Phylobioactive hotspots in plant resources used to treat Chagas disease.

Globally, more than six million people are infected with Trypanosoma cruzi, the causative protozoan parasite of the vector-borne Chagas disease (CD). We conducted a cross-sectional ethnopharmacological field study in Bolivia among different ethnic groups where CD is hyperendemic. A total of 775 extracts of botanical drugs used in Bolivia in the context of CD and botanical drugs from unrelated indications from the Mediterranean De Materia Medica compiled by Dioscorides two thousand years ago were profiled in a multidimensional assay uncovering different antichagasic natural product classes. Intriguingly, the phylobioactive anthraquinone hotspot matched the antichagasic activity of Senna chloroclada, the taxon with the strongest ethnomedical consensus for treating CD among the Izoceño-Guaraní. Testing common 9,10-anthracenedione derivatives in T. cruzi cellular infection assays demarcates hydroxyanthraquinone as a potential antichagasic lead scaffold. Our study systematically uncovers in vitro antichagasic phylogenetic hotspots in the plant kingdom as a potential resource for drug discovery based on ethnopharmacological hypotheses.

Introduction

Chagas disease (CD) or American trypanosomiasis is the disease caused by infection of the hemoflagellate protozoan Trypanosoma cruzi (Pérez-Molina and Molina, 2018). The parasitic T. cruzi is mainly transmitted to humans by hematophagous reduviid bugs of the subfamily Triatominae (Triatoma spp.) (Barrett et al., 2003; Coura and Borges-Pereira, 2010; Pérez-Molina and Molina, 2018). Other means of transmission involve blood transfusion, organ transplant, congenital, and oral contamination (Coura and Borges-Pereira, 2010; Pérez-Molina and Molina, 2018). CD is classified as one of the most neglected tropical diseases, especially among low-income populations (Conteh et al., 2010). The World Health Organization estimates that about 6–7 million people are infected worldwide, primarily in Latin America, where the disease is endemic (WHO, 2018). Bolivia has the highest CD incidence in the world (Organización Panamericana de la Salud, 2006; WHO, 2015), and T. cruzi infection has a high prevalence in different rural areas, primarily affecting the indigenous communities in the inter-Andean valleys and the Chaco (Cassab et al., 1999; Organización Panamericana de la Salud, 2006; de Araújo-Jorge and Medrano-Mercado, 2009).

CD presents two clinical phases. The acute stage of the disease (i.e., infection) often remains unperceived and only very rarely morbidity and mortality occur (Teixeira et al., 2006). In some cases, the sites of bug bites get inflamed and form nodules called chagoma or “Romaña sign” when the protozoan enters via the conjunctiva. After an incubation period of less than two weeks, newly infected individuals may develop fever, chills, myalgia, rash, or meningeal irritation. However, the initial chagoma may be the only symptom of T. cruzi infection for years. At the chronic stage, parasitism and inflammation of the heart and/or enlarged colon can result in severe pathophysiological endpoints, including megacolon, megaesophagus, and cardiomyopathy (Rassi et al., 2010). During the chronic phase, T. cruzi infection can develop an indeterminate form, where, depending on the T. cruzi strain and possibly host immunity, approximately 70% of the infected individuals remain completely asymptomatic (Barrett et al., 2003; Coura and Borges-Pereira, 2010; Ribeiro et al., 2012). Currently, two approved drugs are used to treat acute CD, namely the nitroheterocyclic compounds nifurtimox (Lampit) and benznidazole (Rochagan). However, there is a controversy about their efficacy in chronic CD(Coura and de Castro, 2002; Marin-Neto et al., 2009), and both drugs show adverse effects upon prolonged administration, such as abdominal pain, headache, and neutropenia (Castro et al., 2006; Jackson et al., 2010). Moreover, these antichagasic drugs are hardly accessible for the indigenous communities in Bolivia. Consequently, people living in rural areas of Bolivia rely on traditional medicines for their primary health care (Quiroga et al., 2012; Vandebroek et al., 2008). Botanical drugs contribute significantly to Bolivian folk medicine and they are harvested wild, cultivated, and traded at local markets (Bussmann et al., 2016). Despite reports making general claims about the efficacy of traditional botanical drugs (Cordell and Colvard, 2005; Gertsch, 2009), few studies address their mode of action based on comparative hypothesis-driven ethnopharmacological research. It is unknown whether CD, which has a millennial history in the area (Aufderheide et al., 2004), enforced a selection pressure to prompt ethnomedical strategies among the affected indigenous groups in Bolivia capable of reducing parasitemia or primarily treat the symptoms of CD.

Plants are known to produce a high diversity of secondary metabolites showing numerous pharmacological effects, including antiparasitic and antimicrobial (Wink, 2012; Newman and Cragg, 2020). Medicinal plants may act via different mechanisms of action also against various causative agents of neglected tropical diseases like Mycobacterium ulcerans or T. cruzi (Llurba-Montesino et al., 2015; Zimmermann et al., 2013; Tsouh Fokou et al., 2015). Plant secondary metabolites have led to the development of antiprotozoal chemotherapeutics (Cheuka et al., 2016). Ethnopharmacology continues to inspire bioprospecting as it investigates treatment consensus of natural drugs among ethnic groups as anecdotal indications for presumed pharmacological efficacy (Buenz et al., 2018; Trotter and Logan, 1986). Considering Bolivia's rich biocultural diversity and the endemic nature of CD, ethnopharmacological and bioprospecting studies focusing on antichagasic remedies are surprisingly scarce(Calderón et al., 2010; de Arias et al., 1994; Fournet et al., 1994; Muñoz Ortiz et al., 2010). In parallel, few studies were dedicated to the ethnomedicine and management of CD in Bolivia (Bastien, 1998; Forsyth, 2017). Although numerous plant extracts and natural products have been reported to exert moderate (IC50 values ≤100 μg/mL) to significant (IC50 values 15 ≤ μg/mL) selective toxicity toward different strains of T. cruzi epimastigotes (Izumi et al., 2011; Muschietti and Ulloa, 2016; Schmidt et al., 2012, Schmidt et al., 2012), reports on activity against trypomastigote cellular infection and amastigote replication in vitro are relatively rare. Overall, systematic and comparative antitrypanosomal screenings with comprehensive extract libraries derived from botanical drugs are lacking, thus hindering a direct comparison of efficacy within the same assay.

Here, we report a cross-sectional ethnopharmacological field study among the indigenous ethnic groups Quechua, Izoceño-Guaraní, Ayoreo, and Chiquitano in Bolivia with the aim to investigate their prevalent ethnomedical strategies to treat CD. We document the knowledge about botanical drugs used for the treatment of CD-related symptoms. Fieldwork involved structured and semi-structured interviews, as well as plant collections (drug samples and herbarium vouchers), which was followed by taxonomical identification. An ethyl acetate (EtOAc) extract library of the botanical drugs used to treat CD was generated for in vitro testing against trypanosomes. To validate the ethno-directed bioprospecting approach, we also profiled in vitro antitrypanosomal effects of EtOAc extracts obtained from Mediterranean botanical drugs described in De Materia Medica (DMM), written by Dioscorides in the first century AD (Staub et al., 2016). We hypothesized that the CD botanical drug library would result in a comparatively higher number of extracts showing selective antichagasic activity in vitro. We assembled bioactive (i.e. antichagasic) plant extracts into phylogenetic groups reflecting chemotaxonomic relationships, a process we designated “phylobioactivity”. Employing phylobioactivity-guided grouping in combination with high pressure liquid chromatography (HPLC)-based activity profiling of active extracts, we tested whether bioactive phylogenetic clusters could serve as a basis for more efficient characterizations of antichagasic plant metabolites. The screening of the DMM extract library showed phylogenetic patterns of activity reflecting chemical associations. This multidimensional profiling resulted in a preliminary structure-activity relationship (SAR) study on the antichagasic effects of anthraquinones on parasite release upon host cell infection by T. cruzi trypomastigotes in vitro.

Results

Ethnopharmacological survey in Bolivia

Ethnomedical data were obtained from 361 informants in three different geographic areas representing four ethnic groups (Figure 1). In total, 152 research participants (5 Ayoreo, 68 Chiquitano, 54 Izoceño-Guaraní, 19 Quechua, and six sellers from the city markets) reported ethnomedical knowledge related to the treatment of CD. In our study, the gender of the informants was uniformly distributed, and we were not able to detect gender-specific knowledge related to the treatment of CD (Table 1). When asked about CD and its symptoms, informants of all ethnic groups showed a lack of ethnomedical knowledge related to the pathophysiology of CD. Unlike other disease categories, such as dermatological affections or diseases of the digestive system, chronic CD was recognized unambiguously only after diagnosis by blood tests in the health centers or hospitals. Symptoms related to cardiomyopathies (general fatigue, shortness of breath, dizziness, chest pain, heart palpitations, high blood pressure, and swelling of feet and legs) were generally associated with CD among the Chiquitano and Izoceño-Guaraní who enriched their ethnomedical knowledge by information obtained from medical doctors. Botanical drugs used for CD were most frequently associated with these symptoms.

Figure 1

Ethnopharmacological survey in Bolivia

(A) The three different municipalities (orange), surveyed communities (red dots), and cities (stars) where the field study was conducted are shown.

(B) Graphical summary of the ethnomedical data. The number of informants from the four ethnic groups (Ayoreo, Chiquitano, Guaraní, and Quechua) reporting knowledge of medicinal plants/agents for CD treatment (CD medplant knowledge) and the reported occurrence of CD in the family (CD in family) are shown. See also Figure S1.

Table 1

Sociodemographic characteristics of the informants in the surveyed rural areas reporting knowledge about medicinal plants for CDa, number of use reports, and reported taxa

Ethnicity	Geographical zone	Total	Gender		Age class			Use	Reported
		informants	F	M	<40	41 to 60	>60	reports	taxa
Ayoreo	Chiquitania	5	3	2	1	4	0	5	3
Chiquitano	Chiquitania	68	33	35	22	24	22	199	38
Guaraní	Chaco	54	33	21	19	24	11	93	15
Quechua	Inter-Andean valleys	19	10	9	8	9	2	42	31
Total		146	79	67	50	61	35	339	87

Responses of women sellers in herbal markets in La Paz and Santa Cruz cities are not included in this table.

The survey resulted in more than 350 use reports for 79 plant taxa used in the context of CD. Of these, 69 were identified to the species level, nine plants were identified to the genus level, and one taxon was identified only to the family level. The recorded remedies comprised plant taxa distributed across 37 botanical families and 74 genera (Table 2). From the plants identified to the species level, 80% were native to Bolivia. Only 41 medicinal plant species (52%) were collected from the wild, while the others were obtained through cultivation or from local markets. Decoctions and infusions in cold and hot water, respectively, were the most common preparation forms for oral administration. Informants reported that the chronic symptoms were relieved after taking these plant remedies.

Table 2

List of plant species reported to be used in the treatment of Chagas disease and related symptoms by Ayoreo, Chiquitano, Guaraní, and Quechua informants

Family genus/species	Vernacular name	Used by	Part used	No. use reports	Application	Origin (status)	Habit	Acute or chronic phase	Indication Chagas symptom	Voucher specimen
Amaranthaceae
Dysphania ambrosioides (L.) Mosyakin & Clemants	Caré, paico	C	L, R	17	Or	N (Cul)	Herb	chro	CAR, GAS	ASMP33
Anacardiaceae
Astronium urundeuva (Allemao) Engl.	Cuchi	C	B, L	2	Or	N (W)	Tree	chro	CAR	ASMP56, ASMP6
Schinus molle L.	Molle	Q	L	2	Or, Tp	N (W)	Tree	chro	CAR, GAS	ASMP90
Annonaceae
Annona nutans (R. E. Fr.) R. E. Fr.	Sinini, Aratiku, Sorimimi	G	L	5	Or	N (W)	Shrub	chro	CAR	ASMP11
Duguetia sp.	Sinini	C	L	12	Or	n.d. (Cul)	Tree	chro	CAR	ASMP41
Apocynaceae
Aspidosperma quebracho-blanco Schitdl.	Cacha	C	B	8	Or	N (W)	Tree	chro	CAR, FAT	ASMP35
Vallesia glabra (Cav.) Link	Amarguillo, Arakuarembiu	G	L	3	Or	N (W)	Tree	chro	CAR	ASMP17
Aristolochiaceae
Aristolochia andina F. Gonzáles & I. Vargas	Waje	Q	Br, L	1	Or	N (W)	Vine	chro	CAR	ASMP94
Asteraceae
Acanthostyles buniifolius (Hook. ex Hook. & Arn.) R.M. King & H. Rob.	Romero	Q	AP	2	Or	N (Cul)	Shrub	chro	FAT	ASMP82
Achyrocline alata (Kunth) DC.	Guacanqui, Wira wira	C, LP	AP	6	Or	N (Cul,Pur)	Herb	chro	CAR	ASMP45, ASMP103
Achyrocline hyperchlora S. F. Blake	Wira wira	Q	AP	2	Or	N (W)	Herb	chro	CAR	ASMP91
Ambrosia arborescens Mill.	Altamisa	Q	L	2	Or	N (W)	Subshrub	chro	CAR	ASMP83
Baccharis genistelloides (Lam.) Pers.	Carqueja, Kinsa k'ucho	LP, Q	AP	3	Or	N (Pur,W)	Herb	chro	CAR, GAS	ASMP28, ASMP85
Bidens andicola Kunth	Misuka	Q	AP	1	Or	N (W)	Herb	chro	PUR	ASMP89
Chromolaena connivens (Rusby) R.M. King & H. Rob.	Sunchuj maman	Q	Fr, L	1	Or	N (W)	Shrub	chro	FAT	ASMP75
Culcitium canescens Humb. & Bonpl.	K'ia k'ia	LP	AP	2	Or	N (Pur)	Herb	chro	CAR	ASMP102
Cynara sp.	Alcachofa	LP, SC	AP	2	Or	n.d. (Pur)	Herb	chro	CAR	ASMP13
Gochnatia boliviana S. F. Blake	Melinco	Q	L	1	Or	N (W)	Shrub	chro	CAR, FAT	ASMP87
Mutisia acuminata Ruiz & Pav.	Chinchirkuma	Q	Br, L	1	Or	N (W)	Shrub	chro	CAR	ASMP88
Parthenium hysterophorus L.	Artemisa, Chupurujumo	C	AP	1	Or	N (W)	Herb	acu, chro	FEV, PUR	ASMP59
Pluchea sagittalis (Lam.) Cabrera	4 Cantos	C	Fl, L	1	Or	N (Cul)	Shrub	chro	CAR	ASMP36
Schkuhria pinnata (Lam.) Kunza ex Thell.	Jayaj pichana	Q	AP	2	Or	N (W)	Herb	chro	CAR, FAT	ASMP76
Sonchus oleraceus L.	Diente de Leon	C, LP	AP	3	Or	I (Pur)	Herb	chro	CAR	AMP12, ASMP30
Verbesina sp.	Tabaco	Q	L	1	Tp	n.d. (W)	Tree	acu	BIT	ASMP93
Bignoniaceae
Handroanthus impetiginosus (Mart. ex DC.) Mattos	Tajibo, Tajibo negro	C, G	B, Fl	3	Or	N (W)	Tree	chro	CAR	ASMP18, ASMP57
Jacaranda cuspidifolia Mart.	Paraparau	C	B	6	Or	N (W)	Tree	chro	CAR, FAT	ASMP34
Tabebuia aurea (Silva Manso) Benth. & Hook. F. ex S. Moore	Paratodo, Alcornoque	C, A	B	30	Or	N (Cul)	Tree	chro	CAR, FAT, PUR	ASMP9
Bixaceae
Bixa orellana L.	Uruku	C, G	L	4	Or	N (W)	Tree	chro	CAR	ASMP26
Caricaceae
Carica papaya L.	Papaya	C, G	Fl	3	Or	I (Cul)	Tree	chro	CAR, GAS	ASMP38
Cochlospermaceae
Cochlospermum tetraporum Hallier f.	Kuari, Pela pela	G	B, L	8	Or	N (Cul)	Tree	chro	CAR	ASMP22
Cucurbitaceae
Momordica charantia L.	Balsamina	C	AP	3	Or	I (Cul)	Vine	chro	CAR	ASMP58
Ephedraceae
Ephedra americana Humb. & Bonpl. Ex Willd.	Pisqo simi	Q	AP	1	Or	N (W)	Shrub	chro	FAT	ASMP96
Euphorbiaceae
Croton andinus Muell. Arg.	Taporita, Tupeicha	G	R	5	Or	N (W)	Herb	chro	CAR	ASMP21
Croton sp.	K'uru k'uru	Q	La	1	Tp	n.d. (W)	Herb	acu	BIT	ASMP92
Euphorbia serpens Kunth	Quebra Pedra	C	Wh	4	Or	N (Cul)	Herb	chro	CAR, FAT	ASMP49
Jatropha curcas L.	Piñón	C	L	1	Or	I (Cul)	Treelet	chro	FAT	ASMP39
Fabaceae
Acacia aroma Gillies ex Hook. & Arn.	Tusca	G	B, Fl	2	Or	N (Cul)	Tree	chro	CAR	ASMP24
Acacia sp.	Karikari	C	B	1	Or	n.d. (Cul)	Tree	chro	PUR	ASMP43
Bauhinia sp.	Patecabra	C	L	1	Or	n.d. (Cul)	Shrub	acu, chro	FEV, PUR	ASMP42
Copaifera langsdorffii Desf.	Copaibo	C, SC	B	5	Or	N (W)	Tree	chro	CAR	ASMP51
Crotalaria incana L.	Amorocita	Q	L	1	Or	N (W)	Herb	chro	FAT	ASMP95
Hymenaea courbaril L.	Paquío	C	B	3	Or	N (W)	Tree	chro	CAR, FAT	ASMP55
Pterodon sp.	Pezoe	C	Se	2	Or	N (W)	Tree	chro	FAT	ASMP52
Senna chloroclada (Harms) H. S. Irwin & Barneby	Lanza lanza, Mbuijare, Retama	G	Fl, R, Wh	45	Or	N (W)	Shrub	acu, chro	CAR, FEV	ASMP10
Spartium junceum L.	Retama	A, SC	AP	3	Or	I (Pur)	Shrub	chro	CAR	ASMP100
Gesneriaceae
Gloxinia gymnostoma Griseb.	Ortelón	C	AP	2	Or	N (Cul)	Herb	chro	CAR, FAT	ASMP40
Gramineae
Cymbopogon citratus (DC.) Stapf	Paja de cedrón	C, G, SC	L	25	Or	I (Cul)	Herb	chro	CAR, FAT	ASMP31
Labiatae
Clinopodium axillare (Rusby) Harley	Huayra Muña	Q	Br, Fl, L	2	Or	N (Cul)	Subshrub	chro	CAR	ASMP80
Lepechinia vesiculosa (Benth.) Epling	Raq'acho	Q	L	1	Or	N (W)	Shrub	chro	CAR	ASMP81
Minthostachys ovata (Briq.) Epl.	Muña	Q	AP	3	Or	N (Cul)	Subshrub	chro	FAT, GAS	ASMP69
Ocimum americanum L.	Albahaca	C	Wh	2	Or	I (Cul)	Herb	chro	CAR	ASMP48
Aloe vera (L.) Burm.f.	Sábila, karaguata guasu	C, G	La	5	Tp	I (Cul)	Herb	acu	BIT, PUR	ASMP25
Linaceae
Linum usitatissimum L.	Linaza	Q	Fr	2	Or	I (Cul)	Herb	chro	PUR	ASMP71
Malpighiaceae
Galphimia brasiliensis (L.) A. Juss.	Masiaré	A, C	R	11	Or	N (Cul)	Herb	chro	CAR, FAT	ASMP32
sterile	Azucaró	C	B	4	Or	n.d. (W)	Tree	chro	CAR	ASMP54
Malvaceae
Malva parviflora L.	Malva	LP, Q	AP	2	Or	I (Cul,Pur)	Herb	chro	CAR, FAT, GAS	ASMP77, ASMP15
Meliaceae
Trichilia sp.	Tipa	C, SC	B	11	Or	n.d. (Cul)	Tree	chro	CAR	ASMP53
Myrtaceae
Myrcianthes callicoma McVaugh	Guapurú	Q	L	1	Tp	N (W)	Tree	chro	FAT	ASMP73
Myrcianthes pseudomato (D. Legrand) McVaugh	K'arasacha	Q	L	2	Or	N (W)	Tree	chro	CAR, GAS	ASMP72
Plinia cauliflora (Mart.) Kausel	Guapurú	C	L	1	Or	N (Cul)	Tree	chro	CAR	ASMP60
Oxalidaceae
Hypseocharis pimpinellifolia Remy	Sultaki	Q	R	1	Tp	N (W)	Herb	chro	FAT	ASMP70
Papaveraceae
Argemone subfusiformis G. B. Ownbey	Cardo Santo	C, G	Fl	3		N (W)	Herb	chro	CAR	ASMP20
Bocconia integrifolia Bonpl.	Turumi	Q	L	1	Or	N (W)	Tree	chro	FAT	ASMP97
Passifloraceae
Passiflora cincinnata Mast.	Pachío, Mburukuya	C, G	Fl, L, R	13	Or	N (W)	Vine	chro	CAR, FAT	ASMP16
Piperaceae
Piper rusbyi C. DC.	Matico	C	L	7	Or	N (Cul,W)	Shrub	chro	FAT, PUR	ASMP50, ASMP46
Plantaginaceae
Plantago major L.	Llantén	C	Fl, L	3	Or	I (Cul)	Herb	chro	CAR	ASMP47
Polygalaceae
Monnina wrightii A. Gray	T'ian t'ian	Q	L	1	Or	N (W)	Herb	chro	CAR	ASMP84
Rosaceae
Rubus sp.	Kari kari	LP	AP	2	Or	n.d. (Pur)	Herb	chro	CAR	ASMP101
Rutaceae
Heterophyllaea lycioides (Rusby) Sandwith	Kapi	Q	L	1	Or	N (W)	Shrub	chro	GAS	ASMP86
Ruta chalepensis L.	Ruda	Q	L	1	Or	I (Cul)	Herb	chro	GAS, PUR	ASMP99
Zanthoxylum coco Gillies ex Hook. F. & Arn.	Chirimolle	Q	L	1	Or	N (W)	Tree	chro	CAR	ASMP74
Solanaceae
Cestrum parqui L'Her.	Andrés Huaylla	LP, Q	AP, L	2	Or	N (Cul,Pur)	Shrub	chro	CAR, GAS	ASMP78, ASMP14
Solanum palinacanthum Dunal	Pica Pica	C	L, R	1	Or	N (Cul)	Subshrub	chro	PUR	ASMP44
Urticaceae
Urtica urens L.	Ortiga	LP	AP	1	Or	I (Pur)	Herb	chro	CAR	ASMP29
Valerianaceae
Valeriana potopensis Briq.	Jama jama	Q	B, L	1	Or	N (W)	Herb	chro	CAR	ASMP98
Verbenaceae
Aloysia citriodora Palau	Cedrón	C, Q	L	4	Or	N (Cul)	Shrub	chro	CAR	ASMP79
Aloysia polystachya (Griseb.) Moldenke	Poleo	C	AP	2	Or	N (Cul)	Shrub	chro	CAR	ASMP37
Zingiberaceae
Alpinia zerumbet (Pers.) B. L. Burtt & R. M. Sm.	Colonia	C, G, SC	Fl	26	Or	I (Cul)	Herb	chro	CAR	ASMP19
Zygophyllaceae
Bulnesia bonariensis Griseb	Guayacán, Guayacán Morado	C, G	B	3	Or	N (Cul)	Tree	chro	CAR	ASMP23

A, Ayoreo; C, Chiquitano; G, Guaraní; LP, La Paz; SC, Santa Cruz; AP, aerial parts; B, bark; Br, branches; Fl, flowers; Fr, fruit; La, latex; L, leaves; R, roots; Se, seeds; Wh, whole plant; Or, oral application; Tp, topical application; N, native; I, introduced; n.d., not determined; Cul, cultivated; Pur, purchased at market; W, wild; chro, chronic; acu, acute; BIT, vinchuca bite; CAR, cardiovascular symptoms; FEV, fever; GAS, gastro digestive symptoms; FAT, fatigue; PUR, purifier/blood purifier/to strengthen the blood.

The most cited plant species in the treatment of CD-related symptoms were Senna chloroclada (flowers, root) (Figure S1) and Tabebuia aurea (bark), followed by the introduced and globally known medicinal plant species Alpinia zerumbet (rhizome) and Cymbopogon citratus (herb). Noteworthy, Chiquitano informants also reported a blend of several botanical drugs against CD, such as a combination of the barks of Jacaranda cuspidifolia, T. aurea, and Trichilia sp. Particularly interesting was that few informants stated that certain botanical drugs could cure chronic CD, namely S. chloroclada (flowers, root), Cochlospermum tetraporum (bark, leaves), and Bulnesia bonariensis (bark). However, there was no consensus among independent informants. Even during prolonged stays, the use of these remedies was rarely observed by the authors. Only the use of Galphimia brasiliensis (root) (Figure S1) could be repeatedly observed among the Chiquitano. To treat triatomine bites, topical applications in the form of poultices were mentioned (Table 2). Informants agreed that acute symptoms and inflammatory swellings (chagoma) at the bite sites occurred very rarely. Nocturnal triatomine bites were considered common and harmless. In highly affected communities (Izoceño-Guaraní and Ayoreo), informants stated that the bites were irritating but not painful and therefore remained mostly untreated. Only the taxa Aloe vera, Croton sp., and Verbesina sp were reported to be used for triatomine bites. Bauhinia sp., S. chloroclada, and Parthenium hysterophorus were mentioned as a treatment of fever during first infections. Besides plant-based remedies, eleven Chiquitano informants reported the use of animal products, such as Coragyps atratus (black vulture; blood), Tapirus sp. (tapir; nails), and Equus africanus asinus (donkey; milk) to treat the chronic symptoms of CD. Few Izoceño-Guaraní informants also reported the use of grease of armadillos (Dasypoda spp.) against triatomine bites, and others applied alcohol topically. The plant taxa used for CD obtained in markets were common medicinal plants known to be used for a variety of diseases and ailments (Table 2).

The Quechua and Ayoreo reported relatively fewer remedies to manage CD as compared to the Izoceño-Guaraní and Chiquitano (Table 1), despite the high prevalence of CD in their communities (>40%; Figure 1). The vast majority of the Ayoreo interviewed in this study did not report any botanical drugs used in the context of CD. The visited Ayoreo communities were evangelic or catholic Christians that stopped practicing shamanism decades ago. Generally, they had no or little knowledge about traditional medicinal agents. Although Quechua informants have been educated on the association between triatomine bugs and CD by institutional anti-CD campaigns (Salm and Gertsch, 2019), they had a very limited understanding of the transmission mechanisms and symptoms of the illness. There was poor consensus among Quechua informants, and we obtained only one or two use reports for each plant taxon, except for Minthostachys ovata aerial parts (3 use reports, Table 2). Most of the plants used by the Quechua were native to the region (86% of the total number of identified taxa at species level) and were gathered from the wild (74%). Among the Quechua, traditional healers (curanderos) generally carried out the preparation of botanical drugs. Although there were specialized healers among the Izoceño-Guaraní (paye) and Chiquitano communities, the knowledge and use of botanical drugs were not restricted to them. Guaraní and Chiquitano community members used botanical drugs to alleviate CD-associated symptoms because they are readily accessible and due to lack of alternatives. Especially among the Guaraní communities, there was a general lack of primary health care and traveling to urban centers for seeking medical care, posing significant difficulties for patients.

Identification of selective antitrypanosomal extracts from plant libraries

Biological profiling of the CD ethnobotanical drug library from Bolivia

Crude EtOAc extracts are obtained from 115 botanical drugs, representing 79 plant species and collected based on their use in the management of CD (Table 2). Although the ethnomedical extraction generally proceeded with hot water infusions, we justify the use of EtOAc for extract preparation by the reduced extraction of polar and high molecular weight compounds, such as sugars and tannins, which would potentially interfere with the screening. The extracts were dissolved in dimethyl sulfoxide (DMSO) and tested in vitro for their antitrypanosomal activity against T. cruzi epimastigotes (Y strain) and procyclic T. b. brucei, as well as for general cytotoxic effects in HeLa and Raw264.7 cells. Antiproliferative IC50 values <25 μg/mL were considered biologically significant. To define selective hits, a cutoff of 50% inhibition (HeLa antiproliferative effect) at 25 μg/mL was used. IC50 values were determined only for selective hits. Table S1 summarizes the results obtained in the in vitro bioassays with the plant extracts. Phylogenetic distribution of reported anti-CD plant taxa and screening results is shown in Figure 2. In general, the T. b. brucei strain was very sensitive toward the tested extracts, and, rather surprisingly, 90 extracts (78%) exhibited antitrypanosomal activity against procyclic T. b. brucei. In contrast, only 20 (17%) extracts inhibited T. cruzi epimastigotes with IC50 values ≤25 μg/mL. Seven extracts (6%) showed pronounced (IC50 < 10 μg/mL) and 13 extracts (10%) good (IC50 ≤ 25 μg/mL) antitrypanosomal activity.

Figure 2

Biological profiling of the CD botanical drug library collected in Bolivia

Few EtOAc extracts showed selective toxicity toward T. cruzi epimastigotes (Y strain) with IC50 values below 20 μg/mL (arrows). A high number of extracts were toxic for procyclic T. brucei, and many showed antiproliferative effects in HeLa and Raw 264.7 cells. The Leguminosae was the only subfamily cluster (top) showing no activities up to 25 μg/mL. In the sesquiterpene lactone-rich family Asteraceae, only Acanthostyles buniifolius showed selective antitrypanosomal effects. Data represent profiling values (based on IC50 values) from at least two independent screening assays, each performed in triplicates. See also Table S1 and Figure S2.

The most potent extracts were those of Cynara sp. aerial parts (2 μg/mL), Acanthostyles buniifolius aerial parts (2 μg/mL), and Gochnatia boliviana leaves (4 μg/mL). The extracts from roots and flowers of S. chloroclada, the most frequently cited anti-CD botanical drugs among the Guaraní, were not active against epimastigotes (IC50 > 50 μg/mL), but an extract of the aerial parts of this plant was moderately active. The S. chloroclada flower extract showed significant inhibitory effects on parasite release in the infection assay (15 μg/mL was equally effective as 20 μM of benznidazole) without being cytotoxic to host cells (Table S1). A preliminary assessment of its phytochemical content was thus performed (vide infra). Since many extracts exhibited significant general cytotoxicity, the observed toxicity against epimastigotes could also be due to non-specific effects. Besides, cytotoxic extracts (27.8%) could not be analyzed in the trypomastigote release assay as the response could be erroneous due to the toxicity on the host cells. Particularly, cytotoxic extracts were those from Ambrosia arborescens leaves, G. boliviana leaves, G. boliviana branches, Bocconia integrifolia leaves, B. integrifolia roots, Cynara sp. aerial parts, and P. hysterophorus aerial parts. The only highly active and selective antitrypanosomal extracts were from Pterodon sp. seeds, Sonchus oleraceus leaves, A. buniifolius aerial parts, Aloysia polystachya aerial parts, and Gloxinia gymnostoma aerial parts (Figure 2). The extracts of these taxa also showed significant inhibitory effects in the T. cruzi infection assay (Table S1).

As initial IC50 values were obtained in antiproliferation assays with T. cruzi epimastigotes, we next screened the extracts lacking general cytotoxicity in the cellular infection assay. In order to validate the antitrypanosomal activity on the mammalian stage forms (trypomastigotes) that are relevant for the disease, extracts were tested for their ability to inhibit T. cruzi parasite release from infected chinese hamster ovarian cells (CHO). To that aim, we employed a versatile fluorescence-activated cell scanning (FACS)-based assay to quantify the number of T. cruzi trypomastigotes released (and residual amastigotes in case cells burst prematurely). Using the SYTO9 DNA dye, the released parasites in the supernatant were stained (Figure 3A). Using culturing conditions without heat-inactivated fetal bovine serum (hiFBS) yielded consistently viable host cells with high reactive oxygen species (ROS) but reduced (about 50%) parasite release compared to CHO cells cultured in 0.5% hiFBS (Figures 3A and 3B).

Figure 3

FACS parasite release assay in 0.5% FBS and serum-free host cell conditions

(A) Representative FACS histograms of the FL-1 channel (488/530 nm) showing parasites released into the medium from host cells in no FBS and 0.5% hiFBS cells infected with wild-type trypomastigotes. The vehicle control, benznidazole (BZN) treatment at 20 μM, and no infection control are shown. All samples were stained with the SYTO9 dye. Data are representative of at least six independent experiments.

(B) Measurement of reactive oxygen species in CHO-K1 cells cultured in the different FBS conditions. ROS levels were measured using the indicator DCFDA by FACS after 24 hr of exposure to different medium conditions. Control cells were cultured in complete medium. Bar graphs represent the mean fold change in the geometric mean of fluorescence intensity ±SD and were analyzed using FlowJo from three independent experiments performed in triplicate. Statistical significance was calculated with t test. ∗∗∗∗P < 0.0001.

Serum deprivation is known to increase intracellular ROS (Halliwell, 2003; Tangtrongsup and Kisiday, 2017) and helped to prevent overconfluency of the host cell layers over 6 days of culturing. Dichlorofluorescein diacetate (DCFDA) fluorescence was used to assess the relative amount of ROS during the critical infection time (24 hr). Host cells infected and treated in the presence of 0.5% hiFBS generated a lower level of ROS than cells cultured in the absence of hiFBS (Figure 3B). Amastigotes develop intracellularly, differentiate into trypomastigotes, and leave the host cell. As elaborated in the discussion, ROS is a factor in host cells that is important for parasite infection in vivo. Benznidazole was used to validate this parasite infection assay. Although the assay showed significant variability (±35%), due to the nature of independent T. cruzi infections, reliable IC50 values for benznidazole were obtained (in the range of 5–9 μM), thus somewhat higher than in previously published data (Da Silva et al., 2011; Koovits et al., 2020) Nevertheless, we employed this serum-free condition as it mimicked more closely the pathophysiological condition (see discussion) for our profiling of the CD botanical drug library. The initial testing of the plant extracts was performed at a single concentration (15 μg/mL). As shown in Table S1, extracts which were active against T. cruzi epimastigotes were generally also active in the infection assay. The pre-screening with epimastigotes was thus considered suitable for discovering antichagasic compounds able to inhibit different stages of the infection/replication cycle.

Comparative profiling of the DMM botanical drug library from the Mediterranean

To validate or challenge the ethno-directed approach for bioprospecting antichagasic activity, we compared the results obtained for botanical drugs used in a CD context with those of 660 botanical drugs described in Dioscorides' DMM. Notably, the botanical drugs mentioned in DMM have no traditional association with CD. EtOAc extracts were tested at 25 μg/mL in vitro against T. cruzi epimastigotes and on HeLa cells for cytotoxicity (Figure 4 and Table S2). The extracts were considered active (i.e., hits) when the percentage of inhibition was ≥50% at 25 μg/mL. A total of 59 (8.9%) extracts exhibited antitrypanosomal activity, while 102 (15.5%) extracts were cytotoxic toward HeLa cells. Among the antitrypanosomal hits, only 27 extracts (4.1%) were selectively toxic for T. cruzi epimastigotes over HeLa cells. Selective antitrypanosomal extracts were those obtained from Levisticum officinale fruits, Opopanax chironium roots, Glebionis coronaria flowers, Tanacetum parthenium flowers, Convolvulus scammonia roots, Iris foetidissima seeds, Laurus nobilis fruits, Rheum rhaponticum roots, Rumex crispus roots, and Ruta chalepensis roots. Extracts with selective toxicity for T. cruzi epimastigotes were considered potential hits, and their antichagasic activity was subsequently determined at 15 μg/mL in the trypomastigote release assay in CHO cells. All extracts except those obtained from A. vera resin, L. nobilis leaves, and L. albus roots were also active in the infection assay (>50% inhibition of parasite release at 15 μg/mL).

Figure 4

Comparative profiling of the DMM library from the Mediterranean

Phylobioactivity tree displaying phylogenetic relationships associated with bioactivities of EtOAc extracts. The outer ring shows growth inhibition on T. cruzi epimastigotes, and the inner ring shows growth inhibition of HeLa cells (both at 25 μg/mL; only most active plant part shown). A hypothetical coumarin cluster (aRuta chalepensis root, Levisticum officinale seeds, and Seseli tortuosum root) and the anthraquinone cluster (Rumex crispus and Rheum rhaponticum rhizoma) are visible. Laurus nobilis root and fruits and Sium sisarum root (microfractionated) are indicated in the phylogenic tree. Detailed data on plant species and activities are shown in Table S2.

Comparison of extract libraries derived from disparate ethnopharmacological contexts

In an attempt to pharmacologically validate the ethno-directed approach, hit rates were calculated for both the Bolivian CD and Mediterranean DMM botanical drug libraries. The Pearson Χ2 test was applied for statistics. Results show that there is a significantly higher probability (17.4% vs. 8.9%, P = 0.0057) of detecting antichagasic (T. cruzi epimastigotes) extracts when the plant had a reported use against CD. However, we also found a significantly higher percentage of cytotoxic (IC50 ≤ 25 μg/mL) extracts among CD botanical drugs (27.8% vs. 15.5%, P = 0.0012). Taking the importance of selectivity into account, the hit rate of selective antichagasic extracts in vitro was not considered statistically different (P = 0.079) between the Bolivian CD (7.8%) and the DMM (4.1%) extract libraries. The two libraries shared 20 genera and the species S. oleraceus, Spartium junceum, A. vera, Linum usitatissimum, and R. chalepensis (Tables S1 and S2). Only S. oleraceus and S. junceum (both aerial parts) share also the botanical drug.

Microfractionation of selected extracts and identification of antitrypanosomal natural products

For the isolation of potentially antichagasic metabolites related to the DMM library, active extracts from different phylogenetic clusters (Figure 4) were selected. A major phylogenetic hotspot showing trypanocidal selectivity was the “anthraquinone cluster” with R. crispus (curly dock) and R. rhaponticum (rhapontic rhubarb) rhizomes from the Polygonaceae family. As anticipated, anthraquinones and naphthoquinones were major active principles in R. crispus (Figure 5) and R. rhaponticum (not shown). The latter also yielded the sesquiterpene lactone (SL) (6R, 7S)-costunolide. Since sesquiterpene lactones are known to be generally cytotoxic (Schmidt et al., 2009), we performed the host cell cytotoxicity assays also in serum-free conditions (Table 3). The observed differences (CC50 values) in serum-free conditions versus 0.5% hiFBS culture conditions likely resulted from differences in cell proliferation.

Figure 5

Microfractionation of selected extracts and identification of antitrypanosomal natural products

Bioactivity-guided microfractionation is exemplified with R. crispus using liquid chromatography and photodiode array (PDA) and evaporative light scattering detectors (ELSD). Isolation of antichagasic metabolites was based on epimastigote proliferation inhibition. False negatives are a limitation of this qualitative approach (shown here with chrysophanol) due to low concentrations. Nepodin and emodin/torachrysone were identified and isolated from fully active fractions (0% cell viability). The moderately active chrysophanol was identified in a negative fraction. Controls: BZN, benznidazole; Nifx, nifurtimox.

Table 3

In vitro antiproliferative activity (50% inhibition [IC50]) of compounds isolated from plant extracts on T. cruzi epimastigote stage (72 hr) and trypomastigote release (6 dpi)

Isolated cpd	IC50 epimastigotes [μM]	Percentage inhibition of parasite release (serum-free) at 5 μM	Cytotoxicity CC50 CHO cells [μM]	Cytotoxicity CC50 CHO cells without FBS [μM]
Nepodina	28.7 ± 13.3	34.7 ± 30.9	>100	n.d
Torachrysonea	>50	25.9 ± 15.5	>100	n.d
Emodina	14.1 ± 8.2	61.5 ± 18.2	>100	n.d
Falcarindiolb	>50	0	>100	n.d
Costunolidec	7.4 ± 5.9	0	10.9 ± 3.7	35.4 ± 13.8
Reynosind	>50	50.0 ± 47.1	31.3 ± 8.0	66.1 ± 1.2
Santamarined	19.5 ± 9.3	55.9 ± 46.7	13.2 ± 4.3	39.4 ± 1.5
Zaluzanin Cd	6.7 ± 0.7	71.0 ± 7.5	6.6 ± 1.7	19.2 ± 5.2
3-Acetylzaluzanin Cd	6.3 ± 0.9	83.0 ± 10.1	6.9 ± 2.4	n.d
Dehydrocostus lactoned	1.4 ± 0.4	86.2 ± 4.1	5.8 ± 1.9	12.2 ± 0.5
Eremanthind	1.9 ± 0.3	87.6 ± 9.9	7.3 ± 1.4	11.5 ± 0.2
BZN (20 μM)	13.8 ± 2.9	76.9 ± 15.2	>100	n.d

Cytotoxic (antiproliferative) effects of the compounds were assessed on CHO host cells after 72 hr (50% cytotoxic concentration [CC50]). Data shown are mean values ±SD of at least three independent experiments, each performed in triplicates. n.d, not determined; BZN, benznidazole.

Rumex crispus root.

Sium sisarum root.

Rheum rhaponticum root.

Laurus nobilis leaf.

As illustrated with the R. crispus extract, the most potent antichagasic principle was the anthraquinone emodin, which completely inhibited T. cruzi epimastigote proliferation during the bioactivity-guided isolation (Figure 5). Emodin subsequently also inhibited parasite release in vitro with an IC50 value of 0.72 μM in the serum-free conditions but less potently in the presence of 0.5% hiFBS (IC50 = 9.52 μM) (vide infra). Nepodin from R. crispus inhibited epimastigote proliferation (IC50 = 28.7 ± 13.3 μM) and inhibited the parasite release in the infection assay (approximately 30% inhibition at 5 μM) (Table 3).

Taking the anthraquinone phylobioactive hotspot (Figure 4) and the strong ethnopharmacological consensus for S. chloroclada (vide supra) into consideration, we did a preliminary profiling of the anthraquinone content of this taxon. The presence of the anthraquinones emodin, aloe-emodin, and chrysophanol was confirmed by thin-layer chromatography (TLC) using reference compounds and Borntrager's test, as well as with an electrospray ionization mass spectrometry (ESI-MS) scan of the S. chloroclada aerial part extract (Figure S2).

Impact of 9,10-anthracenedione substitutions on antichagasic effects in vitro

Emodin (2), representative of the anthraquinone cluster, moderately inhibited T. cruzi epimastigote growth but significantly inhibited parasite release in the cellular infection assay (IC50 = 0.72 μM) without being cytotoxic to host cells up to 100 μM (selectivity index [SI] >7 (Table 5)). As shown below, emodin was less potent in the 0.5% hiFBS (low ROS) host cell culture conditions. In order to reduce the variability of the DNA staining and occasional cellular debris and to have a more versatile assay allowing tracking of amastigotes in follow-up experiments, we generated a green fluorescent protein (GFP)-expressing T. cruzi (Y) strain (see transparent methods). As shown in Figure 6A, the number of released trypomastigotes quantified was comparable to the wild type (Figure 3A). Moreover, the IC50 values obtained for benznidazole with the GFP-expressing parasites were not different from those of the SYTO9 staining on wild-type parasites. Intriguingly, benznidazole was significantly less potent to inhibit parasite release from host cells with high ROS, cultured without fetal bovine serum (FBS) (IC50 = 7.5 μM [95% confidence interval {CI} = 6.2–9.0 μM]) as compared to host cells cultured in 0.5% hiFBS with lower ROS (IC50 = 1.4 μM [95% CI = 1.1–1.6 μM]) (Figure 6B).

Table 5

In vitro antiproliferative activity of anthraquinones on T. cruzi epimastigotes (72 hr) and GFP-expressing trypomastigotes release (6 dpi)

ID	IC50 epimastigotesa [μM]	Percentage inhibition of parasite releaseb at 5 μM (low ROS)	CC50 CHO cells [μM]	Selectivity Index (CC50 CHO/IC50 Epi)
1	>50	22.2 ± 9.6	>100	n.d
2	14.1 ± 8.1	24.8 ± 12.9	>100	>7
3	>50	48.3 ± 18.1	59.5 ± 10.1	n.d.
4	>50	0	n.d.	n.d.
5	>50	0	n.d.	n.d.
6	>50	68.5 ± 19.2	38.1 ± 16.0	<1
7	>50	0	n.d.	n.d.
8	5.3 ± 2.3	0	n.d.	n.d.
9	>50	0	n.d.	n.d.
10	6.6 ± 3.5	22.6 ± 17.2	35.7 ± 7.4	5.4
11	14.5 ± 5.6	44.0 ± 17.6	42.0 ± 15.5	2.9
12	6.4 ± 3.7	0	n.d.	n.d.
13	>50	0	n.d.	n.d.
14	>50	20.9 ± 9.7	88.9 ± 8.9	n.d.
15	>50	0	n.d.	n.d.
16	>50	0	n.d.	n.d.
17	32.9 ± 13.9	46.8 ± 25.6	>100	>3
18	>50	0	n.d.	n.d.
BZN	13.8 ± 2.9	84.9 ± 4.8	>100	>7

Compounds that inhibited >20% parasite release in the trypomastigote infection assay at a single concentration of 5 μM when cultured with 0.5% hiFBS were tested for antiproliferative effects on CHO host cells. Data shown are mean values ±SD of at least three independent experiments, each performed in triplicates. n.d., not determined; BZN, benznidazole.

IC50 values of the compounds against epimastigotes were assessed as described in the transparent methods.

The percentage inhibition of parasite release by the compounds was estimated at a single concentration (5 μM) as described in transparent methods.

Figure 6

FACS parasite release assay using the GFP-expressing T. cruzi strain and reduced potency of benznidazole in serum-free culture conditions

(A) Representative FACS histograms of the FL1 channel (488/530 nm) showing parasites released into the medium from host cells in 0.5% hiFBS cells infected with GFP-expressing trypomastigotes. The samples were prepared by fixing the released trypomastigotes in 4% paraformaldehyde as described in transparent methods. The vehicle control, benznidazole (BZN) treatment at 20 μM, and no infection control are shown. Data show mean values ± SD of at least 6 independent experiments.

(B) Dose-dependent inhibition of parasite release by BZN under low and high ROS conditions. Data show mean values ±SD of at least three independent experiments performed in triplicate. See also Figure S3.

Using the GFP T. cruzi assay, we performed a preliminary SAR study on the 9,10-anthracenedione (1) scaffold (Table 4). The anthraquinones were screened for general cytotoxicity on CHO cells, epimastigote proliferation, and trypomastigote release from infected CHO cells (Table 5). For the T. cruzi parasite release assay, GFP-expressing trypomastigotes were used to infect CHO cells cultured with 0.5% hiFBS in the medium. Under these low ROS conditions, the IC50 of emodin was found to be significantly less potent than in the serum-free (high ROS) conditions used in the screening, namely 9.5 μM (95% CI = 8.2–10.9 μM) as compared to 0.7 μM (95% CI = 0.5–1.0 μM) (Figure 7A). Thus, emodin behaved in the opposite manner as benznidazole and was more than ten times less active in the low ROS host cell conditions (0.5% hiFBS). In the low ROS conditions, all active hydroxylated anthraquinones behaved in a similar manner showing in vitro antichagasic IC50 between 5 and 10 μM (Figure 7B).

Table 4

Chemical structures of natural and synthetic anthraquinones tested on T. cruzi

ID		R1	R2	R3	R4	R5	R6	R7	R8
1	Anthraquinone	H	H	H	H	H	H	H	H
2	Emodin	OH	H	OH	H	H	CH3	H	OH
3	Purpurin	OH	OH	H	OH	H	H	H	H
4	Alizarin	OH	OH	H	H	H	H	H	H
5	Alizarin Red S	OH	OH	SO3Na	H	H	H	H	H
6	Quinizarin	OH	H	H	OH	H	H	H	H
7	Anthrarufin	OH	H	H	H	OH	H	H	H
8	Dantron	OH	H	H	H	H	H	H	OH
9	Anthraflavic acid	H	OH	H	H	H	OH	H	H
10	Chrysophanol	OH	H	CH3	H	H	H	H	OH
11	Aloe-emodin	OH	H	CH2OH	H	H	H	H	OH
12	Physcion	OH	H	OCH3	H	H	CH3	H	OH
13	2-Hydroxy-3-methylanthraquinone	H	OH	CH3	H	H	H	H	H
14	2-Hydroxy-1-methylanthraquinone	CH3	OH	H	H	H	H	H	H
15	Rhein	H	COOH	H	OH	OH	H	H	H
16	Diacerein	H	COOH	H	O(C=O)CH3	O(C=O)CH3	H	H	H
17	Disperse Red11	NH2	OCH3	H	NH2	H	H	H	H
18	Aurantio-obtusin	OH	OCH3	OH	H	H	CH3	OH	OCH3

Figure 7

Emodin potency is dependent on serum and host cell ROS concentration

(A) Dose-dependent inhibition of parasite release by emodin under serum and no serum (low and high ROS) conditions.

(B) Dose-dependent inhibition of parasite release by hydroxyanthraquinones which inhibited more than 25% release at 5 μM using 0.5% hiFBS (low ROS): purpurin, aloe-emodin, quinizarin, and disperse Red11. Data represent the average ± SD of three independent experiments each performed in triplicate.

Our profiling data indicate that the position of the hydroxyl groups was important for the antichagasic activity in vitro, with the trihydroxy-substituted derivatives being the most active (Table 5). The canonical anthraquinone (1) was ineffective against epimastigotes and only marginally active against cellular parasite release at 5 μM. Hydroxyl groups at R1/R2 (4, 5) abolished this activity. Hydroxyl groups at R2/R6 (9) did not improve the activity of 1 in the infection assay. Likewise, the additional hydroxymethyl moiety at R3 found in aloe-emodin (11) increased the general cytotoxicity and the compound was found to have a selectivity index (SI) of 2.9.

Interestingly, hydroxylation at R1/R8 showed a trend toward a general increase in toxicity toward epimastigotes (2, 8, 10-12). The carboxylic acid at R2 abolished the antichagasic activity as indicated by the lack of activity of rhein (15) and the clinically used antirheumatic prodrug diacerein (16). The replacement of hydroxyl groups with methoxy and methyl groups generally led to decreased activity, as shown by compounds 12 and 18. Thus, 9,10-anthracenedione with hydroxyl groups at positions R1 and R3 or R4, as exemplified by compounds 2, 3, and 6, proved to be most active in the infection assay but did not inhibit proliferation of epimastigotes. We performed dose-response experiments for compounds that inhibited the parasite release by more than 25%, i.e., 3, 6, 11, and 17 (Figure 7B). We also determined the long-term cytotoxicity (72-hr incubation) toward host cells for the anthraquinones that effectively inhibited T. cruzi parasite release in CHO cells (1, 2, 3, 6, 9, 10, 11, 14, 17). Noteworthy, all anthraquinones were cytotoxic only at high micromolar concentrations, with the exception of quinizarin, which showed a CC50 value of 38 ± 16 μM. In this study, emodin (2) was the most active antichagasic 9,10-anthracenedione in all stages of T. cruzi in vitro, followed by aloe-emodin (11), purpurin (3), and quinizarin (6).

Discussion

Biological profiling of plant extracts and assessment of the ethno-directed approach

In the present study, we document the current knowledge about botanical drugs used to manage CD by the indigenous peoples Ayoreo, Chiquitano, Izoceño-Guaraní, and Quechua in Bolivia. A major aim was to use this information to validate the antichagasic ethnomedical plant resources present among the indigenous groups most affected by CD in Bolivia. Likewise, the multidimensional profiling allowed bioprospecting of anti-CD botanical drugs to inspire ethnopharmacologically driven drug discovery. Previous studies reported superior antimicrobial properties of extracts derived from botanical drugs selected based on popular uses related to microbial infections (Khafagi and Dewedar, 2000; Svetaz et al., 2010; Silva et al., 2013). Such studies lend credit to the ethno-directed approach in bioprospecting for specific bioactive metabolites. In order to challenge this approach, we tested a total of 775 EtOAc extracts from two independent botanical drug libraries generated from 79 taxa selected for their reported use against symptoms of CD in Bolivia and from 389 taxa described in DMM. Our findings indicate that the CD botanical drug library contains a significantly higher percentage of cytotoxic plant taxa. However, hit rates for selective antichagasic plant extracts in the two libraries were not significantly different. The overall higher hit rate of the CD library was possibly due to non-specific cytotoxic effects, which could be conditioned by the ecological factors prevailing in the Chaco and inter-Andean valleys. The extreme atmospheric and ecological conditions (altitude, extreme dryness, and high-temperature ranges) in these regions may favor the production of metabolites with broad-spectrum toxic or general antifeedant properties.

Among the indigenous peoples participating in this study, the biomedical concept of CD was only recently introduced and did not match with any existing traditional disease concept. This seems related to the fact that infection with T. cruzi shows a diffuse and varied disease pattern and is asymptomatic in most cases. We found that most botanical drug preparations intended for oral administration (aqueous decoctions and infusions) were used to relieve symptoms associated with the chronic phase of CD, such as cardiac complications, fever, and fatigue, and not for combatting the (invisible) parasites. In fact, most of the botanical drugs applied for CD-related symptoms were also used for other therapeutic purposes involving inflammatory conditions (Table 2). We thus conclude that tangible ethnomedical concepts about CD were absent until recently and developed only during the last decades, which is in agreement with the cultural perception of T. cruzi vectors (Salm and Gertsch, 2019). This clearly hampers the application and selection of botanical drugs targeting T. cruzi parasitemia and its symptoms and probably explains in part why the majority (>80%) of the extracts derived from botanical drugs with reported use against CD and its symptoms were not active. The use of plant- and animal-based traditional medicine among the Chiquitano, Izoceño-Guaraní, and Quechua was widespread and in agreement with previous reports that these people widely use traditional medicine despite the presence of Western health care (Quiroga and Arrázola, 2013; Vandebroek et al., 2008). Noteworthy, the Ayoreo did not treat CD at all, which agrees with their overall minimal use of herbal medicine. Chiquitano, Izoceño-Guaraní, and Quechua research participants stated that they tried to manage CD with plant-based remedies because Western health care was limited and chemotherapy not accessible during the chronic stage of CD. Another study from a different region in Bolivia reported a similar situation (Forsyth, 2017). Asteraceae was the most dominant family of plants used for CD, likely due to their overall abundance and species richness. In general, Asteraceae is over-proportionally represented in medical florae (Heinrich et al., 1998; Moerman et al., 1999; Thomas et al., 2009), and this may be linked to the high diversity of bioactive secondary metabolites in the family (Heinrich et al., 1998). Quechua informants showed a low consensus regarding the species to be used in the treatment of CD, with only one species mentioned three times. A higher consensus was found among the Chiquitano and Izoceño-Guaraní participants (estimated Trotter and Logan Informants' consensus: Fic >0.8), with relatively few species being used by a large proportion of participants. Among the Izoceño-Guaraní, S. chloroclada (referred to as lanza lanza, mbuijare, or retama) was clearly the most important species for treating CD, with a share of 48% of total use reports. Interestingly, S. chloroclada belonged to a phylogenetic cluster that showed no or little inhibition of epimastigote proliferation in the pre-screening (Figure 2) but exhibited significant in vitro antichagasic effects in the parasite release assay (79.5% inhibition at 15 μg/mL by EtOAc extract of the flower). This discrepancy was also observed with some anthraquinones (purpurin and quinizarin). Since the extract likely contains glycosides that may be hydrolyzed by CHO cells but not by epimastigotes, we cannot exclude metabolic changes induced by host cells. The genus Senna (syn. Cassia) is known to contain anthraquinone, dianthrone, and naphthol glycosides (Dave and Ledwani, 2012). A preliminary ESI-MS scan and TLC confirmed the presence of anthraquinones in this botanical drug (Figure S2). It may not be a coincidence that the anthraquinone phylobioactive hotspot includes S. chloroclada which was one of the most frequently used taxa for CD. In fact, S. chloroclada flowers and roots are the only botanical drug repetitively stated to treat CD among the Izoceño-Guaraní. To date, no in-depth phytochemical study is available on S. chloroclada, and follow-up studies are planned. T. aurea bark, as its Spanish vernacular name “paratodo” indicates, was used for numerous diseases, in agreement with a previous study (Hajdu and Hohmann, 2012). The introduced species A. zerumbet and C. citratus are well known in South America for the treatment of cardiovascular diseases (Hajdu and Hohmann, 2012; Lim, 2016; Luz et al., 1984; Moreira et al., 2010). An apparently specific medicinal indication for CD was also reported for Aloysia citrodora, Baccharis genistelloides, Bixa orellana, Dysphania ambrosioides, Handroanthus impetiginosus, L. usitatissimum, Plantago major, Ruta chalepensis, S. junceum, and Schinus molle (Alonso, 2000; Bastien, 1998; Bourdy et al., 2004; Forsyth, 2017; Grandi et al., 1989; Martínez et al., 2004; Martino, 2012; Quiroga et al., 2012; Report, 2002); . None of the EtOAc extracts obtained from botanical drugs of these species were significantly and selectively toxic against T. cruzi in our infection assay with the exception of S. chloroclada (flowers) inhibiting parasite release >50% in vitro at 15 μg/mL (Table S1) but not showing any effect on epimastigotes.

The extract obtained from the aerial parts of A. buniifolius had an SI > 20 and was the other noteworthy hit obtained with the CD-informed library. It fully inhibited epimastigote proliferation at 25 μg/mL and parasite release at 15 μg/mL. Its antichagasic and antileishmanial activity was previously reported for plant material collected in Argentina, and the flavonoid santin was thought to be the active compound (Sülsen et al., 2007). However, santin was only moderately active against T. cruzi epimastigotes and trypomastigotes (IC50 values >30 μM). Certainly, the presence of additional antichagasic metabolites should not be excluded, and an in-depth phytochemical investigation of A. buniifolius is warranted. The extract of seeds of a Pterodon sp. showed significant and specific antitrypanosomal effects. Whether the very common diterpene alcohol geranylgeraniol, previously identified as the major antichagasic component in Pterodon pubescens seeds (Menna-Barreto et al., 2008), is responsible for this specific antitrypanosomal activity remains to be clarified.

The screening of the DMM extract library resulted in 23 extracts with selective parasite toxicity in the T. cruzi release assay. Of these extracts, eight were from the Apiaceae family and seven from the Asteraceae family showing taxonomic parallels with the CD-informed library. Asteraceae is the largest plant family and characterized by the presence of sesquiterpene lactones (SLs) several of which known to have antiprotozoal properties (Schmidt et al., 2009; Sepúlveda-Robles et al., 2019) (vide infra). It is possible that the use of SL containing botanical drugs could therefore represent an ethnopharmacological strategy to reduce parasitemia in CD. The 660 extracts representing the DMM library were obtained from 389 different plant species. The largest part is from the Mediterranean basin, but central European species and exotic herbal drugs imported from Africa, Arabia, Central Asia, Himalaya, and the Indo-Malayan region are also included. The families with the highest share of botanical drugs and species are Apiaceae with 69 botanical drugs from 37 spp., Asteraceae with 51 botanical drugs from 33 spp., Rosaceae with 37 botanical drugs from 18 spp., Lamiaceae with 33 botanical drugs from 26 spp., and Fabaceae with 29 botanical drugs from 22 spp. This pattern reflects the overall taxonomic composition of DMM for which a total of 536 plant taxa representing 924 botanical drugs were identified and recommended for 5314 medical applications (Staub et al., 2016). The medical categories with the most therapeutic uses are dermatology (1216), gastroenterology (805), gynecology (615), urology (437), respiratory system (374), and neurology (269). The most frequently mentioned parasite treatments are related to lice, scabies, and tapeworms but also applications for malaria causing infections with Plasmodium were described (tertian and quartan fever). This includes parts of Anchusa sp., seed and leaves of Bituminaria bituminosa, root of Dipsacus fullonum, seeds of Heliotropium europaeum, root of Plantago sp., herb of Potentilla reptans, and the herb of Verbena officinalis (Staub et al., 2016). With respect to the overall Mediterranean flora, the Apiaceae and Rosaceae appear to be overrepresented in this library, while the frequency of Asteraceae and Lamiaceae appears to be rather consistent with the overall species diversity and abundance. Apiaceae fruits were frequently used as antidotes and their resins for neurological and musculoskeletal problems. Considering the size of plant families, the Fabaceae seem underrepresented, while Poaceae, Caryophyllaceae, and Orchidaceae are clearly underrepresented in the DMM library (Table S2 Staub et al., 2016; Tutin, 1978). Concerning the treatment of fevers and parasites, the active extracts from plants belonging to the “coumarin cluster” such as the herb of rue (R. chalepensis) were recommended in DMM for the internal use of the treatment of tremor and shivering before fever attacks occur, while the seeds of Seseli tortuosum are suggested against fevers in general. The hits belonging to the “anthraquinone cluster” such as the roots of Rumex species and rhubarb (Rheum spp.) were recommended for the treatments of scabies and fevers, respectively. For L. officinale, L. nobilis, and S. sisarum, no uses related to fever and parasites are recorded in DMM (Staub et al., 2016).

Microfractionation and isolation of bioactive principles of selected antichagasic plant taxa from the DMM library

For the isolation of active compounds from the DMM library, we selected extracts pertaining to three different phylogenetic groups, with taxa whose plant material was readily accessible. A major phylogenetic hotspot showing apparent antitrypanosomal selectivity was the anthraquinone cluster with the rhizomes of R. crispus (curly dock) and R. rhaponticum (rhapontic rhubarb) from the Polygonaceae family. The anthraquinones emodin and chrysophanol were identified in R. crispus and served as a basis for the subsequent preliminary SAR study (vide infra). Since the naphthoquinone derivative nepodin with known antimalarial activity (Alonso, 2000) was isolated together with anthraquinones from R. crispus and likewise inhibited epimastigote proliferation, the potency of the extract may reflect additive effects. In general, naphthoquinones have been studied extensively in T. cruzi epimastigotes (Salas, 2011; Ventura Pinto and Lisboa de Castro, 2009) without leading to a successful translation to clinical trials (Sales Junior et al., 2017). Extracts with antitrypanosomal activity from plants with distinct phylogenetic positions were those from L. nobilis (bay) (Lauraceae) and S. sisarum (skirret; Apiaceae). L. nobilis leaves contain a range of SLs, which may explain the significant antichagasic effects of its extract in epimastigotes. Numerous studies have addressed the selective versus non-selective antitrypanosomal effects of SLs, which are also widely present in Asteraceae (Kimani et al., 2018; Moraes Neto et al., 2019; Muschietti and Ulloa, 2016). SLs like cynaropicrin and others can act as electrophiles and form adducts with biological nucleophiles, such as trypanothione, the parasitic equivalent of glutathione in mammalian cells (Zimmermann et al., 2013, Zimmermann et al., 2014). The reason why L. nobilis (leaves) was ineffective in the infection assay could be due to the SLs reacting with thiols in host cells, e.g., glutathione, without reaching the parasite. It has been shown that CHO cells can produce glutathione upon stress (Orellana et al., 2015). Thus, SLs undergoing a Michael-type addition with thiols are likely poorly bioavailable to infected tissues as they are detoxified by glutathione. Although the polyacetylene falcarindiol from S. sisarum had no inhibitory effect on T. cruzi, we cannot exclude the possibility that other polyacetylenes present in S. sisarum root may be more potent as indicated by the activity profile of the extract.

SAR study of anthraquinones as antichagasic natural products in vitro and importance of ROS in host cells

Anthraquinones are condensed aromatic hydrocarbons found in different plant species known for their medical and dye applications (Malik and Müller, 2016). In Western pharmacopeias, anthraquinone containing botanical drugs such as Rhamnus spp. (Frangulae cortex and Rhamni purshiani cortex) or Rheum officinale (Rhei radix) are used as laxatives. To elaborate on our natural product drug discovery approach, the most active antichagasic secondary metabolite emodin (1,3,8-trihydroxy-6-methylanthracene-9, 10-dione) led us to explore the SAR of differentially substituted 9,10-anthracenediones. Trihydroxylated anthraquinones have already been shown to be trypanocidal (vide infra). However, the present work provides a preliminary SAR study on 9,10-anthracenediones for both T. cruzi epimastigote and parasite release from trypomastigote-infected CHO cells. Different anthraquinones have been shown to exert moderate to good antimalarial, antibacterial, and antiviral effects in vitro at low micromolar concentrations (Li and Jiang, 2018). In our study, emodin showed specific inhibition of T. cruzi parasite release in CHO cells in the nanomolar or low micromolar range, depending on host cell culture condition. Emodin has been tested previously only on epimastigotes at high micromolar concentrations (De Lima et al., 2017) and as weak inhibitor of casein kinase 1 with an IC50 value of 130 μM (Justiniano et al., 2014). Other reported effects of emodin include anti-inflammatory, antiosteoporosis, anti-cardiovascular disease, and antidepressant effects (Li and Jiang, 2018). Purpurin, a natural tri-hydroxylated anthraquinone, inhibited T. cruzi parasite release and showed a selectivity index of 7. Rather unexpectedly, it did not inhibit epimastigote proliferation, which may be due to differences in pH between lysosomes and culture medium related to the trypanosomal uptake of anthraquinones. Purpurin has previously been shown to inhibit blood stream trypomastigotes (De Castro, Pinto and Pinto, 1994) but again has not been studied in cellular T. cruzi infection assays. The 9,10-anthracenedione scaffold is likely to interfere with the parasite redox system as anthraquinones can mediate the production of hydrogen peroxide or ROS via oxygen reduction in situ (Campos-Martin et al., 2006). This is a feasible antichagasic mechanism as anthraquinones have been shown to interfere with redox reactions in cells (Okumura et al., 2019). ROS generation is also the postulated mode of action of the approved antichagasic drug benznidazole (Pedrosa et al., 2001). Interestingly, based on the T. cruzi parasite release assay conducted using different serum conditions in combination with differential ROS amounts in host cells, it was evident that the generation of ROS plays a key role in the mechanisms of action of benznidazole and emodin with opposite effects. The IC50 value of benznidazole reduced 5-fold in the presence of only 0.5% hiFBS, whereas the IC50 of emodin increased by 13-fold, probably as a function of increased ROS. This striking difference could have translational implications for CD treatment, for instance, in the context of the lack of efficacy of benznidazole against residual amastigotes in advanced CD stages. Moreover, benznidazole emodin combinations may be tested for potential synergistic antichagasic effects in vivo. However, as reviewed recently, emodin shows relatively low oral bioavailability in rodents (5–10%) (Li et al., 2020). As some anthraquinones did not inhibit T. cruzi epimastigote proliferation in our study (Table 5), their antichagasic effects in the infection/parasite release assays may be due to host cell ROS generation induced by these compounds. Anthraquinones are primarily present in their glycosylated form in plants. Since anthraquinone glycosides can be deglycosylated and reduced to anthrones and anthranols by gut bacteria (Hattori et al., 1988), resulting in potent laxative effects, the systemic application of these compounds can be challenging. However, as exemplified by the clinical pharmacokinetics on diacerein (prodrug of rhein), low micromolar plasma concentrations can be achieved (Nicolas et al., 1998). Unfortunately, rhein showed no antichagasic effects in our assays, possibly due to the carboxylic acid at C2. It is noteworthy that the most important botanical drug used in the context of CD among the Guaraní, S. chloroclada, also contains anthraquinone aglyca, thus potentially rendering the anthraquinone cluster as relevant in the ethnomedical treatment of CD.

Overall, the comparative phylobioactivity-guided screening for in vitro antichagasic activity is an enabling tool to investigate plant-based ethnomedical resources and for the discovery of natural products that may have potential in CD drug development. Our study demonstrates the potential of plant secondary metabolites as prospective antichagasic principles in a comparative matrix, linking ethnomedical information with phylogenetics and chemotaxonomy. The comparative profiling also facilitated the pharmacological validation of the extract library from botanical drugs currently used in the context of CD in Bolivia, thus challenging the ethno-directed CD bioprospecting approach. We did not find statistical evidence that the CD botanical drug extract library yielded more hits than the DMM drug extract library. Nevertheless, our study led to the identification of significant antichagasic phylogenetic hotspots in the plant kingdom that may serve as a basis for future phytochemical investigations and antichagasic bioprospection.

Limitations of the study

Polar secondary metabolites like glycosides, polymers, or polyphenolics potentially present in the botanical drugs but generally showing poor systemic bioavailability were not present in the EtOAc extracts. In addition, the potential additive or antagonistic effects of natural products present in the plant extracts were not investigated. The in vitro profiling may give false-negative results as secondary metabolites are metabolized in vivo and can serve as inactive pro-drugs. We assessed the activity of the compounds and extracts using a validated FACS-assisted T. cruzi parasite release assay, which allowed us to perform relative quantifications of parasite load using appropriate controls. However, the method did not discriminate between amastigote replication and the release of the parasites from the host cells.

Resource availability

Lead contact

Further information and requests for resources should be addressed to and will be fulfilled by the lead contact Jürg Gertsch (gertsch@ibmm.unibe.ch).

Material availability

This study did not generate any unique reagents.

Data and code availability

The data supporting the findings of this study are available within the paper and its supplemental information.

Methods

All methods can be found in the accompanying transparent methods supplemental file.