Identification of Novel Chemical Scaffolds Inhibiting Trypanothione Synthetase from Pathogenic Trypanosomatids

Background The search for novel chemical entities targeting essential and parasite-specific pathways is considered a priority for neglected diseases such as trypanosomiasis and leishmaniasis. The thiol-dependent redox metabolism of trypanosomatids relies on bis-glutathionylspermidine [trypanothione, T(SH)2], a low molecular mass cosubstrate absent in the host. In pathogenic trypanosomatids, a single enzyme, trypanothione synthetase (TryS), catalyzes trypanothione biosynthesis, which is indispensable for parasite survival. Thus, TryS qualifies as an attractive drug target candidate. Methodology/Principal Finding A library composed of 144 compounds from 7 different families and several singletons was screened against TryS from three major pathogen species (Trypanosoma brucei, Trypanosoma cruzi and Leishmania infantum). The screening conditions were adjusted to the TryS´ kinetic parameters and intracellular concentration of substrates corresponding to each trypanosomatid species, and/or to avoid assay interference. The screening assay yielded suitable Z’ and signal to noise values (≥0.85 and ~3.5, respectively), and high intra-assay reproducibility. Several novel chemical scaffolds were identified as low μM and selective tri-tryp TryS inhibitors. Compounds displaying multi-TryS inhibition (N,N'-bis(3,4-substituted-benzyl) diamine derivatives) and an N5-substituted paullone (MOL2008) halted the proliferation of infective Trypanosoma brucei (EC50 in the nM range) and Leishmania infantum promastigotes (EC50 = 12 μM), respectively. A bis-benzyl diamine derivative and MOL2008 depleted intracellular trypanothione in treated parasites, which confirmed the on-target activity of these compounds. Conclusions/Significance Novel molecular scaffolds with on-target mode of action were identified as hit candidates for TryS inhibition. Due to the remarkable species-specificity exhibited by tri-tryp TryS towards the compounds, future optimization and screening campaigns should aim at designing and detecting, respectively, more potent and broad-range TryS inhibitors.


Introduction
Protozoan parasites from the genus Trypanosoma and Leishmania are responsible for diseases affecting humans and their livestock. The zoonotic character of these diseases, which involve different insect species as vectors and wild animals as reservoirs, hamper the implementation of successful control strategies [1]. Immuno-prophylaxis is not yet available and for some species, such as T. brucei spp. and T. cruzi, appears unfeasible due to complex immune-evasion mechanisms [2,3]. So far, and probably for several decades ahead, chemotherapy remains as the sole choice of treatment. Only a handful of drugs are available to fight Chagas's disease (T. cruzi), sleeping sickness (T. brucei gambiense and T. b. rhodesiense) and the different forms of leishmaniasis (Leishmania spp.). Unfortunately, they suffer from several drawbacks encompassing low efficacy, resistance and route of administration [4][5][6][7]. Moreover, several of these drugs (e.g. nifurtimox, benznidazole and melarsoprol) present a non-specific mode of action that accounts for their high toxicity [8,9]. Thus, the discovery of new chemical entities targeting specific and indispensable components of parasite metabolism is a priority for trypanosomiasis and leishmaniasis.
The thiol-dependent redox metabolism is one of the unique metabolic features that distinguish trypanosomatids from humans and offer reliable molecular targets for selective drug development [10]. An example of genetic hallmarks of trypanosomatids is the lack of genes coding for glutathione reductase and thioredoxin reductase [11][12][13], which fuel the major redox systems of most living organisms (i.e. the glutathione/glutaredoxin system and the thioredoxin system) with reducing equivalents. Instead, trypanosomatids rely on the low molecular mass thiol N 1 , N 8 -bis(glutathionyl)spermidine [trypanothione, T(SH) 2 ] and the flavoenzyme trypanothione reductase for sustaining the intracellular redox homeostasis (Fig 1). T(SH) 2 delivers reducing potential to different redoxin proteins, which, by acting on different targets, control vital functions (for a review see [10]; Fig 1). Proposed functions of T(SH) 2 further comprise neutralization of xeno-and endobiotics (e.g. methylglyoxal, iron and nitric oxide), coordination of iron-sulfur complexes and reduction of ascorbate [14,15].
The biosynthesis of T(SH) 2 is achieved in two consecutive steps each involving the ligation of a glutathione (GSH) molecule by its glycine carboxyl group to the free N 1 and N 8 amine groups of spermidine (SP). Both reactions are catalysed by the C-terminal ligase domain of trypanothione synthetase (TryS; EC 6.3.1.9) at the expense of ATP (Fig 1). Some trypanosomatids species harbour (L. infantum, L. donovani, L. mexicana and T. cruzi) or express (Crithidia fasciculata) a gene coding for glutathionylspermidine synthetase (GspS; EC 6.3.1.8), which synthesizes the reaction intermediate N 8 mono-glutathionylspermidine.
Transition state mimics, previously identified as potent inhibitors of EcGspS [30][31][32], proved to be equally active against recombinant CfGspS. Notably, a Gsp-phosphinate derivative was capable to inhibit recombinant TryS from L. major, T. cruzi and T. b. brucei albeit with apparent K i values 16-40-fold higher than that obtained for CfGspS (K i of 18.6 nM) [36]. Unfortunately, these compounds displayed null biological activity against pathogenic trypanosomatids at 100 μM. Nevertheless, this phosphinate remained the only compound able to target TryS from three different pathogenic trypanosomatids.
Based on the ATP-dependency of TryS, a compound library of protein kinase inhibitors was screened against recombinant CfTryS. This led to the identification of N 5 -acetamide paullones (benzo [2,3]azepino [4,5-b]indol-6-ones) as potent inhibitors of CfTryS [10,41]. Recently, we reported the finding of a related 10-trifluormethylated acetamide derivative of this paullone as nM inhibitor of LiTryS (IC 50 350 nM) [20]. Information on the activity of these paullones against TryS and parasites from other pathogenic species is lacking.
More recently, a HTS campaign against TbTryS with a library of~62000 compounds identified several hits that upon optimization yielded leaders with inhibitory activity against TbTryS in the nM range (IC 50 values of 45, 95 and 140 nM for DDU86439, DDD85811 and DDD86243, respectively) and EC 50 values towards infective T. b. brucei that ranked from 5 to 10 μM. The compounds allowed the chemical validation of TbTryS as drug target [17,18].
We here describe the setting-up of a screening technique to detect TryS inhibitors and report the identification and chemical validation of inhibitors from TryS from three major trypanosomatid species (Trypanosoma brucei brucei: T. b. brucei, Trypanosoma cruzi: T. cruzi and Leishmania infantum: L. infantum). Trypanothione dependent redox metabolism. The chemical structure of trypanothione (N 1 ,N 8 -bis(glutathionyl)spermidine; T(SH) 2 ) is depicted at the center. Synthesis: trypanothione synthetase catalyzes the ligation of two molecules of gluthatione to one of spermidine using the energy provided by two ATP molecules. Regeneration: trypanothione reductase maintains trypanothione in the reduced state at expenses of NADPH, which can be supplied by the oxidative phase of the pentose phosphate pathway via glucose 6-phosphate dehydrogenase. Utilization: reduced trypanothione is involved in multiple functions such as the detoxification of xenobiotics, cell proliferation, defense against oxidants and protein thiol-redox homeostasis. The multipurpose oxidoreductase tryparedoxin plays an important role catalyzing electron transfer from T(SH) 2 to different molecular targets (e.g. peroxidases, ribonucleotide reductase and protein disulfides). G6P: glucose-6-phosphate, 6PGL: 6-phosphogluconolactone, T(SH) 2 : reduced trypanothione, TS 2[ox] : oxidized trypanothione, NDPs: nucleosides diphosphate, dNDP: deoxinucleosides diphosphate, E -: electrophilic species, TS-E: trypanothione-electrophile adduct, ROOH: hydroperoxide, ONOOH: peroxynitrite, NO 2

Organisms and Reagents
Unless otherwise stated all reagents were of analytical grade and purchased from Sigma-Aldrich, J.T. Baker, Carlo Erba Reagents SA, Gibco, Invitrogen, Life Technologies, Enzo Life Sciencies, Roche.
Protein concentration was determined using the Bicinconinic Acid assay with bovine serum albumin as standard. The protocols described above yielded 4-8 mg of recombinant TryS per liter of culture medium with ! 95% purity and homogeneous specific activity.

Kinetic Characterization
The kinetic characterization of His-tagged TryS was performed using the LDH/PK assay which couples ATP regeneration to NADH oxidation. The end-point assay based on detection of inorganic phosphate (Pi) by the BIOMOL GREEN reagent was used to estimate the apparent K i for ADP. All reactions were performed at room temperature (RT, 20-25°C) and a detailed description of both assays is provided in S1 Text. The apparent kinetic parameters (K M and V max ) were calculated by fitting plots of initial velocity (v) vs. substrate concentracion ([S]), determined at saturating concentration of co-substrates, to the Michaelis Menten equation assisted by the software OriginPro 8. For GSH, the K M and K i values were determined using the following equation v = V max / (1 + K M / [GSH] + [GSH] / K i ), which considers the nonproductive binding of GSH to the substituted enzyme [42]. The apparent K i for ADP was estimated from linear fitting of the plots [E]/v vs.
[ADP] at different concentrations of ATP, where [E] is enzyme concentration.

TryS Screening Assay Development
General considerations. The BIOMOL GREEN assay was adapted for the screening of potential TryS inhibitors. All assays (screening and interference test) were performed manually (non-automated) in quadruplicates, at RT and using a total reaction volume of 50 μL in 96-well plates. The colorimetric reaction was initiated by adding 200 μL BIOMOL GREEN reagent and after 20 min incubation at RT the A 650 was measured with a MultiScan EX plate reader (Thermo SCIENTIFIC). The interference of reagents (DMSO, NaCl, glycerol, DTT, SP and GSH) with the colorimetric reaction was evaluated individually using 20 μM K 2 HPO 4 as Pi standard. The assay conditions (e.g. reagents, TryS units, time) were adjusted to yield optimal signal detection (20 μM Pi) taking into account the intracellular concentration reported or estimated for each substrate and/or to avoid interference with the colorimetric reaction.
The linearity range, detection limit (e.g. [Pi] corresponding to 3σ n-1 the A 650nm of the blank) and interference was estimated from plots of blank-corrected A 650nm vs.
[Pi] or the reagent concentration for each condition tested. The plots were prepared and analyzed with OriginPro 8 (OriginLab Corporation). The errors are expressed as 2σ n-1 .
Screening assay. For all TryS, ATP was used at 150 μM because of high background signal at higher concentrations and SP fixed at 2 mM, in agreement with the intracellular levels calculated from data reported in the literature for trypanosomatids [58][59][60]. GSH was adjusted to 0.05, 0.57 and 0.25 mM for TbTryS, TcTryS and LiTryS, respectively, to avoid substrate inhibition or to approach physiological concentrations [14].
A master mix (MM) solution containing all the substrates at 1.25-fold their end concentration in assay was prepared in screening reaction buffer (5 mM DTT, 10 mM MgSO 4 , 0.5 mM EDTA, 100 mM HEPES pH 7.4, 9 mM NaCl, 10% v/v DMSO) and kept on ice until use. Microtiter plate wells were loaded with 5 μL of test compound, DMSO (reaction control) or TrySspecific inhibitor (inhibition control) and 40 μL of MM. The reactions were then started by adding 5 μL of TryS (1, 2 and 3 x 10 −5 μmol.min -1 .mL -1 for TcTryS, LiTryS and TbTryS, respectively) and stopped after 15 min with 200 μL BIOMOL GREEN reagent.
The compounds used as inhibition control were added at concentrations that inhibited 50% TryS activity: 30 μM and 150 nM MOL2008 for TbTryS and LiTryS, respectively, and 30 μM J18 for TcTryS. Blanks were prepared for each condition by adding 5 μL of screening reaction buffer with 150 mM NaCl instead of enzyme.
The TryS activity is calculated as follow: , where A 650 nm Ci is the mean absorbance at 650 nm for the reaction test with compound i, A 650 nm CiB is the mean absorbance at 650 nm for the blank control with compound i, A 650 nm E is the mean absorbance at A 650 nm for the reaction control with DMSO and A 650 nm EB is the mean absorbance at A 650 nm for the blank control with DMSO. The assays yielded an intra-assay coefficient variation (CV = σ n-1 E / E) 2.5%, a signal to background coefficient (E/E B ) of~3.5 and a Z' factor [61] ! 0.85.
Screening work-flow. The solubility in DMSO of each compound was evaluated immediately prior to assay. Compounds insoluble at 3 mM in DMSO were excluded from the screening. Soluble compounds were stepwise diluted in DMSO up to 0.3 mM. For the primary screening all compounds were tested (inhibition and blank) at a fixed concentration of 30 μM.
Compounds that decrease or increase TryS activity 55% or ! 105%, respectively, were tested for interference as described in S1 Text.
For compounds that at 30 μM decreased TryS activity 55%, the IC 50 values were obtained from 10-point dose-response curves (% TryS activity ± 2σ n-1 vs. Log nM compound) fitted to a four-parameter sigmoidal equation (Boltzmann model). The reaction conditions were identical to those described for the screening assay and, whenever corresponding, the TryS activity values were corrected by the interference at each compound's concentration.

Biology
Proliferation assays for parasites and macrophages. The biological activity of selected compounds was evaluated in the infective form of T. b. brucei strain 427, cell line 514-1313 (WT) and cell line RNAi-TryS [16], as well as Leishmania infantum promastigotes (MHOM MA67ITMAP263; kindly provided by Dr. Helena Castro and Dr. Ana Tomas, IBMC, Portugal). The assay conditions to assess the cytotoxicity of compounds towards parasites and murine macrophages (cell line J774) are described in S1 Text. All assays were performed in duplicates or triplicates.
Western blot analysis of TryS expression. Total cell extracts from the WT and RNAi-TryS cell lines of T. b. brucei grown in the presence or absence of oxytetracycline were separated under reducing conditions on an SDS / 10% PAGE. The blotted membranes were incubated with mouse polyclonal antibodies against TbTryS (dilution 1:1000) followed by a horseradish peroxidase-conjugated anti-mouse IgG (dilution 1:10000). Reactive bands were detected by chemiluminescence using ECL plus (GE Healthcare) and quantified by densitometric analysis using the free software ImageJ (http://imagej.nih.gov/ij/).
Analysis of thiols. The method is a modified version of that described in [62]. Briefly, L. infantum promastigotes or bloodstream T. b. brucei (WT and RNAi-TryS cell lines) treated or not with compounds (MOL2008 or EAP1-47, respectively) for 24 h were harvested by centrifugation (2000g, 10 min at 4°C), washed twice with PBS and resuspended at a density of 10 8 cells/ 0.1 mL of 40 mM Hepes pH 8.0 and 2 mM mono-bromobimane in methanol. After incubation at 70°C for 3 minutes, the samples were cooled on ice, then added of 100 μL 4 M methanesulfonic acid (adjusted to pH 1.5 with LiOH) and incubated overnight at 4°C. The supernatant containing the derivatized low molecular mass thiols was centrifuged at 13000g 40 min at 0°C. Labeled thiols were separated using solvent A: 0.25% camphorsulfonic acid pH 2.64 and solvent B: 25% 1-propanol in solvent A and the following conditions: 100 μL of the sample, corresponding to 5x10 7 cells, was injected onto a C18, 5 μm, 4.6 x 150 mm column connected to an HPLC (both from Agilent) pre-equilibrated in 12% solvent B followed by a linear gradient from 12-50% solvent B over 36 minutes at a flow of 1 mL/min. Specific thiol concentrations were determined from peak areas relative to different masses of derivatized standards (i.e. 100, 75, 50, 25, 10, 5 and 0.5 nmol) of GSH and T(SH) 2 , the last produced according to [63].

Kinetic parameters of TryS and its implicancies for enzyme regulation and inhibition
The kinetic studies of TryS employed in this work were performed with three major aims, first to test the quality of the recombinant enzymes and establish optimal assay conditions, second, to disclose kinetic data previously not addressed for TryS from T. cruzi strain Tulahuen 0 and L. infantum (strain JPCM5) and a reaction product (ADP), and third, to compare the kinetic behavior of different TryS and its implication for enzyme regulation and inhibition. The kinetic data available for TryS from earlier [18; 22-27] and the present work are presented in Table 1 and S1 Fig.
LiTryS presented apparent K M values of 166 ± 62 μM for GSH, 1335 ± 167 μM for SP and 42 ± 10 μM for ATP, and a apparent K i for GSH of 680 ± 160 μM, all of which are in the same order of magnitude as those reported for non-tagged TryS from the related species L. major [26].
The apparent K M values for substrates and K i for GSH obtained for His-tagged TbTryS (K M GSH = 135 ± 43 μM, K M SP = 238 ± 51 μM, K M ATP = 18 ± 6 μM and K i GSH = 242 ± 102 μM) were almost 3-to 7-fold higher than those reported for the untagged version of this protein (K M GSH = 23.8-56.2 μM, K M SP = 37.8-92 μM, K M ATP = 6.6-8.6 μM and K i GSH = 36.5-143 μM) by different laboratories [24,27], which may be ascribed to different assay conditions (summarized in S8 Table) as highlighted in previous publications [23,24]. Nonetheless, all values determined here for His-tagged tritryp TryS differed in less than one order of magnitude from those reported for tag-free versions from identical or homologue proteins, hence the recombinant form of the enzymes were rated as suitable for the screening assay.
Inhibition by ADP has been reported to occur for CfGspS with a K i of 80 μM [64] but information is lacking for related enzymes from pathogenic trypanosomatids. As shown here, ADP competed for the ATP-binding site of TbTryS, LiTryS and TcTryS (S1 Fig) with apparent K i The apparent K M and K i values (μM) were determined using the lactate dehydrogenase (LDH)/pyruvate kinase (PK) assay and the K i ADP was assessed using the end-point assay. Whenever available, the values are expressed as the mean ± S.D.. The parameters determined for the N-terminal His-tagged tritryp TryS studied in this work are shown in bold letters and the references to data published earlier in brackets. Please, refer to S8 values of 40 ± 5 μM, 90 ± 8 μM and 60 ± 6 μM, respectively (Table 1). It is worth to note that these K i values represent overall estimates of product inhibititon because the assay conditions used do not allow distinguishing a preferential inhibition of the first or second biosynthetic step catalyzed by TryS. Comparison of the kinetic parameters for TryS from three major trypanosomatid species obtained under similar experimental conditions (this study), shows that the enzyme from African trypanosomes presents K M values for ATP and SP that are remarkable lower (2.3-5.6-fold) than those for TryS from L. infantum and T. cruzi (Table 1). A similar conclusion can be drawn from studies carried out for related TryS in other laboratories [18,22,[24][25][26][27]. At variance with Leishmania spp. and T. cruzi, T. brucei spp. is an extracellular pathogen that fully relies on de novo synthesis of polyamines and glycolysis to fulfil its metabolic and energetic needs [58,65]. Thus, TryS from African trypanosomes presents kinetic parameters that guarantee the production of the indispensable metabolite trypanothione [16,17] under conditions of restricted supply of ATP and SP that the parasite may face during its complex life cycle (e.g. differentiation, which entails drastic reprogramming of energetic metabolism, and different nutrient availability in vector and host). All three tritryp TryS displayed similar K M values for GSH, which contrast with earlier studies reporting a higher affinity of T. b. brucei TryS for this substrate [18,22,24,27]. As previously reported [18,[23][24][25][26][27], tritryp TryS were susceptible to substrate inhibition by GSH, although the inhibitory efficiency varied between species as follows: . This suggests that the T. b. brucei enzyme is particularly sensititive to inhibition by the substrate GSH. However, the physiological role of this inhibition mechanism, if any, remains questionable because substrate accumulation will further enhance TryS inhibition and GSH cannot surrogate T(SH) 2 functions in T. brucei [16]. For the leishmanial and T. cruzi TryS, substrate inhibition will become relevant only at high concentrations of GSH (e.g. > 1.5 mM), which, except for a few examples of L. donovani parasites grown to mid-log phase (e.g. 2.27 mM GSH for axenic amastigotes from the strain BOB and 1.68 mM for promastigotes from the strain LV9) [14], appears to be a non-physiological condition. Indeed, L. infantum parasites harbouring a single trys allele and with a TryS content about 50% lower than that of wildtype cells show no phenotype in vitro and in infected animals [20].
The K i ADP / K M ATP ratios for tri-tryp TryS range from1.5 to 2.2-fold, suggesting that ADP may be a physiological modulator of TryS activity. However, taking into account that the intracellular concentration of ATP for trypanosomatids (e.g. 2-4 mM and 0.58 mM for infective T. b. brucei [65,66] and T. cruzi [67], respectively, and 0.87 mM for L. donovani promastigotes [68]) is > 14-fold in excess with respect to the respective K M values for tritryp TryS (e.g. 18-60 μM; Table 1) and that the ATP/ADP ratio reported or estimated by us for trypanosomatids is between 3 to 10 [65,66,[69][70][71], such regulatory role of ADP in T(SH) 2 biosynthesis may be questioned. A recent kinetic analysis of TbTryS assisted by computational modeling of the enzymatic mechanism [24] highlighted that the activated enzyme is particularly sensitive to inhibition by GSH and T(SH) 2 . Because the experimental setup employed in this study did not consider inhibition by ADP and excluded ADP as an integral component of the enzyme activated complex, the real contribution of this product to TryS inhibition remains to be addressed.

Screening
Assay development and validation. Under reaction conditions resembling the screening assay and using KH 2 PO 4 as Pi standard, the detection limit and linearity range of the colorimetric reaction with BIOMOL GREEN was~1.9 μM Pi and 2-50 μM Pi, respectively (S2 Fig). DTT and NaCl (salt present in enzyme buffer) added up to 5 mM and 30 mM, respectively, do not affect assay's readout (S3A and S3D Fig). DMSO (compound solvent) and glycerol (cryoprotective agent added to TryS) interfered with BIOMOL GREEN signal in an additive fashion. For instance, 10% v/v DMSO (e.g. concentration used in the assay) and 4% v/v glycerol reduced assay sensitivity 8.4 ± 1.5% (n = 8) and~12% (S3C Fig), respectively, whereas addition of both reagents (DMSO 10% v/v and glycerol 4% v/v) decreased~25% assay sensitivity (S3D Fig). In order to reduce the signal interference, assays were performed at glycerol concentrations < 1% v/v. Assay linearity is affected at ATP concentrations > 150 μM due to increase background signal and ADP-mediated inhibition of TryS (S1 Fig; see previous section). Thus, for practical reasons ATP was adjusted to 150 μM, a concentration that is 4-27-fold lower than the estimated physiological concentration of the substrate in different trypanosomatids [66][67][68][69][70] but~3.5 and 8-fold higher than the K M values for TcTryS and LiTryS, and TbTryS, respectively (Table 1).
GSH added up to 2.5 mM does not interfere with the colorimetric signal. For assays with TcTryS, GSH was used at 570 μM, a value within the physiological concentrations reported for infective forms of T. cruzi strains (i.e. 340 to 1180 μM) [14]. For the screening involving LiTryS, GSH was employed at a concentration of 250 μM, which is near the intracellular content estimated for amastigote forms of L. donovani (250 μM) and L. major (310 μM) isolated from hamsters and mice, respectively [14]. In order to avoid substrate inhibition of TbTryS, GSH was adjusted to 50 μM a value close to the corresponding K M (Table 1) and 5-fold lower than the intracellular concentration (i.e. 230-260 μM) reported for this substrate in trypanosomes isolated from infected rats [14], respectively.
Irrespective of the TryS species, SP, a substrate not interfering with the colorimetric reagent (S3B Fig), was fixed to 2 mM for the screening assays. This concentration is 8.4-, 2.9-and 1.5-fold higher than the K M values for TbTryS, TcTryS and LiTryS, respectively and close to that we estimated for the clinically relevant forms of T. b. brucei (SP~2 mM for bloodstream stage), T. cruzi (~3 mM for amastigotes and trypomastigotes) and L. mexicana (~7 mM for amastigotes) according to information available on polyamine content [58][59][60], cell volumes and/or total protein content per cell for each parasite species [14,71,72].
The linearity A 650nm corr vs. time was evaluated for each TryS at substrates concentrations used in the screening assay (see Experimental Procedures, S4 Fig). For LiTryS (2.3 x 10 −5 μmol. min -1 .mL -1 ), the assay was linear for at least 30 min whereas the production of Pi by TcTryS (3.5 x 10 −6 μmol.min -1 .mL -1 ) and TbTryS (1.5 x 10 −5 μmol.min -1 .mL -1 ) displayed a linear relation up to 15 min. As indicated above, the trypanosomal TryS were more sensitive than LiTryS to inhibition by ADP, which may explain the lower turnover of the former at longer reaction times. Therefore, a reaction end-point of 15 min was selected for the screening assays, which corresponds to the formation of 20 μM Pi and ADP for all three enzymes.
The assay conditions reported above for each enzyme yielded a Z' factor [61] of !0.85, a signal background coefficient (E/EB) of~3.5 and a CV (σ n-1 E /E) intra-assay 2.5%, which validate the screening test as highly robust and reproducible.
Compound screening at near physiological concentrations of substrates. The screening work flow is shown in Fig 2 and all data here reported correspond to assays that fulfilled the quality control parameters indicated above.
The initial screening performed with 144 compounds belonging to 7 different families (Fig  3 and S1-S7 Tables) and tested at 30 μM against tritryp TryS, allowed the identification of 22 compounds that lowered Pi signal to 55%. An assay interference counter-screen confirmed 15 of them as true TryS inhibitors (Table 2) and the remaining 7, all being BDA, as false positive readouts (S3 Table).
Early investigations had shown that the N 5 -substituted paullone derivatives MOL2008 and FS-554 are potent inhibitors of TryS from C. fasciculata (IC 50 = 30 nM) [10] and L. infantum (IC 50 = 350 nM) [20], respectively, but information on their activity towards the homologue enzyme from related trypanosomatids was missing. MOL2008 was the most potent inhibitor of LiTryS (IC 50 = 0.15 ± 0.06 μM) identified in the screening that displayed a moderate activity against TbTryS (IC 50~3 0 μM) and TcTryS (40% TryS inhibition at 30 μM; see Table 2). On the other hand, FS-554 showed moderate activity against TcTryS (~50% inhibition at 30 μM) and a weak activity against TbTryS (~50% inhibition at 75 μM) ( Table 2). Substitution of position N 5 appears key to confer anti-LiTryS activity to AI since the inhibitory activity of unsubstituted derivatives was almost abrogated (compound 11a and 11b in S5 Table). Given the amine nature of the N 5 substitutent, a potential SP-competitive mechanism of MOL2008 on LiTryS was investigated. Assays performed at fixed concentration of co-substrates (250 μM GSH and 150 μM ATP) and variable concentrations of SP (0.2 to 16 mM) show that enzyme inhibition was inversely proportional to polyamine concentration (S9 Table) and confirmed the proposed inhibition mechanism. A collection of 4-azapaullones substituted in position 9 and 11 was recently assayed for their anti-T. b. brucei and TbTryS activity [56]. Only derivatives containing in position 9 an α,β-unsaturated carbonyl substituted with a phenyl, phenyl (amino) or benzylamino group showed a minor anti-TbTryS activity (30-40% inhibition at 30 μM). Given the species-specificity displayed by tri-tryp TryS, the complete set of 4-azapaullones was tested against the L. infantum and T. cruzi enzyme (S5 Table). None of the compounds inhibited TcTryS to a significant level (enzyme inhibition 17% at 30 μM compound) whereas LiTryS showed again to be more susceptible to inhibition by the paullone scaffold, with 4-chloro-phenyl > furan-2-yl > 4-methoxy-phenyl analogues in 9-position reducing 32-44% enzyme activity at 30 μM. Identical substitutions (i.e. 4-methoxy-phenyl and 4-chlorophenyl) in 11-position of 4-azapaullones rendered analogues with similar inhibitory activity (30-35%) towards LiTryS. The BDA (S5 Fig, S3 Table) exerted a low to moderate inhibition of TryS that allows performing a SAR analysis. For all TryS, the inclusion of bulky substitutions (R 1 = R 4 = OBn: APC1-87, APC1-99 and APC1-109, or in addition R 2 = R 3 = OMe: APC1-111, APC1-101 and APC1-89) at the benzyl rings yields derivatives with higher activity (40-60% inhibition) than the unsubstituted congeners and analogues containing less bulky groups (R 2 = R 3 = OMe, R 1 = R 4 = OH) (S5A Fig). In contrast to LiTryS, the trypanosomal enzymes were particularly sensitive to the length of the linker, with derivatives bridged by 3-or 10-carbon being less active. This suggests a substituent additive effect to the length of the carbon chain with a possible steric fulfillment demand of the interacting pocket. For the halogenated BDA, the length of the linker appears to be a common determinant of activity against all TryS, whereas each enzyme displayed certain specificity for the nature of the halogen atom (S5B Fig). Thus, derivatives with a 12-carbon linker are more potent TryS inhibitors than those consisting of shorter chains and the halogen substitution yielding the highest activity (~50% inhibition at 30 μM) for this scaffold is F (EAP1-69), Cl (EAP1-67) and Br (EAP1-63) for TcTryS, LiTryS and TbTryS, respectively. On the other hand, the inclusion of an isopropyl group at position R 1 = R 4 of BDA with a 10-carbon linker (EAP1-47) improved up to 40% tri-tryp TryS inhibition when compared to the unmodified compound EAP1-9 and produced the best multi-TryS inhibitor here identified (50% inhibition at 30 μM).
From a group of 35 APPDA tested, three derivatives containing a substituted phenyl or benzyl group in position R3 (i.e. ZEA35, ZEA 40 and ZEA 41) showed selectivity and a moderate potency towards TcTryS (50% inhibition at 30 μM; Table 3). Replacing the R3 benzyl of ZEA41 by a naphthalen group (ZEA10) increased the activity against TbTryS (50% inhibition at 30 μM) and LiTryS (17% increased inhibition) but was detrimental to TcTryS inhibition (18% lower inhibitory activity). TcTryS was also target of selective inhibition by several BZ with J18 being the most active (IC 50~3 0 μM), and the derivatives J20 and J21 reducing about 40% enzyme activity when tested at 30 μM. J18 produced a similar degree of inhibition of LiTryS, whereas TbTryS was almost refractory to inhibition by the BZ (S4 Table). Certain BBHPP (ADMRC150, ADMRC154, ADMRC159 and ADPKN164) showed specificity for targeting LiTryS with an average inhibitory activity of 35% at 30 μM and marginal or, even, enhancing activity of trypanosomal TryS (Table 3 and S2 Table).
Under the assay conditions used for the screening, none of the PD, AOCA and singletons inhibited tri-tryp TryS to a significant level (S6 and S7 Tables). In this respect, prochlorperazine has been early reported as a SP-competitive inhibitor of TbTryS with an IC 50 of 19 μM, when assayed at a subsaturating concentration of the competing substrate (i.e. 25 μM SP) [18]. To investigate the potential of prochlorperazine to target TryS from other species and validate their mode of inhibition, we tested the activity of this compound against different TryS and at different SP concentrations. At 2 mM SP, the estimated intracellular concentration of this polyamine in trypanosomatids [58][59][60], none of the TryS was inhibited by prochlorperazine added at 30 and 300 μM. Lowering the SP concentration in the assay to 0.24 mM SP, 300 μM prochlorperazine reduces the activity of TbTryS, TcTryS and LiTryS by 58%, 34% and 0%, respectively. On the one hand, this result show that high intracellular concentrations of prochlorperazine have to be reached to impair T(SH) 2 synthesis to a significant level and, on the other hand, highlights that more stringent assay conditions, as those representing nearly physiological substrate concentrations, will avoid the detection of hits with low pharmacological potential.
The screening also allowed the identification of compounds that increased 16-69% the activity of the trypanosomal TryS (i.e. for TbTryS: 7 BBHPP and 2 BZ and for TcTryS the APPDA ZVR159 and the PD TC227) (Table 3). Strikingly, none of the compounds tested here produced a similar stimulatory effect on the activity of the leishmanial enzyme. In order to shed light on the mechanism underlying the activation of TryS, the most active compounds were tested against trypanosomal TryS in the presence of different combinations of co-substrates. As shown in S10 Table, enhacement of TryS ATPase activity by non-chemically related compounds (ZVR159, TC227 and ADPKN165) was detected only when both co-substrates, GSH and SP, were present in the reaction mixture. These results are suggestive of an allosteric mechanism of TryS activation. Enzyme activity is expressed as % TryS activity ± 2σ n-1 and in brackets is provided the interference factor used to correct TryS activity of compounds affecting BIOMOL GREEN signal (see S1 Text). All values reported stem from at least triplicates. Biological activity of compounds inhibiting TryS from African trypanosomes and L. infantum The biological activity of the most active inhibitors of TbTryS (i.e. IC 50 Table 2). In comparison to paullone FS-554 (EC 50 112.3 ± 1.1 μM) [20], MOL2008 displayed a 10-fold higher antileishmanial activity (EC 50 12.6 ± 1.6 μM).
In order to get an insight into the on-target activity of promising compounds, the intracellular thiol content of WT parasites exposed for 24 h to EAP1-47 and MOL2008 was determined. For bloodstream T. b. brucei treated with 50 nM EAP1-47, T(SH) 2 content decreases by 28% while GSH level increases by 39% with respect to untreated cells, which overall resembles the metabolite changes observed for parasites with 48 h RNAi-downregulated expression of TryS (S11 Table and [16]). A compound acting on T(SH) 2 metabolism is expected to display an increased cytotoxicity against parasites depleted in TryS. The cytotoxic effect of 100 nM EAP1-47 towards trypanosomes with a TryS content that is 1/3 of that corresponding to non-induced RNAi or wild-type cells was slightly increased (1.3-fold; Fig 5). For comparison, nifurtimox, a drug inducing thiol depletion [73], added at 5 μM (EC 50 determined for WT cells) displayed a 1.8-fold increased potency towards TryS-depleted parasites. Strikingly, identical assays performed with the chemically-related and trypanosome-selective compounds EAP1-63 and APC1-99 did not show differences in the potency of these compounds towards WT or TryS-depleted parasites. Taking together, these results shows that EAP1-47 is interfering with T(SH) 2 biosynthesis, although the almost two order of magnitude difference between TbTryS inhibition and EC 50 indicates that the compound has also other molecular targets in vivo. In contrast, the lack of enhanced cytotoxicity displayed by EAP1-63 and APC1-99 towards the TryS-RNAi induced cell line, is an strong indication that their potent trypanocidal activity is unrelated to interference with T(SH) 2 metabolism.
On the other hand, L. infantum promastigotes in the log growth phase exposed to MOL2008 at its EC 50 (12 μM) for 24 h presented a marked decrease in the intracellular pool of thiols. From two independent experiments, T(SH) 2 and GSH content decreases ! 90% and > 30%, respectively, upon MOL2008 treatment (S11 Table). Interestingly and at variance with the metabolic changes observed for the genetic (RNAi) or chemical (EAP1-47) silencing of TryS in T. b. brucei, inhibition of TryS by MOL2008 did not lead to accumulation of the substrate GSH in L. infantum. Further experiments are required to establish whether the reduction on GSH level is caused by paullone-mediated inhibition of any of the two ATP-dependent enzymes in charge of glutathione synthesis or is a consequence of a species-specific regulatory mechanism triggered by T(SH) 2 depletion.

Discussion
The screening of a compound library consisting of 7 major chemical scaffolds and several singletons against TryS from three major pathogenic species of trypanosomatids at near physiological concentration of substrates led to the identification of 15 inhibitor molecules with μM (APPDA, BDA, BZ, AI) and sub-μM (AI) potency and a remarkable species-specificity for the molecular target. Despite the high sequence identity between tri-tryp TryS (e.g. the amino acid sequence identity is 71.6% for TbTryS/TcTryS, 64.7% for TbTryS/LiTryS and 63.4% for TcTryS/LiTryS) and the almost strict conservation of residues involved in substrate binding [20,21], only 4 compounds were able to target TryS from different species, with only one of them (a BDA) inhibiting all three TryS with moderate potency (50% at 30 μM). This behavior highlights the existence of structural differences between tritryp TryS that determine their specificity for ligands. In support of this observation, single aminoacid substitutions in the homologue enzyme from C. fasciculata were previously reported to produce drastic changes in enzyme activity [23,36] and our comparative analysis of the kinetic parameters of tritryp TryS obtained under similar assay conditions reveals considerable differences between species. Early investigations have shown that substitutions at different positions of the paullone scaffold yield nM inhibitors of TryS from C. fasciculata or L. infantum (N 5 -substituted AI) [10,20] and potent anti-T. b. brucei agents with moderate inhibition towards TbTryS (11-substituted 4-azapaullone) [56]. By extending the analysis of AI activity towards tritryp TryS, we have confirmed that the inclusion of a N- [2-(methylamino)ethyl] acetamide in position N 5 confers SPcompetitive inhibition of LiTryS activity, since compounds lacking this substitution were almost inactive. From the tritryp TryS evaluated here, LiTryS presented the higher K M for SP, which indicates the enzyme binds less efficiently this substrate probably due to an unfavorable conformation of the polyamine binding site. However, the shape adopted by the SP-binding site of LiTryS appears suitable for accommodating the N 5 -substituent present in MOL2008 and FS554. In support of this hypothesis, the trypanosomal TryS displayed a 2-to 6-fold lower K M values for SP and a marked refractoriness to inhibition by N 5 -substituted paullones (> 200-fold higher IC 50 ). The opposite behavior was observed for prochlorperazine, a SP-competitive inhibitor of TbTryS [18]. In this case, the degree of enzyme inhibition was inversely proportional to the SP K M value for each TryS, with TbTryS being the most sensitive to inactivation followed by TcTryS, whereas LiTryS was refractory to inactivation (300 μM prochlorperazine) even at a sub-K M concentration of SP (240 μM vs. K M of 1335 μM). Altogether, these data strongly point to the existence of remarkable differences in the polyamine binding site of tritryp TryS that should be carefully considered for the design of new AI with multi-species activity. In this respect, our study also demonstrated that in spite of the potent anti-T. b. brucei activity [56], AI substituted with α,β-unsaturated carbonyl chains in position 9 or 11 should be disregarded as ligands for future optimization because of their null to marginal anti-TryS activity.
Both N 5 -substituted paullones identified as potent LiTryS inhibitors displayed an IC 50 slope (Hill coefficient) below the unity (~0.35), which suggests binding of the inhibitor to non-equivalent binding pockets or to partitioning of the compounds into an inactive, less potent or inaccessible form at higher concentrations [74]. Since both paullones displayed good solubility in the concentration range tested (~1 nM-50 μM) and produced 100% LiTryS inhibition, we ruled out insolubility or aggregation phenomena as responsible of this behavior. On the other hand, the dose-response plots do not show the presence of a second inflection point that would be indicative of a second binding pocket for paullones. TryS are monomeric but mechanistically complex (trisubstrate) enzymes [23,24] whose substrate binding sites display large conformational changes during catalysis [21,75,76], hence, it is possible that the low Hill slopes reflect an equilibrium between two or more forms of the enzyme that interact differentially with the paullones. Further mechanistic studies are needed to shed light on this issue.
From this screening, MOL2008 was the most potent inhibitor identified for LiTryS that, in addition, shows potential for further optimization of its anti-trypanosomal TryS activity (IC 50 30 μM). The related paullone FS554 was chemically validated in a previous study using transgenic cell lines of L. infantum [20]. Here we show that MOL2008 is almost one order of magnitude more potent towards L. infantum promastigotes than FS554 and, more importantly, that it targets in vivo trypanothione biosynthesis. The future design of MOL2008 analogues should also aim at improving its biological properties, which according to our data are far from optimal (EC50 in the μM range and SI 2) and suggest off-target effects in host cells.
Precedent studies demonstrated that disustituted polyamines are potent antiproliferative agents that interfere with the polyamine metabolism of Plasmodium and African trypanosomes [47][48][49][50]. On the basis that the substituted diamine moiety may eventually be recognized as ligand by TryS, a collection of simplified derivatives was screened for their anti-synthetase activity. Two BDA (EAP1-47 and APC1-111) displayed inhibitory activity, albeit moderate (IC 50~3 0 μM), against multiple TryS. SAR analysis revealed the need to fulfill a specific steric demand on the enzyme interacting pocket. A rational optimization of these compounds seems difficult, since the screening performed here included a wide diversity of derivatives (monoand di-substituted, with bulky groups or halogen atoms, and with linker of different length) that yielded only weak TryS inhibitors. Nevertheless, as in a fragment-based approach, certain moieties of the most active BDA can be selected as substituents of novel or known TryS inhibitors.
With few exceptions, most APC derivatives tested here were previously shown to display low to sub-μM potency towards infective T. b. brucei [46]. Our study confirmed and disclosed the potent (EC 50 15-280 nM) and selective (SI = 10-164) anti-parasitic activity of APC analogues and of two halogenated BDA from the EAP series, respectively, all of which exherted a moderate inhibition of TbTryS (IC 50~3 0 μM). Interestingly, EAP1-47 induced changes in the intracellular pool of low molecular thiols that overall resembled those triggered by RNAi-mediated silencing of TryS [16] and displayed a moderate increase in its cytotoxicity towards TrySdefficient cells that was not paralleled by the more potent and selective molecules APC1-99 and EAP1-63. Nonetheless, as pointed out before, the higher in vivo (anti-trypanosomal activity) vs. in vitro (TbTryS inhibition) activity of EAP1-47 suggests that this compound is targeting other essential molecular target(s) in addition to TryS.
The positive loop of substrate inhibition of TryS by GSH has a direct impact on the design of inhibitors. On the one hand, it suggests that groups competing with GSH can be excluded from the ligand scaffold to avoid unnecessary substrate competition and to reduce the molecular mass of the inhibitor. On the other hand, it shows that full inhibition of TryS is not required because the phenomenon of substrate inhibition will amplify in vivo the inactivation of the enzyme, which represents a pharmacological advantage of this molecular target.
The four APPDA identified as moderate inhibitors of, preferentially, the T. cruzi TryS, are structuraly related to compounds originally designed as inhibitors of bacterial D-Ala:D-Ala ligase. Interestingly, these derivatives presented a 10-fold lower IC 50 towards the trypanosomal enzyme than the analogue compounds against the corresponding bacterial target (best IC 50 = 260 μM) [43]. The APPDA may serve as primary scaffolds to develope analogues with improved anti-TryS activity and, consequently, superior selectivity index. TcTryS, and to minor extent LiTryS, was target of inhibition by a BZ substituted with an imidazolone. A BZ with an additional imidazolone group retained activity against TcTryS but enhanced ATP hydrolysis by LiTryS and TbTryS, indicating that these ligands present a molecular pattern recognized by the active site of the enzymes that should be further investigated.
Our work also disclosed several compounds enhancing ATP consumption by trypanosomal TryS. Adding value to the existence of distinctive structural features between tritryp TryS, TbTryS was targeted specifically by large heterocyclic compounds from the BBHPP whereas TcTryS was the most promiscuous enzyme, being activated by a PD, an APPDA and the BZ mentioned above. A single compound, a bis-imidazolone BZ, increased LiTryS activity. According to our results, the activation of ATPase activity in trypanosomal TryS by these compounds takes place only when both co-subtrates, namely GSH and SP, are present in the reaction. More detailed studies are required to establish whether such allosteric activation of TryS is paralleled by an increased production of T(SH) 2 . Compounds promoting the non-productive consumption of ATP are attractive candidates for biological evaluation, because the metabolic outcome of their action should be similar to that of true enzyme inhibitors.
In summary, this study: (i) led to the identification and on-target validation of a potent inhibitor of leishmanial TryS and of chemical scaffolds that can be further developed into inhibitors of wide spectrum against TryS from different trypanosomatid species, (ii) highlights the existence of remarkable kinetic and structural differences between tritryp TryS that prompt to obtain 3D structures of TryS from different trypanosomatids in order to guide a structure-based rationale design of multi-TryS inhibitors; iii) demonstrates the relevance of running a HTS under near physiological concentration of subtrates to minimize the possibility to detect weak competitive inhibitors with low potential to act in vivo Supporting Information , 10% v/v DMSO and 9 mM NaCl. The A 650 nm corr is expressed as percentage relative to A 650 nm corr at 5 mM DTT; B) SP (mM) for a reaction containing 20 μM K 2 HPO 4 , 150 μM ATP, 2 mM SP, 250 μM GSH, 5 mM DTT, 10% v/v DMSO and 9 mM NaCl. The A 650 nm corr is expressed as percentage relative to A 650 nm corr at SP 2 mM; C) glycerol (%) for a reaction containing 20 μM K 2 HPO 4 , 150 μM ATP, 2 mM SP, GSH 250 μM, 5 mM DTT and 9 mM NaCl. The A 650 nm corr is expressed as percentage relative to A 650 nm corr at 0% v/v glycerol; D) GSH (mM) for a reaction containing 20 μM K 2 HPO 4 , 150 μM ATP, DTT 5 mM, 10% v/v DMSO, 9 or 30 mM NaCl, and 0 or 4% v/v glycerol. The A 650 nm corr is expressed as percentage relative to A 650 nm corr at 0% v/v glycerol, 0 mM NaCl and 2.5 mM GSH (black squares), 4% v/v glycerol and 0 mM NaCl (red circles), 4% v/v glycerol and 30 mM NaCl (blue triangle) and 4% v/v glycerol and 9 mM NaCl (green triangle). (TIF)  LiTryS (a, b), Trypanosoma cruzi trypanothione synthetase, TcTryS (c, d) and Trypanosoma brucei trypanothione synthetase, TbTryS (f, g). The assays were conducted at RT in a total reaction volume of 50 μL containing 150 μM ATP, 2 mM SP, 5 mM DTT, 10 mM MgSO 4 , 0.5 mM EDTA, 100 mM HEPES pH 7.4, 10% v/v DMSO and 9 mM NaCl, with variable concentrations of gluthatione (GSH) and enzyme units according to the TryS species: 250 μM GSH and 2.3 x 10 −5 μmol.min -1 .mL -1 for LiTryS, 570 μM GSH and 3.5 x 10 −6 μmol.min -1 .mL -1 for TcTryS, and 50 μM GSH and 1.5 x 10 −5 μmol.min -1 .mL -1 for TbTryS. Blanks lacking enzyme were prepared for each condition. The TryS reaction was stopped at different time points by adding 200 μL BIOMOL GREEN reagent. The plates were incubated for 20 min at RT and A 650 nm measured using a MultiScan EX plate reader (Thermo SCIENTIFIC). The mean A 650 nm corr ± 2 S.D. of BIOMOL GREEN signal is plotted vs. time. A linear regression model based on the least square method was applied to each plot to estimate the linearity range. (TIF) S5 Fig. Activity profile for N,N'-bis(3,4-substituted-benzyl) diamine derivatives (BDA). All compounds were evaluated at 30 μM against Trypanosoma cruzi trypanothione synthetase, TcTryS (Tc), Leishmania infantum trypanothione synthetase, LiTryS (Li) and Trypanosoma brucei trypanothione synthetase, TbTryS (Tb). TryS activity is expressed as percentage (see S3  Table).