Leishmania infantum arginase: biochemical characterization and inhibition by naturally occurring phenolic substances

Abstract Inhibition of Leishmania arginase leads to a decrease in parasite growth and infectivity and thus represents an attractive therapeutic strategy. We evaluated the inhibitory potential of selected naturally occurring phenolic substances on Leishmania infantum arginase (ARGLi) and investigated their antileishmanial activity in vivo. ARGLi exhibited a Vmax of 0.28 ± 0.016 mM/min and a Km of 5.1 ± 1.1 mM for L-arginine. The phenylpropanoids rosmarinic acid and caffeic acid (100 µM) showed percentages of inhibition of 71.48 ± 0.85% and 56.98 ± 5.51%, respectively. Moreover, rosmarinic acid and caffeic acid displayed the greatest effects against L. infantum with IC50 values of 57.3 ± 2.65 and 60.8 ± 11 μM for promastigotes, and 7.9 ± 1.7 and 21.9 ± 5.0 µM for intracellular amastigotes, respectively. Only caffeic acid significantly increased nitric oxide production by infected macrophages. Altogether, our results broaden the current spectrum of known arginase inhibitors and revealed promising drug candidates for the therapy of visceral leishmaniasis.


Introduction
Leishmaniasis is an infectious-parasitic disease caused by different species of the genus Leishmania and transmitted by the female of the phlebotomine insect. Leishmaniasis is classified as a neglected tropical disease, occupying the ninth position among the most prevalent diseases in the world. It is estimated that 1.5-2 million people are infected annually, and that 350 million people live at risk of infection 1,2 . Visceral leishmaniasis (VL), or kalazar, is the most severe clinical manifestation of the disease, being responsible for high mortality rates in the absence of proper diagnosis and treatment 3 . According to WHO, in 2017, 20,792 new cases (94% of total cases) of VL occurred in seven countries: Brazil, Ethiopia, India, Kenya, Somalia, South Sudan, and Sudan 4 .
There are currently no vaccines for human VL, and treatment consists of the use of chemotherapeutic agents, such as pentavalent antimonials, amphotericin B, and miltefosine. However, those treatments show high toxicity, variable efficacy, and contribute to the emergence of resistant Leishmania strains 5 . Therefore, there is an urgent need for novel antileishmanial agents as well as the discovery of new therapeutic targets that may lead to a safer and more effective treatment of VL.
Arginase (E. C. 3.5.3.1, L-arginine aminohydrolase) is a metalloenzyme that catalyses the hydrolysis of L-arginine into Lornithine and urea, participating in the urea cycle. Arginase exhibits a trimeric structure with one active site present in each monomer. Each active site contains two manganese ions, which are responsible for activating a water molecule forming a metalhydroxide ion that attacks the guanidine carbon of L-arginine 6 . In mammals, two arginase isoforms are found, arginase I and II. They catalyse the same reaction but differ in cellular expression, regulation, and subcellular localisation 7 . In Leishmania species, arginase regulates parasite growth, differentiation, and infectivity 8,9 . Roberts et al. have shown that an arginase knockout mutant of L. mexicana is unable to grow in vitro. Addition of exogenous ornithine and/or the polyamine putrescine restores L. mexicana growth, indicating that growth arrest is probably due to the lack of these substances 8 . Moreover, putrescine is a precursor for the biosynthesis of trypanothione, which is central for parasite protection against reactive oxygen species 10 .
As a strategy for the development of safer and more effective antileishmanial drugs, several efforts have been made to find specific parasite arginase inhibitors. Previously, inhibitors of both synthetic and natural origin have been described. However, most studies have focussed on enzymes from Leishmania species causing the tegumentary form of the disease [11][12][13][14][15][16][17][18][19] . To the best of our knowledge, there is a single report on Leishmania infantum arginase. In this specific study, enzyme inhibition assay was performed with cellular extracts and did not employ the purified enzyme 20 . Here, we biochemically characterised recombinant arginase from L. infantum (herein referred as ARGLi) and evaluated its inhibition by a panel of fourteen naturally occurring phenolic substances. We used molecular docking to gain further insights into the mechanism of inhibition. In addition, we investigated the effects of these substances on parasite biology and the mammalian host cell response to L. infantum infection.

Expression of ARGLi
The plasmid containing the gene encoding L. infantum arginase was commercially obtained from Genscript USA (Piscataway, USA). ARGLi was cloned into the RP1B plasmid 21 and fused in-frame with an N-terminal six-histidine tag (His 6 ) followed by a TEV (Tobacco Etch Virus) protease cleavage site. Escherichia coli BL21 (DE3) cells were transformed with RP1B-ARGLi and cultured at 37 C until reaching optical density at 600 nm of $0.6. ARGLi expression was induced with 1 mM isopropyl-b-D-thiogalactopyranoside (IPTG) for 16 h at 30 C. Cells were harvested by centrifugation and cell pellets were kept at À80 C until protein purification.

Purification of ARGLi
ARGLi-expressing cells were suspended in lysis buffer [50 mM Tris-HCl (pH 8.0), 500 mM NaCl, 5 mM imidazole, 0.1% triton-X 100, 250 lM phenylmethanesulfonyl (PMSF)] and lysed by sonication (15 cycles of 60 s sonication with 60 s intervals). The clarified cell lysate was filtered on 0.22 lm membrane and purified by nickel-affinity chromatography on a HisTrap HP column (GE Healthcare) pre-equilibrated in buffer A [50 mM Tris-HCl (pH 8.0), 500 mM NaCl, 5 mM imidazole, 10 mM b-mercaptoethanol]. Oncolumn immobilised ARGLi was washed with 25 ml of buffer A containing 100 mM MnCl 2 for enzyme activation. Elution was carried out with a linear imidazole gradient ranging from 5 to 500 mM. Fractions containing purified ARGLi, identified by enzyme activity and SDS-PAGE, were pooled, incubated with His 6 -TEV protease for removal of the N-terminal His 6 tag, and dialysed against buffer [50 mM Tris-HCl (pH 7.4), 500 mM NaCl, 10 mM b-mercaptoethanol]. After complete cleavage, the His 6 tag and His 6 -TEV protease were separated from ARGLi by a second nickelaffinity chromatography step, from which ARGLi flowed through the column. Finally, ARGLi was subjected to a final dialysis in buffer [50 mM CHES (pH 9.5), 5 mM DTT, 100 mM NaCl, 250 lM PMSF]. ARGLi purity was determined by SDS-PAGE and the enzyme was stored at À80 C.

ARGLi activity measurements
For the biochemical characterization and inhibition experiments, 0.2 lg/mL ARGLi was incubated with 50 mM L-arginine in 50 mM CHES buffer (pH 9.5) at 37 C for 5 min. The amount of urea produced was determined by the UREA CE kit (Labtest), according to analytical procedures described by the manufacturer. Briefly, 10 lL of reaction mixture was incubated with 500 lL of urease solution (13 KU/L) diluted in buffer [20 mM sodium phosphate (pH 6.9), 62.4 mM sodium salicylate, 3 mM sodium nitroprusside] at 37 C for 5 min. Subsequently, 500 lL of oxidising solution [140 mM sodium hydroxide, 6 mM sodium hypochlorite] was added and the indophenol blue formed was measured at 600 nm. The concentration of product was calculated using a standard curve of urea. Enzymatic assays were performed in triplicate.

Determination of ARGLi optimal temperature and pH
To determine the influence of temperature on enzyme activity, 0.2 lg/mL ARGLi was incubated with 50 mM L-arginine in 50 mM CHES buffer (pH 9.5) for 5 min at different temperatures: 4 C, 25 C, 37 C, 45 C, 55 C, 65 C, 75 C, and 85 C. To determine the influence of pH, 0.2 lg/mL ARGLi was incubated with 50 mM L-arginine at 37 C for 5 min under different buffer conditions:100 mM MOPS at pHs 7.0, 7.5, 8.0, and 100 mM CHES at pHs 8.6, 9.0, 9.5, and 10.0.

Determination of ARGLi kinetic parameters
ARGLi (0.2 mg/mL) was incubated with L-arginine at concentrations ranging from 1 to 50 mM in 50 mM CHES buffer (pH 9.5) at 37 C for 5 min. V max and K m values were determined by the analysis of the initial velocity versus substrate concentration curve using the Michaelis-Menten equation in Graph Prism 6.

Comparative modelling of ARGLi
Leishmania infantum arginase amino acid sequence was obtained from UniProtKB (access code A0A145YEM9) 22 . The template structure was obtained using standard options of BLASTP 23 against the Protein Data Bank (PDB) 24 . Template selection considered the best results for identity, similarity, and gaps. Clustal-Omega (https:// www.ebi.ac.uk/Tools/msa/clustalo/) 25 was used for the alignment step and the three-dimensional model was built using MODELER 9 v19 26 . PROCHECK was used to evaluate the stereochemical quality of the final model 27 .

Molecular docking
In order to establish and validate a protocol for molecular docking, the structure of NOR-N-OMEGA-HYDROXY-L-ARGININE (NNH) was re-docked into the crystallographic complex structure of L. mexicana arginase (PDB: 4IU1) 28 . Molecular docking calculations were performed in AutoDock 4.1. Lamarckian Genetic Algorithm was used and parameters such as initial population, number of energy assessments, mutation rate, crossover rate, elitism, and number of runs were modified. The active site was defined using AutoGrid. The grid size was set to 40 Â 50 Â 38 points with a grid spacing of 0.375 Å centred on the Ne of His139 residue in the crystal structure. The docking of each molecule consisted in a total of 100 runs that were carried out with an initial population of 150 individuals, a maximum of 240,000 energy evaluations, a maximum of 25,000 generations, a mutation rate of 0.02, an etilism of 1, and a crossover rate of 0.8. For the docking, ligand was considered flexible, while protein was treated as a rigid structure. Visual analysis of the generated complexes was performed using AutoDockTools 1.5.6 29 and PyMOL (The PyMOL Molecular Graphics System, Version 1.4.1 Schr€ odinger, LLC.).

In silico ADMET studies
In silico ADMET studies were performed using ADMET 8.5 (Simulations Plus, Inc., Lancaster, CA, USA), in which the Lipinski's Rule of 5, toxicity evaluation, and ADMET risk were estimated for all substances exhibiting percentage of ARGLi inhibition higher than 50% at pH 7.4. As a control, the antileishmanial drug miltefosine was used.

Inhibition of L. infantum promastigotes growth
The parasite growth inhibition assays were performed on 96-well microplates, where the selected ARGLi inhibitors were previously diluted (1-200 lg/mL). Subsequently, promastigote forms of L. infantum collected in late logarithmic phase (96 h) were added to microplates at the final concentration of 10 6 parasites/mL. Microplates were incubated at 26 C for 96 h. After the incubation period, parasites viability was determined using 0.005% resazurin 30 . The 50% inhibitory concentration (IC 50 ) was determined from the dose-response curves generated from the data.
Cytotoxicity in RAW 264.7 macrophages RAW 264.7 macrophages were cultured in polystyrene culture flasks containing DMEM medium supplemented with 10% FBS at 37 C and 5% CO 2 atmosphere. For cytotoxic evaluation, 48 h cultured cells were harvested, washed with culture medium, and distributed into 96-well culture plates at 10 5 cells/well concentration. The cells were allowed to adhere for 2 h prior to the addition of increasing inhibitor concentrations (31-1000 mg/mL). Treated cell cultures were incubated for 48 h under the same experimental conditions. After this period, cell viability was determined by colorimetric assay using MTT solution (5 mg/mL) as an indicator. The 50% cytotoxic concentration (CC 50 ) was calculated by analysing the dose-response curves generated from the data.

Intracellular anti-amastigote activity and nitric oxide production by infected macrophages
The intracellular anti-amastigote activity was performed according to previously described procedures 31 with some modifications. Initially, RAW 264.7 macrophages were distributed in 96-well plates and, after adherence (incubation for 2 h), cells were washed twice with phosphate buffer saline (PBS, pH 7.2). Then, the adherent cells were infected with promastigote forms of L. infantum at the stationary phase of growth in a proportion of 10 parasites/ macrophage. After 4 h of interaction at 37 C and 5% CO 2 atmosphere, free parasites were washed away with PBS and infected macrophages were incubated for another 24 h to allow differentiation into amastigotes. After this time, infected cells were treated for 48 h with increasing concentrations of the selected inhibitor. Subsequently, the culture supernatant was collected for evaluation of nitric oxide production using the Griess reaction 32 . Cells were washed with PBS, then Schneider's medium supplemented with 5% FBS was added and the plate was incubated at 26 C for 72 h. The parasite survival was estimated by viability of differentiated promastigote forms recovered from the cultures using MTT (5 mg/mL). The results were expressed as the percentage of viability in relation to that obtained for the control (100%). The IC 50 for amastigote forms was calculated from the doseresponse curves generated from the data.

Selectivity index
The selectivity index (SI) for promastigotes and intracellular amastigotes of L. infantum was calculated by the ratio between the CC 50 obtained for the host cell and the parasite IC 50 . Inhibitors that showed SI !10 were considered low cytotoxic 33 .

Optimal parameters and kinetic properties of ARGLi
In order to investigate the inhibitory profile of ARGLi by natural substances, we first characterised its optimal parameters and kinetic properties. ARGLi showed the highest activity between 37 C and 55 C, with an activity drop at temperatures above 60 C (Figure 2(A)). Similar temperature conditions have been described for other isoforms of the enzyme 34 . We then selected the temperature of 37 C for all subsequent enzyme activity measurements.
ARGLi displayed a stringent pH dependency centred at pH 9.5 ( Figure 2(B)). A difference in ±0.5 pH units caused a sharp decrease in about 50% of enzyme activity. Similar results were found for arginase from other Leishmania species. Recombinant arginase from L. mexicana showed a broad pH optimum between 8.5 and 9.5 11 . Recombinant L. amazonensis arginase exhibited the highest activity at pH 9.6. A decrease in reaction pH to 7.0 led to an increase in enzyme K m accompanied by a decrease in V max 12 . Recombinant arginase from human liver and arginase isolated from erythrocytes also showed basic pH optima with values ranging from 9.7 to 11 34 . Thus, our results agree well with those previously reported, in which basic pH 9.5 or close is ideal for performing arginase activity measurements.
ARGLi exhibited classical Michaelis-Menten kinetics, a hyperbolic dependency of the reaction rate as a function of L-arginine concentration. Non-linear regression of the Michaelis-Menten plot enabled the determination of kinetic parameters, such as V max (0.28 ± 0.016 mM/min) and K m (5.1 ± 1.1 mM) (Figure 2(C)). ARGLi K m values are more similar to those described for human arginase I (K m ¼7.6 mM) 35 than for native and recombinant L. amazonensis arginase (K m ¼23.9 ± 0.96 and 21.5 ± 0.9, respectively) 12 , recombinant L. mexicana arginase (K m ¼25 ± 4 mM) 11 , and recombinant human arginase (K m ¼13 ± 2 mM) 11 . Our results suggest that ARGLi shows a stronger affinity for L-arginine than arginase from human and tegumentary Leishmania species. From the K m and V max values, we calculated a K cat of 2.55 Â 10 3 s À1 and a specificity constant (K cat /K m ) of 5 Â 10 8 M À1 s À1 . The high K cat and K cat /K m values suggest that ARGLi displays excellent catalytic efficiency, higher than that described for L. mexicana arginase (K cat ¼1.7 s À1 ) 11 .
Interestingly, many of the phenolic substances that exhibited the greatest inhibitory activity against ARGLi (quercetin, catechin, caffeic acid, rosmarinic acid, chlorogenic acid, and rhamnetin) share a common structural characteristic; they all contain a catechol group. The importance of the catechol group for efficient arginase inhibition has been described for other Leishmania species. Chlorogenic acid, (þ)-catechin, (À)-epicatechin, and isoquercitrin, isolated from the ethyl acetate fraction of Cecropia pachystachya leaf extract, inhibited more than 50% of L. amazonensis arginase activity at a final concentration of 20 lM 13 . In addition to these phenolics, orientin, isoorientin, fisetin, and luteolin showed inhibitory activity against L. amazonensis arginase with IC 50 values of 16 ± 2.0, 9.0 ± 1.0, 1.4 ± 0.3, and 9.0 ± 1.0 lM, respectively 15 . The high activity displayed by these substances was attributed to the presence of the catechol group, since other phenolics that do not contain a catechol group in their structures (apigenin, vitexin, and isovitexin) showed little effect. Comparative analysis of the structure and inhibitory activity of quercetin (IC 50 ¼4.3 lM) with galangin (IC 50 ¼100 lM) and kaempferol (IC 50 ¼50 lM) revealed the importance of the phenolic group hydroxylation for increased arginase inhibition 15 . In addition, by investigating the interaction of (þ)-catechin and (À)-epicatechin with L. amazonensis arginase in silico, dos Reis et al. showed that the catechol group makes hydrogen bonds with amino acids from the enzyme active site 14 . These phenolics also showed interaction with the arginase cofactor, acting as manganese chelators 12 . The relationship between the presence of the catechol group and the inhibitory activity was also observed against L. donovani, in which quercetin (IC 50 ¼1 lg/mL) was three times more active than kaempferol (IC 50 ¼2.9 lg/mL) 36 .
Caffeic acid and quercetin (100 lM) have been shown to inhibit bovine hepatic arginase with percentages of inhibition and IC 50 values of 61.3 ± 4.1% and 86.7 lM for caffeic acid, and 64.4 ± 2.5% and 31.2 lM for quercetin 37 . Concomitant inhibition of Leishmania and mammalian arginase may represent a potential therapeutic strategy, since inhibition of host arginase may increase NO production 38,39 .

Structural investigation of rosmarinic acid interaction with ARGLi
An experimentally-derived, atomic-resolution structure of L. infantum arginase has not been reported so far. Thus, we used comparative modelling to build a three-dimensional structural model of ARGLi, enabling us to investigate its interaction with inhibitors. The accuracy of protein structure models depends on the identity between the template and target sequences. We used the crystal structure of L. mexicana arginase in complex with the NNH inhibitor (PDB: 4IU1) 28 , which showed 96% of identity with the target sequence, as a template. The sequence alignment obtained from Clustal-Omega was provided as input in MODELLER to generate the three-dimensional model of ARGLi. The final model displayed 91.1% residues in the most favourable regions, 8.2% in allowed regions, 0.4% in generously allowed regions, and 0.4% in disallowed regions of the Ramachandran plot (Supplemental Figure  S1). As Lys9 was the only residue in the disallowed region of the Ramachandran plot and it is located far from the enzyme active site, the model was considered sufficient for further studies.
In order to gain insights into the mechanism of ARGLi inhibition, we investigated its interaction with rosmarinic acid by molecular docking. Initially, the docking accuracy was evaluated by redocking the NNH inhibitor into the crystal structure of L. mexicana arginase (PDB: 4IU1) 28 . The in silico analysis revealed a conformation similar to the crystal structure with a root mean square deviation (RMSD) between the top docking pose and the original crystallographic geometry of 0.4 Å (Supplemental Figure  S2). This result supported the hypothesis that the docking protocol was able to reproduce the experimental binding mode. Thus, the same docking procedure was employed for rosmarinic acid.
The ARGLi-rosmarinic acid complex showed an estimated binding affinity energy of À45 kcal/mol. Four hydrogen bonds were observed between the inhibitor and ARGLi. The hydroxyl oxygen atom at para position of the ligand ring B (catechol group) with the main chain nitrogen atoms of residues Ala141 (distance OÁÁÁN 2.2 Å) and Asp142 (distance OÁÁÁN 2.4 Å), and the side chain oxygen atom of residue Glu198 (distance OÁÁÁO 2.0 Å) and the main chain oxygen atom of residue Pro259 (distance OÁÁÁO 2.4 Å) with the hydroxyl oxygen atom at para position of the ligand ring A. In addition, cation-p interactions between the aromatic ring of the substance and residue His140 (distance 2.4 Å) were identified (Figure 4). It is worth noting that His140, Ala141, and Asp142 are conserved among arginases and are key residues for ligand interaction with L. mexicana arginase (His139, Ala140, and Asp141, respectively) 28,40 .
Finally, ADMET properties were estimated for caffeic acid, isorhamnetin, rhamnetin, rhaponticin, and rosmarinic acid, since Figure 3. Effect of phenolic substances of natural origin on ARGLi activity. ARGLicatalysed urea production was measured in the absence (control) and presence of fourteen natural phenolics at 100 lM concentration. Data are shown as percentage of ARGLi activity in relation to the control and represent the mean ± SE of three independent measurements. The dashed line highlights 100% ARGLi activity (control). Arrows indicate rosmarinic acid (71.48 ± 0.85%) and caffeic acid (56.98 ± 5.51%) as potent ARGLi inhibitors. these substances inhibited more than 50% of ARGLi activity. Our results suggested that these substances exhibit good oral bioavailability, based on the Lipinski rule of 5 41 . In the toxicity evaluation, these substances displayed low risk of cardiotoxicity. In contrast, except for rosmarinic acid, they showed a high risk of hepatotoxicity and mutagenicity. In addition, pharmacokinetic and toxicological parameters were compiled in ADMET Risk. The substances showed a value in the range of 2.01-5.68. ADMET risk provides a range between 0 and 24. These values indicate the number of potential ADMET risk factors that a compound might possess. Thus, we may infer that all substances exhibit low probability of having ADMET problems.

In vivo activity of ARGLi inhibitors
Substances that inhibited more than 50% of ARGLi activity were selected for the in vivo assays against L. infantum parasites. Except for rhaponticin, which did not inhibit promastigote growth at the highest concentration tested (Table 1), all other ARGLi inhibitors displayed antileishmanial activity. The phenylpropanoids rosmarinic acid and caffeic acid displayed the greatest effects against L. infantum with IC 50 values of 57.3 ± 2.65 lM (20.64 ± 0.96 mg/mL) and 60.8 ± 11 lM (10.97 ± 2.01 mg/mL) for promastigotes, and 7.9 ± 1.7 mM (2.86 ± 0.62 mg/mL) and 21.9 ± 5.0 mM (3.95 ± 0.91 mg/mL) for intracellular amastigotes, respectively (Table  1). Montrieux et al. previously described the leishmanicidal activity of rosmarinic acid and caffeic acid against the dermotropic species L. amazonensis in vitro and in vivo. These phenolic acids displayed IC 50 values of 0.2 ± 0.1 and 0.9 ± 0.2 mg/mL, respectively, against promastigote forms 42 . Caffeic acid showed an IC 50 of 2.9 ± 0.3 mg/mL for intracellular amastigotes of L. amazonensis and a CC 50 of 32.5 ± 2.0 mg/mL for BALB/c mice peritoneal macrophages. Comparatively, rosmarinic acid was more active and showed greater selectivity for L. amazonensis than caffeic acid, exhibiting an IC 50 of 1.7 ± 0.4 mg/mL for intracellular amastigotes and a CC 50 of 33.5 ± 1.4 mg/mL 42 . In addition, these phenolic acids proved to be more effective in reducing lesion size and parasite burden of L. amazonensis-infected BALB/c mice than the reference drug Glucantime V R . Moreover, treatment with these substances did not lead to mice death or significant loss of body weight 42 . Previous reports have demonstrated the anti-L. infantum activity of plant extracts enriched in rosmarinic and caffeic acids [43][44][45] . Here, we demonstrate the inhibitory effect of these phenylpropanoids alone against this viscerotropic species.
Rhamnetin and isorhamnetin, although inhibiting $56% of arginase activity, showed low activity against promastigote forms of L. infantum with IC 50 of 832 ± 6.9 lM (258.7 ± 9.5 lg/mL) and 818 ± 30 lM (263.2 ± 3.9 lg/mL), respectively. Despite the low activity, these substances did not exhibit toxicity for RAW 264.7 macrophages at the highest concentration tested (1000 lg/mL) and thus they are expected to be highly selective. Tasdemir et al. described better results for rhamnetin and isorhamnetin  using axenic amastigote forms of L. donovani 36 . Both isorhamnetin (IC 50 ¼3.8 lg/mL) and rhamnetin (IC 50 ¼4.6 lg/mL) showed greater selectivity for the parasite than for myoblast lineage cells (L9) with SI of 10.7 and >20, respectively. Catechin displayed a CC 50 of 885 ± 2.5 mM (256.9 ± 0.6 lg/mL) and IC 50 of 395 ± 50 mM (114.7 ± 14.6 lg/mL) for L. infantum promastigotes and 286.9 ± 36.5 mM (83.2 ± 10.6 lg/mL) for intracellular amastigotes, leading to SI values of 2.23 and 3.08, respectively. Extraction of Rhus punjabensis leaves in methanol/chloroform originated a catechin-rich extract that showed an IC 50 of 77.75 ± 1.23 lg/mL for promastigotes forms of L. tropica 46 . In addition, catechin isolated from the ethanolic extract of Stryphnodendron obovatum stem bark showed an IC 50 of 43.2 ± 2.1 lg/mL for L. amazonensis promastigotes and a SI of 2.7 47 . These results suggest that catechin displays higher activity against Leishmania spp. responsible for the tegumentary forms of the disease.
Despite presenting the lowest IC 50 for L. infantum amastigotes (0.191 ± 0.02 lM), the reference drug Fungizone V R showed a SI of 62.54, which is quite similar to those found for rosmarinic acid (61.79) and caffeic acid (55.69). Therefore, the pronounced antileishmanial activity of these substances in combination with their greater selectivity for the parasites evidence their potential use as promising candidates for less toxic and more effective drugs in VL treatment.
In order to determine if ARGLi inhibitors were capable of modulating the host cell response, RAW 264.7 macrophages were infected with promastigote forms of L. infantum and treated with different concentrations of inhibitors for 48 h. After treatment, NO production was measured 32 . Among the inhibitors tested (catechin, rosmarinic acid, and caffeic acid), only caffeic acid significantly increased NO production. Treatment with 100 and 200 mg/ mL caffeic acid induced an increase in nitrite levels of 234.6% (p > .01) and 311.5% (p > .001) in comparison to the control, respectively ( Figure 5(A)). This increase is similar to that observed for the reference drug Fungizone V R (Figure 5(B)). The effect of caffeic acid on NO production may be a result of macrophage arginase inhibition and consequent oxidation of accumulated Larginine by inducible NO synthase (iNOS). Caffeic acid has been shown to positively regulate iNOS expression and activity in a TNFa-dependent manner in L. major-infected BALB/c mice 48 . Remarkably, despite their anti-amastigote activity, catechin and rosmarinic acid did not affect NO production. These results suggest that catechin and rosmarinic acid trigger a NO-independent antileishmanial mechanism, which may include ARGLi inhibition in vivo.
Our results reinforce the importance of the catechol group for ARGLi inhibition. Catechol groups qualify as pan assay interference compounds (PAINS), due to their redox activity and reactivity against proteins 49 . Even though discriminating between false positives and true hits is a complex and difficult task, evidence of specific interactions between substance and target as well as a description of the mechanism of action validate compounds inhibitory activity. Previous results have shown that the catecholcontaining norathyriol is a site-specific inhibitor of mitogen-activated protein kinase 1 (ERK2), further supporting its potential activity 50 . Our in silico docking results revealed that rosmarinic acid makes a number of direct contacts with ARGLi, suggesting specific interactions. In addition, we conducted an in-depth in vivo inhibition study that showed that caffeic acid acts by increasing NO production by infected macrophages, even suggesting inhibition of host arginase.

Conclusion
We report on the purification, biochemical characterization, and inhibition of recombinant arginase from L. infantum (ARGLi). The enzyme showed a strong affinity for L-arginine and excellent catalytic efficiency. Among the phenolic substances tested, the phenylpropanoids rosmarinic acid and caffeic acid displayed potent inhibitory activity. Rosmarinic acid and caffeic acid were effective against L. infantum promastigote and intracellular amastigote forms and exhibited high selectivity. Moreover, caffeic acid led to an increase in NO production by infected macrophages. The results presented here further support arginase inhibition by naturally occurring phenolics as a promising strategy for VL treatment. It should be also noted that these natural products show effective inhibition of other enzymes such as for example the carbonic anhydrases 51-55 . As Leishmania spp. also encode for such an enzyme 56-60 , our findings may provide the rationale for using such multi-targeted derivatives for the management of this protozoan infection. , and catechin (dark gray). (B) Treatment with 0.062-0.5 lg/mL of the reference drug fungizone. Control (white) represents the production of nitric oxide by infected and untreated macrophages. Nitrite concentration was measured by the Griess reaction and data represent mean ± SE of two independent experiments. Asterisks indicate treatments that were significantly different compared to the control, in which ÃÃÃÃ p < .0001, ÃÃÃ p < .001, and ÃÃ p < .01.

Disclosure statement
No potential conflict of interest was reported by the authors.

Funding
This work was supported by grants from Fundação Carlos Chagas Filho de Amparo a Pesquisa do Estado do Rio de Janeiro (FAPERJ), Conselho Nacional Cientıfico e Tecnologico (CNPq) and PROEP/CNPq 407856/17, Coordenação de Aperfeiçoamento de Pessoal de N ıvel Superior (CAPES), and by a Brazil Initiative Collaboration Grant from Brown University. ARG is recipient of a Fundac¸ão Carlos Chagas Filho de Amparo a Pesquisa do Estado do Rio de Janeiro (FAPERJ) graduate fellowship.