Anti- Trypanosoma cruzi activity of alkaloids isolated from Habranthus brachyandrus (Amaryllidaceae) from Argentina

caused by the parasite Trypanosoma cruzi , affects over six million people worldwide, mainly in Latin American countries. Currently available drugs have variable efficacy in the chronic phase and significant side effects, so there is an urgent need for safer chemotherapeutic treatments. Natural products provide privileged structures that could serve as templates for the synthesis of new drugs. Among them, Amaryllidaceae plants have proved to be a potential natural source of therapeutical agents due to their rich diversity in alkaloids. Purpose: To identify alkaloids with anti- T. cruzi activity from Habranthus brachyandrus (Baker) Sealy (Amar- yllidaceae, subfamily Amaryllidoideae) collected in Argentina. Methods: An H. brachyandrus alkaloid extract was tested against T. cruzi , and its cytotoxicity profile was evaluated against two mammalian cell lines to ascertain its selectivity against the parasite and potential liver toxicity. It was also assessed by a stage-specific anti-amastigote assay and analysed by GC/MS to determine its alkaloid profile. The isolated alkaloids were also tested using the aforementioned assays. Results: The extract showed high and specific activity against T. cruzi . The alkaloids lycoramine, galanthindole, 8- O -demethylmaritidine, 8- O -demethylhomolycorine, nerinine, trisphaeridine, deoxytazettine, and tazettamide were identified by means of GC-MS. In addition, hippeastidine (also named aulicine), tazzetine, ismine, and 3- epimacronine were isolated. The alkaloid ismine was specifically active against the parasite and had low toxicity against HepG2 cells, but did not show anti-amastigote activity. Conclusion: The extract had specific anti- T. cruzi activity and the isolated alkaloid ismine was partially responsible of it. These results encourage further exploration of H. brachyandrus alkaloids in search of novel starting points for Chagas disease drug development.


Introduction
Chagas disease, caused by the protozoan parasite Trypanosoma cruzi, is an anthropozoonosis affecting over six million people worldwide. It is endemic in Latin America, where it is the parasitic disease with the greatest socioeconomic impact, causing over 800,000 disabilityadjusted life years and ~7500 deaths per year (Kratz, 2019;WHO, 2021). In the last decades, the disease has spread beyond its natural geographical boundaries due to migratory movements and is now a global health issue (WHO, 2021).
The clinical course of Chagas disease usually comprises two phases. The initial acute phase is generally asymptomatic and thus frequently goes undiagnosed and untreated. This is followed by a chronic phase that can be clinically silent for life. Nevertheless, around 30-40% of chronically infected individuals will develop digestive and/or cardiac symptomatology, cardiac tissue alterations being the most frequent and leading cause of morbidity and mortality.
There is no vaccine available for Chagas disease and current specific chemotherapy for T. cruzi infections relies only on benznidazole (BNZ) and nifurtimox (NFX). Both drugs, in use since the early 1970s, have several drawbacks, including variable efficacy in the chronic phase and frequent toxic side-effects that often lead to treatment interruption (Kratz, 2019). Therefore, new drugs with improved efficacy and safety profiles are urgently needed.
In this regard, natural products constitute a source of highly underexplored chemical entities with privileged structures and bioactivity, which could serve as templates for the synthesis of new drugs. In fact, 61% of new chemical entities approved as anti-infective drugs over the period 1981-2019 were of natural origin (Newman and Cragg, 2020).
Almost 400 species belonging to more than 100 plant families have been studied for a possible Chagas disease treatment (Santos et al., 2020;Varela et al., 2018). Several compounds with anti-trypanosome activity have been extracted from plants, including alkaloids, terpenoids, flavonoids, and quinonoids (Salem and Werbovetz, 2006). Alkaloids are of particular interest in biomedicine and drug discovery research due to their structural diversity and specific biological potential (Daley and Cordell, 2021). The Amaryllidaceae is a plant family that contains an exclusive, large and still expanding alkaloids group, which is characterized by unique skeleton arrangements and a broad spectrum of biological activities (Berkov et al., 2020).
Amaryllidaceae are distributed in tropical and subtropical regions, coinciding with areas conventionally associated with a rich culture in folk medicine. For instance, they have been used for their therapeutic activity in cancer, fever, swelling, and some parasitic diseases (Berkov et al., 2020). Indeed, around 15 species of Amaryllidaceae have been indicated for their use against parasitic infections (Nair and van Staden, 2019). Among them, extracts or alkaloids isolated from the genera Crinum, Hippeastrum, Rhodophiala, and Zephyranthes have been previously reported for their anti-parasitic activity (Martinez-Peinado et al., 2021a, 2020bNair and van Staden, 2019). In Argentina, the Amaryllidaceae family includes around 61 species belonging to ten genera (Chlidanthus, Crinum, Habranthus, Haylockia, Hieronymiella, Hippeastrum, Phycella, Rhodophiala, Stenomesson and Zephyranthes).

Alkaloid extraction and isolation
Dry powdered bulb material (200 g) was macerated in H 2 SO 4 2% for 4 h in an ultrasonic bath (3 × 1000 ml). Subsequently, samples were centrifuged at 5000 x g (10 min), and the supernatant was transferred to another flask where it was defatted with di-ethyl-ether (3 × 500 ml). The aqueous solution was led to pH 11-12 with 10% NaOH and the alkaloids were extracted with dichloromethane (3 × 500 ml). The organic phase was dried with anhydrous sodium sulfate and then evaporated to obtain the basic alkaloid extract, named AE-Hbr (194 mg).

GC-MS analysis
The alkaloids were identified by comparing their GC-MS spectra and Kovats retention index (RI) values against authentic Amaryllidaceae alkaloids previously isolated and identified. Spectral data were processed with AMDIS 2.64 software. Alkaloids were identified by comparing their fragmentation pattern and RI with those of the Amaryllidaceae alkaloids of our laboratory library, in which the isolated compounds were identified by NMR and other spectroscopic techniques (UV, CD, MS), as well as the NIST database and literature data. RI values were calibrated with an n-hydrocarbon calibration mixture (C9-C36), and compounds were semi-quantified according to Torras-Claveria et al. (2014) and expressed as µg GAL/mg AE-Hbr as well as µg GAL/100 mg of bulb dry weight.

T. cruzi growth inhibition assay
The anti-T. cruzi assay was performed as previously described (Martinez-Peinado et al., 2020b). The assay test plates were prepared with starting concentrations of 754 µg/ml of extract and 500 µM of alkaloid, which were diluted following a dose-response pattern. The percentage of DMSO was kept below 0.5% in all the wells. Vero cells and purified trypomastigotes were diluted at a concentration of 1 × 10 6 per ml and mixed at a volume/volume (1:1). Then, 100 μl of the solution was added per well (50,000 Vero cells and trypomastigotes), the multiplicity of infection being 1 (MOI) (Martinez-Peinado et al., 2020b). Each run contained the reference drug BNZ as well as positive and negative controls, as described (Martinez-Peinado et al., 2020b). Plates were incubated for 4 days at 37 • C. Assay read out was performed by adding 50 μl per well of a PBS solution containing 0.25% NP40 and 500 μM chlorophenol red-β-D-galactoside (CPRG) substrate, incubating for 4 h at 37 • C and recording the absorbance at 590 nm, as described (Buckner et al., 1996). All experiments were performed at least in triplicate.

Anti-amastigote specific activity
An anti-amastigote assay was performed as previously reported (Martinez-Peinado et al., 2021b). Briefly, 5 × 10 6 Vero cells were seeded in a T-175 flask and cultured for 24 h. Then, cells were washed and infected with 1 × 10 7 trypomastigotes (MOI ~ 1). After 18 h, infected cell monolayers were washed with PBS and detached from the flask. Cells were counted and diluted to a concentration of 5 × 10 5 cells per ml, before adding 100 µl per well to test plates already containing the extract or alkaloids as previously described. We included BNZ, positive and negative controls, as described (Martinez-Peinado et al., 2021b).

Cell toxicity assays
Vero and HepG2 cell toxicity assays were performed as described (Martinez-Peinado et al., 2020b). Briefly, Vero and HepG2 cells were respectively diluted at a concentration of 5 × 10 5 and 3.2 × 10 5 cells per ml, before adding 100 µl per well. Each run contained its own negative and positive controls as described (Martinez-Peinado et al., 2020b). Plates were incubated at 37 • C for 4 days in the case of Vero cells, and 2 days for HepG2 cells. Then, 50 µl of a PBS solution containing 10% AlamarBlue was added to each well and the plates were incubated for another 6 h at 37 • C before reading the fluorescence intensity (excitation: 530 nm, emission: 590 nm) as described (Martinez-Peinado et al., 2020b). All experiments were performed at least in triplicate.

Statistical analysis
The absorbance and fluorescence values were normalized to the controls as described (Martinez-Peinado et al., 2020b). IC 50 and TC 50 values were determined with GraphPad Prism 7 software (version 7.00, 2016) using a non-linear regression analysis model ([Inhibitor] versus normalized response -Variable slope) (Martinez-Peinado et al., 2020b). Values provided are means and standard deviation (mean ± SD) of at least three independent experiments.

Alkaloid profile analysis
To elucidate the alkaloid profile, AE-Hbr was analyzed by GC/MS, comparing the fragmentation patterns of the compounds in the extract against those of the alkaloids belonging to the GC-MS library (Fig. S1). Twelve alkaloids were identified from AE-Hbr by GC-MS analysis, involving nine Amaryllidaceae skeleton types (Berkov et al., 2020). The GC-MS results are presented in Table 1, which shows the alkaloid composition of the extract, as well as the retention index and semi-quantitation expressed as µg GAL/mg AE-Hbr and µg GAL/100 mg of bulb dry weight for each alkaloid.
Based on the semi-quantitation analysis, the alkaloids found in highest abundance were hippeastidine (also named aulicine), tazettine and, to a lesser extent, ismine and 3-epimacronine (see µg GAL/mg AE-Hbr in Table 1). Alkaloid structures are shown in Fig. 1 (MS data of the alkaloids are provided in Table S1).
AE-Hbr was then evaluated in a Vero cell toxicity assay to discern whether the activity observed was specific against the parasite. Compared to BNZ, the extract displayed higher toxicity against Vero cells, showing a TC 50 value of 14.24 ± 4.53 µg/ml (Table 2, Fig. 2B). As previously described, a selectivity index (SI, or TC 50 to IC 50 ratio) > 10 was considered for progression of the extract (Martinez-Peinado et al., 2020b). AE-Hbr committed to this threshold with a SI value of 25.42, so it was tested in a HepG2 cell toxicity assay, a model used to predict acute liver toxicity. Digitoxin was included in every round and an average TC 50 value of 0.19 ± 0.06 µM [i.e., 0.15 ± 0.04 µg/ml] was obtained. As in the Vero cell toxicity assays, AE-Hbr was more toxic than BNZ against HepG2 cells, with a TC 50 value of 16.20 ± 3.56 µg/ml (Table 2, Fig. 2C).
Amastigotes, the intracellular replicative forms infecting mammals in the parasite life cycle, are likely the main target for any prospective drug to treat chronic T. cruzi infections. Thus, AE-Hbr was evaluated by means of a biological assay specifically targeting amastigotes and found to have a potent IC 50 value of 0.66 ± 0.02 µg/ml and a SI window > 10 (SI = 21.57), indicating specific anti-amastigote activity ( Table 2, Fig. 2D).
Ismine exhibited an IC 50 value of 31.13 ± 1.10 µM (Table 2), which was less potent than that of BNZ (Fig. 3A). After progressive testing in the Vero and HepG2 cell toxicity assays, it showed TC 50 values ≥ 300    Fig. 3B and C). However, it was found to be inactive against the amastigotes (Table 2, Fig. 3D). Hippeastidine and 3-epimacronine, tested for the first time against T. cruzi in this study, were both found to be inactive (IC 50 values > 150 µM) (Table 2, Fig. S2). Tazettine, which has been evaluated previously, was also inactive against T. cruzi (Martinez-Peinado et al., 2020b).

Discussion
The chemical profile of AE-Hbr showed the presence of twelve alkaloids, four of which, ismine (1), hippeastidine (8), tazettine (9) and 3epimacronine (11), were isolated and characterized by means of NMR data. In other phytochemical studies of H. brachyandrus, using samples obtained in the Netherlands, Boit and Döpke (1959) and Wildman and Brown (1968) found a different alkaloid profile. Likewise, Jitsuno et al. (2009) reported a different alkaloidal composition of a H. brachyandrus bulb alkaloidal extract obtained from a sample purchased from a garden  N. Martinez-Peinado et al. center in Heiwaen, Japan. Causes of alkaloid diversity can be genetic or environmental, or an interaction between both factors, as well as the geographical origin of the species (Berkov et al., 2011). Finally, a mistake in taxonomic characterization is possible.
In this study, AE-Hbr was found to specifically inhibit the parasite growth. It also showed low toxicity against HepG2 cells and Vero cells, and was active against amastigote forms, with an IC 50 value of 0.66 ± 0.02 µg/ml and an SI value of 21.57 (Table 2). The low toxicity observed against both mammalian cell lines is of interest and could be related to the absence of lycorine-and crinine-type alkaloids, which are reported to have cytotoxic properties (Nair et al., 2012;Nair and van Staden, 2014).
To determine their individual contributions to the biological activity of the extract, the alkaloids hippeastidine (8), ismine (1), and 3-epimacronine (11) were isolated and tested against the parasite. Ismine had specific activity against the T. cruzi mammalian forms with IC 50 values equal to 31.13 ± 1.10 µM. Notably, it showed very low toxicity against Vero and HepG2 cells, which contributed to a good selectivity index. Gasca et al. (2020) conducted in silico studies with Amaryllidaceae alkaloids, showing that ismine is highly hydrophobic, and therefore able to penetrate biological membranes such as the human intestinal and blood-brain barrier, and a substrate of the human P-glycoprotein. These pharmacokinetic properties could allow effective distribution of the compound if administrated orally, which is an important requirement for new antichagasic drugs.
Information on the mechanisms of action involved in the antiprotozoal effects of Amaryllidaceae alkaloids is scant in the literature. Amaryllidaceae alkaloids have shown potent antioxidant activities due to the presence of enol or phenol groups able to stabilize reactive oxygen species and reactive nitrogen species (Cortes et al., 2018). Since the 1980s, it has been known that several synthetic antioxidants inhibit the respiration and growth of T. cruzi in culture (Aldunate et al., 1986). Some aphorphine alkaloids with antioxidant properties have also shown anti-T. cruzi activity (Barbosa et al., 2021;Morello et al., 1994). For instance, Morello et al. (1994) found that boldine and other aporphine alkaloids were able to inhibit T. cruzi epimastigote growth. The most active compounds also inhibited cell respiration, suggesting that these alkaloids may act by blocking mitochondrial electron transport. Moreover, their anti-T. cruzi activity appears to be correlated with their antioxidative properties (Morello et al., 1994). Recently, another aporphine alkaloid, dicentrine-β-N-oxide, was described to have specific activity against T. cruzi trypomastigotes (IC 50 = 18.2 µM, SI = 11) and the mechanism of parasite death was mitochondrial depolarization (Barbosa et al., 2021). Aporphine and Amaryllidaceae alkaloids, which belong to the isoquinoline class, are characterized by a benzene ring fused to a pyridine ring, and share antioxidant properties. Taking these previous works into consideration, it could be hypothesized that ismine is acting through a similar mechanism of action. In addition, genes encoding enzymes involved in mitochondrial oxidative phosphorylation are upregulated in amastigotes compared to trypomastigotes, suggesting that the respiratory capacity differs between these two parasite life stages (Li et al., 2016). Thus, this mechanism of action would also account for the lack of activity of ismine against amastigote forms. However, other mechanisms of action reported for other Amaryllidaceae alkaloids in cancer cell lines, such as apoptosis induction and/or interference with DNA, RNA, or protein synthesis, should not be discarded (Nair and van Staden, 2019). The identification of the molecular mode of action or potential targets of ismine is a major challenge that might be explored with in silico or chemical genomics approaches.
In any case, ismine was non-active when tested specifically against T. cruzi amastigotes and thus it would only be partially responsible for the anti-parasitic effects observed in AE-Hbr. Its lack of activity against amastigotes suggests that the anti-parasitic properties of the extract are caused by other less abundant alkaloid/s, or by a synergic action among them. Hippeastidine and 3-epimacronine were found to be inactive. Regarding the alkaloids with a lower content (< 1 µg GAL/mg AE), only 8-O-demethylmaritidine and 8-O-demethylhomolycorine have been tested against T. cruzi and both proved inactive (de Andrade et al., 2012;Kaya et al., 2011). Thus, the composition of AE-Hbr deserves further study, focusing on the anti-T. cruzi activity of the remaining alkaloids. Lycoramine is a GAL-type alkaloid that structurally differs from GAL in the absence of a double bond between C4 and C4a. GAL has been found to be inactive against T. cruzi (Osorio et al., 2010). Thus, hypothesizing that this small structural difference would not have a modifying effect, lycoramine is unlikely to contribute to the AE-Hbr anti-T. cruzi activity. Similarly, we would expect nerinine to be a poor antiprotozoal agent, based on previous work by de Andrade et al. (2012), in which three homolycorine-type alkaloids (homolycorine, 8-O-demethylhomolycorine and 6-O-methyllycorenine) were inactive against the T. cruzi strain Tulahuen C2C4. Deoxytazettine would also be expected to lack anti-T. cruzi activity, as tazettine and 3-epimacronine are reported to be inactive (Martinez-Peinado et al., 2020a). Therefore, the anti-T. cruzi effects of AE-Hbr may be due to the presence of trisphaeridine, galanthindole and/or tazettamide. Structure-antiprotozoal activity relationships have been poorly studied in Amaryllidaceae alkaloids. However, some results suggest that the presence of a methylendioxy group, a feature common to the three aforementioned alkaloids, and a tertiary non-methylated nitrogen, found in trisphaeridine, induces a higher anti-parasitic activity (Osorio et al., 2008). In any case, the alkaloids found in AE-Hbr in low amounts (<1 µg GAL/mg AE) warrant testing against T. cruzi to elucidate their potential contribution to the extract activity. To the best of our knowledge, this is the first time AE-Hbr and its isolated alkaloids have been evaluated for their anti-T. cruzi properties.

Conclusions
H. brachyandrus specifically inhibited T. cruzi growth, showed low toxicity to HepG2 cells and was active against amastigote forms. The chemical profile of AE-Hbr showed the presence of twelve alkaloids and four of them were isolated. Among them, ismine was shown to be partially responsible for the anti-parasitic effects of AE-Hbr. The results obtained in the present work encourage us to continue exploring the alkaloids that are also present in AE-Hbr in different amounts, including those that are present in minor proportions.