, among the newly tested triazole-based analogues, the propylamine biphenyl derivative 12i and the C-2 symmetric 22 displayed the best leishmanicidal activity and the lowest in cellulo cytotoxicity with SI values that exceeded those obtained for the imidazoles 2 and 3 by 5−6-fold. It is interesting to point out that, despite the relatively poor IC 50 value obtained for 12i in the oxidoreductase activity assay, its behavior in the dimerization assay is similar to that of the most potent compounds from these series. This finding suggests that their ability to disrupt the dimer may be largely responsible for the leishmanicidal activity of these compounds. As IC 50 values were calculated taking into consideration only the initial velocities of the reactions and because of the time-dependent behavior of these molecules, compounds such as 12i may show good activities in the dimerization assay while displaying poor IC 50 values.
Leishmanicidal Activity in Intracellular Amastigotes. The most potent and selective compounds were further tested for their leishmanicidal activity in amastigote-infected THP-1 cells. The EC 50 values obtained are summarized in Table 5. Miltefosine, edelfosine, amphotericin B, the R 9 -conjugate of the prototype peptide 1, and compounds 2 and 3 from the precursor imidazole series were included as reference compounds. In our view, the fact that the EC 50 values cover a range from 22 to >75 μM indicates that not all of the compounds are able to cross the plasma and phagolysosomal membranes in the macrophages. Among the monomeric triazole-based compounds, only the phenyl derivative 12a, with an EC 50 value of 32.4 μM, and the biphenyl derivatives 11.0 ± 1.3 12b 5.9 ± 1.1 9.7 ± 2.1 4.8 ± 1.9 12c 10.9 ± 1.8 10.0 ± 2.5 4.8 ± 1.8 12n 9.0 ± 0.2 12.5 ± 1.9 9.3 ± 0.8 12r 11.9 ± 0.6 15.0 ± 2.

■ CONCLUSIONS
The development of new and more efficient antileishmanial drugs is a priority in the field of neglected tropical diseases. From a small library of 26 triazole-phenyl-thiazole compounds, we have identified the most potent small-molecule dimerization disruptors of LiTryR described to date. Similar to prototype peptide 1, they behave as slow-binding, noncompetitive inhibitors with overall K i * values of 0.5 μM (19a). These molecules are also endowed with high antileishmanial activity against both extracellular and intracellular parasites and improved selectivity indexes when compared to those obtained for previous in-house imidazolebased compounds. Molecular modeling studies carried out for the most potent LiTryR triazole-containing dimerization inhibitors 12b, 12c, and 19a identified a previously unexplored but apparently druggable binding site consisting of a large central cavity at the dimer interface of the enzyme. In the absence of crystallographic evidence, despite extensive efforts, this putative binding mode provided a rationale for the observed SAR and a basis for the subsequent structure-based design of a new symmetrical triazole analogue 22 that displays improved LiTryR inhibitory activities over those of its parental monomeric triazole. This rationally designed compound shows a dramatic enhancement in leishmanicidal potency, in particular against intracellular amastigotes.
All in all, our results provide the first evidence for ligand binding to a so-far unexplored region of LiTryR and lay the ground for further design of innovative TryR inhibitors that may become valuable candidates in the process of discovering new drugs against leishmaniasis.

■ EXPERIMENTAL SECTION
Chemical Procedures. Experiments dealing with air-and watersensitive reagents/compounds were performed in an argon atmosphere. Hygroscopic species were previously dried under vacuum for 24 h by P 2 O 5 as the drying agent. Unless otherwise noted, analyticalgrade solvents and commercially available reagents were used without further purification. Anhydrous CH 2 Cl 2 was dried by refluxing over CaH 2 . EtOH was dried with 4 Å molecular sieves previously activated in a microwave oven. THF was dried by reflux over sodium/ benzophenone. Pressured reactions were carried out on a Sigma-Aldrich Ace Glass flask (50 or 100 mL) supplied with a Teflon cap. Microwave-assisted experiments were performed on a Biotage Initiator 2.0 single-mode cavity instrument from Biotage (Uppsala). Experiments were carried out in sealed microwave reaction vials at the standard absorbance level (400 W maximum power). Lyophilizations were carried out on a Telstar 6-80 lyophilizer. Miltefosine, linear peptide 1 conjugated to R 9 , and imidazole-containing compounds 2 and 3 were included as in-house reference compounds. b Results are representative of three independent experiments, each performed in triplicate. EC 50 ± SE values are indicated. c The half-maximal effective concentration (EC 50 ) is defined as that causing a 50% reduction of proliferation in L. infantum promastigotes (EC 50p ) and amastigotes (EC 50a ) or human hepatocellular carcinoma (HepG2) cells. Dead parasites were identified by their increased permeability to propidium iodide (PI). d Cytotoxicity in HepG2 cells was measured using the crystal violet method. e The selectivity index (SI) is the ratio of EC 50 against human cells to EC 50 against L. infantum promastigotes (SI p ) and amastigotes (SI a ). f ND = not determined.
Monitoring of the reactions was performed by analytical thin-layer chromatography (TLC) on silica gel 60 F 254 (Merck) or by highperformance liquid chromatography coupled to mass spectrometry (HPLC−MS) using a Waters Alliance 2695 Separation Module with a Waters Micromass ZQ Detector.
The purity of the compounds was checked by analytical HPLC on a Waters 600 system equipped with a C18 Sunfire column (4.6 mm × 150 mm, 3.5 μm) and a selectable wavelength UV detector. Alternatively, an Agilent Technologies 1120 Compact LC provided with a reverse-phase column ACE 5 C18-300 (4.6 mm × 150 mm, 3.5 μm) and integrated diode detector PDA 996 was used. As the mobile phase, mixtures of A (CH 3 CN) and B (H 2 O) (0.05% TFA) on isocratic or gradient mode with a flow rate of 1 mL/min were used. The following gradients were used: from 10% A to 100% A in 10 min (gradient A) and from 1% A to 100% A in 10 min (gradient B). The purity of the final compounds was also determined to be >95% by elemental analysis with a LECO CHNS-932 apparatus. Deviations of the elemental analysis results from the calculated value were within ±0.4%.
Melting points (m.p.) were measured on a Mettler Toledo M170 apparatus and are uncorrected. Infrared (IR) spectra were recorded on a PerkinElmer Spectrum One instrument and KBr pellet as a sample support. Mass spectrum (MS) on electrospray ionization (ESI) mode was recorded on a Hewlett-Packard 1100SD in positive mode. High-resolution mass spectra (HRMS) were obtained on an Agilent 6520 Accurate-Mass Q-TOF LC-MS equipped with an HP-1200 (Agilent) liquid chromatograph coupled to a mass spectrometer with a Q-TOF 6520 hybrid mass analyzer. Usually, an error equal to or lower than 5 ppm was allowed. Nuclear magnetic resonance (NMR) spectra were recorded on a Varian INNOVA-300 operating at 300 MHz ( 1 H) and 75 MHz ( 13 C), a Varian INNOVA-400 operating at 400 MHz ( 1 H) and 100 MHz ( 13 C), a Varian MERCURY-400 operating at 400 MHz ( 1 H) and 100 MHz ( 13 C), and a Varian SYSTEM-500 operating at 500 MHz ( 1 H) and 125 MHz ( 13 C). The assignments were performed by means of different standard homonuclear and heteronuclear correlation experiments (gradient heteronuclear single quantum coherence (gHSQC), gradient heteronuclear multiple bond coherence (gHMBC), and nuclear Overhauser enhancement spectroscopy (NOESY)) when required.
TbTryR and TcoTryR Purification. The DNA coding sequences for Homo sapiens GR, T. brucei TryR, and T. congolense TryR were polymerase chain reaction (PCR)-amplified and cloned in the pRSETA plasmid. The DNA coding for hGR (transcript variant 1; NCBI Reference Sequence: NM_000637.5) was purchased from GeneScript. Genomic DNA from T. congolense parasites was kindly provided by Dr. Stefan Magez (University of Vrije, Brussels). Genomic DNA from the T. brucei S16 cell line was kindly provided by JoséM. Peŕez-Victoria. Instituto de Parasitología y Biomedicina "LoṕezNeyra" CSIC. HIS-tagged recombinant T. brucei and T. congolense TryRs (TbTryR and TcoTryR, respectively) were purified from E. coli as already explained for LiTryR.
TryR Oxidoreductase Activity. TryR oxidoreductase activity was determined spectrophotometrically using a modified version of the DTNB-coupled assay described by Hamilton et al. 52  TryR oxidoreductase activity was monitored at 26°C by an increase in absorbance at 412 nm in an EnSpire Multimode Plate Reader (PerkinElmer, Waltham, MA). 2-Nitro-5-meta mercaptobenzoic acid (TNB) concentration was obtained by multiplying the absorbance values by 100 (50 μM TNB generates 0.5 arbitrary units of absorbance at 412 nm). In this DTNB-coupled assay, one molecule of T(SH) 2 reduces one molecule of DTNB, producing two TNB molecules. All of the assays were conducted in at least three independent experiments. Data were analyzed using a nonlinear regression model with GraFit 6 software (Erithacus, Horley, Surrey, U.K.).
Dimer Quantitation Assay. The stability of the LiTryR dimeric form in the presence of 1,2,3-triazole-phenyl-thiazole-based compounds was evaluated using the novel enzyme-linked immunosorbent assay (ELISA) developed in our laboratory. 25 Briefly, HIS-FLAGtagged LiTryR (400 nM) was incubated in dimerization buffer (1 mL of 300 mM NaCl, 50 mM Tris, pH 8.0) for 16 h at 37°C with agitation and in a humid atmosphere in the presence of the different compounds ranging from 75 to 3.12 μM. Next, the different solutions were centrifuged at 18 000g for 15 min at room temperature. The supernatants (200 μL/well) were added to the α-FLAG-coated plates (Sigma-Aldrich, St. Louis, MO) and incubated for 30 min at 37°C with agitation in a humid atmosphere. The plates were washed five times with TTBS buffer (2 mM Tris,pH 7.6,138 mM NaCl, and incubated with diluted monoclonal α-HIS horseradish peroxidase (HRP)-conjugated antibody (200 μL, 1:50 000, Abcam, Cambridge, U.K.) in fatty acid-and essentially globulin-free bovine serum albumin (BSA) (5%) prepared in TTBS buffer for 1 h at room temperature. The plates were washed five times, and o-phenylenediamine dihydrochloride (OPD) substrate prepared according to the manufacturer's instructions was added. The enzymatic reaction was stopped after 5 min with H 2 SO 4 (100 μL, 0.5 μM), and the absorbances were measured at 490 nm in an EnSpire Multimode Plate Reader (PerkinElmer, Waltham, MA). All of the assays were conducted in triplicate in at least three independent experiments. Data were analyzed using a nonlinear regression model with GraFit 6 software (Erithacus, Horley, Surrey, U.K.).
Determination of 19a Inhibitory Constants. The inhibitory constants K i and K i * for 19a were calculated following a slightly modified version of the standard DTNB-coupled assay. 52 TS 2 was serially diluted (sixfold dilution from 1666 to 52 μM) in a buffer containing 40 mM HEPES (pH 7.5) and 1 mM EDTA. The different TS 2 solutions were dispensed (15 μL) in a 96-well microplate. 19a was serially diluted in DMSO (sixfold dilution from 1875 to 445 μM). The different aliquots of 19a and the equivalent amount of the vehicle (DMSO) were added to a preassay mixture, yielding different mixtures containing 40 mM HEPES (pH 7.5), 1 mM EDTA, 416.67 μM NADPH, 208.33 μM DTNB, 1.45% DMSO, and different 19a concentrations ranging from 6.2 to 26 μM. These mixtures (180 μL) were subsequently added to the appropriate wells previously filled with different TS 2 solutions. The assay was initiated by addition of 55 μL of buffer containing 40 mM HEPES (pH 7.5), 1 mM EDTA, 272.73 μM NADP + , 0.01% glycerol, and 3.5 nM recombinant LiTryR using an automated dispensing system (PerkinElmer, Waltham, MA). The order of addition was essential to avoid enzyme preincubation with the inhibitor or the substrates. The final 250 μL assay contained 40 mM HEPES buffer (pH 7.5) 1 mM EDTA, 300 μM NADPH, 60 μM NADP + , 150 μM DTNB, 6.25−50 μM TS 2 , 1.04% DMSO, 0.002% glycerol, and 0.8 nM recombinant LiTryR. TryR oxidoreductase activity was monitored at 26°C by an increase in absorbance at 412 nm in an EnSpire Multimode Plate Reader (PerkinElmer, Waltham, MA). Coupling an automated dispenser module to the spectrophotometer enabled real-time detection of the enzymatic reaction, which was crucial for high-quality nonlinear fits of the experimental data and, especially, to estimate the initial velocity of the progress curves in the presence of 19a. To guarantee linearity of the enzyme reaction in the absence of an inhibitor, the analysis was performed with data obtained during the first 20 000 s to avoid TS 2 depletion. Under this condition, all uninhibited reactions followed straight lines until the generation of 230 μM TNB ( Figure S10). This product concentration was never reached in the inhibited reactions during the first 20 000 s. All of the assays were conducted in three independent experiments.
Data and Statistical Analysis. GraFit 6.0 software (Erithacus, Horley, SRY, U.K.) was used to perform linear and nonlinear regressions. All experiments were undertaken in triplicate to ensure the reliability of single values.
k obs values were estimated for every 19a concentration at all TS 2 concentrations by fitting the values of the TNB concentration produced during each enzymatic reaction to eq 1.
where v i and v s are the initial and the steady-state velocities, respectively, k obs is the apparent first-order rate constant for the conversion of v i into the steady-state velocity (v s ), and d is a parameter that indicates the displacement of the curve on the vertical ordinate and takes into account any nonzero value of the measured signal at time zero caused by some of the reagents. The apparent inhibition constants (K i app and K i * app ) for each TS 2 concentration were determined by fitting the k obs values obtained at different 19a concentrations to eq 2 where K i app is the apparent value of the K i for the initial encounter complex (EI) and K i * app is the apparent value of the overall inhibition constant K i * to form a higher-affinity complex (E*I). k 6 is the firstorder rate constant that defines the reverse isomerization step that returns E*I to EI. 42a, 53 The K i value for the formation of the EI complex was determined by fitting the K i app values obtained at different TS 2 concentrations to eq 3. In addition, this graph was used to assess the inhibition modality of 19a during the first equilibrium of its two-step binding mechanism. The value obtained for α defines the mode of inhibition (competitive, α → infinite; uncompetitive, α → 0; pure noncompetitive, α = 1; mixed, 1 < α < 10). 42a The overall inhibition constant K i * value was determined by fitting the K i * app values obtained at different TS 2 concentrations to eq 4. In addition, this graph was used to assess the overall inhibition modality of 19a during its two-step binding mechanism. As explained above, the value obtained for α defines the mode of inhibition. 42a Supporting information). Counter-screening against related (glutathione reductase; GR) and unrelated (superoxide dismutase) targets discarded any nonspecific activity of the compounds. Reaction curves for TryR and GR in the presence of compounds 12b and 12c are shown in Figure S15. Nifurtimox is included as a well-characterized inhibitor of GR. SAR, considered the most relevant criterion that distinguishes a PAIN from a non-PAIN, is deeply discussed in the manuscript. 54 Computational Methods. Automated Ligand Docking. A three-dimensional cubic grid consisting of 65 × 65 × 65 points with a spacing of 0.375 Å centered midway between the two active sites of LiTryR was defined for ligand docking. Electrostatic, desolvation, and affinity maps for the atom types present in the ligands of this series were calculated using AutoGrid 4.2.6. Then, the Lamarckian genetic algorithm implemented in AutoDock4 49 was used to generate up to 100 feasible binding poses of the ligands studied. The best poses were selected on the basis of results from intra-and intermolecular energy evaluations.
Molecular Dynamics Simulations. The leaprcff14SB AMBER force field and the graphics processing unit (GPU)-based implementation of the pmemd.cuda module of Amber16 in the single-precision−fixed-precision (SPFP) mode were used, as described before. 30 The molecular dynamics trajectories were analyzed using the cpptraj module of AmberTools18, and the binding energy analysis was carried out with our in-house tool MM-ISMSA. 55 The molecular graphics program PyMOL 56 was employed for molecular editing, visualization, and figure preparation.

* sı Supporting Information
The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.jmedchem.1c00206. Additional synthesis; biological and computational assays; screening of representative compounds for PAINS; 1 H NMR, 13 C NMR, and HRMS spectra and HPLC chromatograms for all of the target compounds (PDF) Molecular formula strings of the prepared compounds and biological data (CSV) Movie of LiTryR-19a complex (Movie S1) (AVI)