Non-competitive inhibitor of nucleoside hydrolase from Leishmania donovani identified by fragment-based drug discovery

Abstract

Nucleoside hydrolase is an important target for the development of new leishmanicidal agents due to its role in parasite proliferation. Using the principles of fragment-based drug discovery, a library of 111 fragments was assembled for screening by nuclear magnetic resonance using the saturation transfer difference. Five fragments were selected as ligands of Leishmania donovani nucleoside hydrolase (NHLd) and kinetics studies revealed that fragment 3 acts as a new non-competitive inhibitor. Its binding mode was proposed on the basis of molecular docking, using a structural model of NHLd constructed by homology. An intermolecular interaction between fragments 3 and 5 was found by NOESY and docking studies. These data could be used to design more potent NHLd inhibitors and to aid the development of leishmanicidal drugs.

---

1. Introduction

Trypanosomatids represent a family of protozoa that are unable to perform purine synthesis, as other species of microorganisms can. The uptake of this auxotrophic nutrient is essential for the biosynthesis and proliferation of parasite DNA and RNA. This dependence has opened an avenue for the development of new antitrypanosomatid agents.^1–3

Enzymes involved in the uptake of purines in parasites include purine nucleoside kinases (HGPRTs) and purine ribohydrolases (i.e. nucleoside hydrolases, NHs). Among these enzymes, NHs are the most abundant. They have been identified and characterized in several microorganisms, including Trypanosoma brucei, T. cruzi, T. gondii, Leishmania donovani, L. mexicana, Entamoeba histolytica, and Escherichia coli. In these organisms, NHs are glycosidases which act by hydrolyzing the N-glycosidic bonds of β-ribonucleosides, forming nucleic ribose and free bases to be used in purine synthesis.^4–6 Since they are not found in mammals, NHs are considered important targets for the design of selective antiparasitic agents.^4,7–9

The NH of L. donovani (NHLd) is a protein with 314 amino acids that shows 96.2% similarity with the NH of L. major.⁵ Different ribohydrolase subtypes are found in trypanosomatids and they differ in their preference or specificity for inosine/uridine substrates. The NH subtypes are highly homologous in their catalytic site, and share similar mechanisms of catalysis.

The early stages of a drug discovery program generally focus on the identification and optimization of new ligands or leads that can promote the inhibition or activation of a protein or pathway, essential for the treatment or cure of a target disease.^10,11 Among the strategies to identify new ligands, fragment-based drug discovery (FBDD) has been spotlighted in recent years.¹² FBDD is based on the evaluation of the interaction of small molecules (e.g. molecular fragments) with macromolecules (enzymes or receptors), aiming to identify low and medium affinity ligands.

The application of the FBDD strategy is supported by the use of methodologies such as surface plasmon resonance,¹³ thermophoresis,¹⁴ mass spectrometry¹⁵ and nuclear magnetic resonance,¹⁶ which permit a rapid screening of thousands of compounds (i.e. fragments), and allow the measurement of their interaction with target proteins.^17–19

This paper describes the identification of a new non-competitive inhibitor of L. donovani nucleoside hydrolase, identified by the FBDD strategy, and reports its binding mode as proposed by docking studies.

2. Results and discussion

Construction of the fragment library

The molecular fragments were selected according to the Rule of Three,^20–22 and based on the structure of inosine, the natural substrate of the enzyme. Accordingly, the selected library was constituted by fragments able to mimic the interactions performed by nitrogenous base or ribose subunits, present in the inosine structure. As exemplified in Fig. 1, most chemical fragments selected in this study are in accordance with the Rule of Three, anticipating a broad scope for chemical transformation into compounds of greater structural complexity.

Fig. 1 Profile of fragment library. Molecular weight (MW); calculated partition coefficient (clog P); polar surface area (PSA); number of hydrogen bond donor atoms (NDHB); number of hydrogen bond acceptor atoms (NAHB); number of aromatic rings (NAR).

All 111 samples of fragments were prepared in DMSO-d₆ (1 mM) and analyzed by ¹H NMR. The assignment of all hydrogens was performed for each fragment. Based on the ¹H NMR chemical shift data, one spectral database of all fragments was generated. Fragments were separated into groups containing 3–5 compounds based on the absence of overlapping signals. The simultaneous saturation transfer difference (STD) analysis of groups of fragments in one NMR tube leads to a great economy of time and cost, with less consumption of the enzyme.

Ligand selection by STD NMR

For the interaction analyses, groups of fragments (1 mM) were prepared in 20 mM phosphate buffer pH 7.4, 300 mM NaCl, and 10% D₂O. The target protein NHLd was added for a final concentration of 100 μM.

Adapting the basic sequence described by Meyer and Meyer²³ and using the sequence DPFGSE²⁴ to suppress the water signal, parameters such as pulse power and saturation time were adjusted in the STD control spectra (in the absence of protein) to minimize peaks due to artifacts. By doing so, we ensured that all signals in the STD spectra in the presence of the protein corresponded to the hydrogens in the structure of the ligand involved in the interaction with the protein. These hydrogens are susceptible to interference from protein magnetization, induced by the NMR radio frequency pulse, due to interaction with the macromolecule, a process that allows the characterization of epitope groups.

Fig. 2 shows an example of the ligand selection process by STD NMR from a mixture of fragments. Based on the chemical shifts of the hydrogens of each fragment contained in the mixture, it was possible to identify the ligand fragment in the STD spectra (Fig. 2C), and to determine the epitope groups interacting with the target protein. STD NMR spectra in the absence of the protein were used as controls (Fig. 2B), avoiding the selection of false positives due to artifact peaks with the same chemical shift as the epitope groups. Fig. 2 illustrates how fragment 1 and fragment 2 were selected as ligands of NHLd and indicates their epitope groups. In order to characterize more precisely the signals of the epitope groups, the STD spectra were repeated separately for each active fragment, as shown in Fig. 3.

Fig. 2 Identification of ligands in a mixture of fragments by STD in phosphate buffer saline pH 7.4 with 10% D₂O. (A) ¹H NMR spectrum of the mixture of fragments; (B) STD spectrum of fragment mixture without protein (control); (C) STD spectrum of the fragment mixture with protein. Epitope groups are highlighted in red.

Fig. 3 Characterization of epitope groups of fragment 3 with NHLd protein. (A) ¹H NMR spectrum in phosphate buffered saline pH 7.4 with 10% D₂O and fragment 3 (1 mM); (B) STD NMR spectrum in phosphate buffered saline pH 7.4 with 10% D₂O and fragment 3 (1 mM) without protein; (C) STD NMR spectrum in phosphate buffered saline pH 7.4 with 10% D₂O and fragment 3 (1 mM) in the presence of the protein. The epitope groups are highlighted in red.

Using this methodology, it was possible to identify five fragments that interact with NHLd (Fig. 4). Knowing that the shorter the length of the connection between ligand and protein, the stronger the signal found in the STD spectrum, the intensity of this signal was normalized, adopting the value of 100% for the strongest signal. This procedure allowed us to map the chemical space of the hydrogens of the ligands participating in the interaction and to classify these epitopes quantitatively, in percentage terms.

Fig. 4 Mapping of epitope groups (in red) of the fragments selected by STD NMR study as ligands of NHLd and STD factor.

NHLd inhibition assays

Determination of fragments inhibitory activity on NHLd. To establish whether the fragments selected by STD were able to inhibit NHLd activity, we optimized the enzyme inhibition assay described by Rennó and collaborators²⁵ and used a fixed concentration of 1 mM fragment and 2 μg mL⁻¹ NHLd. The uric acid formed by catalysis of the substrate inosine (0.5 mM) after conversion of hypoxanthine by the auxiliary enzyme xanthine oxidase was measured. The percentages of inhibition obtained at the screening concentration (1 mM) are shown in ESI (Table S1†) for the five fragments selected by the STD methodology. Compounds that were able to inhibit the NHLd activity by a percentage ≥50% (fragments 1 and 3) were further analyzed by performing full concentration–effect curves in order to determine the IC₅₀ and K_i values. The results indicate that fragments 1 and 3 have significant potencies characterized by IC₅₀ values of 525 (K_i = 410) and 294 (K_i = 260) μM, respectively. These fragments could be used as starting points for further design of more potent NHLd inhibitors.

Kinetics studies. Drugs or compounds that act as enzyme inhibitors can interact in a reversible or irreversible manner with a specific conformational form of the target enzyme.²⁶ The reversibility of the inhibition produced by the most potent NHLd inhibitor (i.e. fragment 3) was studied for the two possible equilibrium complexes: enzyme–inhibitor complex (EI) and enzyme–substrate–inhibitor (ESI). After an equilibrium period favoring the formation of complex ESI or EI, a jump dilution was performed and the recovery of the enzymatic activity was evaluated for a period of 10 min and compared to the control in the absence of the inhibitor during the equilibrium period. The results are depicted in Fig. 5. Note that in both cases, the speed of the enzymatic reaction after dilution of the preformed complex with fragment 3 was similar to the rate of the control reaction, performed in exactly the same conditions but without the inhibitor, indicating a very fast recovery of the enzyme and the reversibility of enzyme inhibition.

Fig. 5 Recovery of NHLd activity after a “jump dilution” of the EI and ESI complexes. Fragment 3 (2.9 mM; 10 × [IC₅₀]) was present or not (control) during the pre-incubation period (20 min for EI complex and 2 min for ESI complex) preceding the dilution of the inhibitor (see Methods for details). In both cases, time zero indicates the starting time of the enzymatic reaction. For study of the EI complex, the reaction was started by addition of the substrate 20 min after pre-incubation of fragment 3 with the enzyme. For study of the ESI complex, the reaction was started by addition of the substrate to the medium containing fragment 3 and the enzyme. After 1 min (peak of absorbance in the control), the mixture was diluted with the same medium but without fragment 3, in order to promote a 10-fold dilution of the inhibitor. Note that the product of the reaction was also diluted, explaining the decrease in absorbance measured at 2 min (control: first measure after dilution).

In order to further characterize the mechanism of inhibition of NHLd by fragment 3, we determined which of the four classical models of inhibition better describes the data, obtained using six different concentrations of the substrate inosine, in the absence or in the presence of three different concentrations of fragment 3.²⁷ Visual analysis of the data exhibited in Fig. 6 shows a decrease of V_max in the presence of high concentrations of the inhibitor, so that a competitive mechanism of inhibition was discarded. We used a non-linear regression method considering all the data simultaneously in order to discriminate between the possible mechanisms and to allow a quantitative analysis of the different parameters. The model of non-competitive inhibition was better than the models of uncompetitive and mixed inhibition, as statistically determined by the Aikake criteria. The global fitting of the data using the model of non-competitive inhibition (Fig. 6) led to the determination of the following parameters: V_max (mM min⁻¹): 234 ± 0.5; K_m (mM): 0.166 ± 0.002 and K_i: 251 ± 1 μM. Note also that such methodology is free of the statistical bias inherent in linear transformations of the Michaelis–Menten equation such as in the Lineweaver–Burk plot.^27,28 Based on these results fragment 3 was identified as a new non-competitive inhibitor of NHLd.

Fig. 6 Concentration-effect curves for the activation of NHLd by inosine in the presence of fragment 3. The experiment was performed in the absence (control, ●) or presence of increasing concentrations of fragment 3 (■ 0.1 mM; ▲ 0.3 mM; ♦ 0.6 mM).

NOESY and molecular docking studies

To obtain evidence that could further help in the design of compounds with greater structural complexity and better affinity for NHLd, we considered the possibility of combining fragment 3, which displayed the best inhibitory activity (higher potency), with fragment 5, a polyhydroxyl ligand able to mimic the ribose subunit present in the inosine structure. Therefore, NOESY experiments were conducted to evaluate the spatial proximity between the selected fragments in the presence of NHLd.

In addition to the correlations for the spatial region of intramolecular interactions, two spatial regions of intermolecular interactions were also identified and are represented in ESI (Fig. S1–S3†). These results indicate that fragments 3 and 5 could interact in regions in close proximity, usually less than 5 Å, in the NHLd protein.

Following this, a molecular docking study was performed, aiming to establish the binding mode of these fragments with NHLd. A structural model of NHLd was constructed by homology using the SWISS-MODEL program.²⁹ Among the templates of nucleoside hydrolases (NH) containing inosine as substrate, as suggested by the server, the crystal structure of the NH of Escherichia coli (NHEc) was selected, due to its higher confidence range (0.97 between changes 1–0.09), and its high sequence similarity with NHLd (i.e. 89%).

The shape complementarity of the experimental structure and the NHLd model was verified through their superimposition in Pymol 1.5.0.5 and obtainment of an RMS value of 0.143. The Ramachandran³⁰ plot of the best model showed that 93.5% of the residues were in the most favored region, 2.6% were in additional allowed regions, 3.9% were in generously allowed regions and no residues were in disallowed regions (data not shown).

Based upon the NMR results, we used a model of the monomeric structure of NHLd to identify the molecular contact surface between NHLd and fragments 3 and 5. The molecular docking analyses were divided according to the two main purposes. First, to determine where fragment 3 interacts with the NHLd monomer, both in inosine-bound and apo form (Fig. 7). Second, to evaluate the interactions between fragment 5 and the fragment 3-NHLd complex both in inosine-bound and inosine-free forms (Fig. 8).

Fig. 7 Docking of fragment 3 into NHLd and NHLd-substrate complex. (A) Cartoon representation of NHLd including substrate and fragment 3 (stick representation). The backbone is colored with a rainbow scheme, ranging from blue (N terminus) to red (C terminus). Circle highlights the best binding mode of fragment 3. (B) The best docked conformation of fragment 3 in inosine-bound NHLd. (C) The best docked conformation of fragment 3 in free NHLd. Fragment 3 (Frag. 3) and residues of NHLd within 4 Å of fragment 3 are indicated by arrows and displayed using stick representation in B and C. Dashed lines represent hydrogen bonds.

Fig. 8 Docking of fragment 5 with NHLd. (A) Best pose obtained by docking onto NHLd/inosine/fragment-3 complex. (B) Best pose obtained by docking onto NHLd/fragment-3 complex. The enzyme in the background is shown as a surface map representation. Inosine and fragments 3 and 5 are indicated by arrows and displayed using stick representation. NHLd and fragments are colored by element (carbon – gray, hydrogen – white, nitrogen – blue, oxygen – red, sulfur – yellow).

The docking results obtained in the first step indicated that the intermolecular energy (ligand–protein interaction energy based on dispersion–repulsion, hydrogen bonding, electrostatics, desolvation, and electrostatics term) for the complex between NHLd and fragment 3 would range between −4.35 and −3.9 kcal mol⁻¹, and can be found in ESI (Table S2†). In about 90% of the poses, there were hydrogen bonds involving the side chain of residue Arg76 of NHLd, regardless of whether the enzyme was complexed or not with the inosine molecule (Fig. 7B and C). Furthermore, in both complexes, the residues that interacted with fragment 3 were similar (Arg76, Arg77, Ala78, Phe167), the only exception being Ser79, which made contact solely in the presence of inosine (Fig. 7B and C).

Since there is at least partial occupancy by fragment 3 in the active site gate (Fig. 8A and B), the pocket opening volumes for both apo NHLd and the fragment 3-bound form were calculated using the MOLE 2.0 server.³¹ The volume of the cavity for the apo form was 4003 ų and for the fragment 3-bound form was 3955 ų. The decrease in the volume suggests that fragment 3 blocks the pocket opening, preventing the substrate from entering.

Our NMR results indicate that fragments 3 and 5 could bind to NHLd close to each other in the active site. Furthermore, the docking studies suggest that whereas fragment 3 binds in the gate region, fragment 5 is able to establish interactions inside the active site. Based on these results, our research group is now working on the design of different connectors to link fragments 3 and 5, aiming at the discovery of new potent inhibitors of NHLd.

3. Experimental section

Materials

Reagents used in this work, such as inosine and purified xanthine oxidase (XO), were commercially obtained from Sigma-Aldrich®, and were of high purity. Fragments included in the library used here were selected from our laboratory or provided by our collaborators. GenScript® synthesized the plasmid used for NHLd expression.

Methods

Fragment library. According to the Rule of Three,²⁰ fragments should be structurally simple molecules, with few functional groups, must have a molecular weight less than 300, a number of donors and hydrogen bond acceptors not higher than three, and a clog P less than or equal to three.

In addition to these criteria, fragments with some structural similarity with the enzyme substrate inosine were preferentially selected. Fragments were selected for having functional groups that mimicked donor and acceptor groups of hydrogen bonds, aromatic subunits, and for being polyhydroxylated, mimicking nitrogenous bases and riboside present in the enzyme substrate.

Expression and purification of L. donovani nucleoside hydrolase (NHLd). The L. donovani nucleoside hydrolase gene (GenBank: AY007193.1) was cloned in pET-28b(+) by NcoI and XhoI to express the recombinant enzyme nucleoside hydrolase L. donovani (NHLd) containing a tail of six histidine residues at the C-terminus. The recombinant protein was expressed in E. coli BL21(DE3) cells and purified by chromatography using a nickel affinity column (Ni Sepharose 6 Fast Flow, GE Healthcare®), eluted with 20 mM phosphate buffer pH 7.4, 500 mM NaCl and 300 mM imidazole.

Imidazole was removed (<0.01 mM) by washing with 20 mM phosphate buffer at pH 7.4 and 300 mM NaCl in a Stirred Ultrafiltration Cell System model 8200 (Millipore) or Amicon® Ultra Centrifugal, with 3000 MWCO filters. Protein samples were concentrated and quantified by the Lowry method,³¹ before storage at −80 °C.

NMR analysis. All NMR spectra were acquired at 25 °C on a 500 VNMRS (Agilent) at 499.78 MHz for ¹H and processed with the MestReNova 9.0.1 program. STD spectra were acquired using the basic pulse sequence described by Mayer and Meyer,²³ with the inclusion of the double pulsed field gradient spin echo (DPFGSE)²⁴ for water suppression. Spectra subtraction was performed on alternate scans, producing the difference spectrum. Two seconds were used for saturation transfer with irradiation on-resonance at −2020.5 Hz and off-resonance at 15 790.9 Hz, and spectra were acquired with 1024 scans. 2D NOESY spectra were acquired with the DPFGSE pulse sequence for water suppression, 1 second mixing time, 64 scans, 1202 points in f2 and 128 in f1.

NMR samples. All 111 molecular fragments at 1 mM in DMSO-d₆ were first analyzed by ¹H NMR. Mixtures of fragments (3 to 5) were prepared at 1 mM in 20 mM phosphate buffer pH 7.4 containing 300 mM NaCl and 10% D₂O. ¹H NMR spectra were acquired for each fragment mixture sample. For STD binding assays, the protein, NHLd (100 μM), was added to the medium to obtain a molar ratio of 100 : 1 (fragment : protein).

Analysis of steady-state enzyme inhibition. Kinetics parameters of NHLd were determined by adapting the methodology described by Rennó and collaborators²⁵ to 96-well plates and measuring the enzyme kinetics using the spectrophotometric system SpectraMax M5 Molecular Devices® Microplate Reader. We used a coupled optical assay with the auxiliary enzyme xanthine oxidase (XO, 93 μg mL⁻¹) in 20 mM phosphate buffer pH 7.4 containing 300 mM NaCl and 0.5 mM inosine. The reaction was started by the addition of 2 μg mL⁻¹ NHLd. The product (uric acid) formation was monitored at 293 nm in a final volume of 300 μL. All experiments were performed in triplicate at room temperature. The kinetics parameters Michaelis–Menten constant (K_m) and maximum velocity (V_max) were evaluated by increasing the concentration of inosine from 0.1 to 1 mM, to obtain data above and below the K_m, while maintaining constant the concentration of NHLd (2 μg mL⁻¹). The saturation curve was obtained by plotting the values of initial rate versus substrate concentration and analyzed by nonlinear regression (GraphPad Prism 5.0 software) using the Michaelis–Menten equation: V₀ = V_max × [S]/([S] + K_m), where [S] is the substrate concentration, V_max is the maximum velocity and K_m is the Michaelis–Menten constant.

Fragments selected by STD were first screened at 1 mM for their inhibition of NHLd activity using the procedure described above with fixed concentrations of inosine (0.5 mM) and enzyme (2 μg mL⁻¹). Fragments that inhibited more than 50% of the enzyme activity were selected for IC₅₀ determination. IC₅₀ values were determined using at least 6 concentrations of inhibitor and fixed concentrations of enzyme (2 μg mL⁻¹) and inosine (0.5 mM). IC₅₀ values were obtained by nonlinear regression analysis (GraphPad Prism 5.0). K_i values were determined using the Cheng–Prussof equation: K_i = [IC₅₀/(1 + [S]/K_m)].²⁶ The reversibility of the enzymatic inhibition was analyzed for the complexes EI and ESI.²⁶

The complex EI was evaluated by pre-incubating the fragments at concentrations 10 times higher than their IC₅₀ together with 0.2 mg mL⁻¹ NHLd (100 times higher than the usual concentration for incubation) in phosphate buffer (20 mM pH 7.4, 300 mM NaCl). After 20 minutes, a “jump dilution” was performed and the reaction was started by adding the phosphate buffer containing XO (93 μg mL⁻¹) and inosine (0.3 mM) in order to promote a 100-fold dilution of both fragments and enzyme.

The complex ESI was evaluated by adding the usual concentrations of enzyme, inosine, XO, and the fragment (at a concentration equal to its IC₅₀). After the formation of complex ESI (∼1 min), the medium was diluted 10 times with the same solution but without the fragment, in order to promote a 10-fold dilution of the inhibitor. In both experiments, the reversibility of the inhibition was assessed by measuring the recovery of the enzymatic activity in comparison with control samples without inhibitor during the pre-incubation step.

The same experimental conditions were used to determine the type of enzymatic inhibition. In this case, five concentrations of inosine (0.03, 0.1, 0.3, 1, and 3 mM) were used in the absence (control) and presence of three fixed concentrations of the inhibitor (100, 300 and 600 μM), selected according to the IC₅₀. The complete data (from the four curves) were simultaneously analyzed by non-linear regression in order to compare different classical models of inhibition and to choose the best one using the Aikake criteria (GraphPad Prism version 5.0).

Molecular modeling. A three-dimensional model of NHLd was obtained based on the structure of the nucleoside hydrolase of E. coli (access code 3B9X, 2.3 Å resolution, Protein Data Bank – PDB),³² which has 89% sequence similarity³³ with NHLd. This structure was chosen because it was solved with the substrate inosine, differently from the model previously described by Tanos and collaborators³⁴. We used the T-Coffee server³⁵ to align the template and the NHLd. Then, the model building was performed by submitting the alignment to the SWISS-MODEL server.²⁹ Model validation using the Ramachandran plot was performed by the server Rampage.³⁰

Molecular docking. Fragments were drawn using the PC package Spartan'14 and their binding modes (ligand–target) were investigated by the docking methodology implemented in AutoDock4.2.³⁶ The graphical interface AutoDock Tools was used to prepare and analyze the docking.³⁶ The enzyme was kept fixed and the docking calculations were performed by using a grid map of 60 × 50 × 60 points centered on the calcium atom, with a grid spacing of 0.375 Å. Since the ligands were not peptides, the Gasteiger–Marsili charge was assigned³⁷ and then non-polar hydrogens were merged. Docking runs were performed using the Lamarckian genetic algorithm³⁸ and the remaining parameters were set as default values.

4. Conclusions

We performed a screening analysis of 111 fragments on NHLd using saturation transfer difference (STD) NMR spectroscopy. From this analysis, five fragments were selected as ligands of the target enzyme, and their ability to inhibit the enzymatic activity of NHLd was investigated. Fragments 1 and 3 stood out, showing IC₅₀ values of 525 μM and 294 μM, respectively. The mechanism of inhibition of fragment 3 was determined, revealing the identification of a new non-competitive inhibitor of NHLd. Its binding mode was proposed by molecular docking, using a structural model of NHLd constructed by homology. NOESY and comparative docking studies between fragments 3 and 5 were established and the results will be used in the design of more potent NHLd inhibitors.

Author contributions

Conceived and designed the experiments: MAM, LML, FN EJB, PGP, LWT. Performed the experiments: MAM, CN, MMM, ROS. Analyzed the data: MAM, FN, CMRS, LWT. Contributed reagents/materials/analysis tools: EJB, LML, PGP, PRRC, LWT. Wrote the paper: MAM, ROS, CMRS, FN, LML, LWT.

Conflict of interest

Authors report no conflicts of interest in this work.

Acknowledgements

Authors would like to thank CNPq (BR), CAPES (BR), FAPERJ (BR) and INCT-INOFAR (BR, 573.564/2008-6 and E-26/170.020/2008) for fellowship and financial support.

References

1. P. Y. Nicola, S. Carter, A. S. Cassandra, B. M. Jan and U. Buddy, Drug Targets in Kinetoplastid Parasites, Springer, New York, 2008, pp. 141–154.
2. M. Berg, P. V. D. Veken, A. Goeminne, A. Haemers and K. Augustyns, Curr. Med. Chem., 2010, 17, 2456–2481.
3. A. K. Datta, R. Datta and B. Sen, Drug Targets in Kinetoplastid Parasites, ed. H. K. Majumder, Springer New York, New York, NY, 2008, pp. 116–132.
4. M. Santana, G. P. Borja-Cabrera, E. Paraguai de Souza, N. R. Sturm, C. B. Palatnik de Sousa and D. A. Campbell, Mol. Biochem. Parasitol., 2002, 120, 315–319.
5. L. Cui, G. R. Rajasekariah and S. K. Martin, Gene, 2001, 280, 153–162.
6. W. Versées and J. Steyaert, Curr. Opin. Struct. Biol., 2003, 13, 731–738.
7. N. Carter, P. Yates, C. Arendt, J. Boitz and B. Ullman, Adv. Exp. Med. Biol., 2008, 625, 141–154.
8. D. Nico, D. C. Gomes, I. Palatnik-de-Sousa, A. Morrot, M. Palatnik and C. B. Palatnik-de-Sousa, Front. Immunol., 2014, 5, 273.
9. W. Barnali, G. Ambarnil, B. Dhananjay, G. Nanda and B. Rahul, BMC Res. Notes, 2014, 7, 802.
10. R. V. C. Guido, A. D. Andricopulo and G. Oliva, Estudos Avançados, 2010, 24, 81–98.
11. J. P. Hughes, S. Rees, S. B. Kalindjian and K. L. Philpott, Br. J. Pharmacol., 2011, 162, 1239–1249.
12. O. Rabal, M. Urbano-Cuadrado and J. Oyarzabal, Future Med. Chem., 2010, 3, 95–134.
13. S. Perspicace, D. Banner, J. R. Benz, F. Muller, D. Schlatter and W. Huber, J. Biomol. Screening, 2009, 14, 337–349.
14. S. A. I. Seidel, C. J. Wienken, S. Geissler, M. Jerabek-Willemsen, S. Duhr, A. Reiter, D. Trauner, D. Braun and P. Baaske, Angew. Chem., Int. Ed., 2012, 51, 10656–10659.
15. V. Vivat Hannah, C. Atmanene, D. Zeyer, A. Van Dorsselaer and S. Sanglier-Cianférani, Future Med. Chem., 2009, 2, 35–50.
16. M. J. Harner, A. O. Frank and S. W. Fesik, J. Biomol. NMR, 2013, 56, 65–75.
17. G. de Kloe, D. Bailey, R. Leurs and I. de Esch, Drug Discovery Today, 2009, 14, 630–646.
18. P. Hajduk and J. Greer, Nat. Rev. Drug Discovery, 2007, 6, 211–219.
19. S. J. Pfaff, M. S. Chimenti, M. J. S. Kelly and M. R. Arkin, in Protein-Protein Interactions: Methods and Applications, ed. L. C. Meyerkord and H. Fu, Springer New York, New York, NY, 2015,  DOI:10.1007/978-1-4939-2425-7_39, pp. 587–613.
20. M. Congreve, R. Carr, C. Murray and H. Jhoti, Drug Discovery Today, 2003, 8, 876–877.
21. H. Jhoti, G. Williams, D. C. Rees and C. W. Murray, Nat. Rev. Drug Discovery, 2013, 12, 644.
22. G. M. Keserű, D. A. Erlanson, G. G. Ferenczy, M. M. Hann, C. W. Murray and S. D. Pickett, J. Med. Chem., 2016 DOI:10.1021/acs.jmedchem.6b00197.
23. M. Mayer and B. Meyer, J. Am. Chem. Soc., 2001, 123, 6108–6117.
24. C. Dalvit, J. Biomol. NMR, 1998, 11, 437–444.
25. M. N. Rennó, T. C. C. França, D. Nico, C. B. Palatnik-de-Sousa, L. W. Tinoco and J. D. Figueroa-Villar, Eur. J. Med. Chem., 2012, 56, 301–307.
26. R. A. Copeland, Evaluation of Enzyme Inhibitors in Drug Discovery: A Guide for Medicinal Chemists and Pharmacologists, John Wiley & Sons, 2005.
27. F. Noel and J. Cumps, Braz. J. Med. Biol. Res., 1989, 22, 433–445.
28. H. J. Motulsky and L. A. Ransnas, FASEB J., 1987, 1, 365–374.
29. M. Biasini, S. Bienert, A. Waterhouse, K. Arnold, G. Studer, T. Schmidt, F. Kiefer, T. G. Cassarino, M. Bertoni, L. Bordoli and T. Schwede, Nucleic Acids Res., 2014, 42, W252–W258.
30. I. W. D. S. C. Lovell, W. B. Arendall III, P. I. W. de Bakker, J. M. Word, M. G. Prisant, J. S. Richardson and D. C. Richardson, Proteins: Struct., Funct., Genet., 2002, 50, 437–450.
31. D. Sehnal, R. S. Vareková, K. Berka, L. Pravda, V. Navrátilová, P. Banás, C.-M. Ionescu, M. Otyepka and J. Koca, J. Cheminf., 2013, 5, 39.
32. O. H. Lowry, N. J. Rosebrough, A. L. Farr and R. J. Randall, J. Biol. Chem., 1951, 193, 265–275.
33. B. G. E. Lovane, L. Muzzolini, V. Matafora, A. Fornili, C. Minici, F. Giannese and M. Degano, Biochemistry, 2008, 47, 4418–4426.
34. T. C. C. França, M. R. M. Rocha, B. M. Reboredo, M. N. Rennó, L. W. Tinoco and J. D. Figueroa-Villar, J. Braz. Chem. Soc., 2008, 19, 64–73.
35. D. G. H. C. Notredame and J. Heringa, J. Mol. Biol., 2000, 302, 205–217.
36. G. M. Morris, R. Huey, W. Lindstrom, M. F. Sanner, R. K. Belew, D. S. Goodsell and A. J. Olson, J. Comput. Chem., 2009, 30, 2785–2791.
37. J. Gasteiger and M. Marsili, Tetrahedron, 1980, 36, 3219–3228.
38. G. M. Morris, D. S. Goodsell, R. S. Halliday, R. Huey, W. E. Hart, R. K. Belew and A. J. Olson, Comput. Chem., 1998, 19, 1639–1662.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra15143d

---