Improving the drug-likeness of inspiring natural products - evaluation of the antiparasitic activity against Trypanosoma cruzi through semi-synthetic and simplified analogues of licarin A

Neolignan licarin A (1) was isolated from leaves of Nectandra oppositifolia (Lauraceae) and displayed activity against trypomastigote forms of the etiologic agent of American trypanosomiasis, Trypanosoma cruzi. Aiming for the establishment of SAR, five different compounds (1a – 1e) were prepared and tested against T. cruzi. The 2-allyl derivative of licarin A (1d) exhibited higher activity against trypomastigotes of T. cruzi (IC50 = 5.0 μM and SI = 9.0), while its heterocyclic derivative 1e displayed IC50 of 10.5 μM and reduced toxicity against NCTC cells (SI > 19.0). However, these compounds presented limited oral bioavailability estimation (<85%, Papp <1.0 × 10−6 cm/s) in parallel artificial membrane permeability assays (PAMPA) due to excessive lipophilicity. Based on these results, different simplified structures of licarin A were designed: vanillin (2), vanillyl alcohol (3), isoeugenol (4), and eugenol (5), as well as its corresponding methyl (a), acetyl (b), O-allyl (c), and C-allyl (d) analogues. Vanillin (2) and its acetyl derivative (2b) displayed expressive activity against intracellular amastigotes of T. cruzi with IC50 values of 5.5 and 5.6 μM, respectively, and reduced toxicity against NCTC cells (CC50 > 200 μM). In addition, these simplified analogues showed a better permeability profile (Papp > 1.0 × 10−6 cm/s) on PAMPA models, resulting in improved drug-likeness. Vanillyl alcohol acetyl derivative (3b) and isoeugenol methyl derivative (4a) displayed activity against the extracellular forms of T. cruzi (trypomastigotes) with IC50 values of 5.1 and 8.8 μM respectively. Based on these results, compounds with higher selectivity index against extracellular forms of the parasite (1d, 1e, 3d, and 4a) were selected for a mechanism of action study. After a short incubation period (1 h) all compounds increased the reactive oxygen species (ROS) levels of trypomastigotes, suggesting cellular oxidative stress. The ATP levels were increased after two hours of incubation, possibly involving a high energy expenditure of the parasite to control the homeostasis. Except for compound 4a, all compounds induced hyperpolarization of mitochondrial membrane potential, demonstrating a mitochondrial imbalance. Considering the unique mitochondria apparatus of T. cruzi and the lethal alterations induced by structurally based on licarin A, these compounds are interesting hits for future drug discovery studies in Chagas disease.

Antitrypanosomal activity. The antitrypanosomal potentials of the semi-synthetic analogues 1a, 1b, 1d and 1e were superior to the natural licarin A (1), as showed in Table 1. However, these compounds showed appreciable activity only against the trypomastigote form and were not active against the intracellular amastigote. The DNDi states that a good antichagasic agent should present activity against both forms of the parasite 5,9 . Efficacy against trypomastigotes also contributes on the clinical outcome, since this avoids the increases in parasitaemia that can contribute to the spreading of the parasites into healthy cells 9 . On the other hand, several compounds showed no important mammalian cytotoxicity to the maximal tested concentration of 200 μM, and even though some analogues showed some toxicity to the cells, the selectivity index (SI) were >19 for the intracellular amastigotes stage of the parasite, making these compounds interesting hits for further studies.
The most active semi-synthetic compound in the series was 1d, which showed an IC 50 of 5.0 ± 0.8 µM against the trypomastigotes, followed by compound 1e (IC 50 of 10.5 ± 5.7 µM). This suggests that the presence of an additional substitution in the aromatic ring of licarin A contributed to the antitrypanosomal activity. However, the presence of the phenolic hydroxyl led to increased cytotoxicity to mammalian cells, as can be observed in the results for compound 1d (CC 50 of 45.5 ± 16.7 μM). Conversely, moderate cytotoxicity was detected to acetyl derivative 1b (CC 50 67.2 ± 24.6 μM). This can be attributed to the hydrolysis by cellular esterases, which can hydrolyse 1b to the phenol 1 after its penetration into the cells, thus exerting cytotoxic effect.
Compounds with more stable substituents in the hydroxyl group (such as 1a, 1c and 1d) showed no important cytotoxic effect against NCTC cells. This corroborates to our hypothesis on the toxicophoric role of the phenolic hydroxyl, and thus this functional group should be avoided to increase the selectivity of such compounds. However, some of the cytotoxicity can also be attributed to the excessive lipophilicity of such compounds. It is known that lipophilicity affects not only the water solubility of the drugs, but also the ADME properties and accordingly the toxicity 24,25 . Highly lipophilic compounds (log P > 5) tend to bind to hydrophobic sites in the cells, increasing the promiscuity and the cytotoxicity. Considering this aspect, analogues with reduced lipophilicity but maintaining the probable pharmacophore of licarin A were designed. Our pharmacophore hypothesis regards on the presence of the vanillin-like (Fig. 3, red) or the isoeugenol-like (Fig. 3, blue) motifs. Thus, simplified analogues from vanillin (2), vanillyl alcohol (3), isoeugenol (4), and eugenol (5) may present the pharmacophore motifs with less lipophilicity and improved water solubility indeed. The results presented in Table 1 show that analogues 2, 2b, 3b, 4, 4a, 4b and 4d kept the pharmacophore to exert antiparasitic activity against the trypomastigote form, and some of them (2, 2b and 4d) also presented activity against the amastigote form of the parasite. Moreover, almost all compounds showed no relevant toxicity for the mammalian cells (except for compound 3b), reinforcing the hypothesis of the contribution of excessive lipophilicity on the cytotoxic effect.
Isoeugenol derivatives (4a-4d) showed the best activity profile against trypomastigote forms, with IC 50 values ranging from 54.6 ± 10.2 to 8.8 ± 2.0 µM and thus yielding good selectivity towards the parasites. Among them, compound 4d can be highlighted, since it also presented activity against the amastigotes (IC 50 of 10.4 ± 0.9 µM) and a high SI value (>9.5). This result suggests that a possible pharmacophore fragment on licarin A structure is the isoeugenol moiety, as showed in Fig. 4. Furthermore, this data is corroborated by the poor activity showed by the eugenol derivatives 5 , on which the isomerization of the unsaturation of the terminal carbon was detrimental to the activity and thus this double bond seems to be part of the pharmacophore. Additionally, the substitution of the hydroxyl on isoeugenol (4) by a methyl group (4a) and the ortho-substitution in the aromatic ring (4d) seems to increase the antiparasitic activity. As shown in Table 1, compound 4c was the unique isoeugenol derivative that did not display activity, suggesting that the presence of the allyloxy group is detrimental to the activity. The same is also observed for the compound 1c.
Vanillin (2) and its acetyl derivative 2b showed interesting activity against the amastigotes of T. cruzi. Despite the high activity presented by such compounds, this effect may be due the presence of the aldehyde, which is a reactive group and may exert antiparasitic effect through covalent interaction with parasitic proteins. Moreover, compound 2b reinforces the hypothesis of a hydrolysis-dependent effect in the intracellular environment, since compounds 2 and 2b showed the same activity. The explanation to the activity of such compounds only at amastigotes can be related to the higher activity of aldehyde reductases in the extracellular form of the parasite. Sanchez-Moreno and co-workers showed that epimastigotes (but not amastigotes) of T. cruzi, release ethanol to the environment 26 . Cazzulo et al. reported differences in the metabolism of the different forms of T. cruzi, suggesting that the extracellular forms present higher aldehyde reductase activity than the amastigotes 27 , showing that benzaldehydes are reduced faster than the benzyl alcohols are oxidized by the parasitic enzymes 28 . Considering that the corresponding alcohols were inactive in both forms of the parasite, the higher aldehyde reduction rate in the trypomastigotes may explain the reason for the activity of the aldehydes only against the amastigotes 29 . Mechanism of action studies. Compounds 1d, 1e, 3b, and 4a exhibited higher activity against T. cruzi trypomastigotes and were selected for studies of mechanism of action, to understand the possible alterations  Table 1. Anti-T. cruzi activity, cytotoxicity in mammalian cells, selectivity index, log P estimation, and apparent permeability for the licarin A, semisynthetic derivatives, simplified analogues, and positive control benznidazole (Bnz).
www.nature.com/scientificreports www.nature.com/scientificreports/ caused by these compounds in the plasma membrane permeability and the mitochondria of T. cruzi trypomastigotes. The fluorescent probe Sytox Green enters damaged cells and binds to nucleic acid, increasing 500-fold the fluorescence levels. Alterations in the plasma membrane can lead to a total parasite breakdown affecting morphology, fluidity, ion transport, and consequent cell death 30 . In our study, no changes in the fluorescence levels were observed for all the studied compounds, with exception of 1e after 120 min of incubation (Fig. 5) when compared to untreated parasites. The production of reactive oxygen species (ROS) was determined with the cell permeant fluorescence probe H 2 DCFDA. After 1 h of treatment with all studied compounds, it was possible to observe an increased ROS generation by the trypomastigotes, followed by a drop of the levels after 2 h (Fig. 6A,B), showing the cellular apparatus controlling the oxidative stress caused by these highly toxic radicals. The redox imbalance occurs when the endogenous antioxidants fail to remove the excessive ROS produced, which leads to oxidative stress 31 . The bioenergetic system of the parasite was highly compromised, as a result of the increased ATP production after 2 h of incubation with all compounds. This energy expenditure to control the homeostasis was clearly observed in the parasites incubated with all compounds, with exception of compound 1e, which  www.nature.com/scientificreports www.nature.com/scientificreports/ induced no significant alterations in ATP levels (Fig. 7). The stability of ATP levels and mitochondrial membrane potential is a requisite for a normal cell functioning 32 . Mitochondria is a single organelle in trypanosomatids and is directly involved in redox status of the parasite 33 and plays a central role in energy metabolism, being the site of the oxidative phosphorylation that drives the ATP synthesis and represent the main sources of ROS. Additionally, it participates in the nutrient oxidation, calcium homeostasis and apoptosis 34 . With the exception of compound 4a, the compounds induced a mitochondrial hyperpolarization, but with no statistical significance (Fig. 8), suggesting that they probably lead to alterations in the respiratory chain.

Permeability of licarin A and analogues in PAMPA models. To estimate the intestinal absorption
and permeability of the semi-synthetic and simplified analogues of licarin A through the BBB, two PAMPA models were employed, as summarized in Table 1. Accordingly, the semi-synthetic derivatives 1a-1e present low intestinal permeability since their Papp values are lower than 1.0 × 10 −6 cm/s, suggesting that the excessive lipophilicity may impair the oral bioavailability of these compounds. This result is in agreement with the low permeability observed in the BBB model for 1b -1e. The Lipinski's rule-of-five 35 defines that compounds with high molecular mass (>500 Da) and high log P (>5) may present low oral bioavailability, so the compounds fulfil the rule-of-five. However, they already present Clog P values close to 5, and further modifications would raise this value over 5. The rule-of-three 36 is used as guide for designing compounds identified from screening tests because it considers that medicinal chemists will modify the compound, increasing the molecular weight and the lipophilicity. Accordingly, the threshold of rule-of-three is more rigid, limiting the log P until 3 and molecular mass until 300 and thus compound 1 do not fulfil this criterion. Considering this point, the simplified compounds 2-5 are smaller, less lipophilic compounds that fulfil the rule-of-three and, as can be noted in the results Figure 5. Plasma membrane permeability analysis on T. cruzi trypomastigotes with the probe Sytox Green treated with compounds 1d, 1e, 3b and 4a at the respective IC 50 values. As positive (C+) and negative (C−) controls were used trypomastigotes treated with TX-100 at 0.5% (maximum permeabilization) and untreated T. cruzi parasites (minimum permeabilization), respectively. One representative experiment of two assays is shown. **p < 0.005. Figure 6. Evaluation of reactive oxygen species (ROS) generation in T. cruzi trypomastigotes treated with compounds 1d, 1e, 3b and 4a for 1 h (A) and 2 h (B). The H 2 DCFDA fluorescent probe was analyzed spectrofluorimetrically (excitation 485 nm and emission 520 nm). Untreated trypomastigotes and treated with azide (10 mM) were used to achieve minimal and maximal ROS production, negative (C−) and positive control (C+), respectively. One representative experiment of two assays is shown. **p < 0.001.
from Table 1, these compounds presented increased permeability through GIT and limited permeability through BBB. This data suggests that these compounds present better drug-likeness, improved oral bioavailability and less CNS-related off-target toxicity. Therefore, the most promising was the compound 4, which presented adequate permeability on GIT model, and did not seem to cross the BBB, reinforcing its promising pharmacological profile. In addition, the PAMPA-GIT model also estimated that compound 4 has a protein plasma binding percentage lower than 90% (since Papp is <1 × 10 −5 cm/s) 37 , which is desirable from the clinical point-of-view. In counterpart, lipophilic compounds such as 2a, 2b, 4a, 4b and 5 seem to have a plasma protein binding percentage higher than 90% (Table 1) which can impair the bloodstream availability of the free compound and their distribution into the tissues, where the amastigotes are present, compromising drug efficacy.

conclusions
In summary, although the semi-synthetic derivatives of licarin A showed activity against T. cruzi, their low drug-likeness limit their exploitation as prototypes for designing novel compounds with improved pharmacological profile. Therefore, the molecular simplification approach increased the lead-likeness of the set and generated fewer complex compounds with interesting antiparasitic activity that can be considered better prototypes for further modifications, aiming improved activity allied with promising ADME profile.

Methods
General experimental procedures. 1 H and 13 C NMR spectra were recorded on a Bruker Advance 300, operating at 300 MHz for 1 H and 75 MHz for 13 C, using CDCl 3 as solvent and TMS as internal standard. Chemical shifts (δ) are given in ppm and the coupling constants (J) are presented in Hz. HRESIMS spectra were measured on a Bruker Daltonics MicroTOF QII spectrometer. Starting materials were acquired from commercial suppliers  Mitochondrial membrane potential analysis in T. cruzi trypomastigotes treated with compounds 1d, 1e, 3b and 4a for two hours labeled with JC-1 probe (0.2 μM). The fluorescence was measured in a flow cytometer (ATTUNE). Minimum (untreated -negative control, C−) and maximum (treated with CCCP-100 µg/mL -positive control, C+) alterations in the mitochondrial membrane potential were obtained. Fluorescence was quantified by calculating the ratio between the channels BL2/BL1. One representative experiment of two assays is shown.

Scientific RepoRtS |
(2020) 10:5467 | https://doi.org/10.1038/s41598-020-62352-w www.nature.com/scientificreports www.nature.com/scientificreports/ with purity higher than 98% and used without further processing. Claisen rearrangements were performed on a Discovery microwave reactor (CEM Inc.) using a sealed reaction vial with a high-pressure accessory. Column chromatography (CC) procedures were performed using silica gel 60 while progress of the reactions was monitored through TLC in silica gel plates with fluorescence indicator and visualized at 254 nm. Isolation of licarin A from N. oppositifolia. After being dried and powdered, the leaves of N. oppositifolia (332 g) were extracted using n-hexane (6 × 1 L) at room temperature. After evaporation of the solvent at reduced pressure, 32 g of crude n-hexane extract were obtained. Part of this material (20 g) was chromatographed over a silica gel column eluted with n-hexane containing increasing amounts of EtOAc to afford five groups (I-V). Part of the group IV (3280 mg) was chromatographed over a silica gel column eluted with mixtures of n-hexane:EtOAc

Molecular simplification of licarin A and derivatives.
The molecular simplification approach was employed to licarin A and semi-synthetic derivatives considering that licarin A is comprised by a vanillin-like moiety (represented in red -part A, in Fig. 3) and an isoeugenol-like moiety (represented in blue -part B, in Fig. 3). The parent licarin molecule was then broke apart to generate the vanillin analogues (2 and 3) and the isoeugenol analogues (4 and 5), with the same substitution pattern from the semi-synthetic derivatives (a -4-methoxy; b -4-acetoxy; c -4-allyloxi; d -5-allyl). It must be in mind that simplification strategy is very intuitive and based on a medicinal chemist's hypothesis from the pharmacophore units of the parent molecule, that is usually supported by preliminary SAR data 22,23 , as those obtained with the derivatives of licarin A.
General procedure for the preparation of compounds 1a-5a. Using individual flasks containing licarin A (1), vanillin (2), vanillyl alcohol (3), isoeugenol (4), or eugenol (5), two equivalents of K 2 CO 3 , two equivalents of methyl iodide and 10 mL of acetone were added. The reaction mixtures were stirred under reflux for 7 h, and thus the volatiles were evaporated under reduced pressure. The residue was taken up in CH 2 Cl 2 and washed with H 2 O (2 × 25 mL). The organic layers were separated, dried over anhydrous Na 2 SO 4 and the solvent was evaporated under reduced pressure. Crude products were purified through silica gel column chromatography using n-hexane:EtOAc (9:1) as eluent. www.nature.com/scientificreports www.nature.com/scientificreports/ General procedure for the preparation of compounds 1b -5b. Using individual flasks containing compounds 1-5, one equivalent of triethylamine, two equivalents of acetyl chloride and 5 mL of CH 2 Cl 2 were added. The reaction mixtures were stirred under an ice bath for 4 h, when NaHCO 3 solution (10%) was added for neutralization of acids. The organic layers were separated, washed with H 2 O (2 × 25 mL), dried over anhydrous Na 2 SO 4 and evaporated under reduced pressure. Crude products were purified through silica gel column chromatography using n-hexane:EtOAc (9:1) as eluent.
[ General procedure for the preparation of compounds 1c -5c. Using individual flasks containing compounds 1-5, two equivalents of K 2 CO 3 , two equivalents of allyl bromide and 15 mL of acetone were added. The reaction mixtures were stirred under reflux for 19 h, and thus the volatiles were evaporated under reduced pressure. The residues were dissolved in CH 2 Cl 2 and washed with H 2 O (2 × 25 mL). The organic layers were separated, dried over anhydrous Na 2 SO 4 and evaporated under reduced pressure. Crude products were purified through silica gel column chromatography using n-hexane:EtOAc (9:1) as eluent.  mmol of I 2 were added and the mixture was stirred for 24 h at 25 °C. Afterwards, 15 mL of CH 2 Cl 2 were added and treated with Na 2 S 2 O 3 20% solution. The organic layer was separated, dried over anhydrous Na 2 SO 4 and evaporated under reduced pressure. The crude product was then adsorbed on basic alumina (Brockmann I) and submitted to heating at 150 °C for 30 min. After extraction with CH 2 Cl 2 and evaporation of the solvent under reduced pressure, the crude product was purified on a silica gel column using n-hexane:CH 2 Cl 2 (1:1) as eluent to give 7% yield of 1e as a white amorphous solid. www.nature.com/scientificreports www.nature.com/scientificreports/ Trypomastigotes and mammalian cell lines maintenance. Rhesus monkey kidney cells (LLC-MK2-ATCC CCL 7) were used for the maintenance of trypomastigotes of T. cruzi (Y strain) using RPMI-1640 medium supplemented with 2% fetal bovine serum (FBS). The cells and parasites were kept at 37 °C in a humidified atmosphere containing 5% CO 2 . Peritoneal macrophages, used in the experiments of anti-amastigote assay, were obtained by washing the peritoneal cavity of BALB/c mice, with RPMI-1640 medium supplemented with 10% FBS and kept at 37 °C in a 5% CO 2 humidified incubator. Murine conjunctive cells (NCTC clone 929, ATCC) and LLC-MK2 were kept in RPMI-1640 supplemented with 10% FBS at the same conditions described above.
Anti-amastigote assay. To obtain 50% inhibitory concentration (IC 50 ) values against intracellular amastigotes, peritoneal macrophages collected from the peritoneal cavity of BALB/c mice were used. The cells were plated on a 16-well chamber slide -NUNC (Thermo Fisher Scientific) at 1 × 10 5 cells/well and incubated for 24 h at 37 °C in a 5% CO 2 humidified incubator. Next, free trypomastigotes-LLC-MK2 derived, were washed in RPMI-1640 medium, counted and used to infect the macrophages previously plated (10:1, parasite: macrophage ratio). After an incubation of 2 h at 37 °C (5% CO 2 humidified incubator), residual free parasites were removed by washing with RPMI-1640 medium. Tested compounds were subsequently incubated with infected macrophages for 48 h at 37 °C (5% CO 2 humidified incubator) in different nontoxic concentrations. Benznidazole was used as standard drug. At the end of the assay, slides were fixed with MeOH and stained with Giemsa, counted under a light microscope (EVOS M5000, Thermo, USA) and IC 50 values were determined by the infection index 39 .
Statistical analysis. IC 50 and CC 50 values were calculated using a sigmoid dose-response curves in Graph-Pad Prism 5.0 software (GraphPad Software -San Diego, CA, USA). For the mechanism of action studies one-way ANOVA (Turkey's Multiple Comparison test) was used for significance (p < 0.05). The assays were repeated at least twice and the samples were tested in duplicate.
Assessment of the apparent permeability through PAMPA intestinal model. Intestinal permeability of tested compounds was estimated applying the PAMPA model previously developed and validated 37 . Briefly, stock solutions were prepared in dimethyl sulfoxide (DMSO) at the concentrations of 10 mM and then diluted with Tris buffer to give the final concentration donor solution at 300 µM and 5% DMSO. The assay procedure was initiated by filling each well of the microtiter plate (MultiScreen, catalogue no. MATRNPS50, Millipore Corporation, Bedford, MA, USA) with 300 μL of each donor drug solution. Carefully, and avoiding the pipette tip contact with the filter, the hydrophobic filter (0.45 μm) of each acceptor well of the 96-well microfilter plate (MultiScreen-IP, catalogue no. MAIPNTR10, Millipore Corporation, Bedford, MA, USA) was adsorbed with 6 μL of the artificial lipid solution which was composed of 2% of L-α-phosphatidylcholine from soybean dissolved, by sonication, in n-dodecane. Immediately after this application, 150 μL of Tris buffer containing 5% DMSO was added to the receiving well. The receiving well was mounted on the donor plate, keeping the underside of the membrane in contact with the donor solution. The assembled donor-acceptor plates were incubated under constant stirring (3 g) at 25 °C for approximately 16 h. Subsequently, the quantity of each compound presented at the receptor solution (150 μL) was determined by UV/VIS spectrophotometrically. The experiments were performed in hexaplicates (n = 6) and the apparent permeability coefficient (Papp) calculated in centimeters per second (cm/s), together with the standard deviation (SD). Compounds with Papp equal to or higher than 1.0 × 10 -6 cm/s are classified as with high intestinal absorption (>85%) but if it is higher than 1.0 × 10 −5 cm/s they also exhibit a plasma protein binding higher than 90% 41 .
Assessment of the apparent permeability through PAMPA-BBB model. The methodology herein applied was similar to that described in the previous section, however with the purpose of assessing the permeability of the compounds through the blood brain barrier (BBB). Thus, stock solutions (10 mM) of each test compound were prepared in DMSO and diluted with phosphate buffered saline (PBS) at pH 7.4. The final concentration of donor solutions was 300 μM and DMSO of 5%. Artificial membrane lipid solutions were prepared daily by dissolving, in n-dodecane, the porcine brain lipid extracted as described 40 at the final concentration of 2% (m/v). PAMPA procedure was similar to that described in the previous section although, in PAMPA-BBB, the donor and acceptor solutions were prepared using phosphate buffer saline (PBS) at pH 7.4 and the Papp was obtained through the equation previously reported 42 . Accordingly, compounds with values of Papp equal to or higher than 2.0 cm/s are classified as permeable through BBB while those with compounds with lower values of Papp are classified as compounds that do not cross BBB.