Venom alkaloids against Chagas disease parasite: search for effective therapies

Chagas disease is an important disease affecting millions of patients in the New World and is caused by a protozoan transmitted by haematophagous kissing bugs. It can be treated with drugs during the early acute phase; however, effective therapy against the chronic form of Chagas disease has yet to be discovered and developed. We herein tested the activity of solenopsin alkaloids extracted from two species of fire ants against the protozoan parasite Trypanosoma cruzi, the aetiologic agent of Chagas disease. Although IC50 determinations showed that solenopsins are more toxic to the parasite than benznidazole, the drug of choice for Chagas disease treatment, the ant alkaloids presented a lower selectivity index. As a result of exposure to the alkaloids, the parasites became swollen and rounded in shape, with hypertrophied contractile vacuoles and intense cytoplasmic vacuolization, possibly resulting in osmotic stress; no accumulation of multiple kinetoplasts and/or nuclei was detected. Overexpressing phosphatidylinositol 3-kinase—an enzyme essential for osmoregulation that is a known target of solenopsins in mammalian cells—did not prevent swelling and vacuolization, nor did it counteract the toxic effects of alkaloids on the parasites. Additional experimental results suggested that solenopsins induced a type of autophagic and programmed cell death in T. cruzi. Solenopsins also reduced the intracellular proliferation of T. cruzi amastigotes in infected macrophages in a concentration-dependent manner and demonstrated activity against Trypanosoma brucei rhodesiense bloodstream forms, which is another important aetiological kinetoplastid parasite. The results suggest the potential of solenopsins as novel natural drugs against neglected parasitic diseases caused by kinetoplastids.


Results
The solenopsins isolated from the venom of the fire ants S. invicta and S. saevissima. After extraction and fractionation, the composition of the solenopsins from the venom of S. invicta and S. saevissima was assessed by gas chromatography (GC-MS). Total ion chromatograms (Fig. S1) illustrate the diversity of solenopsin analogues found in the venom of each fire ant species, and their chemical structures and distribution (compounds I-VI) are presented in Fig. 1 and Table 1, respectively. By averaging the approximate composition analyses of Table 1, we estimated the approximate molecular masses of 288 g mol −1 and 253 g mol −1 for solenopsin extracts from S. invicta and S. saevissima, respectively.

Solenopsins inhibit proliferation but not alter epimastigotes of T. cruzi.
To quantify the toxic effect of solenopsins on the parasites, we tested their effects against the proliferation of T. cruzi epimastigote forms of two different strains: Dm-28c and CL-Brener. As summarised in Table 2 (also Fig. S2A), after 48 h of incubation, solenopsins presented IC 50 values against Dm-28c strain epimastigotes of 0.87 μM for S. invicta and 0.64 μM for S. saevissima, which are considerably lower than the IC 50 values of the usual treatment drugs miltefosine (3.15 μM) 28 (Fig. 1, compound VIII) and benznidazole (36.80 μM) 29,30 (Fig. 1, compound VII). A similar range of IC 50 values were also obtained against CL-Brener epimastigotes cultured in the presence of solenopsins of S. invicta (0.73 μM) and solenopsins of S. saevissima (0.58 μM; see Table 2; Fig. S2B). Parasites treated with either solenopsin extracts within the range of 0.25-0.5 × IC 50 values for up to 8 days later recovered culture growth capacity when the solenopsins were removed (Fig. S2C), indicating that the replication inhibition induced by solenopsins is reversible.
To quantify the possible effect of solenopsins on the cell cycle, Giemsa-stained epimastigotes were surveyed for the development of multiple nuclei (n) and kinetoplasts (k) over time. A total of 500 epimastigotes were evaluated by light microscopy over 7 days of culture (Fig. 2). As epimastigote forms cannot be cultured in precise synchrony, an arbitrary zero time-point was set when > 95% cells presented 1k and 1n. Following a stabilisation period, observed cultures underwent division cycles until no observable significant difference existed between treated and untreated parasites (days 1-3, Fig. 2A, B). Following 5-7 days of culture, there was still no observable difference between controls and treated cells in terms of 1k/1n, 2k/1n or 2k/2n proportions (Fig. 2C, D). Fluorescent intercalator displacement assays (Fig. S3) indicated that solenopsins are unable to intercalate in DNA, unlike berberine and emetine 31,32 alkaloids, indicating that the activity to epimastigotes is likely not linked to direct DNA interaction.
Solenopsins affect the morphology and long chain polyphosphate levels of T. cruzi epimastigotes. The inhibition of T. cruzi epimastigote replication by solenopsins was followed by a cumulative concentration-dependent increase in aberrant rounded morphology (Fig. 3). While untreated cells maintained normal, fusiform shape (Fig. 3A), the treated epimastigotes showed an increase in the density of cytoplasmic vacuoles associated with altered rounded shapes, typical of cells undergoing osmotic stress, autophagy, and/or apoptosis (Fig. 3B, C). No significant increase in bi-or multi-flagellate forms was observed.
It is known that polyphosphate (polyP) concentration changes drastically during the lifecycle of T. cruzi, especially when parasites are exposed to osmotic or alkaline stress 33,34 . Therefore, we compared polyP levels among treated and untreated parasites (Fig. 4) and observed that solenopsin-treated parasites consistently showed increased long-chain polyP concentrations but not short-chain polyP concentrations (Fig. 4A, B). www.nature.com/scientificreports/ It has been reported that the enzyme T. cruzi TcVps34-a homologue of phosphatidylinositol 3-kinase (PI3K)-is actively involved in the recovery of parasites submitted to hypo-osmotic stress 35 and participates in the regulation of autophagy 36 . However, it has been previously shown that solenopsins suppress PI3K activation 21 . We therefore hypothesised that solenopsins might similarly induce osmotic stress and/or autophagy in epimastigotes by inhibiting TcVps34. To test this hypothesis, we compared the growth curves of wild-type and CL-Brener epimastigotes overexpressing TcVps34 (TcPI3K) and exposed to different solenopsin concentrations. TcPI3K overexpression did not affect the IC 50 of solenopsins from S. invicta (0.56 μM) and S. saevissima (0.57 μM; Table 2; Fig. S1B). TcPI3K overexpression typically leads to enlarged contractile vacuoles and hypertrophic alterations near the cytostome and flagellar pocket of epimastigotes 35 (see Fig. 3D). Nonetheless, TcPI3K overexpression Figure 1. Chemical structure of all tested compounds used in the present study. Chemical structure of the solenopsins (I-VI), benznidazole (VII) and miltefosine (VIII). Additional information about the solenopsin alkaloids I-VI and their relative abundance in the venom of Solenopsis invicta and S. saevissima can be found in Table 1.  Solenopsins induce autophagy and programmed incidental cell death in T. cruzi epimastigotes. Solenopsin-treated epimastigotes presented increased numbers of cytoplasmic vacuoles and cytosolic concentric double-membrane inclusions symptomatic of autophagy and apoptosis (Fig. 5). We searched for evidence of autophagy by incubating cells with monodansylcadaverine (MDC), a marker of autophagic vacuoles in vivo 37 (Fig. 6). Epimastigotes show more intensive MDC staining when incubated under low-nutrient conditions (Fig. 6B, E) than under control conditions (Fig. 6A, E); similar intense MCD labelling was observed in solenopsin-treated parasites ( Fig. 6C-E).
The capacity of solenopsins to induce programmed cell death was also analysed by measuring DNA nicks and fragmentation (Fig. 7). Relative to untreated controls (Fig. 7A) and cells necrotized with Triton X-100 (Fig. 7B), solenopsin-treated epimastigotes stained positively for DNA 3′-OH free ends (Fig. 7D); the difference was statistically significant (Fig. 7E). Taken together, these results strongly suggest that solenopsins induce autophagy and incidental cell death in epimastigote forms of T. cruzi 38,39 . Solenopsins are trypanocidal to peritoneal T. cruzi-infected macrophages and toxic to T. brucei rhodesiense. The activity of the solenopsins was assessed against cultured mammalian BMDM, CHO, and LLC-MK 2 cell lines using 3-(4,5-dimethyl-2-thiazolyl)-2,5-diphenyl-2H-tetrazoliumbromide (MTT) 40 and lactate dehydrogenase (LDH) 41 assays. Solenopsins proved toxic to CHO cells at concentrations ≥ 7.5 μM ( Fig. S4A-B) based on MTT and LDH assay methods. Solenopsins were more toxic to primary bone marrowderived macrophages (BMDMs) than to other cell lineages, possibly as a result of the immortalized phenotype of lineages (Fig. S4C). The effects of solenopsins on T. cruzi-infected murine peritoneal macrophages were evaluated in vitro under similar settings and quantified as the infectivity index (i.e., number of infected macrophages x number of intracellular amastigotes/total number of macrophages) following staining and counting (Fig. 8). Solenopsins reduced the number of infected macrophages and the number of amastigotes inside infected macrophages (Fig. 8A) in a concentration-dependent manner, yielding an estimated IC 50 value of 2.59 µM for S. invicta solenopsins and 2.47 μM for S. saevissima solenopsins (Table 2), which was significantly lower than the IC 50 of benznidazole (Fig. 8B) with a value of 7.30 μM (Table 2). However, based on the toxic effects observed against BMDM cells, the selective index values of the solenopsin alkaloids for intracellular amastigote forms (2-3.5) were significantly lower than those observed for the reference compounds benznidazole (112.7) or miltefosine (93.5) ( Table 2). Solenopsins also proved toxic against T. brucei rhodesiense ( Figure S5), the aetiological agent of human sleeping sickness. Solenopsins from S. invicta inhibited the growth of T. brucei rhodesiense (Fig. S5) bloodstream forms in a concentration-dependent manner, yielding an estimated IC 50 value of 0.42 µM ( Table 2).

Discussion
Arthropod venoms are a rich, essentially untapped source for bioactive molecules. Among venomous insects, ants are remarkably chemically diverse 42 , as illustrated by a recent increase in the number of studies surveying the biomedical applications of ant toxins, such as the Brazilian giant ant Dinoponera quadriceps 43 . Among several other biomedical activities, the venom peptides of D. quadriceps were reported to be toxic to T. cruzi 44 . Notwithstanding, the venom alkaloids of ants have remained untested against trypanosomatids. www.nature.com/scientificreports/ The fire ant alkaloids known as solenopsins abound in the venom of these ants following typical speciesspecific configurations 45,46 . The two fire ant species selected for this study represent extremes in solenopsin chemical diversity; S. invicta venom includes almost all known solenopsin analogues ( Fig. 1, compounds II-VI), and the cryptic species of S. saevissima abounds in analogues of solenopsin A 45 (Fig. 1, compounds I-II), which is to date, the best-studied solenopsin alkaloid 18 . Despite conspicuous differences in the solenopsin analogue proportions isolated from both species (Table 1), similar IC 50 growth inhibition values were obtained against the T. cruzi epimastigote and amastigote forms ( Table 2). This contrasts with the fact that different solenopsins exhibit a diverse cytotoxic effect on different bacterial species 9,18 , suggesting that different mixtures of solenopsin analogues have similar effects against T. cruzi. A number of piperidine alkaloids can pass the blood brain barrier, such as nicotine 4 . Such properties, if present in solenopsins, could prove useful against some protozoan infections, e.g., in the brain, such as sleeping sickness caused by Trypanosoma brucei species 47 , and where treatment is impaired by a low tissue penetration by current medications, e.g., in leishmaniasis caused by Leishmania species 48 or in Chagas disease caused by T. cruzi 49 . www.nature.com/scientificreports/ The negative effects of solenopsins on the proliferation of T. cruzi epimastigote forms were reversible, similar to previous reports on other alkaloids [50][51][52][53][54] . Additionally, the observation that epimastigotes become rounded with increased intracellular vacuolization and complex membrane invaginations has been reported for most studies involving the treatment of T. cruzi with alkaloids. A remarkable difference herein, however, was the absence of a clear swelling of the mitochondrial matrix, as observed with camptothecin 48,50 or piperine 52 . Additionally, solenopsins delayed cell cycle progression but did not block the synthesis or segregation of nuclear and kinetoplast DNA, as reported with Taxol 50 or colchicine 54 . Finally, unlike the effects caused by vinblastine and vincristine that induce cytokinesis arrest in T. cruzi epimastigotes 51 , exposure to solenopsins did not induce the formation of multinucleate cells.
Either the molecular interactions of solenopsins with cellular components are reversible, or parasites can somehow compensate for the toxic effects. The growth inhibition caused by solenopsins-with no arrest at  www.nature.com/scientificreports/ any specific cell cycle stage-might be associated with the inhibition of the total synthesis of macromolecules, indirectly leading to a larger G 0 "stationary" phase. In fact, the cell cycle is extremely sensitive to any kind of chemical or physical stress, due to several control mechanisms that ensure the capacity for progression to the next phase 55 . Therefore, several environmental stresses can generate a delay in cell cycle progression of different cells. Antimicrobial alkaloids are typically tested for DNA intercalation 56,57 . The fluorescent DNA intercalator assay suggested that solenopsins either present weak DNA interactions, e.g., weak ionic interactions, or merely bind to DNA-associated histones or enzymes. Concerning histone interactions, it has been shown that the overexpression of histone (H1) in Leishmania generates a delay in the cell cycle from increased histone interactions with nuclear DNA 58 . Chaetocin, a fungal toxin possessing an epipolythiodioxopiperazine alkaloid moiety and a nonspecific inhibitor of histone lysine methyltransferases 59,60 , has been shown to impair proliferation, arrest cell cycle progression and induce nucleolar disassembly in T. cruzi 53 . The inhibition of other cell functions could disrupt the cell cycle and morphology, as illustrated by the alkaloid vinblastine impairing transcriptional and www.nature.com/scientificreports/ post-transcriptional regulatory levels and regulating tubulin expression in T. cruzi 54 . Similar effects could also result from solenopsin alkaloids affecting membrane transporters, as has been already described in mammalian cells 13,17,18 . Nevertheless, the observations reported here delineate the scope of the mechanisms of action of solenopsins acting on T. cruzi.   15 , Na + -K + ATPase activity in chickens and catfish 16 , and suppress PI3K activation and/or associated downstream phosphorylation in mammals-for example, protein kinase B (PKB/Akt) and its substrate forkhead box 01a (FOXO1A) 21 . The functional inhibition of Akt activity has been linked to reduced 3-phosphoinositide-dependent protein kinase 1 (PDK1) activation and increased mitochondrial reactive oxygen species (ROS) and autophagosome formation, lethal to several malignant tumour cell lines 19 .
Solenopsins share the long alkyl side chains observed in ether-phospholipids, and sphingolipids have a structural resemblance to miltefosine (Fig. 1, compound VIII), edelfosine, perifosine 21 and ceramide 19,20 in that both have a positively charged amine moiety. Indeed, like solenopsins, ether-lipid analogues interact with Akt and on other potential alternative signalling targets, such as mitogen-activated protein kinase (MAPK) and protein kinase C (PKC) in cancer cells 61 . Miltefosine has been used to treat leishmaniasis in humans 62 and has already been shown to be toxic against T. cruzi both in vitro and in vivo 28 . Based on the present ultrastructural observations, solenopsins presented similar effects previously ascribed to ether-lipid analogues, as illustrated by autophagy via autophagosome formation, the onset of membranes around organelles and cytosolic structures, and apoptosis-like cell death resulting in DNA fragmentation and formation of apoptotic bodies 63 . At the molecular level, despite the intrinsic pro-inflammatory effects 28 , the main targets of miltefosine are (1) the initial biosynthesis enzymes of ether lipids involved in the synthesis of glycosylphosphatidylinositol anchors 64 , www.nature.com/scientificreports/ (2) membrane Na + -ATPases 65 , and (3) PKC 65 . Notwithstanding such hypothetical structure-function correlations between solenopsins and ether-lipid analogues, additional studies are still needed to elucidate the pathways targeted by solenopsins in T. cruzi. As mentioned, solenopsin-treated parasites typically become rounded, with multiple cytoplasmic vacuoles (Figs. 3, 5). Several adverse environmental conditions (e.g., osmotic stress) can induce cell rounding in T. cruzi epimastigotes, and some altered molecular targets have been associated with this morphological alteration, such as protein phosphatase type 1 66 . Treatment of T. cruzi epimastigotes with calyculin A-an inhibitor of type 1 phosphatases-can induce cell rounding and arrest the cell cycle, but it has not been clarified whether these effects are derived from phosphatase inhibition 66 . An anterior hypertrophied vacuole close to the flagellar pocket was observed in solenopsin-treated parasites (Fig. 5), which could be a response to hypo-osmotic stress. The same contractile vacuole is considered essential to the parasite cell cycle, as it is exposed to drastic environmental alterations (involving osmotic and pH stress) in switching between different hosts 37,67 . However, this interpretation that solenopsins cause osmotic imbalance was not in principle supported by the observed greater susceptibility of the CL-Brener strain, as it overexpresses PI3K that should provide augmented adaptability to osmotic stress.
Cellular polyP levels are known to vary in T. cruzi with stress and are associated with a reduction in polymeric phosphate levels 67,68 . Exposure to solenopsins increased long-chain polyP levels; however, no alterations in the levels of short chain polymers were observed. This altered polyP chain phenotype had not yet been previously described for T. cruzi, even when epimastigotes were subjected to different stresses 30 . This result further argues against an activity mechanism of solenopsins in which the compounds act as osmotic stress inducers (e.g., via membrane interaction). A direct interaction of the alkaloids with the factors regulating polyP levels cannot, however, be ruled out yet. The altered proportions between the short and long polyP levels induced by solenopsin exposure could be associated with diverse pathways. For instance, it is believed that polyP short chains are preferentially consumed in energy metabolism through the cleavage of phosphodiester bonds, leaving long-chain polyphosphates available to regulatory functions 69 . Several studies with bacteria and fungi suggested that longchain polyP plays an important role in the response to stress, such as in bacteria surviving under very low nutrient availability 70 . In this scenario, the accumulation of polyP during the logarithmic growth phase seems central to bacterial survival, but this accumulation normally does not take place at the same phase in T. cruzi 36 . We therefore interpret the increased long-chain polyP levels as a physiological stress response to exposure to solenopsins.
The morphological changes observed in solenopsin-treated parasites are typical of autophagy (cytoplasmic vacuolization and increased MDC binding) and cell death (i.e., retracted cytoplasm and fragmented DNA). Similar effects were reported on protists 71,72 , including T. cruzi 39,[73][74][75][76] and T. brucei 77 exposed to other alkaloids like piperidine analogues. The collected evidence indicates that solenopsins act on interconnected pathways. Therefore, it seems worthwhile to explore whether ROS and Ca +2 levels (e.g., in the mitochondria) are also affected by exposure to solenopsins and would further indicate whether the alkaloids can mediate T. cruzi apoptosis 54,76 .
We have demonstrated that solenopsins are also active against the intracellular infectious amastigote T. cruzi. The obtained IC 50 values that induced a reduction in the number of intracellular amastigotes were approximately 2.5 µM for the different solenopsin extracts, which is considerably lower than the values reported for other tested compounds, such as benznidazole (Table 2), piperine 52 and naphtoquinolones 63 . These values are also lower than the drug concentrations that suppress PI3K activation in SVR murine endothelial cells 19 or the concentrations used to reduce respiration and increase reactive oxygen species to kill cancerous cell lines or even to elevate Akt phosphorylation 19 as a biochemical barrier restoration agent to enhance inflammation in mice challenged with psoriasis 20 . The solenopsins isolated from S. saevissima proved slightly stronger against amastigote T. cruzi than the solenopsins from S. invicta, perhaps as a result of greater macrophage penetration because of the shorter alkyl chain length. This is an interesting hypothesis pending further experimentation.
In short, solenopsin alkaloids proved toxic to T. cruzi amastigotes and epimastigotes, inducing morphological and biochemical alterations in the latter. This was congruent with pilot tests against T. brucei rhodesiense. Taken together, the results suggest that solenopsin alkaloids could be used in designing novel treatments against T. cruzi and other kinetoplastids that cause neglected parasitic diseases in humans. Although toxic at higher concentrations, synthetic solenopsins might be used candidates to increase treatment effectiveness and to decrease treatment toxicity in mammals, as illustrated by strategies adopted with other antibacterial compounds 78 .
In vitro cultures of the bloodstream forms of T. brucei rhodesiense were performed in HMI9 supplemented with 10% FCS and 10% serum plus (Gibco, USA) 82 .
Extraction, purification and characterization of solenopsin alkaloids. Colonies of the fire ants S. invicta and S. saevissima were identified and venom alkaloids extracted as described previously 83 . In brief, fire ants' mounds were collected at Ilha do Fundão, Rio de Janeiro, Brazil, separated from soil by slow flooding, and extracted with hexane (Merck Brasil, Brazil). The organic extract was further purified with hexane-acetone silica column (Sigma-Aldrich, USA) 83 . After solvents evaporation, relative proportion of solenopsin analogues was determined by gas chromatography-mass spectrometry (GC-MS) with a Shimadzu GCMS-QP2010 plus system using a fused silica RTX-5MS column (30 m, ID = 0.25 mm, d r = 0.25 µm) (Restek, Bellefonte, PA, USA) 48 . Mass spectra were obtained using electron impact (EI) at 70 eV. The alkaloids were identified based on comparison with published GC profiles and mass spectra 6,84 . The alkaloids were further quantified by GC-MS using spiked known amounts of the methylxantine alkaloid caffeine (Sigma-Aldrich, USA) as an internal standard. Sample quantification was determined by dividing the total peak area of each alkaloid by the peak area of the internal standard, multiplied by the amount of standard added to the sample. 39 and by the lactate dehydrogenase (LDH) assay 40 . In the MTT assay, cell viability was quantified by the ability of living cells to reduce the yellow dye 3-(4,5-dimethyl-2-thiazolyl)-2,5-diphenyl-2H-tetrazoliumbromide (MTT) to a purple Formazan product. The cells were plated in 24 well plates (7 × 10 4 cells mL −1 ) and alkaloids dissolved in dimethyl sufoxide (DMSO; Pierce, USA) stock solution were tested in triplicate at 1:1 ascendant dilution (varying from 1 to 80 µM) in DMSO (final DMSO concentration was 0.1%). After 72 h of incubation, the supernatant was replaced by fresh medium containing MTT (0.5 mg mL −1 ) and 3 h later, the Formazan product was dissolved in DMSO and absorbance was measured at 595 nm in a DTX-880 spectrophotometer (Beckman Coulter, USA). In the LDH assay, cultures were incubated in 24 well plates for 72 h in the absence or presence of alkaloids as above, and the supernatants were collected (~ 900 µL), centrifuged at 5,000g for 10 min, and used for the LDH enzymatic test. Reactions were prepared in a 96 well plate containing 80 28 . Numbers of parasites in each culture were estimated daily by direct counts in a Neubauer chamber, and their viability estimated by the Trypan blue exclusion method 33 . Control conditions were established with parasites cultured likewise but containing either PBS or with equivalent amounts of DMSO. The highest concentration of solvent used (0.1% DMSO) had no significant effect on the growth of the T. cruzi epimastigote forms or T. brucei rhodesiense bloodstream forms (not shown). The 50% inhibitory concentration (IC 50 ) and its 95% confidence interval values were calculated plotting the inhibition (%) against the log of drug concentration fitted to a sigmoidal curve determined by non-linear regression 85 . Reversibility of the solenopsin effects on the growth of epimastigotes was evaluated by incubating the parasites with 0.16 µM and 0.30 μM of solenopsins, which are the approximate 0.25 × and 0.50 × 48 h IC 50 values (described in "Results" section) of the alkaloids derived from S. invicta and S. saevissima, respectively. After incubation for up to 8 days in BHI-FResultsCS medium, the parasites were collected by centrifugation to 3,000g for 10 min at 4 °C, washed 2 × in PBS, suspended into fresh media free of solenopsins, and cultured for additional 8 days. Daily counts and viability evaluations of the parasites in the cultures were done as above.

Effects of solenopsin on the viability of mammalian cells. Cytotoxicity of solenopsins was evaluated against CHO and BMDM cells by the MTT assay
Effects of solenopsins on the proliferation of amastigote forms inside murine peritoneal macrophages. Murine peritoneal macrophages seeded onto round glass coverslips slides in 12 well plates were co-cultured with TCTs at a ratio of 1:3. After 12 h interaction, the macrophages were washed 5 × with 1 mL PBS to remove the non-internalized parasites and incubated in the absence or presence of solenopsin for 2 days as described above for the mammalian cell viability tests. After this period, the macrophages were fixed with methanol, stained with InstantProv hematological stain (NewProv, Brazil) and the slides observed under a light microscope. The infection index values (the number of infected macrophages X number of intracellular amastigotes/ total number of macrophages) were estimated by direct counting of at least 300 fields. saevissima or S. invicta for up to 8 days as described above. Cell growth was estimated daily by Neubauer hemocytometer counts, and culture smears were Giemsa-stained 86 . At least 500 randomly-chosen microorganisms of each culture were evaluated and classified according to the number of kinetoplasts (k) and nuclei (n) per cell 35  Extraction and quantification of long-and short-chain polyphosphates. Aliquots of epimastigote forms (10 7 -10 8 ) cultured for 2 days in the absence or presence of 0.3 μM of solenopsins from S. saevissima or S. invicta were centrifuged to 3,000g for 15 min at 4 °C, washed with PBS, and processed for the extraction of either long-chain 69 or short-chain polyphosphates (polyP) 35 . The long-and short-chain polyP levels were then determined as a function of the amount of orthophosphate (Pi) released upon treatment with an excess of recombinant exopolyphosphatase purified from Saccharomyces cerevisiae (scPPX) 32 . The released Pi was measured by the malachite green assay.
Fluorescent intercalator displacement assay-Evaluation of the affinity of solenopsins with DNA. The DNA fluorescent intercalator displacement assay was based on the protocol as described before 87 .
Briefly, samples were prepared in each well of a Costar black 96-well containing 10 mM Tris-HCl pH 7.5, 100 mM NaCl Evaluation of epimastigotes cell death. In order to evaluate the presence of autophagic vacuoles, parasites (2 × 10 6 mL −1 ) were cultured in the absence or presence of 0.3 μM of solenopsins for 48 h at 28 °C, centrifuged for 10 min at 1,500g, suspended in PBS containing 0.05 mM monodansylcadaverine (MDC, Sigma-Aldrich, USA), and incubated for 1 h at 28°C 33,88 . Positive controls were obtained through similar incubation of parasites in PBS buffer (nutrient deprivation), which induces autophagy in T. cruzi 41 . An aliquot of 10% of each cellular suspension was collected, washed in PBS, fixed in 4% formaldehyde-PBS solution (PBS/formaldehyde) for 10 min at room temperature, washed again, and images of the parasites were acquired using an epifluorescence Zeiss AxioPlan II microscope (Oberkochen, Germany). The rest of the cell suspensions (90%) were then washed four times with PBS and suspended in 10 mM Tris-HCl, pH 8 containing 0.1% Triton X-100. Intracellular incorporated MDC was measured by fluorescence photometry (λ Ex 380 nm, λ Em 525 nm) in a Spectra Max 250 micro plate reader (Molecular Devices, USA) and expressed as arbitrary units per number of cells. To normalize the measurements to the number of cells present in each well, a solution of EtBr was added to a final concentration of 0.2 mM and the DNA fluorescence was measured (λ Ex 530 nm, λ Em 590 nm). Evaluation of programmed cell death induced by solenopsins was performed using the ApopTag® Peroxidase In Situ Apoptosis Detection Kit (EMD Millipore, Germany) following manufacturer's instructions. In brief, a total of 10 8 epimastigotes (10 7 cells mL −1 in 10 mL) were cultured in BHI-FCS medium in the absence or presence of solenopsins (2.5 µM of S. invicta and 2.5 μM of S. saevissima). After 24 h incubation cells were centrifuged at 3,000g for 10 min at 4 °C, washed once in PBS buffer and fixed in 1% formaldehyde-PBS solution (PBS/formaldehyde) for 10 min at room temperature. A drop of the cell suspension was dispensed on slides pre-coated with 0.01% poly-l-lysine and, after 10 min, the liquid excess was removed, and the slide washed with PBS buffer. Then, the endogenous peroxidase was inactivated by covering the sections with 3% H 2 O 2 for 5 min at room temperature. The sections were rinsed with PBS buffer and immersed in terminal deoxynucleotidyl transferase (TdT) buffer (30 mM Trizma base, pH 7.2, 140 mM sodium cacodylate, 1 mM cobalt chloride). TdT (0.3 U µL −1 ) followed by reaction buffer to cover the sections. After incubation in humid atmosphere for 60 min at 37 °C, the reaction was terminated by transferring the slides to TB buffer (300 mM sodium chloride, 30 mM sodium citrate) for 15 min at room temperature. Then the slides were incubated for 30 min with anti-digoxigenin serum (coupled with peroxidase). The slides were washed four times in PBS (for two min each) and the 3,3′-diaminobenzidine (DAB) reagent was added to cover the slides (diluted 50 times in DAB buffer). After incubation for 4-5 min at room temperature, the slides were rinsed with distilled water and counterstained with methyl green Scientific RepoRtS | (2020) 10:10642 | https://doi.org/10.1038/s41598-020-67324-8 www.nature.com/scientificreports/ (Sigma-Aldrich, USA) for 10 min. The slides were rinsed with water and the samples dehydrated in three washes of butyl alcohol (Sigma-Aldrich, USA) followed by three washes in xylene (Sigma-Aldrich, USA) and processed for light microscopy as above. Results were quantified using CellProfiler image analysis software 89 .
Statistical analysis. Results presented are from two or three separate experiments, performed in duplicate or triplicate, as indicated. Statistics and plots were generated with R v. 3.0.0 85 , incremented with the open packages "plyr", "reshape2", "ggplot2", "conover.test", "drc". Numeric raw data are provided as Supplementary Information 1 with the R scripts ("Supplementary_R_Script_File_Costa_Silva") to ensure output reproducibility and peer verification of details. Mostly due to limited numbers of repetitions no parameters for data distribution were assumed. Statistical differences using non-parametric Kruskal-Wallis followed by Dunn's Multiple Comparison Test (in comparing multiple treatments) or by Wilcoxon-Mann-Whitney Test (in comparing two treatments). Our conclusions were also compatible with general patterns obtained by parametric methods (not shown).