Antioxidant properties and phenolic composition of “Composed Yerba Mate”

Abstract

Yerba mate contains bioactive compounds, and is widely consumed as a decoction beverage in several Southern American countries. At present, the consumption of mate with added herbal blends and flavors, called “composed yerba mate”, has increased; however, no studies on the antioxidant characteristics of these products have been published. In this sense, the main objective was to assess the antioxidant characteristics of “composed yerba mate” compared to “traditional yerba mate”, in the form it is traditionally consumed. Total polyphenols content ranged from 15 to 45 mg/g GAE in all decoctions analyzed. Seventeen phenolic compounds were identified and quantified by HPLC–DAD-MS/MS, mainly belonging to the caffeoylquinic acids group. The antioxidant capacity was measured using in vitro assays, Ferric reducing ability of plasma (FRAP) and Trolox equivalent antioxidant capacity (TEAC), and with Saccharomyces cerevisiae as the in vivo model organism. All decoctions displayed antioxidant activity and were capable of rescuing yeast cells between 10.68 and 18.38% from oxidative stress. Multiple regression analysis showed a high correlation between phenolic composition and activity of samples, where different compounds indicate a significant contribution to the observed activity. Significant differences were found in the content, profile and antioxidant activity of polyphenols when “traditional yerba mate” and “composed yerba mate” were compared. In some cases, the antioxidant capacity was similar or higher in composed yerba mate; while the rest displayed lower biological activity. Based on these findings, it would be possible to assume that the addition of herb mixtures modifies the antioxidant and biological properties of mate.

Supplementary Information

The online version contains supplementary material available at (10.1007/s13197-020-04961-x).

Introduction

Indigenous plants have special importance around the world, and add value to the earth's diversity. Their different organs -leaves, roots, shoots, fruits, etc.- have a high content of non-nutritive and bioactive compounds, such as polyphenols and phenolic acids, in addition to nutritive compounds, such as sugars, essential oils, carotenoids, vitamins, and minerals (Zia-Ul-Haq et al. 2014; Senica et al. 2019). Ilex paraguariensis Saint Hilaire, best known as “yerba mate”, a plant native to South America, is a crop of great regional socioeconomic importance in Argentina, and the only species of yerba mate authorized for consumption. Their dried, minced leaves and twigs are used for the elaboration of a beverage rich in antioxidant compounds, named green mate. It is highly consumed in several Southern American countries.

The national consumption has remained stable in the last decade (6.34 kg/person/year), (http://faostat3.fao.org); however, consumers, motivated by changes in their consumption habits, are now demanding higher quality, nutritional characteristics and genuineness. Consequently, new flavors are being proposed for the traditional yerba mate through the addition of different herbs, thus creating a “composed yerba mate”. The most commonly used herbs include Mentha x piperita L. (peppermint), Mentha pulegium (pennyroyal), Minthostachys verticillata (peperina), Peumus boldus (boldo), Melissa officinalis (melissa) and Aloysia citrodora (lemon verbena). Given that these added herbs have medicinal (digestive, diuretic, etc.) or simply flavoring properties, producers seek to provide other benefits than those already recognized in yerba mate.

da Silva et al. (2008) reported that the yerba mate extract was rich in phenolics having in vitro antioxidant activity, and contained important constituents like vitamins and minerals. Furthermore, Bixby et al. (2005) found that I. paraguariensis is more antioxidant than red wine, green tea and black tea antioxidants. In this regard, the antioxidant activity of yerba mate is particularly attributed to the presence of phenolic compounds (Bixby et al. 2005; Bracesco et al. 2003; Chandra and Gonzalez de Mejía 2004). Based on its phytochemical composition, it could be suggested that the consumption of yerba mate is an effective and economical way of incorporating antioxidants to the diet, providing benefits to the defense system and protection against harmful free radicals. Due to its phenolic composition, consisting mainly of hydroxycinnamic acid derivatives (Bravo et al. 2007), yerba mate provides health benefits, since it has anti-carcinogenic and antidiabetic effects, reduces inflammation, and prevents cardiovascular diseases (Burris et al. 2012; Colpo et al., 2016; Mateos et al. 2018; Gan et al. 2018).

In vitro antioxidant activity assays (such as FRAP and TEAC) are widely used to evaluate the potential bioactivity of plant foods. Many authors suggest that the evaluation of antioxidant capacity requires the use of several methodologies in parallel, because different methods can produce dissimilar results (Niki 2011; Prior et al. 2005; Tabart et al. 2009). In this regard, several cellular models have been developed to gather more information about possible mechanisms of action to confer tolerance to oxidative stress. The evaluation of antioxidant activity using animal models is an alternative; however, a large number of animals are required to obtain statistically significant results. On the contrary, tests with microorganisms are easy and fast, and a large number of cells with the same genetic characteristics can be used (Martorell et al. 2011; Stinco et al. 2015; Lingua et al. 2016; Meng et al. 2017). Saccharomyces cerevisiae (S. cerevisiae) is a model organism for basic studies of eukaryotic cell biology. This last model is used to understand the cellular response to oxidative stress and the antioxidant defense system, and applies mainly to antioxidants from food plants (Bracesco et al., 2003; Baroni et al. 2012; Stinco et al. 2015; Lingua et al. 2016; Meng et al. 2017). The main advantages are that these models reflect the physiologic cellular conditions and consider bioavailability and metabolic issues, affecting the net response of the phenolic compounds in the samples.

While there is a history on the characterization of the polyphenols in yerba mate (Bravo et al. 2007; Dugo et al. 2009; Plagiosa et al. 2010; Mateos et al. 2018) and in some of the herbs included (Hurrell et al. 2011,s 2013), to the extent of our knowledge, there are no data regarding the polyphenolic composition in composed yerba mate, as well as their contribution to the antioxidant capacity of the product. For this reason, the objective was to evaluate the polyphenolic profile of composed yerba mate, as it is consumed, from various commercial brands compared with the profile of a traditional yerba mate, and their antioxidant activity measured by chemical in vitro and in vivo methods. The objective of using different complementary techniques is to obtain a more complete evaluation of its bioactivity. Furthermore, Multiple Regression Analysis was applied to associate polyphenols compounds with the antioxidant activity observed in in vitro and in vivo methods. This characterization is important for future trials since this product, in addition to traditional yerba mate, could be considered a functional food.

Materials and methods

Chemicals and reagents

HPLC-grade methanol and formic acid (puriss. p.a. for mass spectroscopy) were purchased from J.T. Baker (Mexico City, Mexico) and Fluka (Buchs, Germany), respectively. Ultrapure water (< 5 lg/L TOC) was obtained from an Arium 61,316-RO plus Arium 611 UV (Sartorius, Göttingen, Germany) purification system. Filters (0.45 µm, HAWG04756) were obtained from Millipore (São Paulo, Brazil).

Gallic acid was obtained from Riedel-de-Haën (Shanghai, China). Caffeic acid, chlorogenic acid, ferulic acid, quinic acid, kaempferol and rutin (quercetin-3-rutinoside) standards were obtained from Sigma-Aldrich Co. (Missouri, USA).

The agar–agar, yeast extract and meat peptone were obtained from Laboratorios Britania (Buenos Aires, Argentina). TPTZ (2,4,6-tripyridyl-s-triazine), Trolox, ABTS and Folin-Ciocalteu reagent were purchased from Sigma–Aldrich (Buchs, Switzerland). All other reagents were of analytical grade.

Preparation of decoctions from yerba mate samples

Composed yerba mate and traditional yerba mate (without added herbs) of different commercial brands were purchased at supermarkets in Córdoba (Argentina) (Table 1). Each commercial sample was obtained in duplicate. Samples were processed following one of the traditional ways of consuming this product in our region, which involves heating the yerba mate in water until the first boil, and then percolating it to obtain the decoction for consumption. Yerba mate decoctions were prepared according to the protocol previously described by Bravo et al. (2007), using 1.5 g of samples with 150 mL of ultrapure water, and boiled for 5 min. The sample was allowed to cool down to room temperature, filtered, and subsequently adjusted to 150 mL with ultrapure water.

Table 1.

Composition of analyzed samples

Sample	Composition
1	Yerba mate
2	Yerba mate, pennyroyal, peperina and peppermint
3	Yerba mate, boldo and peppermint
4	Yerba mate and lemon verbena
5	Yerba mate, pennyroyal, peperina and peppermint
6	Yerba mate, peppermint, pennyroyal, peperina and melissa
7	Yerba mate, boldo and peppermint
8	Yerba mate, boldo and peppermint

All decoctions were performed in duplicate and stored at − 80 °C until the time of analysis.

Determination of total polyphenols

Total polyphenols (TP) content was determined in decoctions using the Folin Ciocalteau method, according to Baroni et al (2012), with minor modifications. The absorbance of properly diluted samples was read at 750 nm. TP was calculated by linear regression from a calibration plot constructed with gallic acid as standard. All samples were assayed in triplicate, and results were expressed as milligrams of gallic acid equivalents (GAE) per 1 g of yerba mate.

Determination of phenolic profile

The polyphenol profile from yerba mate samples was determined using an HPLC (Agilent Technologies 1200 Series) equipped with a gradient pump (Agilent G1312B SL Binary), solvent degasser (Agilent G1379B) and autosampler (Agilent G1367D SL + WP). HPLC conditions were in accordance with the method described by Lingua et al. (2016). In brief, chromatographic separation was performed on a C18 reverse phase column (LUNA, Phenomenex, Torrance, CA, USA; 5 µm, 250 mm, 4.60 mm i.d.), thermostated at 35 °C. The mobile phase consisted of a gradient solvent (solvent A: 0.5% formic acid v/v and solvent B: 0.5% formic acid v/v in methanol). The flow rate was 0.4 mL/min and the injection volume was 40 µL. The HPLC system was coupled to a diode array detector (Agilent G1315 C Starlight DAD) followed by a QTOF mass spectrometer (MicroTOF-Q11 Series, Bruker) equipped with an electrospray ionization source (ESI). Preferred wave lengths were 280, 320 and 350 nm, while the UV–Vis spectra were recorded from 200 to 600 nm. The working conditions of the ionization source were as follows: capillary voltage of 4500 V, gas pressure 4.0 bar atomization, drying gas flow 8.0 L/min, and drying temperature 180 °C. N₂ and Ar gases were used as nebulizer and collision, respectively. MS spectra were recorded in the negative mode in the range of 100 to 1000 m/z. The MS detector was programmed to perform an MS/MS sweep of the three most abundant ions, using 13.0 eV collision energy. Data acquisition and processing were performed using the Compass Version 3.1 software and DataAnalysis Version 4.1 software, respectively.

Polyphenol quantification was based on external calibration curves from available phenolic standards (quinic acid, chlorogenic acid, caffeic acid, ferulic acid, rutin and kaempferol), by using the mass peak areas obtained from the extracted ion chromatograms (see Fig. 1S of Supplementary data). The calibration curves were prepared at concentrations between 3.12 and 100 µg/mL. When the corresponding standards were not available, the quantification was performed using an external standard structurally related to the compound in question.

In vitro antioxidant activity

The in vitro antioxidant capacity was measured using both ferric reducing ability of plasma (FRAP) assays and Trolox equivalent antioxidant capacity (TEAC), according to Benzie and Strain (1998) and Re et al. (1999), respectively, with minor modifications. Results are expressed in µmol Trolox equivalents per g of yerba mate. All samples were analyzed in triplicate.

FRAP assay

The working reagent consists of a mixture of acetate buffer (pH = 3.6), 10 mM TPTZ in HCl 40 mM and 20 mM FeCl₃.6H₂O, in a 10: 1: 1 dilution. Then, 3 mL of working reagent were mixed with 100 µL of properly diluted sample. After 6 min in dark reaction, the absorbance at 593 nm was measured.

TEAC assay

To measure antioxidant activity, 3 mL of working reagent (ABTS radical) were taken in a cuvette and its absorbance was measured at 734 nm (A0). Subsequently, 100 µL of properly diluted sample were added, mixed for 10 s, and the absorbance was measured exactly after 4 min (A1) at 734 nm. For each determination, the ∆Abs (A1—A0) was then calculated.

In vivo antioxidant activity

The in vivo antioxidant activity was evaluated using a S. cerevisiae model regarding their capacity to confer tolerance to oxidative stress, in accordance with Lingua et al. (2016), with minor modifications. Before stress with H₂O₂, cells were treated with different yerba mate decoctions for 15 min. Subsequently, the oxidant (2 mM H₂O₂) was added and incubated for 1 h at 28 °C. Optimal TP doses from decoctions (dried and dissolved in 35% DMSO, Sigma-Aldrich, Buenos Aires, Argentina) were determined in adaptive treatments, exposing cells to increased concentrations of TP from these samples (data not shown). The concentration chosen was 0.015 mg TP/mL for decoctions. This final concentration was the lowest, showing the highest survival rate as compared to yeast exposed to H₂O₂ (2 mM) without the addition of sample.

Two control groups were used: a control plate (yeast exposed to vehicle of phenolic compounds in samples—35% DMSO) and a decoction control plate (yeast exposed to a mix of decoctions in equal parts, without the addition of H₂O₂).

Cell viability was analyzed by plating on solid YPD medium, after proper dilution. Plates were incubated at 30 °C for 72 h. One hundred percent survival was considered the number of colonies observed in the control plate (yeast exposed to vehicle of phenolic compounds in samples). The number of colonies in each plate was between 150 and 200. All assays were carried out in triplicate.

Statistical analysis

Data were expressed as ± SD obtained from at least two independent experiments. The data were analyzed using an ANOVA test with p < 0.05.

Multiple regression analyses (MRA) were performed to determine the relative contribution of each phenolic compound to the in vivo and in vitro antioxidant activity. The regression (Beta) coefficients were analyzed to evaluate key compounds in order to predict the antioxidant activity in the studied samples. The magnitude of these Beta coefficients allows the comparison of the relative contribution of each independent variable with the prediction of the dependent variable. The Infostat Software Package (Di Rienzo et al. 2013) was used for all statistical analyses.

Results and discussion

Total polyphenols content

The mean contents of TP in composed and traditional yerba mate decoctions, shown in Fig. 1, ranged from 15 to 45 mg/g GAE, with statistically significant differences between all samples analyzed (p < 0.05). As it can be observed, sample D6, containing mint, pennyroyal, peperina and melissa, presented the highest value; while D2 and D5 showed significantly lower values. It is noteworthy that these last two decoctions have the same added herbs except melissa, which could indicate that they would contribute to the TP content.

Fig. 1.

Total Polyphenol Content (TP) and Antioxidant Activity (AA) of different infusions. Different letters and numbers indicate statistically significant differences (p < 0.01) in each parameter between different decoctions

Comparing our results with previously published ones, the TP content was slightly lower than that found by other authors, who reported values close to 79 mg gallic acid/g of dry matter (Bravo et al. 2007). Plagiosa et al. (2010) determined the TP content in decoctions prepared with leaves and barks of yerba mate separately, finding values between 70.1 and 125.0 mg/g, respectively; while Chandra and Gonzalez de Mejía (2004) reported values of 94.91 mg/g for dried yerba mate leaves. Differences found may be due to differences in the samples production processes as well as in the different extraction techniques used.

Identification and quantification of polyphenols in yerba mate decoctions

The identification and quantification of individual phenolic compounds in yerba mate decoctions were carried out by HPLC–DAD-MS/MS. Retention times, elution order, UV–Vis spectra, exact mass (≤ 10 ppm) and MS fragmentation spectra were the criteria used to identify each compound, and compared to standards when available (Table 2). One organic acid and 16 phenolic compounds were identified in samples, being the hydroxycinnamic acids with caffeoyl derivatives the main constituents, followed by flavonols. Figure 2S of Supplementary data show extracted ion chromatogram of different compounds identified. All detected compounds have already been reported for yerba mate (Bravo et al. 2007; Mateos et al., 2018), with the exception of quinic acid, which is probably released by the hydrolysis of its derivatives.

Table 2.

Chromatographic, MS and UV–Vis spectrum data of compounds identified in the decoctions

	rt	λmax	Molecular Formula	[M − H] − (m/z) calcd	[M − H] − (m/z) exp	Error (ppm)	MS2 (m/z)	Proposed structure
1	7.1	224	C7H11O6	191.0561	191.0560	0.8	–-	Quinic acid
2	11.7	327, 297sh	C16H17O9	353.0878	353.0863	4.4	191, 179	Neochlorogenic acid
3	12.1	323, 295sh	C15H17O9	341.0878	341.0868	2.9	179, 281	Caffeoyl-glucoside
4	12.6	327, 297sh	C16H17O9	353.0878	353.0870	2.5	191, 179	Chlorogenic acid
5	13.7	273	C16H15O8	335.0772	335.0773	-0.1	161, 173, 179	Caffeoylshiquimic acid
6	14.1	324, 296sh	C17H19O9	367.1035	367.1037	-1.3	191, 173	Feruroylquinic acid I
7	15.1	326, 296sh	C25H23O12	515.1195	595.1244	-9.4	353, 191, 179	Dicaffeoylquinic acid I
8	15.4	326, 296sh	C25H23O12	515.1195	515.1195	0.0	353, 191	Dicaffeoylquinic acid II
9	15.5	327, 297sh	C16H17O9	353.0878	353.0890	-3.4	191, 179	Criptochlorogenic acid
10	18.1	327, 297sh	C16H17O9	353.0878	353.0885	2.0	191, 179	Caffeoylquinic acid
11	18.1	326, 296sh	C25H23O12	515.1195	515.1195	0.0	353, 173, 179	Dicaffeoylquinic acid III
12	18.3	331	C27H29O16	609.1461	609.1435	4.3	301	Quercetin-rutinoside
13	18.6	330	C21H19O12	463.0882	463.0872	2.2	301	Quercetin-glucoside
14	20.4	324, 296sh	C17H19O9	367.1035	367.1029	1.4	193	Feruroylquinic acid II
15	20.4	332, 298sh	C26H25O12	529.1351	529.1348	0.6	367, 173	Caffeoylferuloylquinic acid I
16	20.6	332, 298sh	C26H25O12	529.1351	529.1356	0.9	353, 367, 173	Caffeoylferuloylquinic acid II
17	20.8	266	C21H19O11	447.0933	447.0912	-4.7	285	Kaempferol-glucoside

Organic Acid. Compound 1 was identified as quinic acid based on exact mass calculations and standard retention time.

Hidroxycinamic acid derivatives. The UV–Vis spectra of compound 3 showed a λ_max at 323 nm, characteristic of caffeic acid and a shoulder at 295 nm, consistent with quinic acid derivatives. The MS spectra showed a molecular ion [M-H]⁻ at m/z 341 and fragments at m/z 281 and 179. The first of these fragments is probably generated by the loss of a molecule of acetic acid (CH₃COO⁻), suggesting that at least one of the two cinnamoyl groups is not involved in a chemical union. In addition, caffeic acid was detected as a fragment (m/z 179). Taking into account exact mass calculations, compound 3 could be proposed as caffeoyl-glucoside, which is consistent with the literature (Correa et al. 2017; Mateos et al. 2018).

Compounds 2, 4, 9 and 10 belong to the caffeoylquinic acid isomers group. They showed similar UV–Vis spectra, with an absorption maximum at 327 nm and shoulder at 297 nm, characteristic of caffeic acid derivatives (Bravo et al. 2007; Mateos et al. 2018). Furthermore, the mass spectra of these compounds were similar, showing an ion [M-H]⁻ at m/z 353 and fragments at m/z 191 and 179, which correspond to deprotonated caffeic and quinic acid, respectively. Compound 4 was identified as chlorogenic acid (3-O-caffeoylquinic acid) by comparison of its retention time with the corresponding standard. Taking into account the elution profile of the compounds and compared with literature data (Bravo et al. 2007; Mateos et al. 2018), compounds 2 and 9 were identified as 5-O-caffeoylquinic acid (neochlorogenic acid), and 4-O-caffeoylquinic acid (criptochlorogenic acid), respectively.

Compound 5 showed a UV–Vis spectrum with an absorption maximum at 273 nm. The mass spectrum showed an [M-H]⁻ ion at m/z 335 coincident with the loss of a molecule of H₂O from caffeoylquinic acid and a majority fragment at m/z 161, which would correspond to the dehydrated caffeic acid. The dehydrated quinic acid (shikimic acid) might be esterified with caffeic acid producing the caffeoylshikimic acid (Bravo et al. 2007; Mateos et al. 2018).

Compounds 6 and 14 showed a molecular ion at m/z 367, yielding three molecular fragments at m/z 193, 191 and 173. Fragments at m/z 193 correspond to ferulic acid, while fragments at m/z 191 and 173 correspond to quinic acid and dehydrated quinic acid, respectively. Based on this information, these compounds were identified as feruloylquinic acid isomers.

Compounds 7, 8 and 11 showed UV–Vis spectra similar to that of chlorogenic acid, although the MS spectra showed a molecular ion [M-H]⁻ at m/z 515 and fragments at m/z 179, 191 and 353. Fragments at m/z 179 and 191 correspond to caffeic and quinic acids, respectively. The fragment at m/z 353 corresponds to the loss of a dehydrated caffeic acid moiety, so these compounds were tentatively identified as dicaffeoylquinic acids.

Compounds 15 and 16 showed a signal at m/z 529 and both generated a fragment at m/z 367, which is caused by the loss of one dehydrated caffeic acid molecule. They presented another fragment at m/z 353 which might derive from the loss of a feruloyl residue, while the fragment at m/z 173 might correspond to the loss of dehydrated quinic acid moiety. Based on this information, and by comparison with literature data (Dugo et al. 2009), both compounds may be proposed as caffeoyl-feruloylquinic acid isomers.

Flavonols. Compound 12 was identified as quercetin-3-O-rutinoside (rutin), since its spectrum shows a signal at m/z 609 and a fragment at m/z 301, consistent with the loss of the sugar moiety (rhamnose-glucose). The identification was confirmed by comparison with the standard rutin, with identical retention times, UV–Vis spectrum and mass spectrum generated under the same chromatographic conditions.

Compound 13 showed an [M-H]⁻ ion at m/z 463 and fragment at m/z 301, consistent with the loss of a sugar moiety (dehydrated hexose) to produce the quercetin aglycon. The identification of this compound was carried out by comparing the UV–Vis and mass spectrum with a commercial standard of isoquercetin.

Compound 17, which has a maximum absorbance at 266 nm, is probably glycosylated kaempferol, because it has a mass spectrum with a deprotonated molecular ion at m/z 447 and fragment at m/z 285. By comparison with the mass spectrum and fragmentation pattern in the literature, it can be identified as kaempferol-glucoside (Dugo et al. 2009).

Finally, we could not detect compounds arising from the herbs added to yerba mate, probably because of the low amount added, and the detection limit of the technique used.

Table 3 shows the results obtained from the quantification of individual polyphenolic compound in different decoctions. Caffeoylquinic acids were the major constituents of the phenolic fraction of yerba mate, representing 70% of the total polyphenols. These values were similar to the results previously reported by others authors (Bravo et al. 2007; Mateos et al. 2018). The content of chlorogenic acid obtained in the present work (between 10.42 and 18.43 mg/g) is in accordance with that reported by Plagiosa et al. (2010), who found values between 16 and 29.30 mg/g of chlorogenic acid in methanolic extracts of leaves and barks of yerba mate, respectively. On the other hand, Monteiro and Farah (2012) and Rodrigues and Bragagnolo (2013) quantified caffeoylquinic acids in four species of Coffee arabica, obtaining lower values than those obtained in this study for yerba mate.

Table 3.

Quantification of polyphenols in different decoctions

	Compounds	Decoctions (mg/g)
		D1	D2	D3	D4	D5	D6	D7	D8
1	Quinic acid	2.32 ± 0.11b	1.19 ± 0.04e	2.81 ± 0.04a	1.76 ± 0.04c	1.48 ± 0.04d	1.79 ± 0.02c	1.42 ± 0.09c	1.57 ± 0.15d
2	Neochlorogenic acid	14.73 ± 0.44a	10.65 ± 0.25b	13.61 ± 1.31a	12.45 ± 0.13b	10.95 ± 1.78b	14.04 ± 1.66a	11.84 ± 0.30a	14.34 ± 0.32a
3	Caffeoyl-glucoside	< LOD	< LOD	< LOD	0.22 ± 0.22a	< LOD	0.26 ± 0.26a	0.06 ± 0.06b	< LOD
4	Chlorogenic acid	17.32 ± 1.12a	13.08 ± 0.17b	15.68 ± 1.41b	14.82 ± 0.25a	12.37 ± 1.95a	16.78 ± 2.03a	13.59 ± 0.44a	15.54 ± 0.37a
5	Caffeoylshiquimic acid	0.87 ± 0.09a	0.16 ± 0.03e	0.49 ± 0.01c	0.35 ± 0.06d	0.22 ± 0.05e	0.68 ± 0.10b	0.26 ± 0.01e	0.55 ± 0.01c
6	Feruroylquinic acid I	4.75 ± 0.52a	2.04 ± 0.09b	4.59 ± 0.91a	3.31 ± 0.63a	1.84 ± 0.69b	4.08 ± 1.07a	2.65 ± 0.32a	3.51 ± 0.38a
7	Dicaffeoylquinic acid I	< LOD	< LOD	2.21 ± 0.23a	1.38 ± 1.38a	< LOD	1.17 ± 1.17a	0.78 ± 1.35a	1.24 ± 1.24a
8	Dicaffeoylquinic acid II	2.06 ± 0.12a	1.76 ± 0.16a	< LOD	0.72 ± 0.72c	1.31 ± 0.42b	0.79 ± 0.79c	1.96 ± 0.27b	1.06 ± 1.06c
9	Criptochlorogenic acid	15.00 ± 0.50a	11.73 ± 1.02b	17.11 ± 1.13a	13.52 ± 0.09a	10.67 ± 2.40b	15.50 ± 2.29a	13.27 ± 0.45a	15.28 ± 0.34a
10	Caffeoylquinic acid	2.26 ± 0.33a	1.82 ± 0.02a	2.50 ± 0.03a	2.04 ± 0.12a	1.53 ± 0.26b	1.92 ± 0.52a	1.94 ± 0.01a	2.09 ± 0.23a
11	Dicaffeoylquinic acid III	1.65 ± 0.12a	1.38 ± 0.15a	1.63 ± 0.24a	1.60 ± 0.50a	0.98 ± 0.26a	1.39 ± 0.31a	1.42 ± 0.12a	1.60 ± 0.12a
12	Quercetin-rutinoside	6.05 ± 0.04a	4.18 ± 0.10b	5.49 ± 0.02a	5.05 ± 1.09a	3.78 ± 1.53b	5.61 ± 0.44a	4.71 ± 0.01a	5.29 ± 0.04a
13	Quercetin-glucoside	0.10 ± 0.02a	0.03 ± 0.03b	0.12 ± 0.02a	0.12 ± 0.03a	0.04 ± 0.04b	0.14 ± 0.02a	0.09 ± 0.00a	0.10 ± 0.01a
14	Feruroylquinic acid II	1.18 ± 0.17a	0.66 ± 0.02b	1.20 ± 0.16a	0.99 ± 0.17a	0.58 ± 0.19b	1.05 ± 0.35a	0.77 ± 0.07a	0.92 ± 0.15a
15	Caffeoyl-feruloylquinic acid I	< LOD	< LOD	0.29 ± 0.07a	0.12 ± 0.10a	< LOD	0.23 ± 0.13a	0.06 ± 0.06b	0.06 ± 0.06b
16	Caffeoyl-feruloylquinic acid II	0.23 ± 0.07a	0.04 ± 0.04b	< LOD	0.05 ± 0.05b	< LOD	< LOD	0.09 ± 0.09b	0.13 ± 0.13a
17	Kaempferol-glucoside	0.36 ± 0.03c	0.38 ± 0.03c	0.46 ± 0.00b	0.42 ± 0.07b	0.30 ± 0.90c	0.50 ± 0.08b	0.57 ± 0.02a	0.59 ± 0.06a

Results are expressed as mean ± SD. Different letters indicate significant differences (p < 0.01). LOD: limit of detection. Compound 1 was quantified using quinic acid as reference compound; compounds 2, 4, 7, 8, 9, 10 and 11 using chlorogenic acid; compounds 3, 5, 15 y 16 using caffeic acid; compounds 6 and 14 using ferulic acid; compounds 12 and 13 using rutin; and compound 17 using kaempferol

In accordance with previous observations, flavonols were the second most abundant group of phenolic compounds identified in yerba mate (up to 10% of total phenolics). Among them, rutin was the most abundant one (90%), even more than other hydroxicinnamic derivatives (Bravo et al. 2007; Dartora et al. 2011; Mateos et al. 2018). Rutin presented higher values than those obtained by Dartora et al. (2011) in decoctions of yerba mate.

Feruroylquinic acids and dicaffeoyl quinic acids were the other derivatives of importance found, representing 7 and 5.5% of total polyphenols, respectively. In addition, lower amounts of other hydroxycinnamic derivatives were determined, including caffeoyl-feruloylquinic, caffeoyl glucosides and caffeoyl shikimc acids, with proportions ranging from 4 to 0.25% of total polyphenols. Instead, values obtained for feruloylquinic acids were higher than those reported by Marques and Farah (2009).

Unlike other researchers (Bravo et al. 2007; Mateos et al. 2018), we could not quantify free caffeic acid, even though a minimal amount of quinic acid was quantified, accounting for 2.5% of total polyphenols.

Statistical differences were observed for all compounds between the different decoctions. In general, D2 and D5 presented lower values than the rest of the decoctions, for example, for quinic acid, caffeoylquinic acids, dicaffeoylquinic acids and feruroylquinic acids. On the other hand, D6, D3, D7 and D1 were, for almost all compounds, the decoctions having the highest concentrations. These results are in accordance with those obtained for total polyphenols content.

It is important to remark that the phenolic composition should be properly identified and quantified to better understand their contribution to the expected physiological effects on humans. If positive and consistent results are confirmed over time, this product could be recognized as a functional food (Riachi and Bastos de Maria 2017).

In vitro antioxidant activity

The antioxidant activity was evaluated by two different in vitro methods, FRAP and TEAC, results are shown in Fig. 1. Statistical analysis of the data showed significant differences between decoctions in both methods (FRAP and TEAC). Samples D2 and D5 showed lower values in FRAP (236.65 and 244.33 µmol Trolox/g, respectively); while only D5 had the lowest result in TEAC (192.82 µmol Trolox/g).

D3 showed significantly higher antioxidant activity in both methods, with values of 459.02 and 514.19 umol Trolox/g in FRAP and TEAC, respectively. This yerba mate is flavored with boldo and peppermint. On the other hand, D6, which contains peppermint, pennyroyal, peperina and melissa, also presented higher values in the FRAP assay.

Comparing the results obtained with other common beverages, FRAP values were higher than those reported in three varieties of coffee (Sánchez-González et al. 2005). Furthermore, the in vitro antioxidant activity evaluated in nine commercial samples of black Mauritian tea presented FRAP values between 325 and 927 micromol Trolox/g, and TEAC values between 426 and 1147 micromol Trolox/g (Luximon-Ramma et al. 2005), demonstrating that yerba mate contributes with as much antioxidants to a normal diet as different and common beverages consumed worldwide.

In vivo antioxidant activity

The antioxidant activity of different samples was also evaluated using S. cerevisiae yeast cells as a biological model exposed to oxidative stress caused by H₂O₂ (Baroni et al. 2012). To assess the biological activity of samples in preserving S. cerevisiae cells from induced oxidative stress, cell viability determination was conducted with or without samples as chemoprotectors (exogenous antioxidants). As shown in Fig. 2, selected concentrations from decoction (0.015 mg/mL) were non-cytotoxic to S. cerevisiae, compared to control yeast. This is in accordance with previous reports that state that yerba mate is non-cytotoxic to S. cerevisiae (Bracesco et al. 2003; Piovezan-Borges et al. 2016). When the oxidative stress was induced, yeast cells showed sensitivity to H₂O₂ (2 mM), and only 36% was able to survive the oxidative damage (Fig. 2) (Baroni et al. 2012; Stinco et al. 2015; Di Paola Naranjo et al. 2016; Piovezan-Borges et al. 2016). On the contrary, pretreatment with different decoctions partially suppressed the damage triggered by the oxidant. All yerba mate decoctions used in this study were able to protect yeast from damage by H₂O₂, increasing the survival rate of yeast with respect to H₂O₂ control between 10.68 and 18.38%. The yerba mate sample without addition of herbs presented the highest survival rates (D1), whereas D3, D4 and D6 samples showed the lowest percentages of survival. The remaining decoctions showed intermediate values of survival.

Fig. 2.

Survival rates of S. cerevisiae treated and untreated with different decoctions and/or H₂O₂ (mean ± SD). Survival percentage with respect to control cells. Different letters indicate significant differences (p < 0.01)

Our results are in line with those published by Brasesco et al. (2003) and Piovezan-Borges et al. (2016), who also observed that antioxidants from traditional yerba mate were capable of reducing the damage produced by H₂O₂ to yeast. On the other hand, Baroni et al. (2012) and Lingua et al. (2016) evaluated the ability of Argentinian red wine and pomaces to rescue yeast from oxidative stress, finding similar results to those obtained for the yerba mate samples studied in this work. Furthermore, a study by Martorell et al. (2011), who used flavonoids-enriched cocoa powder, found that it was able to increase the survival rate in 15.39%, similar to the observed percentage in cells treated with Vitamin C (19.28%). These results demonstrate that both traditional and composed yerba mate are rich in antioxidant compounds similar to those obtained in other foods and beverages with recognized antioxidant activity.

Matching between antioxidant activity and phenolic composition

To better understand the possible contributions of polyphenolic compounds to the antioxidant activity, total polyphenol content and antioxidant capacity were correlated. FRAP and TEAC values showed a positive relationship with the TP content for decoctions, with R² coefficients of 0.76 and 0.72, respectively. These results confirm that polyphenolics are the major contributors to the antioxidant capacity of yerba mate (Bravo et al. 2007).

On the other hand, in vivo antioxidant activity did not correlate with the TP content of decoctions. It is known that the antioxidant capacity of polyphenols is determined by its chemical structure (Rice-Evans et al. 1996). For example, flavonoids are able to donate hydrogen atoms through their hydroxyl groups to inactivate a free radical, but they also have the ability to chelate metal ions involved in the oxidation process (Pietta et al. 2000). Various mechanisms could be involved in the in vivo antioxidant activity observed, and it would depend on the composition of the polyphenolic profile rather than on the total content. In this respect, the next step was to study the association of the antioxidant activity with the complete phenolic profile by MRA. This analysis allowed us to identify those compounds that could better explain the observed activity.

Strong and significant correlations were observed in TEAC and FRAP assays with the phenolic profile, obtaining R² of 0.95 and 0.83, respectively. The analysis of the beta coefficients showed those compounds with higher contribution to the antioxidant activity. In TEAC, cryptoclorgenic acid, caffeoylshikimic acid and dicaffeoylquinic acid III showed the highest positive contribution, while quercetin-glucoside, feruroylquinic acid II, neoclorogenic acid and caffeoyl-feruloylquinic acid were those with the most important negative contribution. In the case of FRAP assay, only quinic acid contributed positively to this activity.

A positive correlation between in vivo antioxidant activity and phenolic profile was observed (R² = 0.78), being kaempferol-glucoside, caffeoylshikimic acid and rutin the metabolites that most contributed to it. On the other hand, quercetin-glucoside, dicaffeoylquinic acid I, dicafeoylquinic acid II, and chlorogenic acid contributed negatively to the in vivo antioxidant activity. It is important to remark that the compounds that mainly contribute to the in vivo activity are not the most recognized in yerba mate (eg. caffeoyl quinic acid), instead flavonols were emphasized (Lara et al. 2016; Burris et al. 2012). Results show that different compounds contributed in various ways to the different antioxidant activities. It is known that compounds, according to their chemical structure, will react in a different way in the in vitro assays, since different mechanisms are involved, such as hydrogen atom transfer, single electron transfer, reducing power, and metal chelation (Prior et al. 2005). Thus, taking in account the possibility of polyphenols acting as antioxidants or pro-oxidants, and analyzing results obtained by MRAs, polyphenols with positive contribution to antioxidant activity probably have synergism between them and, therefore, a high antioxidant effect. On the other hand, those compounds with a negative contribution show antagonistic effects, and they probably have pro-oxidant effects in the samples analyzed. Results obtained from the antioxidant assays are the sum of the antioxidant/pro-oxidant effect of each compound.

Conclusion

In the present study, a total of 17 compounds have been identified and quantified by HPLC–DAD-MS/MS (Q-TOF) in decoctions of seven commercial varieties of composed yerba mate and one traditional yerba mate. Most of these compounds belong to the family of the caffeoylquinic acids. Also, decoctions showed high antioxidant activity compared to other foods, which was assessed by two in vitro methods, as well as an in vivo model of S. cerevisiae, indicating that the antioxidants present in I. paraguariensis are able to reduce the damage caused by a strong oxidant such as H₂O₂. It is important to note that the use of multivariate statistical tools allowed us to detect those compounds associated with antioxidant activity.

Significant differences in content, phenolic profile and antioxidant activity were found when comparing yerba mate with and without the addition of herbs. D2 and D5 were the decoctions with the lowest antioxidant characteristics, and both have exactly the same composition (yerba mate plus the addition of pennyroyal, peperina and peppermint), which may suggest that this mixture has somehow antagonistic effects. However, thorough studies need to be conducted in order to confirm this.

It was not possible to relate those decoctions with higher antioxidant characteristics to the addition -or not addition- of herbs to the yerba mate, since different decoctions were more active depending on the assay. However, we can say that D1, without addition of herbs, together with D3 and D6 are generally the most active samples. Further work is necessary to fully understand the synergist or antagonist effect of added herbs on yerba mate activity, as well as the effect of processing that caused the lower biological activity in many samples.

Finally, it could be concluded that the herbs added do not substantially modify the composition and antioxidant activity of the samples under study, and all of them maintain their biological properties.

Author contributions

GC carried out the experiments and wrote de original draft of manuscript; MVB provided the resources and methodology, supervised the work and edit the manuscript; DAW provided the resources and projects administration; RDD provided the resources and methodology, supervised the work and edit the manuscript.

Funding

This work was mainly supported by CONICET [PIP2015-11220150100684]; FonCyT [PICT-2015–2817, PICT 2017–1637, PICTO COVIAR 0123] and SECyT, Universidad Nacional de Córdoba [33620180100522CB (2018–2021)].

Availability of data and material

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Compliance with ethical standards

Conflict of interest

The authors declare that they have no known competing financial interest or personal relationships that could have appeared to influence the work reported in this paper.