Plant materials

Dried yerba mate (I. paraguariensis A.St. Hil) with leaf grinded and stems (1 kg package of Playadito con palo) was obtained from Cooperativa Agricola de la Colonia Liebig Ltda. (Colonia Liebig’s, Corrientes, Argentina). Dried jarilla (L. divaricata Cav.) leaf grinded (0.5 kg package) was purchased from Homeopathic Argentinian Pharmacy (60th avenue and 10th corner, La Plata, Buenos Aires, Argentina).

Polyphenolic extracts

Ultrasound-Assisted Extraction (UAE) (Branson, St. Louis , MO, USA) was performed according to ^{Cárdenas-Hernández et al. (2020)}. Briefly, a 1 g sample of the plant was weighed and diluted in 20 mL of ethanol (90 %, v/v). The extraction was performed for 20 min at 35 °C (frequency: 60 Hz and power: 42 KHz). Samples were transferred to conical tubes and taken to a centrifuge (SORVALL: Biofuge primo™ R) at 3500 x g at 20 °C (10 min). Subsequently, filtration of each sample was carried out (11 µm through filter paper) and the extracts were stored at -5 °C. The extraction yield was calculated by weight difference. Briefly, an extract of 5 mL was evaporated at room temperature (28 °C). The weight of the residue was multiplied by 4 to normalize the real volume of the experiment (20 mL).

Antioxidant activity

The antioxidant activity in the ethanol extracts was determined by: DPPH• (^{Molyneux, 2004}), ABTS•+ (^{Re et al., 1999}), and FRAP (^{Benzie and Strain, 1996}). Each sample was measured in triplicate. The results of DPPH• and ABTS•+ tests were expressed as a percentage reduction. FRAP assay was expressed as Trolox Equivalents (TE) with a calibration curve of Trolox (0.1-1.0 mg/mL).

Analytical RP-HPLC-ESI-MS

A Varian High-Performance Liquid Chromatography coupled to a mass spectrophotometer (Varian 500-MS IT Mass Spectrometer, USA; PDA detector Varian Pro Star 330, USA) was used to identify the main components in the extract according to ^{Torres-León et al. (2017)} . Samples (5 µL) were injected into a Denali C18 column (150 mm × 2.1 mm, 3 µm, Grace, USA). The thermostat temperature was main-tained at 30 °C. The eluents were as follows: formic acid (0.2%, v/v; A solvent) and acetonitrile (B solvent). The gradient was characterized by using initial, 3 % B with retention time of 0 - 5min, 9 % B linear with retention time of 5 - 15min, 16 % B linear with retention time of 15 - 45min, 50 % B linear.

Determination of physicochemical parameters of the main protease (Mpro)

Main Protease (Mpro; 6WQF_A) was taken from the NCBI database (https://www.ncbi.nlm.nih.gov). Analysis of the physical parameters of main protease (Mpro) is done by studying the number of amino acids and their composition, molecular weight, aliphatic index, theoretical pI, extinction coefficient, instability index, and grand average of hydrophobicity using ExPASy (http://web.expasy.org/protparam).

Molecular docking

A docking Server was used to predict the binding between the phytochemicals of yerba mate and jarilla extracts and the main COVID-19 protease (Mpro) (^{Torres-León et al., 2020}) (https://ncov.schanglab.org.cn/index.php), which is also named chymotrypsin-like protease (3CLpro) (^{Kong et al., 2020b}). The structures of bioactive compounds identified by the HPLC-MS analysis (used as ligands) were downloaded in SDF file format from: https://pubchem.ncbi.nlm.nih.gov/. The verified inhibitors were evaluated as positive controls (Remdesivir and Ribavirin). The main aim of this molecular interaction study was to identify the phytochemicals interacting with the crystal structure of protease (Mpro) in silico. All interaction visualization analysis studies were performed with the PyMol molecular visualization tool.

Statistical analysis

Results were expressed as mean values ± SD. The student’s t-test was run to determine significant differences (p < 0.05). All statistical determinations were performed using Statistica 7.0 software (StatSoft, Tulsa OK, USA).

Antioxidant activity

The results of the DPPH assay (Table 1) showed that both extracts have a high antioxidant activity (> 90 %); without presenting a significant difference between the plants. These results were similar (92 %) to those reported in Larrea genus (^{Aguirre-Joya et al., 2018}) and higher than those reported in yerba mate leaves from Brazil, Argentina, and Uruguay (< 80 %) (^{Colpo et al., 2016}). The results in the ABTS•+ assay also showed high antioxidant activity in both samples (> 90 %). Extracts of jarilla (93 %) were significantly higher than those of yerba mate (90.5 %). Finally, the FRAP analysis showed that the extracts were able to reduce the ferric ion to ferrous ion. The jarilla extract showed a high antioxidant activity, significantly higher than yerba mate (p > 0.05). This trend was similar in the ABTS•+ test; these assays are associated with its ability to scavenge free radicals. The results showed that yerba mate and jarilla leaves have high antioxidant activity. This biological activity is attributed to the presence of bioactive compounds such as polyphenols. The antioxidant activity of phenolic compounds depends on the number of hydroxyl and methoxy groups present in the ring and their dispositions in its structure. Generally, glycosylation of 3-hydroxyl groups reduces the antioxidant properties of compounds. For this reason, yerba mate has less antioxidant activity due to the presence of rutin (^{Olszowy, 2019}). The consumption of these plants can increase the antioxidant defenses in people (^{Schinella et al., 2000}).

Analytical RP-HPLC-ESI-MS

The identification of phenolic compounds of yerba mate and jarilla extracts was performed by comparing the retention time by HPLC analysis with MS. The molecular weight was compared with the literature and the database of the Food Research Department (DIA-UAdeC). In this study, four phenolic compounds were separated and identified in each extract (Table 2). In general, yerba mate extract contained neochlorogenic acid, chlorogenic acid, rutin, and 3,4-dicaffeoylquinic acid. Chlorogenic acid is a compound that derives from caffeic acid and is present in higher concentrations in yerba mate extract. Thus corroborating the results presented in the literature (^{Agüero et al., 2011}; ^{Zapata et al., 2019}).

The HPLC-MS analysis showed that jarilla leaves are a source of 3'-O-methyl-nordihydroguaiaretic acid, p-hydroxybenzoyl glucoside, meso (rel7S,8S,7’R,8’R)-3,4,3’,4’-tetrahydroxy-7,7’-epoxylignan and nordihydroguaiaretic acid. These lignans and others have been previously reported in Larrea spp. extracts (^{Agüero et al., 2011}; ^{Vargas-Arispuro et al., 2005})nordihydroguaiaretic acid (NDGA. Variations in the phenolic profile may be due to factors such as the extraction method or growing conditions of the plants.

Determination of physicochemical parameters of the main protease (Mpro)

The main protease (Mpro) structure is composed of the I, II, and III domains (^{Kneller et al., 2020}); residues 8 - 10, 102 - 184, and 201 - 303, respectively. The number of amino acids is 306. In the present study, the theoretical isoelectric point (pI) was 5.92 - 5.95. The molecular weight of the protein is 33796.64 Da. The instability index is 27.65; a value < 40 indicates that the protein structure is stable. The average extinction coefficient is 33640. The aliphatic index is 82.12. The GRAVY value was - 0.019, this value indicates that the protein is hydrophilic and non-polar.

Molecular docking

The application of the COVID-19 docking server elucidated the interactions between the lead-likeness compounds and Mpro (Table 2).

All the identified compounds in the yerba mate and jarilla leaves extracts showed an inhibitory potential against Mpro. According to in silico results, all the compounds had a higher affinity against COVİD-19 protease (< - 7.20 kcal/ mol) than Ribavirin (- 6.40 kcal/mol). Additionally, the yerba mate compounds, rutin (-9.60 kcal/mol), and 3,4-dicaffeo-ylquinic acid (- 8.20 kcal/mol) had a high affinity that the broad spectrum anti-viral drug Remdesivir (- 8.10 kcal/mol). This molecule has anti-Covid19 activity (^{Singh et al., 2020}) and is currently used in COVID-19 management (^{Hung et al., 2020}). The docking energy is lower, and the binding is more stable, which indicates that yerba mate molecules probably hinder the binding of substrate proteins more effectively. ^{Das et al. (2020)} caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2, reported that rutin could bind Mpro protein (PDB: 6Y84) with the highest affinity among 33 molecules screened (including natural products, antivirals, antifungals, and antiprotozoal agents). Furthermore, the docking energy of the rutin derivatives was determined for isoquercetin (monoglycosylated) and quercetin (deglycosylated), resulting in - 8.70 kcal/mol and - 7.80 kcal/mol, respectively. The docking energy decreases when the flavonoid loses glucose molecules.

Molecular docking predicted that polyphenols from yerba mate and jarilla have inhibitory potential against SARS-CoV-2 infection by interacting with the Mpro protein. The binding between Mpro and Remdesivir, Ribavirin, the bioactive compounds rutin and 3,4-dicaffeoylquinic acid (which presented the highest docking score) as a potential inhibitor of COVID-19 are shown in Figure 1. They have been represented in distinct colors for better visualization. Figure 1 corroborates the high-affinity potential listed in Table 1. There are two parameters to evaluate a better docking, a docking score and the type of interaction. A better docking score is one with more negative binding energy. The stability between ligands and the receptor is attributed to dispersion forces, hydrogen bonds, π-π interactions, and hydrophobic interactions (^{Mpiana et al., 2020}). ^{Yu et al. (2020)} ETCM database and document mining methods were used to collect active compounds. Swiss TargetPrediction and SuperPred server were used to find targets of compounds with smiles number. Drugbank and Genecard database were used to collect antiviral drug targets. Then the above targets were compared and analyzed to screen out antiviral targets of Mongolia medicine. Metascape database platform was used to enrich and analyze the GO (Gene ontology, demonstrated that molecules present in traditional Mongolian medicine, such as chlorogenic acid, combine with SARS-CoV-2 protein in the form of hydrogen bonds.

Fig. 1 2D representation of the interaction of Mpro COVID-19 with the anti-viral drug (A) Remdesivir, (B) Ribavirin, (C) Rutin, and (D) 3,4-dicaffeo-ylquinic acid.  

Fig. 1 Representación 2D de la interacción de Mpro COVID -19 con el fármaco antiviral (A) Remdesivir, (B) Ribavirina, (C) Rutina y (D) ácido 3,4-di-cafeoilquínico. 

The results confirmed that rutin and 3,4-dicaffeoylquinic acid interacted with His41 and Cys145, which are catalytic residues of Mpro by hydrogen bonding. Both rutin and 3,4-dicaffeoylquinic acid formed hydrogen bonds with Cys145 amino acid residues of Mpro. The rutin derivatives (quercetin and isoquercetin) form hydrogen bonds with the His41 amino acid residues of Mpro. In addition to these two key residues, several other Mpro active site amino acid residues were involved in hydrogen bonding (Glu166, Thr190, Gln189, Asn142, and Leu141) (^{Ghosh et al., 2020}; ^{Kneller et al., 2020}). The distance between the ligands (≤ 3 Å) in relation to His41 and Cys145 within Mpro (active site) presents higher interaction. Given the affinity and interaction of the catalytic amino acids, they could inhibit/decrease the catalytic activity of the Mpro protease. These are important candidates for an alternative treatment of SARS-Cov-2, applying them in infusion or purified phenolic compounds. In silico experiments predict promising results.