INTRODUCTION
Among the vector-borne diseases caused by protozoa, leishmaniasis is especially widespread, since it is endemic in many countries of Latin America, in the Indian subcontinent, in East Africa and in the Mediterranean Basin (Sundar and Chatterjee, Reference Sundar and Chatterjee2006; WHO, 2010). Furthermore, travel to and from endemic areas has contributed to spreading the disease to previously safe countries (Amato et al. Reference Amato, Tuon, Siqueira, Nicodemo and Neto2007). Leishmaniasis may be quite diverse, because it is caused by many different parasitic species, transmitted by different vectors and animal reservoirs, and there are different clinical syndromes (Chappuis et al. Reference Chappuis, Sundar, Hailu, Ghalib, Rijal, Peeling, Alvar and Boelaert2007). Among the latter, cutaneous leishmaniasis (CL), which produces ulcers on exposed parts of the body, mucosal leishmaniasis (ML), characterized by non-self-healing destructive ulcerations of the mucosa, and visceral leishmaniasis (VL), presenting with anaemia, systemic infection, bleeding and, if untreated, usually leading to death, are the predominant clinical forms. Although CL is the most common form, ML and VL are the most severe, so we have selected representative species for both of them. Leishmania braziliensis is one of the most usual species that cause ML, and is mainly active in Brazil, Colombia, Venezuela and the Andes range (Guerra et al. Reference Guerra, Prestes, Silveira, Coelho, Gama, Moura, Amato, Barbosa and Ferreira2011). Leishmania infantum is the primary agent responsible for VL in Europe, North Africa and Latin America (Lukes et al. Reference Lukes, Mauricio, Schönian, Dujardin, Soteriadou, Dedet, Kuhis, Tintaya, Jirku, Chocholová, Haralambous, Pratlong, Obornik, Horák, Ayala and Miles2007).
The preferred treatment strategies against leishmaniasis include pentavalent antimonials and amphotericin B in most regions of the world, although the use of new compounds such as miltefosine or paromomycin for the treatment of ML, and allopurinol or itraconazole for VL, is slowly being introduced, with controversial results. However, adverse effects, relative efficacy and the elevated costs involved encourage the design of new effective drugs. Treatments of ML-affected individuals have to deal with widespread increasing drug resistance and moderate cure rates (Croft et al. Reference Croft, Sundar and Fairlamb2006; Woster, Reference Woster, Taylor and Triggle2007; Palumbo, Reference Palumbo2009).
Assuming the need for new drugs against leishmaniasis, different parasite targets have been proposed. One of the most striking characteristics of the Leishmania parasites consists of its ability to survive in the presence of reactive oxygen species generated by the macrophages in which they are hosted. Therefore, it is not surprising that the therapeutic targets most investigated are enzymes involved in antioxidant defence that are either exclusive to the parasite or present significant differences with respect to their homologues in humans (Van Assche et al. Reference Van Assche, Deschacht, Inocencio da Luz, Maes and Cos2011; Deschacht et al. Reference Deschacht, Van Assche, Hendrickx, Bult, Maes and Cos2012). Within this context, one enzyme that plays an essential role in the struggle of the parasite against oxidative stress is iron superoxide dismutase (Fe-SOD), which is present in all trypanosomatids (Miller, 2004), but not in humans, so that compounds able to selectively inhibit Fe-SOD are potential drugs for the treatment of leishmaniasis (Turrens, Reference Turrens2004).
In search of structures with the ability to interact with the active site of the enzyme, we focused our interest in macrocyclic polyamines containing highly complexing pyrazole units and presenting different protonation degrees at physiological pH (compounds 1–4, Fig. 1). Macrobicyclic polyamine 1 (Lamarque et al. Reference Lamarque, Navarro, Miranda, Arán, Ochoa, Escartí, García-España, Latorre, Luis and Miravet2001), macrocyclic polyamines 2 and 3 (Arán et al. Reference Arán, Kumar, Molina, Lamarque, Navarro, García-España, Ramírez, Luis and Escuder1999; Kumar et al. Reference Kumar, Bhalla, Kumar, Singh and Singh2001), with N-methyl- and N-benzyl-substituted pyrazole rings, and macrocyclic analogue 4 (Miranda et al. Reference Miranda, Escartí, Lamarque, Yunta, Navarro, García-España and Jimeno2004), with bulky octyl groups attached to the amine nitrogens on the side-chains had been previously tested by us against Trypanosoma cruzi (Sánchez-Moreno et al. Reference Sánchez-Moreno, Marín, Navarro, Lamarque, García-España, Miranda, Huertas, Olmo, Gómez-Contreras, Pitarch and Arrebola2012a ). In those assays it was found that compounds 1–3 were more active against the parasite and less toxic against Vero cells than the reference drug benznidazole (BZN), whereas the behaviour of 4 was similar to that of BZN or even worse. When inhibition of SOD enzymes by 1–4 was also tested, it was found that all four compounds selectively inhibited parasitic Fe-SOD in comparison with mammal CuZn-SOD, but the greatest selectivity and inhibitory activity were found with compounds 1 and 2, with the worst results again being those obtained with the bulky compound 4. On the basis of these data a potential relationship between the anti-parasitic activity observed and the ability of the compounds to interact with the enzyme was tentatively suggested.
Fig. 1. Pyrazole-containing polyamine macrocycles tested against Leishmania infantum and Leishmania braziliensis in the present study.
Taking into account the very close structural similarity of the Fe-SOD enzymes in T. cruzi and Leishmania spp., it was considered of interest to investigate the leishmanicidal activity of 1–4. In this work, their antiproliferative activity against both L. infantum and L. braziliensis and non-specific mammalian cytotoxicity were evaluated in vitro, and these measures were complemented by infectivity assays on macrophage cells. Furthermore, the treated parasites were submitted to a thorough study of the possible behaviour of the compounds assayed, as follows: (i) alterations caused in the cell ultrastructure of the parasites were recorded using transmission electronic microscopy (TEM); (ii) a 1H-NMR study concerning the nature and percentage of metabolite excretion was performed in order to obtain information on the effect of the polyamine macrocycles on the glycolytic pathway, since it represents the primary source of energy for the parasite; and (iii) inhibition of the parasitic Fe-SOD and human CuZn-SOD enzymes was tested and compared, and the mode of interaction with the enzyme's active site was simulated by molecular modelling.
MATERIALS AND METHODS
Chemistry
3,5-Dimethyl-1H-pyrazole, tris-2-(aminoethyl)amine, and 1,5-diamino-3-azapentane were purchased from commercial sources (Sigma-Aldrich) and used without further purification. 1,5-diamino-3-octylazapentane is not commercial and was obtained from 1,5-diamino-3-azapentane (Miranda et al. Reference Miranda, Escartí, Lamarque, Yunta, Navarro, García-España and Jimeno2004). 1H-pyrazole-3,5-dicarbaldehyde and 1-benzyl-1H-pyrazole-3,5-dicarbaldehyde were obtained from 3,5-dimethyl-1H-pyrazole according to a procedure previously described by the research group of one of the authors (Arán et al. Reference Arán, Kumar, Molina, Lamarque, Navarro, García-España, Ramírez, Luis and Escuder1999). 1-Methyl-1H-pyrazole-3,5-dicarbaldehyde was obtained from diethyl 1H-pyrazole-3,5-dicarboxylate as described by Kumar et al. (Reference Kumar, Bhalla, Kumar, Singh and Singh2001). The solvents were dried using standard techniques (Perrin et al. Reference Perrin, Armarego and Perrin1980). All reactions were monitored by thin layer chromatography (TLC) using DC-Alufolien Silica gel 60PF254 (Merck; layer thickness, 0·2 mm). The compounds were detected using UV light (254 nm), iodine or phosphomolybdic acid. Melting points were determined using a Reichert-Jung hot-stage microscope and are uncorrected. The 1H- and 13C-NMR spectra were recorded on Varian Inova-300, Varian Inova-400 and Varian Unity-500 spectrometers at room temperature employing CD3OD, CDCl3 or D2O as solvents. The chemical shifts were reported in parts per million (ppm) from tetramethylsilane (TMS), but were measured against the solvent signals. All assignments were made on the basis of 1H-13C heteronuclear multiple quantum coherence (gHMQC) and heteronuclear multiple bond coherence (gHMBC) experiments. The electrospray ionization (ESI) mass spectra (MS) were registered on a Hewlett-Packard 1100 MSD spectrometer in the electrospray positive mode. The fast atom bombardment (FAB) MS were obtained on a VG AutoSpec spectrometer using an m-nitrobenzyl alcohol matrix. Elemental analyses were performed by the Departamento de Analisis, Centro de Química Orgánica ‘Manuel Lora-Tamayo’, CSIC, Madrid, Spain. Elemental analysis was used to confirm a purity higher than 95% for all the biologically tested compounds.
Preparation of the pyrazole-containing bicyclic polyamine 1
1H-Pyrazole-3,5-dicarbaldehyde (Arán et al. Reference Arán, Kumar, Molina, Lamarque, Navarro, García-España, Ramírez, Luis and Escuder1999) (372 mg, 3 mmol), was dissolved in warm methanol (180 mL) and added dropwise under an argon atmosphere to a stirred solution of tris (2-aminoethyl)amine (292 mg, 2 mmol) in methanol (120 mL). After stirring overnight at room temperature, the intermediate Schiff base was reduced in situ with solid sodium borohydride (340 mg, 9 mmol) by stepwise addition. After 2 h, a standard work-up of the reaction mixture (Sánchez-Moreno et al. Reference Sánchez-Moreno, Marín, Navarro, Lamarque, García-España, Miranda, Huertas, Olmo, Gómez-Contreras, Pitarch and Arrebola2012a ) afforded 345 mg (59%) of 1,4,7,8,11,14,17,20,21,24,29,32,33,36-tetradecaazapentacyclo-[12.12.12.6,9.119,22.131,34]-hentetraconta-6,9(41),19(40),21,31,34(39)-hexene in the monohydrate form (1, melting point (mp) 244–245 °C (lit mp: 245 °C; Lamarque et al. Reference Lamarque, Navarro, Miranda, Arán, Ochoa, Escartí, García-España, Latorre, Luis and Miravet2001), MS (FAB): 569 (M+ +1).
Preparation of the pyrazole-containing monocyclic polyamine 2
A solution of 1-methyl-1H-pyrazole-3,5-dicarbaldehyde (Kumar et al. Reference Kumar, Bhalla, Kumar, Singh and Singh2001) (276 mg, 2 mmol) in MeCN (10 mL) was added, under argon, to a solution of freshly distilled 1,5-diamino-3-azapentane (206 mg, 2 mmol) in the same solvent (20 mL). After stirring overnight at room temperature, the intermediate Schiff base was reduced in situ by addition of sodium borohydride (304 mg, 8 mmol) in MeOH (40 mL). After 2 h, a standard work-up of the reaction mixture (Sánchez-Moreno et al. Reference Sánchez-Moreno, Marín, Navarro, Lamarque, García-España, Miranda, Huertas, Olmo, Gómez-Contreras, Pitarch and Arrebola2012a ) afforded 267 mg (64%) of 13,26-Dimethyl-3,6,9,12,13,16,19,22,25,26-decaazatricyclo-[22.2.1.111,14]-octacosa-1(27),11,14(28),24-tetraene (2, mp 134–135 °C) (lit.: thick oil; Kumar et al. Reference Kumar, Bhalla, Kumar, Singh and Singh2001), MS (ESI+, MeOH): 419 (M++1).
Preparation of the pyrazole-containing monocyclic polyamine 3
A solution of 1-benzyl-1H-pyrazole-3,5-dicarbaldehyde (Arán et al. Reference Arán, Kumar, Molina, Lamarque, Navarro, García-España, Ramírez, Luis and Escuder1999) (428 mg, 2 mmol) in MeCN (30 mL) was added under argon, to a solution of freshly distilled 1,5-diamino-3-azapentane (204 mg, 2·0 mmol) in the same solvent (30 mL). After stirring overnight at room temperature, the intermediate Schiff base separated as a white solid that was crystallized from EtOH (405 mg, 72%, mp 161–162 °C) (lit. 160–162 °C; Arán et al. Reference Arán, Kumar, Molina, Lamarque, Navarro, García-España, Ramírez, Luis and Escuder1999). A stirred suspension of the above Schiff base (394 mg, 0·7 mm) in EtOH (20 mL) was reduced by stepwise addition of solid sodium borohydride (303 mg, 8·0 mm). After stirring for 2 h at room temperature, a standard work-up of the reaction mixture (Sánchez-Moreno et al. Reference Sánchez-Moreno, Marín, Navarro, Lamarque, García-España, Miranda, Huertas, Olmo, Gómez-Contreras, Pitarch and Arrebola2012a ) afforded 299 mg (75%) of 13,26-Dibenzyl-3,6,9,12,13,16,19,22,25,26-decaazatricyclo-[22.2.1.111,14]-octacosa-1(27),11,14(28),24-tetraene (3, mp 137–138 °C) (lit. 136–138 °C; Arán et al. Reference Arán, Kumar, Molina, Lamarque, Navarro, García-España, Ramírez, Luis and Escuder1999). MS (FAB): 571 (M++1).
Preparation of the pyrazole-containing monocyclic polyamine 4
1H-Pyrazole-3,5-dicarbaldehyde (Arán et al. Reference Arán, Kumar, Molina, Lamarque, Navarro, García-España, Ramírez, Luis and Escuder1999) (248 mg, 2 mmol) was dissolved in hot MeOH (130 mL). This solution was then cooled to room temperature and added dropwise under argon to a stirred solution of 1,5-diamine-3-octyl-3-azapentane (Miranda et al. Reference Miranda, Escartí, Lamarque, Yunta, Navarro, García-España and Jimeno2004) (431 mg, 2 mm) in MeOH (220 mL). After stirring 12 h at room temperature, the intermediate Schiff base was reduced in situ with solid sodium borohydride (454 mg, 12 mmol) by stepwise addition. After 2 h, a standard work-up of the reaction mixture afforded the dihydrated monocyclic polyamine 6,19-Dioctyl-3,6,9,12,13,16,19,22,25,26-decaazatricyclo-[22.2.1.111,14]-octacosa-1(27),11,14(28),24-tetraene (4, 299 mg, 46%), as a colourless syrup (TLC: Rf = 0·54, MeOH 10:1). MS (ESI+, MeOH): 616 (M + +1).
Parasite strain and culture
Leishmania infantum (MCAN/ES/2001/UCM-10) and L. braziliensis (MHOM/BR/1975/M2904) were cultured in vitro in Medium Trypanosomes Liquid medium (MTL) together with 10% inactive fetal calf serum (FCS) kept in an air atmosphere at 28 °C, in Roux flasks (Corning, USA) with a surface area of 75 cm2, according to a methodology previously described (González et al. Reference González, Marín, Rodríguez-González, Hitos, Rosales, Reina, Díaz, González-Coloma and Sánchez-Moreno2005).
In vitro activity assays: extracellular forms
Promastigote assay
The compounds were dissolved in DMSO (Panreac, Barcelona, Spain) at a concentration of 0·1% and were assayed as non-toxic and without inhibitory effects on parasite growth, as previously described (González et al. Reference González, Marín, Rodríguez-González, Hitos, Rosales, Reina, Díaz, González-Coloma and Sánchez-Moreno2005). The compounds were dissolved in the culture medium at concentrations from 0·1 to 100 μ m. The effects of each compound against the promastigote forms and its concentrations were tested at 72 h using a Neubauer haemocytometric chamber. The leishmanicidal effect was expressed as the IC50 value, i.e. the concentration required to result in 50% inhibition, calculated by linear regression analysis from the K c values of the concentrations employed. Values are the means of three separate determinations.
Cell culture and cytotoxicity tests
The macrophage line J774.2 [European collection of cell cultures (ECACC) number 91051511] was derived in 1968 from a tumour in a female BALB/c mouse, and macrophages were grown in minimal essential medium (MEM) plus glutamine (2 mm) and 20% inactive FCS, with a humidified atmosphere of 95% air and 5% CO2 at 37 °C. The cytotoxicity test on macrophages was performed by flow cytometric analysis according to a method previously described (Marín et al. Reference Marín, Ramírez-Macías, López-Céspedes, Olmo, Villegas, Díaz, Rosales, Gutiérrez-Sánchez and Sánchez-Moreno2011). The percentage of viability was calculated with respect to the control culture. The IC50 was calculated using linear regression analysis from the K c values of the concentrations employed. Values are the means of three separate determinations.
Infectivity assay
J774.2 macrophage cells were grown as described above and seeded at a density of 1×104 cells well−1 in 24-well microplates (Nunc) with rounded coverslips on the bottom and cultured for 2 days. The cells were then infected in vitro with promastigote forms of L. infantum or L. braziliensis at a ratio of 10 : 1. The drugs (IC25 concentrations) were added immediately after infection and incubation was performed for 12 h at 37 °C in 5% CO2. Non-phagocytosed parasites and drugs were removed by washing, and then the infected cultures were grown for 10 days in fresh medium. Cultures were washed every 48 h and fresh culture medium was added. Drug activity was determined on the basis of both the percentage of infected cells and the number of amastigotes per infected cell in treated and untreated cultures in methanol-fixed and Giemsa-stained preparations. The percentage of infected cells and the mean number of amastigotes per infected cell were determined by analysing more than 200 host cells distributed in randomly chosen microscopic fields. Values are the means of three separate determinations.
In vitro activity assays: intracellular forms
Adherent macrophage cells were infected with promastigotes in the stationary growth phase of Leishmania spp. at a ratio of 10 : 1 and maintained for 24 h at 37 °C in air containing 5% CO2. Non-phagocytosed parasites were removed by washing, and the infected cultures were incubated with the testing compounds (concentrations from 0·1 to 100 μ m) and then cultured for 72 h in MEM plus glutamine (2 mm) and 20% inactive FCS. Compound activity was determined from the percentage reductions in amastigote number in treated vs untreated cultures in methanol-fixed and Giemsa-stained preparations. Values are the means of three separate determinations (González et al. Reference González, Marín, Rodríguez-González, Hitos, Rosales, Reina, Díaz, González-Coloma and Sánchez-Moreno2005).
Ultrastructural alterations
The parasites were cultured at a density of 5×105 cells mL−1 in the corresponding medium, each of which contained the compounds being tested at the IC25 concentration. After 72 h, the cultures were centrifuged at 400 g for 10 min, and the pellets produced were washed in PBS and then incubated with 2% (v/v) formaldehyde/glutaraldehyde in 0·05 M cacodylate buffer (pH 7·4) for 24 h at 4 °C. The pellets were then prepared for TEM (Zeiss model, Barcelona, Spain) using a technique previously described (González et al. Reference González, Marín, Rodríguez-González, Hitos, Rosales, Reina, Díaz, González-Coloma and Sánchez-Moreno2005).
Metabolite excretion
Cultures of L. infantum and L. braziliensis promastigotes (initial concentration 5×105 cells mL−1) received the IC25 of the compounds (except for control cultures). After incubation for 72 h at 28 °C the cells were centrifuged at 400 g for 10 min. The supernatants were collected to determine the excreted metabolites by 1H-NMR, and chemical shifts were expressed in ppm, using sodium 2,2-dimethyl-2-silapentane-5-sulphonate as the reference signal. The chemical displacements used to identify the respective metabolites were consistent with those described previously by some of the authors (Fernández-Becerra et al. Reference Fernández-Becerra, Sánchez-Moreno, Osuna and Opperdoes1997).
Fe-SOD enzymatic inhibition
Parasite cells were collected at the logarithmic growth phase by centrifugation (400 g for 10 min at room temperature) and then the pellet of cells (0·5–0·6 g wet weight mL−1) was suspended in 25 mL of MTL not enriched with fetal bovine serum (FBS) and cultured at 26 °C for 24 h.
After that, the culture of promastigotes was centrifuged (400 g for 10 min) and the supernatant filtered (Minisart®, 0·2 μm). The filtered supernatant was subjected to ice-cold ammonium sulphate precipitation between 35–85% salt concentration and the resulting precipitate was dissolved in 2·5 mL of distilled water and de-salted by chromatography in Sephadex G-25 column (GE Healthcare Life Sciences®, PD 10 column), previously equilibrated with 25 mL of distilled water, bringing it to a final volume of 3·5 mL (Longoni et al. Reference Longoni, Sánchez-Moreno, Rivera-López and Marín2013). The protein content were quantified using the Sigma Bradford test, which uses BSA as a standard (no trace ability was certified for the BSA standard, Bradford, Reference Bradford1976).
Iron and copper–zinc superoxide dismutases (Fe-SOD and CuZn-SOD) activities were determined using a method previously described (Beyer and Fridovich, Reference Beyer and Fridovich1987) which measures the reduction in nitroblue tetrazolium (NBT) by superoxide ions. Into each bucket, 845 μL of stock solution [3 mL of L-metionine (300 mg, 10 mL−l), 2 mL of NBT (1·41 mg, 10 mL−1) and 1·5 mL of Triton X-100 1% (v/v)] were added, along with 30 μL of the parasite homogenate fraction, 10 μL of riboflavine (0·44 mg, 10 mL−l), and an equivalent volume of the different concentrations of the compounds tested. Five different concentrations were used for each product, from 1 to 100 μ m. In the control experiment the volume was made up to 1000 μL with 50 mm potassium phosphate buffer (pH 7·8, 3 mL), and 30 μL of the parasite homogenate fraction were added to the mixtures containing the compounds. Then, the absorbance (A0) was measured at 560 nm in a spectrophotometer. Following this, each bucket was illuminated with UV light for 10 min under constant stirring and the absorbance (A1) was again measured. Human CuZn-SOD and substrates used in these assays were obtained from Sigma Chemical Co. Resulting data were analysed using the Newman–Keuls test.
Molecular modelling
Molecular modelling studies were carried out using the AMBER (Case et al. Reference Case, Cheatham, Darden, Gohlke, Luo, Merz, Onufriev, Simmerling, Wang and Woods2005) method implemented in the HyperChem 8.0 package (Hypercube Inc., FL, USA), modified by the inclusion of appropriate parameters (Miranda et al. Reference Miranda, Escartí, Lamarque, Yunta, Navarro, García-España and Jimeno2004). Starting structures for compounds 1–4 were built using HyperChem's capabilities. Their geometry was minimized to a maximum energy gradient of 0·1 kcal/(Å mol) with the AMBER force field, using the Polak-Ribière (conjugate gradient) minimizer, and a ‘simulated annealing’ procedure was used to cover all conformational space. The most stable extended geometry was always used in all calculations of interaction with the enzyme. To mimic the conditions used in the activity measurements, i.e. water as solvent, all calculations were carried out in vacuo with a distance-dependent dielectric constant value. In the absence of explicit solvent molecules, a distance-dependent dielectric factor qualitatively simulates the presence of water, as it takes into account the fact that the intermolecular electrostatic interactions should die off more rapidly with distance than in the gas phase (Reviriego et al. Reference Reviriego, Navarro, García-España, Abelda, Frías, Domenech, Yunta, Costa and Ortí2008). The same results can be obtained using a constant dielectric factor greater than 1. We chose to use a distance-dependent dielectric constant (ε = 4 Rij) as this was the method used to develop the AMBER force field (Cornell et al. Reference Cornell, Cieplak, Bayly, Gould, Merz, Ferguson, Spellmeyer, Fox, Caldwell and Kollman1995). Charge assignations for all atoms were done by means of ab initio calculations using the STO-3G basis set, as it is compatible with the AMBER force field, prior to energy minimization using AMBER. The Fe-SOD enzyme structure was obtained from the Brookhaven protein data bank (2gpc entry) and its energy minimized in the same way. Interaction studies were done starting from structures having the compound positioned in the border of the enzyme cavity. To study different possible interactions of the enzyme's iron atom with any of the nitrogen atoms, entry into the cavity was forced using a restraint on the selected N-Fe distance, starting from 0·15 nm and slowly decreasing this distance, and allowing the complex to achieve the minimum energy conformation with no restraints for all the small driving steps, using the same conditions mentioned above.
RESULTS AND DISCUSSION
We report now on the results obtained concerning the toxic activity of the selected macrocyclic polyamines against two species of Leishmania (L. infantum and L. braziliensis).
In vitro antileishmanial evaluation
We measured the in vitro biological activity of compounds 1–4 on both extra- and intracellular forms of the parasites. Extracellular forms are more commonly used due to ease of working with them, but are less indicative of leishmanicidal activity. Use of intracellular forms is more cumbersome but gives more accurate results, as they are converted to amastigotes in vertebrate host cells (Sánchez-Moreno et al. Reference Sánchez-Moreno, Marín, Navarro, Lamarque, García-España, Miranda, Huertas, Olmo, Gómez-Contreras, Pitarch and Arrebola2012a ). Intracellular assays were performed by infecting macrophage cells with promastigotes, which transformed into amastigotes within 1 day after infection. Tables 1 and 2 show the IC50 values obtained after 72 h of exposure for both L. infantum and L. braziliensis. Toxicity values against J774.2 macrophage after 72 h of culture were also calculated, and the selectivity indices (SI = IC50 macrophages toxicity/IC50 activity of extracellular or intracellular forms of the parasite) are shown in the last two columns of Tables 1 and 2. Results obtained for the reference drug glucantime were included for comparison.
Table 1. In vitro activity, toxicity and selectivity index for the macrocyclic polyamines 1–4 on extra- and intracellular forms of L. infantum
Results are averages of three separate determinations.
a IC50 is the concentration required to give 50% inhibition, calculated by linear regression analysis from the K c values at the concentrations employed (1, 10, 25, 50 and 100 μ m).
b Against J774.2 macrophages after 72 h of culture.
c Selectivity index = IC50 macrophages toxicity/IC50 activity on extracellular or intracellular forms of the parasite. In parentheses: number of times the compound SI exceeded the reference drug SI.
Table 2. In vitro activity, toxicity and selectivity index for the macrocyclic polyamines 1–4 on extra- and intracellular forms of L. braziliensis
Results are averages of three separate determinations.
a IC50 = the concentration required to give 50% inhibition, calculated by linear regression analysis from the K c values at the concentrations employed (1, 10, 25, 50 and 100 μ m).
b Against J774·2 macrophages after 72 h of culture.
c Selectivity index = IC50 macrophages toxicity/IC50 activity on extracellular or intracellular forms of the parasite. In parentheses: number of times the compound SI exceeded the reference drug SI.
It was shown that the leishmanicidal activities against both the extra- and intracellular forms of the parasites by the bicyclic polyamine 1 and the N-pyrazole methylsubstituted monocyclic polyamine 2 were, in all cases, higher than those seen with glucantime, while more random values were found for the N-benzylsubstituted analogue 3 and the monocyclic polyamine octylsubstituted on the side-chain nitrogen 4. However, more interesting are the toxicity data in mammalian cells, since all four compounds tested were found to be much less toxic for macrophages than the reference drug. Thus, compounds 1 and 2 were 12-fold and 14-fold respectively less toxic than glucantime, and even the most toxic among them, 4, was 3-fold more benign. Toxicity values substantially influence the more informative selectivity index (SI) data, so the best values were again obtained for compounds 1 and 2. In tests performed on L. infantum species, SI exceeded that of the reference drug by 32- and 25-fold for the extra- and intracellular forms in the case of 1, and by 23- and 20-fold with 2, while on L. braziliensis, the respective values obtained were 22- and 31-fold with 1 and 37- and 75-fold with 2. It should be noted that a SI value exceeding 20-fold that of the reference drug is one of the usual basic criteria for considering primarily screened compounds as candidates for more advanced testing in vitro and in animal models (Nwaka et al. Reference Nwaka, Besson, Ramirez, Maes, Matheeussen, Bickle, Mansour, Yousif, Townson, Gokool, Cho-Ngwa, Samje, Misra-Bhattacharya, Murthy, Fakorede, Paris, Yeates, Ridley, Van Voorhis and Geary2011). On the contrary, the N-octylsubstituted compound 4 was the least efficient (2·5- and 3-fold on L. infantum and 2- and 10-fold on L. braziliensis), whereas 3 gave intermediate values between the 1/2 pair and 4. Leishmania braziliensis showed a higher susceptibility for 2, since its SI was 75-fold that of glucantime in the more indicative intracellular amastigote test, a much better result than that found for L. infantum.
The tests described above represent only a first crude approach to the leishmanicidal properties of the compounds assayed. In order to gain better insight into the activity of compounds 1–3, their effect on the infectivity and intracellular replication of amastigotes was subsequently determined. In accordance with the usual working procedure, polyamine 4 was not included in this second stage, due to the poor SI results obtained. Macrophage cells were grown and infected with promastigotes in the stationary phase. The parasites invaded the cells and underwent morphological conversion to amastigotes within 1 day after infection. On day 10, the rate of host cell infection reached its maximum (control experiment). We used the IC25 of each product as the test dosage.
As shown in Fig. 2a, when macrocycles 1–3 were added to macrophages infected with L. infantum promastigotes, the infection rate decreased significantly with respect to the control and, furthermore, the three compounds were also more effective in decreasing infectivity than glucantime (72, 64 and 47% for 1, 2 and 3 respectively, vs only 21% for the reference drug). The bicyclic polyamine 1 and the monocyclic N-methylsubstituted pyrazole polyamine 2 were clearly more efficient than the N-benzylsubstituted compound 3. A measure of the average number of amastigotes per infected macrophage (Fig. 2b) led to similar conclusions: all three compounds were more effective than glucantime (with only a 23% decrease), and their order of effectiveness followed the same pattern as that seen in the infectivity measures, although the differences between them were less pronounced (60, 52 and 42%) for 1, 2 and 3, respectively).
Fig. 2. Effect of pyrazole-containing polyamine macrocycles 1–3 on the infection and growth rates of Leishmania spp.: (a): rate of infection of L. infantum, (b): mean number of amastigotes of L. infantum per infected J774 A.2 macrophage cells (at IC25 conc). Values are means of three separate experiments.
The same experiment was performed with L. braziliensis, and the results obtained concerning infection rates (a) and amastigote numbers (b) can be observed in Fig. 3. In both cases, the three compounds were also more effective than glucantime, and also in both cases the order of effectiveness was now 2>1>3, since the infectivity rates calculated from Fig. 3 were: 64, 58, 47 and 22%; and the decrease in amastigote numbers was: 70, 65, 55 and 28% for 2, 1, 3 and glucantime respectively.
Fig. 3. Effect of pyrazole-containing polyamine macrocycles 1–3 on the infection and growth rates of Leishmania spp.: (a): rate of infection of L. braziliensis, (b): mean number of amastigotes of L. braziliensis per infected J774 A.2 macrophage cells (at IC25 conc). Values are means of three separate experiments.
If we now compare the results obtained for both Leishmania species, it can be concluded that, in accordance with the SI values displayed in Tables 1 and 2, the N-methylsubstituted polyamine 2 is more efficient in L. braziliensis with respect to L. infantum, and the N-benzylsubstituted polyamine 3 is the least active in both species.
The leishmanicidal activity shown by compounds 1–3 should lead to relevant damage to parasite cells. Therefore, we performed a transmission electron microscopy (TEM) study on the promastigote forms of L. infantum and L. braziliensis treated with 1–3. As expected, significant morphological alterations were observed compared with untreated control cells. Figure 4 displays structural features observed in control and treated cells of both Leishmania species. Treatment with 1 and 2 led to remarkable morphological alterations involving deep vacuolization and loss of cytoplasmic content (Fig. 4, plates b and c in both charts). In most cases, much of the cytoplasm was filled with huge empty or granulated vacuoles, some of them electron-dense, that were occupying nearly all the cytoplasmic space of the promastigotes, and generated very swollen cytoplasmic forms. In fact, the usual cytoplasmic organelles could hardly be found. Some parasites were dead, or appeared as very swollen or were ready to burst. Parasites treated with compound 3 presented scarcer alterations (Fig. 4, plate d), and many of them coalesced to rosettes, although some were also vacuolated, presented small electron-dense vesicles or showed modified kinetoplasts. On the basis of these data, macrocyclic polyamines 1 and 2 were confirmed to be especially harmful to both L. infantum and L. braziliensis species.
Fig. 4. Chart 1: Ultrastructural alterations observed by TEM in promastigotes of L. infantum treated with compounds 1–3. (a) Control parasite (scale bar: 1·00 μm). (b) Comp 1 (scale bar: 1·00 μm), (c) Comp 2, (scale bar: 1·59 μm), (d) Comp 3 (scale bar: 2·33 μm). Promastigotes after treatment with the respective compounds. Abbreviations: dead parasites (D), flagellum (F), glycosomes (G), kinetoplast (K), mitochondrion (M), nucleus (N), vacuoles (V), electron dense vesicles (VE), star-shaped promastigotes (single arrow). Chart 2: Ultrastructural alterations observed by TEM in promastigotes of L. braziliensis treated with compounds 1–3. (a) Control parasite (scale bar: 1·00 μm). (b) Comp 1 (scale bar: 1·00 μm), (c) Comp 2, (scale bar: 1·59 μm), (d) Comp 3 (scale bar: 1·00 μm). Promastigotes after treatment with the respective compounds. Abbreviations: flagellum (F), glycosomes (G), kinetoplast (K), mitochondrion (M), nucleus (N), swollen promastigotes (S), vacuoles (V), electron dense vesicles (VE).
Metabolite excretion
Since trypanosomatids are unable to completely degrade glucose to CO2, they excrete a considerable portion of its hexose skeleton as partially oxidized fragments in the form of fermented metabolites, whose nature and percentage depend on the pathway used for glucose metabolism (Ginger, Reference Ginger2005; Bringaud et al. Reference Bringaud, Riviere and Coustou2006). The catabolism products in Leishmania are usually CO2, succinate, acetate, pyruvate, D-lactate, L-alanine, and, to a lesser extent, ethanol (Cazzulo, Reference Cazzulo1992). Detection of large amounts of succinate as a major end product is a usual feature, because it rules on glycosomal redox balance, enabling reoxidation of NADH produced in the glycolytic pathway. Succinic fermentation requires only half of the phosphoenolpyruvate produced to maintain the NAD+/NADH balance, and the remaining pyruvate is converted inside the mitochondrion and the cytosol into acetate, L-lactate, L-alanine or ethanol according to the degradation pathway followed by each species (Michels et al. Reference Michels, Bringaud, Herman and Hannaert2006).
In order to obtain some information concerning the effect of the tested compounds on glucose metabolism in the parasites, we registered the 1H-NMR spectra of promastigotes from L. infantum and L. braziliensis after treatment with compounds 1–3, and the final excretion products were identified qualitatively and quantitatively. The results were compared with those found with promastigotes maintained in a cell-free medium (control) for 3 days after inoculation with the parasite. The characteristic presence of acetate, D-lactate, succinate, L-alanine and ethanol was confirmed in the control experiments performed on both species. As expected, succinate and acetate were the most abundant end products identified. However, after treatment of the parasites with 1–3, the excretion of catabolites was substantially altered at the dosages employed. Table 3 displays these modifications with respect to the control observed at the height of the spectral peaks corresponding to the most representative final excretion products. Marked differences in the catabolic pathway appeared, and that seemed to be connected with the leishmanicidal activity commented on above. The main features found for both species were substantial increases in the formation of pyruvate and glycine with the most active macrocyclic polyamines 1 and 2, accompanied by corresponding decreases in succinate and acetate and, to a lesser extent, D-lactate. In L. infantum, the most active compound 1 increased pyruvate and glycine by 100%, and triggered widespread decreases in succinate, acetate, D-lactate and L-alanine by 100, 80, 37 and 46% respectively. Similar results were obtained with compound 2, although with slightly smaller variations. In contrast, the less active compound 3 did not increase pyruvate nor glycine at all (they were not detected), but slightly increased succinate and acetate by 39 and 20% respectively, in sharp contrast with 1 and 2. If we turn now to the results obtained with the L. braziliensis species, compounds 1 and 2 also substantially increased the formation of pyruvate (73 and 100% respectively) and glycine (100 and 66%), and decreased the production of succinate (100% in both cases), acetate, D-lactate and L-alanine, whereas with the less active compound 3 pyruvate and glycine were not detected and succinate, acetate, L-lactate and L-alanine did not experience any modifications with respect to the control. All these data could be interpreted on the basis of a change in the succinate and acetate pathways occurring in the presence of the most active compounds 1 and 2. It is well known that acetate, D-lactate, L-alanine and ethanol originate from the transformation of PEP in pyruvate in the presence of pyruvate kinase or pyruvate phosphate dikinase (Bringaud et al. Reference Bringaud, Riviere and Coustou2006). Therefore, it seems possible that compounds 1 and 2 were interacting with the pyruvate kinase enzymes and modifying the glucose metabolism of the parasite at the pyruvate stage. On the other hand, it is interesting to note that the decrease in succinate with compounds 1 and 2 is accompanied by the appearance of glycine, which is not present in the normal catabolism of the parasite. Those catabolic changes could also be related to a malfunction of the mitochondria, due to the redox stress produced by inhibition of the mitochondrion-resident Fe-SOD enzyme (Kirkinezos and Moraes, Reference Kirkinezos and Moraes2001), which should result in decrease of pyruvate metabolism and a consequent decrease of the succinate and acetate produced in mitochondria. These data appear to confirm that the severe modifications generated in organelles such as glycosomes or mitochondria by 1 and 2 are the ultimate reasons for the alterations observed in the excretion products of both species.
Table 3. Variation in the height of peaks corresponding to catabolites excreted by L. infantum and L. braziliensis promastigote forms in the presence of macrocyclic polyamines 1–3 with respect to the control test a
a Eth, ethanol; Lac, D-lactate; Ala, L-alanine; A, acetate; S, succinate; Pyr, pyruvate; Gly, glycine; (–), decreasing peak; (+), increasing peak; (=), no difference detected; (ND), peak undetected.
SOD enzymatic inhibition in the parasites and in human erythrocytes
Since we had found that the anti-T. cruzi activity of compounds 1–4 was accompanied by a remarkable inhibitory activity on the essential antioxidant enzyme Fe-SOD (Sánchez-Moreno et al. Reference Sánchez-Moreno, Marín, Navarro, Lamarque, García-España, Miranda, Huertas, Olmo, Gómez-Contreras, Pitarch and Arrebola2012a , Reference Sánchez-Moreno, Gómez-Contreras, Navarro, Marín, Olmo, Yunta, Sanz, Rosales, Cano and Campayo b ), we tested the effect of 1–4 on Fe-SOD isolated from L. infantum and L. braziliensis over a range of concentrations from 1 to 100 μ m. We used promastigote forms of both species, which excreted Fe-SOD when cultured in a medium lacking inactive FBS (Longoni et al. Reference Longoni, Sánchez-Moreno, Rivera-López and Marín2013). The inhibition data obtained are displayed in Fig. 5, parts a and b, and the corresponding IC50 values are included in order to make interpretation of the results easier. For comparison, part c of Fig. 5 shows the effect of the same compounds on CuZn-SOD obtained from human erythrocytes. The most remarkable result was the significant inhibition of Fe-SOD activity found for 1–3, whereas their inhibition of human CuZn-SOD was clearly lower. If we consider the IC50 calculated for L. infantum, inhibition of Fe-SOD by compounds 1–3 was respectively 4·6-, 6·2- and 2·8-fold higher than inhibition of CuZn-SOD, and respective values of 5·7-, 13·1- and 5·2-fold higher were obtained in the case of L. braziliensis. On the contrary, the N-octyl derivative 4 was not selective at all, and resulted in even less inhibition of Fe-SOD than of the CuZn enzyme, in concordance with the poor in vitro activity results noted before. It can also be concluded that compound 2 is a more efficient Fe-SOD inhibitor in L. braziliensis than in L. infantum (IC50 5·6 and 11·7 μ m, respectively). It is also interesting to note that the Fe-SOD inhibition rankings are 2>1>3>>>4, suggesting a relationship between the inhibitory ability and the hindrance caused by the substituents in the neighbourhood of the macrocycle side-chain nitrogens, since the methyl groups in 2 are less bulky than the benzyl groups of 3, and much less bulky than the octyl groups attached to 4. Compound 1, as an unsubstituted bicyclic macrocycle, should presumably approach the active site core of the enzyme with more difficulty than the monocyclic compound 2. That could be in accordance with some kind of interaction of the macrocycle nitrogens with the active sites of the enzyme in both Leishmania species. As has been pointed out, the modification of Fe-SOD activity by a particular compound should take place by modifying the coordination geometry of the iron atoms, reorganizing the active sites H-bonding network (Yikilmaz et al. Reference Yikilmaz, Rodgers and Miller2006) or reducing the cooperative interaction between the two monomers of the enzyme (Muñoz et al. Reference Muñoz, Morán, Becana and Montoya2005).
Fig. 5. In vitro inhibition (%) of Fe-SOD of: (a) Leishmania infantum and (b) Leishmania braziliensis promastigotes by compounds 1–4. (c) In vitro inhibition of CuZn-SOD in human erythrocytes by compounds 1–4. Values are the average of three separate determinations. Differences between the activities of the control homogenate and those incubated with compounds 1–4 were obtained according to the Newman–Keuls test. aIC50 = the concentration required to give 50% inhibition was calculated by linear regression analysis from the K c values at the concentrations employed (1, 10, 25, 50 and 100 μ m).
In order to obtain more information, molecular modelling calculations on the mode of interaction of the tested compounds with the enzyme were performed. Fe-SOD enzymes are dimeric forms in which either of the two monomers contains an iron atom. While the CuZn-SOD enzyme of mammals has the metal active sites located deeper inside the monomers, the active sites of the parasite Fe-SOD are closer to the interface between the two subunits, so that the channel opening in the inner part of the dimer is the obvious place to locate a molecule in search of interaction with the iron environment (Abreu and Cabelli, Reference Abreu and Cabelli2010). Therefore, the macrocycles were positioned inside the cavity. However, there are three different types of nitrogen atoms able to interact: that of the pyrazole ring and those located in the β and ε positions of the connecting side chains. Table 4 displays the Milliken net charges obtained for the different nitrogens by means of ab initio calculations using the STO-3G basis set (Leite, Reference Leite1998). It was shown that in all four compounds the pyrazole nitrogens exhibited the lowest basicity whereas the β−nitrogens were the most basic in 1 and 4, but not in 2 and 3. Consequently, the approach mode of each molecule to the active centre should take place through the polyamine side chains, instead of the pyrazole units. On that basis, and using the AMBER force field implemented in HyperChem, we applied a methodology previously developed by us (Sánchez-Moreno et al. Reference Sánchez-Moreno, Gómez-Contreras, Navarro, Marín, Olmo, Yunta, Sanz, Rosales, Cano and Campayo2012b ) to the structure of a Fe-SOD enzyme model obtained from the Brookhaven Protein Data Bank 2gpc entry. First, the most favoured distance between the iron atom of one subunit and the three different nitrogens under consideration was calculated for compounds 1–4 (Table 5). In all four cases the pyrazole nitrogens were located further from the metal than those at the side-chains, in accordance with the charge data detailed above. On the other hand, all the nitrogen atoms of the bulky N-octyl substituted macrocycle 4 stood too far from the metal environment to effectively modify its geometry, and this fact should be connected with the very poor inhibitory activity of compound 4 against Fe-SOD measured experimentally. The unsubstituted bicyclic polyamine 1 showed the closest approach to the iron atom (0·361 nm), which took place through the N β nitrogen, and the Npyrazole substituted compounds 2 and 3 afforded their best results through their N ε nitrogens. These distances are too large to enable a direct Fe·····N interaction, but they are suitable to modify the arrangement of the aminoacids involved in the active site core. From these data, it seems that the side-chain nitrogens are better able to affect the iron atom environment, and that the presence of substituents on the macrocycle influences its interaction with that environment.
Table 4. Ab initio calculated charges on the nitrogen atoms of compounds 1–4
Table 5. Calculated distances between the iron atom in the active site of the Fe-SOD enzyme and the different nitrogen atoms of compounds 1–4
A further step should involve consideration of the effect caused by the tested compounds on the hydrogen bonding network wrapping the metal atom. In Fe-SOD, the metal appears as a five-coordinated active site ligated by three histidine units, an aspartate group and an axial H2O or HO− ligand supported by an essential hydrogen bonding network comprising diverse amino acids (Miller et al. Reference Miller, Sorkin and Padmakumar2005). Specifically, the coordinated molecule engages in a H-bond with Asp156, and another H-bond engages with Gln69 (Yikilmaz et al. Reference Yikilmaz, Rodgers and Miller2006). The H2O/HO− ligand and its hydrogen-bonding partners are essential for tuning the antioxidant activity of Fe-SOD (Stallings et al. Reference Stallings, Metzger, Pattridge, Fee and Ludwig1991; Yikilmaz et al. Reference Yikilmaz, Xie, Brunold and Miller2002), since it acts as a proton donor/acceptor and facilitates the release of peroxide generated by the reduction of the substrate (Han et al. Reference Han, Lovell and Noodleman2002). On that basis, we modelled the active site moiety of the Fe-SOD enzyme as shown in Fig. 6a, in which the hydrogen bonding system is displayed as dashed lines, the H2O ligand appears in the foreground and the key interactions with Gln69 and Asp160 are clearly seen.
Fig. 6. (a) Molecular model of the free Fe-SOD enzyme active site with its supporting network of hydrogen bonds displayed as dashed lines. (b) and (c) Molecular models of the Fe-SOD enzyme active site with the Npyrazole methyl substituted macrocycle 2 (b) and the Npyrazole benzyl substituted macrocycle 3 (c) embedded in the proximity of the iron atom, and the modified hydrogen-bonding networks.
The best Fe-SOD inhibitor in both L. infantum and L. braziliensis, 2, was selected for estimating its effect on the active site environment. The approach of 2 to the active site confirmed that the most basic β-nitrogen pointed towards the metal ion in the most favoured disposition (Fig. 6b). As a consequence, the amino acids surrounding the iron atom were displaced apart from their initial positions, the hydrogen bonding pattern appeared completely distorted, and Gln69 was located too far from the HO− group to form a stable hydrogen bond. Instead, the β-nitrogen of the macrocycle was linked through two hydrogen bonds to the imidazole nitrogen of the His160 unit and to one oxygen atom of the Asp156 molecule, two amino acids involved in the coordination sphere of the iron atom. We think that these results could explain in part the enzyme deactivation found experimentally. As a term of comparison, the same treatment was given to the poorly inhibitory compound 3, in which the only difference is replacement of the pyrazole methyl group for the more bulky benzyl substituent. The most favoured disposition is shown in Fig. 6c. It can be seen that, although the H-bonding net is also distorted and Gln69 is moved away from its usual location, the drug molecule is placed further away from the metal environment than in the case of 2 and the two hydrogen bonds that link compound 2 with the coordination sphere are missing. Therefore, steric hindrance originating from the macrocycle's substituents should affect its ability to inhibit the Fe-SOD enzyme. Those catabolic changes could be also related to a malfunction of the mitochondria, due to the redox stress produced by inhibition of the mitochondrion-resident Fe-SOD enzyme (Kirkinezos and Moraes, Reference Kirkinezos and Moraes2001), which should result in decrease of pyruvate metabolism and a consequent decrease of the succinate and acetate produced in mitochondria.
From all the results detailed above, we conclude that the macrocyclic polyamines 1 and 2 show interesting in vitro leishmanicidal activity which seems to be related to their inhibitory ability against the Fe-SOD of the parasite. Furthermore, as discussed above, redox stress derived from Fe-SOD inhibition could also explain the catabolic changes observed. On that basis, we think that they fulfil the requirements needed to justify a more detailed investigation into the nature of the mechanisms involved in its activity patterns, and furthermore, that they could be considered as candidates for study of antiparasitic activity at a higher level.
ACKNOWLEDGEMENTS
The authors thank the Departamento de Analisis of the Centro Nacional de Química Orgánica Manuel Lora-Tamayo (C.S.I.C.) for the analysis performed concerning the purity of the compounds tested. We are also grateful to the transmission electron microscopy and nuclear magnetic resonance spectroscopy services of the CIC-University of Granada for their contribution to the studies on ultrastructural alterations and catabolism.
FINANCIAL SUPPORT
Financial support from the Spanish Ministerio de Ciencia e Innovacion (CTQ2009-14288-C04-01 and Consolider Ingenio CSD2010-00065), and from the Santander-Universidad Complutense Research Programme (GR35/10-A-921371) is also acknowledged.
