Solvents used for extractions were ethanol and methanol (>99% purity; VWR France), and ultrapure water prepared with an Integral 5 Milli-Q system (Merck Millipore). Ethyl acetate (Fisher), acetic acid and formic acid (Carlo Erba) were used for TLC.

Dulbecco’s modified Eagle’s medium (DMEM), Opti-MEM, phosphate buffered saline (PBS), glutamax-I, and fetal bovine serum were purchased from Invitrogen (Carlsbad, CA). 4′,6-Diamidino-2-phenylindole (DAPI) was from Molecular Probes (Thermo Fischer Scientific, Waltham, United States).

Delphinidin chloride was from Extrasynthèse (Lyon, France) and was >96% pure. Boceprevir was kindly provided by Philippe Halfon (Hôpital Européen, Laboratoire Alphabio, Marseille, France). Procyanidins A1, A2, B1, B2 and B3 were >96% pure and from Phytolab (Germany). All these compounds were tested on Huh-7 cells. In addition, several other compounds were used as standards for TLC or UPLC: quercetin and quinin were from Merck (Germany), gallic acid, stigmasterol and glycyrrhetic acid were from Extrasynthése (France), (−)-epicatechin, (+)-catechin, aloin and ellagic acid were from Sigma (Germany), (−)-EGC was from Chromadex (United States), (−)-EGCG was from Selleckchem (United States).

The anti-E1 monoclonal antibody (A4) (Dubuisson et al., 1994) was produced in vitro by using MiniPerm apparatus (Heraeus, Hanau, Germany). Cy3-conjugated goat anti-mouse IgG was from Jackson Immunoresearch (West Grove, PA, United States).

The ethnobotanical survey was conducted in the Bafing region, which is situated in North-West Côte d’Ivoire and corresponds to the Touba department. It covers a total land area of 8,650 km² and shares boundaries with the Odienné department in the North and the Séguéla and Biankouma departments in the East and the South respectively. In the West, it shares boundaries with the Republic of Guinea. It is one of the most distant Ivorian regions from the capital city Abidjan. Therefore, people are still more dependent on traditional medicine compared to people living closer to cities. In this region, traditional healers could be assigned in at least one of these 3 groups: 1) “Bouêti”: they are plain healers, who know medicinal plants and treat common illnesses; they are the first medical resort in this region; 2) “Kamon” or marabout: they heal with natural products according to the Coran; 3) “Komanti”: they heal with natural products and talismans. Traditional healers could also belong to two of these groups: “Kamon-Bouêti” and “Komanti-Bouêti”. Seven of them (5 “Bouêti”, 1 “Komanti-Bouêti” and 1 “Kamon-Bouêti”) were met three times during one month of non-structured interviews, in the city of Touba, the Mahanan village and the Ouintoulo camp. Religion in Côte d’Ivoire is diverse and the main religions are Islam, Christianity (primarily Roman Catholic) and various indigenous religions. The Survey considered the use of plants against a group of diseases called “yellow malaria”, which is a rough translation for diseases that cause icterus. This nosological category include malaria but also other illness characterized by symptomatic jaundice as hepatitis, obstruction of the bile duct, hemolysis. Traditional healers were asked about the illness they cured with plants, the name of these plants, the part used, how they prepared them and the mode of application.

The 15 plants were collected in the wild during the morning. They were then cleaned and air-dried at constant temperature (26°C) for 1–2 weeks at the Nangui Abrogoua University (Abidjan). They were authenticated by a botanist at the Centre National de Floristique (CNF), University of Félix Houphouët Boigny de Cocody (Abidjan), by comparing their morphological characters to a flora. Voucher specimens were deposited in an Herbarium at the CNF. The voucher numbers and the traditional uses of the investigated species are presented in Table 1. Their names have been checked with “The World Flora Online” (http://www.worldfloraonline.org/- accessed October, 2021). Each plant was then powdered and stored in the dark until extractions. For each plant, 20 g of dried powder were mixed with 100 ml methanol for 24 h. After filtration, the grounds were extracted again twice in the same way. The 3 resulting filtrates were combined and dried under vacuum at 40°C.

C. procera and P. laxiflora were also extracted with water-ethanol and water-methanol mixtures (0–100%) as solvent. For each plant, extractions were carried out in 100 ml flasks with 2 g of dried powder and 40 ml of solvent. Flasks were placed for 2 h in a shaking incubator (Multitron II Infors HT) in order to maintain agitation at 160 rpm and temperature at 20°C. These extractions were carried out in triplicates. Final extracts were then centrifuged for 10 min at 10,000 tr. min⁻¹ and 4°C and the supernatants were stored at −18°C in the dark until use. 10 ml of supernatant were used to determine dry mass yields (measurements made in duplicate, each one with 5 ml of supernatant) with an infrared moisture analyzer (Precisa XM60). The remaining volume of supernatant was freeze-dried (freezedryer Heto PowerDry PL9000 Thermo Fisher Scientific) so as to get dehydrated extract for further analysis. Evaporation under vacuum was carried out prior to freeze-drying in the case of supernatants containing ethanol or methanol (rotary evaporator Heidolph instrument, GmbH and Co. KG).

All the dried crude extracts or isolated compounds were resuspended in DMSO before antiviral assays. Plant extracts were dissolved in DMSO to reach a final concentration of 25 mg/ml. Delphinidin and boceprevir were also resuspended in DMSO at a final concentration of 100 mM. Serial dilutions in DMEM cell culture medium were then performed to reach the final concentration mentioned for each experiment.

Phytochemical screening was performed with thin layer chromatography using pre-coated silica gel plates (Alugram Xtra Sil G/UV254, Machery-Nagel). The solvent system was EtOAc/Acetic acid/Formic acid/H₂O (100:11:11:20 v/v). For each extract, 10 µL (5 mg/ml) were spotted on the TLC plate. The same revealers as in Santos et al. (Santos et al., 2018) were used, except for total and condensed tannins, for which ferric chloride (Oliveira et al., 2015) and DMACA (Lea, 1978) (Sigma-Aldrich) were used. Folin-Ciocalteu reagent (VWR chemicals) was also used to reveal all the polyphenols. The following standards were used as controls on TLC plates: quercetin, (+)-catechin, gallic acid, aloin, stigmasterol, glycyrrhetic acid and quinin.

To quantify tannins and other phenolics in our extracts, Wiesneth and Jürgenliemk, (2017) that is based on the Folin-Ciocalteu method of the European Pharmacopeia (2.8.14) was adapted. Briefly, 1.5 ml of water was first added to 5 mg of each extract to dissolve them in water bath (95°C) during 30 min. These were centrifuged at 2,500 rpm for 5 min, and the supernatants were the “total phenolic solutions” (TPS). For each TPS, 10 mg hide powder was added to an aliquot of 0.5 ml. These preparations were stirred and further incubated without light for 60 min while shaking. The supernatants after centrifugation (2,500 rpm, 5 min) resulted in the solutions containing phenols without tannins (TPS-T). The removal of tannins was checked with TLC revealed by the DMACA reagent by comparing TPS (blue spots) and TPS-T (no spot).

For total phenolics and tannins quantification, Wiesneth and Jürgenliemk microplate method was used. Briefly, except for the outer wells filled with water, 20 µL of sample (TPS, TPS-T), or gallic acid (Extrasynthese, France) for calibration curve were added in each well. Then 100 µL of a 1/10th dilution of Folin-Ciocalteu’s reagent was added and shaken without light for four minutes. Finally, 80 µL of a 10.6%-solution of Na₂CO₃ anhydrous in water was added to the samples. Measurement of absorbance at 690 nm at room temperature was achieved after 90 min using a Spectrostar Nano (BMG Labtech) spectrophotometer.

Ultra-High Performance Liquid Chromatography (UPLC) analyses were performed on an Acquity UPLC H-Class system (Waters, Guyancourt, France) coupled with a Diode Array Detector (DAD) and a QDa ESI-Quadrupole Mass Spectrometer. All data were acquired and processed using Empower 3 software. Separation was achieved using an ACQUITY UPLC^® BEH phenyl 1.7 µm (2.1 × 100 mm) column (Waters, Milford MA). Gradient elution was performed with (A) 0.1% formic acid in water and (B) 0.1% formic acid in acetonitrile at a flowrate of 0.3 ml/min. The amount of B was increased from 5% for the initial 1 min to 30% at 9 min, and to 100% at 9.5 min. The condition was then kept at 100% B until 10.5 min. Finally, the column was re-equilibrated to 5% B in 0.5 min and maintained for an additional 2 min, resulting in a total run time of 13 min. Column oven and auto-sampler were set at 30°C and 15°C respectively, and 4 µL of each sample (1 mg/ml in MeOH/water) were injected. Analytes were monitored using UV detection (l range: 190–790 nm, resolution: 1.2 nm) and MS-Scan from 100 to 1,000 Da (both in positive and negative mode). Cone voltage was set at 15 V. Probe temperature was 600°C. Capillary voltage was 0.8 kV. The main tannins and anthocyanins were identified based on the retention time of the commercial standards and their mass spectra (Supplementary Table S1).

During our ethnobotanical survey, 15 plants were identified as commonly used to treat “yellow malaria” and were thus selected (Table 1). Indeed, this symptom could correspond to icterus, which could be caused by hepatitis. Since HCV is quite common in Côte d’Ivoire, and can cause icterus in some cases, some of these 15 plants might have anti-HCV activity. However, not to mention other important considerations such as cultural aspects in traditional medicine, jaundice can be caused by other hepatitis (either viral or not) or many other diseases. For instance, malaria, which is widespread in Côte d’Ivoire, can cause hemolysis which in turn can lead to icterus. Other symptoms such as anemia can be observed and described in “white malaria” (Traore et al., 2021). In fact, many of the plants screened here for putative antiviral activity have been reported to be used against malaria and/or have shown antiplasmodial activity (Table 1). For instance, leave or bark decoction of Alstonia boonei De Wild. are used to treat malaria by Ehotile people in Côte d’Ivoire (Malan et al., 2015) and also in Nigeria (Ishola et al., 2014) and Cameroon (Tsabang et al., 2012). Anogeisus leiocarpa (DC.) Guill. and Perr. was quoted as one of the most used plant to treat malaria in an ethnobotanical survey in Mali (Diarra et al., 2015). Carapa procera DC. extracts had antiplasmodial activity in vitro (Sarpong et al., 2016), and in a more recent study, they had anti-malaria activity on mice (Koama, 2021). Other plant parts were also reported to be used against malaria or have shown antiplasmodial activity: Combretum collinum Fresen, Momordica charantia L., Paulinia pinnata L., Pericopsis laxiflora (Benth. ex Baker) Meeuwen, Rauvolfia vomitoria Afzel., Sarcocephalus latifolius (Sm.) E.A.Bruce and Terminalia macroptera Guill. and Perr. (Table 1). In our sustainable approach, only plant leaves were used for further studies, although other parts of the plants such as bark were traditionally used in some cases. For instance, the stem bark of Pericopsis laxiflora (Benth. ex Baker) Meeuwen was already reported to be traditionally used in Côte d’Ivoire, and the activity of various extracts were tested in vitro on various strains of Plasmodium (Koffi et al., 2020). The methanolic extract was the most active one. However, in Ghana, it was the decoction of P. laxiflora leave that was used against malaria (Asase et al., 2005). Although a putative antiviral activity may be due to different compounds and in different organs, our antiviral assays were carried out with leave extracts only.

Crude methanolic extracts of the 15 selected plants were prepared and tested in vitro on HCV infection (Figure 1). After removing methanol, the dried extracts were dissolved in DMSO at 25 mg/ml stock solution. Huh-7 cells were inoculated with HCV in the presence of the extracts at 25 μg/ml and infection was quantified using an immunofluorescent assay. DMSO was used as negative control, and delphinidin and boceprevir were used as positive controls. Indeed, delphinidin is a natural product with strong anti-HCV activity (Calland et al., 2015), and boceprevir is a commercial protease inhibitor used to treat hepatitis C. The cell number was also quantified at the end of the experiment to determine the toxicity of the extracts.

Among the 15 plants tested, Carapa procera DC. and Pericopsis laxiflora (Benth. ex Baker) Meeuwen extracts were generally the most active on HCV, with less than 10% infected cells when these were in contact with the corresponding crude extracts (25 μg/ml). In addition, these two extracts were not toxic on hepatic cells at those concentrations with a CC₅₀ of 203.3 μg/ml and 196.6 μg/ml at 24 h, and 93.3 μg/ml and 88.7 μg/ml at 72 h for C. procera and P. laxiflora, respectively (Figures 2A,B). A dose-response concentration experiment was performed to determine the half-median inhibitory concentration IC₅₀. A dose-response inhibitory activity was observed for the two extracts confirming their antiviral capacity against HCV. The IC₅₀ was estimated at 0.71 and 0.23 μg/ml for C. procera and P. laxiflora, respectively (Figures 2C,D).

The chemical composition of C. procera and P. laxiflora extracts were evaluated by using TLC and conventional reagents to reveal major groups of secondary metabolites. Both extracts were rich in polyphenols, including flavonoids and tannins (Table 2). This is qualitatively consistent with previous results showing that leave extracts of C. procera were rich in polyphenols including anthocyanins and flavonols (Adjé et al., 2019). Polyphenol were also reported in P. laxiflora bark, and to a lesser extend in leave aqueous extracts (Alhaji et al., 2014). However, none of our extracts contained alkaloids, contrary to a previous study revealing some in P. laxiflora leave (Fadipe et al., 2019). This might be due to different extraction protocols.

Since many tannins have antiviral activity, especially against HCV (Zuo et al., 2007; Calland et al., 2012b; Lin et al., 2013; Ajala et al., 2014; Liu et al., 2015; Chowdhury et al., 2018; Tamura et al., 2019), they were then specifically removed in our two most active extracts (C. procera and P. laxiflora) in order to check whether these “tannin-free” extracts were still active against HCV or not. Results are presented in Figure 3.

The IC₅₀ of crude extracts after heating were 1.89 and 0.53 μg/ml for C. procera and P. laxiflora, respectively. Although slightly higher, these were similar to the IC₅₀ of the same extracts before heating (Figure 2C,D): 0.71 and 0.23 μg/ml for C. procera and P. laxiflora, respectively. Some active molecules might have been lost during this process, either by heat degradation (95°C), or because they were less soluble in water and therefore lost in the pellet (see material and methods section).

After heating and tannins removal (w/o tannin), both extracts lost completely their activity, regardless of their concentration (Figures 3A,B). Therefore, tannins are most likely responsible for C. procera and P. laxiflora activity against HCV in vitro.

Our most active methanolic extracts were from C. procera and P. laxiflora. According to our ethnobotanical survey, C. procera and P. laxiflora leaves are traditionally used in water extracts. However, water may not be the best solvent to extract the active tannins of these plants. Acetone and alcoholic solvents as well as their mixtures with water have been commonly used to extract tannins from various vegetal sources (Wei et al., 2011; Metrouh-Amir et al., 2015). To avoid toxic solvents, water, ethanol and their mixtures were used here since they are considered as eco-friendly and traditionally used solvents. Hence, C. procera was extracted with hydro-alcoholic solvents containing 0–100% ethanol, and total polyphenols and tannins were quantified in these extracts. These results were compared to those obtained using similar water/methanol mixtures as solvents. The effect of these various hydro-alcoholic solvents was first studied on dried mass yields. Figure 4A shows that best yields were obtained with 50% alcohol (ethanol or methanol). The lowest yields were obtained with water or ethanol 100%. Moreover 50% ethanol extraction allowed a better yield than any (hydro-)methanolic extraction.

Then total polyphenols and tannins were quantified in these extracts (Figures 4B,C). For C. procera eco-extracts (Figure 4B), polyphenols and tannins contents increased when ethanol was added to water for extractions and reached a plateau from 50 to 100% ethanol. Quite similar solvent effect was reported for tannins extraction from Arum maculatum leaves (Farahmandfar et al., 2019). Since tannins are not very hydrophilic, the addition of ethanol in the solvent enhanced their extraction. Polyphenols and tannins contents were higher in these hydro-ethanolic extracts compared to methanolic extract. Similarly, for P. laxiflora (Figure 4C), the polyphenols and tannins contents were higher in the 50:50 hydro-ethanolic extract compared to methanolic and aqueous extracts.

Since the dried mass yields were better with the 50% hydro-alcoholic solvents compared to water or alcohol extractions (Figure 4), the ethanol/water (50:50) extracts were chosen for further antiviral assays. Figure 5 shows that for both C. procera and P. laxiflora, these crude hydro-ethanolic extracts were very active against HCV (IC₅₀ = 1.18 and 2.11 μg/ml, respectively), even after heating (IC₅₀ = 2.69 and 5.92 μg/ml, respectively). These results are similar compared to those of Figures 3A,B, suggesting that some active molecules might have been lost during the heating process before tannin removal. In addition, the ethanol/water (50:50) extract and crude methanolic extracts (Figure 3) had a similar anti-HCV activity, whereas the activity of the aqueous extracts was lower (Supplementary Figure S1). This is consistent with the tannin contents of all these extracts (Figures 4B,C). Moreover, the eco-extracts lost completely their activities after tannins removal, as the methanolic extracts did (Figure 3). Hence, the antiviral activities of C. procera and P. laxiflora are similar in eco-extracts compared to methanolic extracts, and they are also due to tannins. Those tannins are in much lower quantity in aqueous extracts that are less active against HCV. It is important to note that according to our survey, these plants were used in decoction, whereas our aqueous extracts were obtained at 20°C. Therefore, the compositions of these extracts were probably different. In this study our aim was to compare different solvents without heating.

The obtained results show that very active extracts could be obtained using eco-friendly solvents such as ethanol-water mixtures in a simple extraction and without heating or any further purification steps. Such extraction method respects the principles of the green extraction (Chemat et al., 2019) and could be easily applied for the preparation of antiviral extracts from C. procera and P. laxiflora.

Several extracts were analyzed by UPLC-MS especially for their tannins and anthocyanins that are known to have antiviral activities. For instance, epigallocatechin-3-gallate (EGCG) has been identified to inhibit HCV entry and cell-to-cell transmission (Ciesek et al., 2011; Calland et al., 2015). (−)-epicatechin-3-O-gallate (ECG) were found to inhibit NS3 protease in a study (Zuo et al., 2007), although it exhibited weak antiviral activity on HCV entry in another study (Calland et al., 2012a). Ellagic acid is also known to have anti-HCV activity (Reddy et al., 2014). Procyanidin B1, a dimer of (−)- epicatechin and (+)- catechin with a procyanidin type B structure, extracted from Cinnamomi cortex, could inhibit HCVpp entry in a dose-dependent manner (Li et al., 2010). More recently, a procyanidin with a type A structure was shown to inhibit HCV entry (Fauvelle et al., 2017). Results presented in Table 3 confirmed that all crude extracts contained tannins in various amounts, that were largely reduced—if not completely removed—after treating these samples with hide powder.

Some of these tannins or related molecules (precursors or degradation products) could contribute to the antiviral activity of these extracts. For instance, ellagic acid is present in all crude extracts, but absent after tannin removal. Conversely, other tannins (EGC and EGCG) or monomers (catechin and epicatechin) are either totally absent or present in all extracts of at least one plant, even after tannin removal. Thus, their contribution to the anti-HCV activity is unlikely here. More interestingly, at least traces of procyanidins A1 and A2 are present in most crude extracts, except procyanidin A2 which is absent in the C. procera aqueous extract (Table 3), but these compounds are not detected or in much smaller amounts after removing tannins with hide powder (“treated” extracts). Therefore, these tannins may explain the anti-HCV activity of P. laxiflora extracts at least, and to a lesser extend this of C. procera extracts, where those procyanidins are found in much smaller quantities. On the other hand, the activity of C. procera extracts might be due to tannins such as procyanidin B1, known for having effects on HCV replication (Li et al., 2010), and which is present in higher amounts for these crude extracts compared to those treated with hide powder. Similarly to the other dimers, procyanidin B2 was usually more concentrated in the most active (methanolic and hydro-ethanolic) extracts compared to the other ones (Table 3), and it has already shown antiviral activities on other viruses (Yang et al., 2014; Liu et al., 2018). Therefore, it might also contribute to the anti-HCV activity.

Hence, these four procyanidins were tested against HCV. Their structures are shown in Supplementary Table S2. Figure 6 shows that procyanidin A2 is clearly the most active one, which is consistent with above results (Figures 3, 6; Table 3). For instance, extracts of P. laxiflora are antiviral before tannin removal, but not after. However, procyanidin A2 cannot explain the anti-HCV activity in all extracts. Indeed, this molecule is in much lower (trace) amount in C. procera extracts compared to P. laxiflora extracts, although the former plant has anti-HCV activity as well. Together with ellagic acid, procyanidins A1 and B1 may contribute to the activity of various extracts, although they are less active compared to procyanidin A2. Thus, the antiviral activity of all tested extracts is likely due to several tannins. Those that were found to be active were present in higher amounts in the most active extracts (methanolic and hydro-ethanolic), and also in aqueous extract in lower amount. In addition, they were absent from all these extracts after hide powder treatment. These results are correlated with the anti-HCV activity of these extracts: lower in aqueous extracts compared to alcoholic ones, and null after tannins removal.

Some of the tannins were identified and semi-quantified here, but many others were probably present in the same extracts. Indeed, the chromatograms in Supplementary Figure S2 shows that most peaks found in C. procera and P. laxiflora crude extracts can also be seen after tannin removal and therefore probably inactive on HCV. But, for both plants, the baseline is increased between 5- and 10-min elution for crude extracts, whereas it is close to zero after tannin removal. Thus, the difference is probably due to various tannins that were not all identified here and that may contribute to antiviral activity. Nonetheless, procyanidin A2 is one of the major compounds of P. laxiflora eco-extract, and was mostly removed after hide powder treatment (Supplementary Figure S2).

Tannins including procyanidins are generally known to have low oral biodisponibility. However, this depends on numerous factors, such as their degree of polymerization (DP). For example, procyanidins with DP < 4 can be absorbed and distributed in the organism, at least under certain conditions (Zhang et al., 2016). Depending on other factors (e.g. gastric pH, food intake), tannins can be degraded into smaller molecules, and metabolized by the colonic flora, leading to other compounds that may have a biological activity as well (Zhang et al., 2016; Smeriglio et al., 2017). In our survey, P. laxiflora was used in bath, whereas C. procera was also used in tea (Table 1). Therefore, digestion related factors might lead to other active molecules after drinking this plant extract. It would be interesting to test procyanidins and other putative anti-HCV tannins in vivo, before extrapolating any of our data in terms of clinical relevance for treating hepatitis C.