The first aspect of this study was to test and to select the best formulation of the herbal mate fermented probiotic beverage. Preliminary tests included changing HME concentrations and decoction conditions (data not shown). Using sensorial and physical criteria, 14 formulations were screened (Table 1) after fermentation using Lactobacillus acidophilus, Lactobacillus sake, Lactobacillus casei subsp. rhamnosus and Lactobacillus casei. After the analysis of appearance, bacterial development, precipitate formation, taste and odor, unacceptable formulations were eliminated and selected the formulation N (Table 1) fermented by L. acidophilus, which contained only HME and honey as the nitrogen and carbohydrate source. This formulation resulted in a sweet taste, moderate acidity, without bottom precipitate and was, thus, demonstrated to be a good substrate for the development of the probiotic strain L. acidophilus ATCC 4356.

Among LAB, L. acidophilus has attracted attention for its potential probiotic effects in human health [16–18] and ability to use prebiotic compounds [19]. When selecting a probiotic culture for use as a dietary adjunct in human food, a number of factors should be considered. One of these is the use of a microorganism originating from the human intestinal tract that exhibits host specificity [20]. Additional factors important in selecting a strain that can survive and grow in the intestinal tract include bile and temperature tolerance, adherence capability to the intestinal wall, cholesterol assimilation and antioxidant ability [18,21,22].

L. acidophilus ATTC 4356, a human isolate, is used as a dietary adjunct in various cultured dairy products. In the present work, strain ATCC 4356 was used to ferment a non-dairy formulation, and showed to not only meet the strict criteria of viability but also capable of surviving storage. L. acidophilus cultivability was evaluated during fermentation and following shelf-life. Figure 1 showed the growth of L. acidophilus, where it was demonstrated to grow exponentially during 10 h of fermentation. During the shelf life period, the microbial count remained similar and was estimated at 10⁸ colony-forming units (CFU) mL⁻¹. The presence of sufficient numbers of viable bacterial cells (10⁹ CFU) is necessary to provide therapeutic benefits [23]. Therefore, in order to call a product “probiotic”, such as in a new beverage, the viability of probiotic bacteria must be maintained. The formulated beverage maintained microbial stability during the 28 days of shelf life at 4 °C. The stable logarithmic probiotic value of 10⁸ CFU ml⁻¹ found during the shelf life may prevent the development of deteriorative/poisoning microorganisms and deliver a sufficient concentration of probiotics to the consumer.

L. acidophilus was shown to easily ferment a substrate rich in glucose and fructose (honey) and to produce lactic and acetic acid (Figure 2) resistant to acidic conditions after fermentation. The successful use of L. acidophilus resulted in a complete fermentation process. The optimum fermentation period of 10 h was determined after longer fermentation tests (data not shown). pH and bacterial growth analyses after 12 h showed a substantial decrease of viability of the cells in an acidic condition at pH 3.67.

Virtanen et al. [16] and Lin and Chang [24] demonstrated that L. acidophilus ATCC 4356 had the ability to scavenge DPPH free radicals. In the present study, antioxidant activities measured directly in the beverage varied from 51 to 62% (Table 2). No positive trend was observed for the confirmation of summation activity from bacterial or lysed cells in the HME.

HPLC analysis for caffeine content in the formulation for 10 h and in the beverage during shelf life indicated almost the same quantitative composition. The average values for each sample are presented in Table 2, based on chromatography using a caffeine standard.

The concentration of caffeine in relation to consumer consumption has been found to be approximately 6.7 mg in one regular dose (100 mL) of beverage (Table 2). Comparing coffee and herbal mate tea (approximately 56 and 52 mg per dose, respectively) [25] with this beverage, 6.7 mg of caffeine was detected, a low amount for ingestion of this alkaloid. Caffeine is one of the most studied alkaloids in terms of physiological effects in human beings [26,27]. Some physiological effects associated with caffeine include central nervous system stimulation, acute elevation of blood pressure, increased metabolic rate, and diuresis [28]. The values observed in the present product indicated that consumption of it provided a little intake of caffeine, an amount so small that not affect consumer preference.

The consumption of herbal mate significantly contributed to the overall antioxidant intake with potentially beneficial biological effects for human health [5]. A number of therapeutic applications have been claimed for herbal mate infusions. I. paraguariensis was shown to contain high antioxidant activity which positively correlated with the concentration of caffeoyl-derivatives [1,3]. Newell et al. [29] demonstrated that herbal mate tea possessed a much higher antioxidant capacity than green tea. According to Bixby et al. [3], herbal mate tea showed greater inhibition against cytotoxicity compared to green tea and or red wine. The ability to quench reactive oxygen species (responsible for cell structure damage after environmental oxidative stress) was examined and correlated to peroxidase-like activity which was related to the polyphenol concentration in herbal mate extract [30]. Higher polyphenol concentrations demonstrated greater peroxidase-like activity [31].

Quenching of the DPPH free radical by the formulations used in this study is shown in Table 2. The formulation during fermentation and beverage during shelf life showed similar amounts of antioxidant activity, approximately 56%. Comparison of the beverage with the study conducted by Ramadan-Hassanien [32] found that the anti-radical performance towards DPPH radicals of the fermented herbal mate beverage was just below those of mango and red grape juice, both rich in phenolic compounds. In other antioxidant beverages, such as hot drinks, especially those rich in caffeine [32], the product demonstrated a large antioxidant capacity, slightly below that of tea with lemon, green tea, black tea, soluble coffee and Turkish coffee.

The acceptance level for the probiotic and herbal mate Tea beverages is presented in Table 3 and supported by variance analysis (Table 4). Results for the sensorial analysis are expressed as a mean value. The mean value of some attributes presented significant differences and could be observed in the frequency distribution for the values of acceptability factors shown in Figure 3. Sweetness, acidity and astringency attributes for all beverages showed a similar acceptance by the testers. Significant differences for transparency and precipitation were expected when comparing fermented to an unfermented beverage, and these differences did not affect the final product acceptance (AF 71 and 82%, respectively).

According to Dutcosky [33], AF ≥ 70% represented good acceptability for the attribute analyzed in a sensorial analysis (Figure 3), and excluding color that presented an AF of 68%, this beverage presented a minimum AF of 71%.

For an overall evaluation, the average values for the sensory analysis and acceptability factors were kept within the acceptance range. No emergent alteration in the values at the final storage period (28 days) was found when compared to the product after fermentation (data not shown). Therefore, product acceptance could be considered as good and comparable to that of commercial beverages available at the local market.

Roasted, milled and sifted leaves and stems of herbal mate (Ilex paraguariensis A.St.-Hil.) were obtained commercially from the local markets in Santa Catarina State (Brazil). HME was prepared by decoction of 15 g in 300 mL of water at 95 °C for 3 min. The extract was filtered under vacuum using a filter paper (Whatman, UK).

The formulations were composed by 300 mL of HME and one or two carbohydrate substrates in the proportions described in the Table 1. Honey (Mel de Abelha Superbom, São Paulo, Brazil), malt extract (Acumedia, USA), yeast extract (Acumedia, USA) and sugar cane molasses (local bioethanol producing company) were used with the aim to prepare formulations for the LAB fermentation.

L. acidophilus ATCC 4356, L. sake ATCC 15521, L. casei ATCC 393 and L. casei subsp. rhamnosus DEBB H-19 (obtained from the Culture Bank of the Bioprocess Engineering and Biotechnology Division, Federal University of Paraná, Brazil) were used in this work. They were grown overnight in MRS (De Man, Rogosa and Sharpe) broth (Acumedia, USA) at 35 °C. The strains were inoculated at a standard concentration of 10⁶ colony-forming units (CFU) mL⁻¹ in 300 mL (3%) of the formulations indicated in Table 1.

The pH was measured directly using a model HI9321 pH meter (Hanna Instruments, Portugal).

Acidity was measured by titrimetric analysis using the fermented broth diluted 1:50 (v/v) in distilled water. The diluted solution was neutralized with 0.1 M NaOH (Merck, Germany) using an alcoholic solution of phenolphthalein (1% v/v) as an indicator.

The fermented broths were diluted 1:10 in MilliQ^® water and filtered using a 0.22 μm pore filter (Millipore, UK). Sugars (glucose and fructose) and organic acids (lactic and acetic) content were determined by HPLC. Chromatographic analyses were performed using a LC-10AD chromatographic system (Shimadzu, Japan) equipped with a model RID-10A detector (Shimadzu, Japan) and an Aminex HPX-87H cation exchange column (300 mm × 7.8 mm i.d., Bio-Rad Laboratories, CA, USA). Analyses were performed using filtered, degassed 5 mM reagent grade H₂SO₄ (Carlo Erba, Italy) as the mobile phase at a flow rate of 0.6 mL·min⁻¹. Eluates were monitored at 215 nm. Calibration curves were obtained by preparing a standard mix of the sugars and organic acids (Sigma, USA). The resulting peaks area were calculated for duplicate 25 μL injections and plotted against concentration. Each assay was carried out in duplicate and the average values expressed in g·L⁻¹ for glucose, fructose, lactate and acetic acid during fermentation were calculated. The estimated error for the chromatographic assays was less than 3%.

Formulations were serially diluted tenfold in 0.05 mol·L⁻¹ sodium citrate (Sigma, USA) buffer, pH 7.5. In order to quantify the cultivable LAB population, MRS agar (Acumedia, USA) was used and incubated at 35 °C for 48 h under anaerobic conditions. Bacterial counts were carried out in triplicate and the standard deviation of mean values was calculated. The estimated error was less than 10%.

After fermentation, the herbal mate probiotic beverage was stored in Scotch flasks at 4 °C during 28 days and the parameters as above were analyzed after 0, 7, 14, 21 and 28 days.

The caffeine content, present in the beverage during shelf life, was analyzed using an Agilent 1100 HPLC system (Hewlett Packard, Inc., USA) with a diode array detector (DAD). Chromatographic separation was accomplished using a Zorbax RX Reversed-Phase C18 column, 150 mm × 4.6 mm (Agilent, USA). An isocratic system (MeOH-H₂O 30:70) was used as the mobile phase at a flow rate of 1.0 mL·min⁻¹ at 35 °C. Detection was performed at 272 nm. Broth and beverage were both diluted (5 mL in 250 mL MilliQ water), filtered through 0.22 μm (Millipore, UK) and injected in triplicate (CV < 5%). Peak areas were compared to standard curves. Suitable amounts of caffeine standard were dissolved in MeOH-H₂O (4:6). Standard solutions were injected in triplicate and peak areas were measured. Linearity was evaluated by linear regression analysis, and the precision and accuracy were determined by the coefficient of variation (CV < 4%). The correlation coefficient was r = 0.988.

The antioxidant activity (free radical scavenging capacity) present in the beverage, during the shelf life, was examined by the reduction of the 1,1-diphenyl-2-picrylhydrazyl radical (DPPH) as described by Ramadan et al. [34] with modifications. Five millilitres of a 20 mg·mL⁻¹ DPPH solution in methanol were added to 5 mL of a methanolic solution of the fermented beverage (1:10, v/v). Absorbance was determined spectrophotometrically (UV-1601PC Spectrophotometer, Shimadzu, Japan) at 515 nm after 30 min., and scavenging activity was calculated as percent of radical reduction. The percent of inhibition was defined as [(A517 blank – A517 sample) A517 blank⁻¹]·100⁻¹ (%). Ascorbic acid was used as a reference compound. The analysis was carried out in triplicate and the standard deviation of mean values was calculated.

Sensorial evaluation of the beverage was performed on the product after 28 days of shelf-life. Sensorial characteristics of the beverage were compared with two commercial herbal mate lemon flavored teas. The commercial products were purchased from a local market. Mate teas A and B correspond to two different commercial brands.

Sensorial hedonic test of the beverage was carried out by a group of 15 non-trained testers who judged the color, transparency, precipitate formation, sweetness, acidity, astringency, saltiness and distasteful after taste using a hedonic rating scale from 1 to 9 (1: dislike extremely; 2: dislike very much; 3: dislike moderately; 4: dislike slightly; 5: neither like nor dislike; 6: like slightly; 7: like moderately; 8: like very much; 9: like extremely). Sensory tests were performed in individual booths, under white light during the morning shift (9:00 a.m.–11:30 a.m.). Refrigerated (5 °C) samples were served in transparent glass cups. The data obtained were analyzed by ANOVA and Tukey’s test according to Monteiro [35] using Assistat version 7.5 software (Assistat, Brazil).

To verify the acceptability of the tested beverages, an acceptability factor (AF) [33] was calculated as the criteria to evaluate each sensorial attribute analyzed:

where A is the mean value obtained for each attribute; B is the maximum value ascribed by the testers for each attribute.