Chemical, microbiological and sensory viability of low-calorie, dairy-free kefir beverages from tropical mixed fruit juices

ABSTRACT The use of non-dairy matrices for probiotic delivery can be beneficial for vegetarian, vegan and lactose intolerant consumers. Stevia rebaudiana extract can be used as sweetener to replace sucrose for glucose intolerants. This research evaluated the use of kefir grains to prepare a low-calorie, fruit-based kefir beverage. Fermentation was carried out by inoculating kefir grains into a mixture of mango and umbu pulps. Two beverages, one containing sucrose, and the other, stevia, were characterized according to their proximate composition and submitted to a 30-day shelf-life period at 5°C, evaluating lactic acid bacteria, yeasts, pH, acidity, color parameters, soluble solids, ascorbic acid, sensory acceptance, and purchase intention every 10 days. Lactic and acetic acids were analyzed at the beginning and end of the shelf-life. Both beverages showed probiotic potential. The stevia-sweetened beverage had lower calorie content and good sensory acceptance. These results support the possibility to develop low-calorie, fruit-based kefir beverages.


Introduction
Kefir is a fermented, probiotic beverage from eastern Europe (Arslan, 2015). Kefir grains are an aggregation of yeasts, lactic acid and acetic acid bacteria that coexist in symbiosis and are enveloped by a matrix of polysaccharides (Alsayadi et al., 2014). The most common substrates utilized for kefir are water with brown sugar and cow's milk. On the other hand, few studies have evaluated the viability of fermenting alternative substrates such as fruit juices for those that suffer from intolerance to dairy products (Laureys & De Vuyst, 2017). Umbu (Spondias tuberosa) and mango (Mangifera indica L.) are common fruits in the northeast region of Brazil, are rich in nutrients such as carotenoids and ascorbic acid, and are used to produce juices for their highly appreciated sensory characteristics (Jahurul et al., 2015;Teixeira et al., 2019). The use of such tropical fruits as fermentation substrates has the potential to insert new kefir beverages onto the food market.
The development of new beverages by combining prebiotic carbohydrates with probiotic microorganisms shows potential on the food market. Fructooligosaccharides (FOS) are considered as prebiotic carbohydrates for their potential to serve as substrates for gut microbiome and for the probiotic microorganisms present in food products, when used in combination (Freire et al., 2017). Elaborating a synbiotic beverage by combining FOS with kefir shows the promise to provide the food market with a product rich in beneficial effects.
The high consumption of sugar-sweetened beverages is a result of the abundant use of sucrose in the Brazilian diet and is associated with the risk of developing obesity. Studies have found that the use of artificially sweetened beverages could reduce weight and promote overall health (Borges et al., 2017) since artificial sweeteners usually have a zerocalorie content.
Stevia rebaudiana is a plant found in South America that contains glycosides, which has sweetening potential up to 300 times higher than that of sucrose and a zero-calorie content (Narayanan et al., 2014). Weber and Hekmat (2013) evaluated the influence of stevia extract on the survival of lactobacillus in yogurt and found that the presence of the sweetener did not result in impairment of the probiotic potential of the beverage. On the other hand, Denina et al. (2014) found that stevia had inhibitory potential in the multiplication of Lactobacillus reuteri, which leads to the need for further evaluation of the use of stevia in probiotic foods.
To ensure the probiotic effects of a product, a minimum value of viable cells is recommended, ranging from 10 6 to 10 7 CFU −1 at the expiry date (Shori, 2016), which can be a challenge when using a dairy-free food matrix. Therefore, the objective of the present work was to evaluate the viability of producing low-calorie, synbiotic mixed fruit flavored kefir beverages with umbu and mango.

Materials and methods
Pasteurized umbu and mango pulps were bought from Doce Mel (Brazil), Fructooligosaccharide was bought from New Nutrition (Brazil), sucrose (União, Brazil) was bought from the local market, water kefir grains were donated by the Federal University of Bahia Recôncavo (Brazil), stevia powder extract (96.50% total stevioside, 60% rebaudioside A) was donated by Sweetmix (Brazil), xantham gum and ascorbic acid were donated by Sigma-Aldrich (United States of America) and Granolab (Brazil), respectively.

Preparation of the kefir beverages
Each beverage formulation was prepared according to the fruit pulp proportions (Table 1). The frozen pasteurized fruit pulps were thawed and homogenized with the water volume defined for each beverage (Table 1). The substrates were packed in sterile glass vials, sealed with screw caps and refrigerated for further fermentation. Cultivation of the kefir grains was carried out according to Tavares et al. (2018). During a seven-day period, the water kefir grains were activated at room temperature in a solution containing water and brown sugar at 12 °Bx. After this period, the grains were inoculated in a proportion equivalent to 10% of the volume of the beverages and allowed to ferment at 25°C in a biological oxygen demand (BOD) incubator (CE-300/350 model, Cienlab, Brazil) for a period of 24 hours. After the fermentation process, the water kefir grains were removed from the fermented beverages and the remaining ingredients (sucrose/stevia, xanthan gum, fructooligosaccharide and ascorbic acid) were added and homogenized. The xanthan gum was added to avoid phase separation, fructooligosaccharide was added to provide the beverages with synbiotic potential (Brazilian Health Regulatory Agency (Anvisa), 2019), and ascorbic acid was added as a preservative. Stevia was added considering its maximum daily ingestion of 4 mg/kg of body weight, as proposed by Food and Agriculture Organization of the United Nations (FAO) (2019).

Acceptance test
The sensory analysis of the kefir beverages was carried out by 65 untrained tasters, both male and female. The tasters were asked to indicate how much they liked or disliked each beverage on a hybrid 10 cm hedonic scale (Villanueva et al., 2005). The samples were served in clear, 50 mL tulip-shaped glasses. The two beverages with the highest acceptance scores were chosen to carry out the following analyses as shown below (one sample from each group).

Proximate composition
The two most accepted kefir samples were characterized for their moisture, ash, carbohydrate, protein, fat, and total calories contents, according to the techniques established by Association of Official Analytical Chemists (AOAC) (2019). The moisture content was determined using an infrared moisture analyzer (IV3100 model, Gehaka, Brazil). For the ash content, the samples were incinerated and then weighed. The protein content was determined by the Kjeldahl method and the fat content was determined using the Bligh and Dyer (1959) method. The carbohydrates were measured as the difference from 100% of the sum of the other macromolecules and the calories were calculated considering 4 kcal per 1 g of carbohydrates and proteins and 9 kcal per 1 g of fat.

Shelf-life analysis
The two most accepted kefir beverages were evaluated for their physicochemical, microbial, and sensory stability during a 30-day period. Each beverage was produced, packed in sterile glass vials and maintained under refrigeration at 5°C in a BOD incubator Cienlab,Brazil). Samples were taken every 10 days to carry out the following analyses: pH, titratable acidity, vitamin C, total soluble solids, color (AOAC, 2019), lactic acid bacteria count, yeast count (American Public Health Association [APHA], 2015), sensory acceptance and purchasing intent (Meilgaard et al., 2007). The titratable acidity was determined by titration with 0.01 N NaOH solution (AOAC, 2019). The pH was determined by direct reading using a potentiometer (K39-1014B model, Kasvi, Brazil). The vitamin C content was determined by a redox titration using an iodine solution (AOAC, 2019). The total soluble solids were determined using an analogical refractometer (DR 201-95 model, Kruss, Germany). The color was analyzed using a benchtop colorimeter (CR 5 model, Konica Minolta, Japan) in the transmittance mode, obtaining values for L*, a* and b*. The total color difference (ΔEordE) was also calculated, using the following equation: ΔE ¼ ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi For the Lactic Acid Bacteria (LAB) count, serial dilutions were prepared in saline solution, inoculated onto MRS agar plates (Kasvi, Brazil) supplemented with cycloheximide (100 mg/L), incubated for 72 hours at 30°C and the colony formation units (CFU) counted. For yeasts, serial dilutions were prepared, inoculated onto Saboraud Agar (Kasvi, Brazil) containing chloramphenicol (500 mg/L) and incubated for 120 hours at 30°C.
Lactic and acetic acids were identified and quantified by High Performance Liquid Cromatography (Series 200 model, Perkin Elmer, United Kingdom) using a 220 mm-× 4,6 mm × 10 μm polypore H column (Perkin Elmer, United Kingdom), injection volume: 10 μL, Uv-Vis detector at 220 nm, flow: 0.8 mL/min and mobile phase: ultra-pure water acidified with H 2 SO 4 to pH 2.0. The peaks corresponding to each acid were identified from the retention times according to the standards.
A sensory acceptance test was carried out every 10 days by the same tasters who took part in the preliminary analysis, again using a hybrid 10 cm hedonic scale (Villanueva et al., 2005), and also determining the purchasing intent of the products using a hedonic scale ranging from 1 (would definitely not buy the beverage) to 5 (would definitely buy the beverage). The sensory analysis was approved by the Ethic Committee of the Nutrition School -Federal University of Bahia (project number 1.759.169).

Statistical analysis
A one-way ANOVA was carried out to compare the means of the 12 samples in the preliminary acceptance test, and to compare the mean results for each beverage individually during the shelf-life. The student T test was used to compare the mean values obtained for the samples in the physicochemical characterization, and to compare the means obtained for the two beverages at the same shelf-life time. Pearson's correlation test was also applied between all shelflife analyses. The software STATISTICA 7.0 was used considering a 95% significance level. Figure 1 shows the mean acceptance scores awarded to each sample in the preliminary sensory analysis. The beverages with the lowest fruit pulp and sweetener concentrations showed the worst results. Samples 6 and 12 had the highest concentrations of fruit pulp and sweetener and the highest scores amongst the samples, both with scores above 6.50 points and were therefore chosen for the characterization and shelf-life analyses.

Results and discussion
The beverages sweetened with stevia obtained lower scores when compared with the beverages sweetened with sugar. This finding suggests that the sweetening capacity of stevia may incur undesirable effects such as a slight bitter taste, that could lower the overall score of the beverage (Singla & Jaitak, 2016). Table 2 reports the results of the proximate composition for the two beverages with highest acceptance. The results for moisture and carbohydrate showed statistical differences between the two samples, probably due to the addition of sucrose to sample 6, which impacted on the total calorie count.
The low number of calories in sample 12 is of commercial interest since the demand for sweetened beverages and low-calorie products has been increasing on the international market. Studies have also shown that a high consumption of sucrose-sweetened beverages may be associated with an increased chance of acquiring type 2 diabetes mellitus and other chronic diseases (Singh et al., 2015). In addition, the evidence suggests a beneficial effect on the body mass index when sucrose-sweetened beverages are replaced by lower-calorie beverages, which may contribute to a reduction in the risk of developing chronic noncommunicable diseases (Zheng et al., 2015).
The use of stevia shows the promise of contributing to a lower energy intake and the promotion of a balanced calorie consumption, while being a relatively safe sweetener considered as GRAS (Generally Recognized as Safe) (Borges et al., 2017). Both samples had low fat and protein contents, since fruits are usually not rich in these nutrients (O´Shea et al., 2015). Kefir fermentation is known to produce peptides, which have been associated with some beneficial effects in the organism, such as the inhibition of lipogenesis and modulation of oxidative damage (Tung et al., 2017). Regarding the ash content, the results found were similar to those described by Natal et al. (2017), while working with mango juices. Figure 2 shows the physicochemical behavior of the beverages during the 30-day shelf-life period. Regarding the pH and acidity, as expected, these parameters showed opposite behaviors during the period, with a strong negative correlation between them for both the sucrose sweetened kefir (−0.97, p < .05) and the stevia sweetened kefir (−0.81, p < .05). The microorganisms present in the kefir grains are responsible for the production of acetic and lactic acids. Tavares et al. (2018) reported that the production of lactic acid during the fermentation of kefir is of great importance, due to its inhibitory effect on both spoilage and pathogenic microorganisms.
Regarding the total soluble solids (TSS), there was a significant difference between the samples, which can be explained by the fact that stevia is considered a non-caloric sweetener. On the other hand, no statistical difference was observed between different shelf-life periods for the same sample. Different results were identified by Khatoon and Gupta (2015), who found a lower TSS count after 7 days of storage. A strong positive correlation with lactic-acid bacteria count (0.89, p < .05) was only observed for the stevia sweetened kefir which could be associated with the fact that the soluble solids serve as an important substrate for the survival of these microorganisms.
Regarding the ascorbic acid, Oliveira et al. (2013) suggested that a loss of more than 50% of the initial content of ascorbic acid at the end of a product's shelf-life is a determinant of low-quality storage. In the present work, the loss of ascorbic acid was lower for both beverages (23.63% for the sucrose sweetened beverage and 15.57% for the stevia sweetened beverage), which is a good predictor.
Negative correlations were observed between ascorbic acid, ΔE (−0.94 for both samples, p < .05) and total acidity (−0.92 for sucrose and −0.95 for stevia sweetened kefirs, p < .05). A positive correlation was found with pH (0.95 for sucrose and 0.90 for stevia sweetened kefirs, p < .05). These correlations are justified since ascorbic acid can degrade into browning compounds due to both aerobic and anaerobic reactions in aqueous acidic solutions (Oliveira et al., 2013). Results are shown as means ± standard deviation. Equal lower-case letters indicate that there was no significative difference between results in columns at p ≤ 0.05.

Figura 2.
Concentraciones de pH (a), acidez (b), sólidos solubles totales (TSS) (c) y ácido ascórbico (d) durante los 30 días de vida útil. Las barras de error representan las desviaciones estándar. Las mismas letras minúsculas señalan que no hay diferencias significativas para la misma muestra en distintos periodos de vida útil a p ≤ 0.05. La misma letra mayúscula indica que no hay diferencias significativas para las distintas bebidas durante el mismo periodo de vida útil analizado a p ≤ 0.05. Representa la bebida endulzada con stevia y representa la bebida endulzada con sacarosa. Figure 3 shows the microbial counts during the 30-day shelflife period and the lactic/acetic acid concentrations. Regarding the lactic acid bacteria count, for both samples, the T30 count was statistically lower than that at the other periods analyzed. The increased acidity level (correlation of −0.97 for the sucrose and −0.82 for the stevia sweetened beverages, p < .05) and the decrease in pH (correlation of 0.99 for both samples, p < .05) could justify the reducing levels of microorganisms. The counts for all the periods analyzed showed more than 7.00 log 10 CFU per recommended portion of the product (250 mL), and thus the products can be considered as potentially probiotic up to the end of the shelf-life period, according to Minelli and Benini (2008). The main benefits associated with the consumption of the probiotic microorganisms in kefir are improved digestion, a hypocholest-erolemic effect, glycemic control, an antihypertensive effect, anti-inflammatory effect, antioxidant activity, anti-carcinogenic activity and anti-allergenic effect .
The period found for the shelf-life of the probiotics in the present work can be considered in accordance with data found in the literature, with results ranging from 7 (Nualkaekul et al., 2013) to 28 (Da Costa et al., 2017) or more days in similar studies. The use of stevia as a sweetener apparently did not affect the overall shelf-life period of the beverage studied, even though a statistical difference was observed for the LAB counts on the 20 th and 30 th days of analysis. Esmerino et al. (2013) obtained similar results.
With respect to the behavior of the yeasts, the results obtained at the different times of analysis were similar to those obtained by other authors Randazzo et al., 2016). The T30 yeast counts were also statistically lower than at the other times analyzed for both samples.
Saccharomyces cerevisiae is the most common yeast found in kefir beverages and has been associated with some beneficial effects, such as inhibition of the growth of Shigella sp. and the cytotoxicity of Clostridium difficile toxins (Bolla et al., 2013). Strong correlations were found with pH (0.91, p < .05) and acidity (−0.96, p < .05), but only for the sucrose-sweetened kefir, which could be due to the fact that other variables, such as the soluble solids content could have a more important correlation with the yeast count in the stevia sweetened kefir. Figure 3(d) shows the behavior of the lactic and acetic acids at the beginning and end of the shelf-life period. The lactic acid concentration was highest at the end of the fermentation period and decreased at the end of the shelflife and similar results were found by Puerari et al. (2012). Acetic acid showed the opposite behavior, with its concentration increasing at the end of the shelf-life for both samples. Lactic and acetic acids are responsible for the unique kefir flavor and they also protect the beverage from spoilage by microorganisms (Puerari et al., 2012;Viana et al., 2017). The modification in the concentration of these acids could be justified by the fact that some lactic acid bacteria are not only capable of producing lactic acid, but of using it as a substrate for fermentation, producing acetic acid (Elferink et al., 2001).
As expected, the LAB count had a strong positive correlation with the lactic acid concentration (0.99, p < .05), since and represent results for stevia and sucrose-sweetened beverages, respectively, regarding acetic acid concentration.
It must be pointed that the fruit juices were also evaluated before fermentation in relation to the pH, lactic acid bacteria and yeast counts. The medium value for pH was 3.81 ± 0.17 and there was no lactic acid bacteria nor yeast found. Immediately after the innoculation of the water kefir grains, the beverages, whilst not fermented, already showed microbial counts, with a 5.21 ± 0.07 log 10 CFU/mL of LAB and 5.79 ± 0.10 log 10 CFU/mL of yeast. These findings may indicate a very mild fermentation process, in comparison with results reported by Randazzo et al. (2016). Figure 4 shows the results for the color parameters and total color difference for both samples during the shelf-life period. The sugar sweetened sample showed different overall color values than the stevia beverage, probably due to the different formulations, with the stevia beverage containing more water and no added sucrose. The samples presented a decrease in all the color parameters during the shelf-life period and the same result was observed by M.G.M. Costa et al. (2013). Table 3 shows the results obtained in the sensory analyses carried out during the shelf-life of the beverages. There was a statistical difference between the samples, with the sucrose beverage presenting greater acceptance and higher purchase intention scores, which suggests more familiarity with its taste as compared to stevia. Similar results were found by Alizadeh et al. (2014).
The longer the shelf-life, the lower the score found for both beverages in both analyses. K.K.F.D. Costa et al. (2017) obtained similar results studying a probiotic beverage, and some tasters described an acidic taste in both samples in the last shelf-life analysis.
The changes observed on the 10 th day of the shelf-life in the instrumental color analyses did not appear to influence acceptance of the beverage, since a statistical difference was only observed for global acceptance on the last day of analysis and a noticeable difference in ΔE was already present on the 20 th day. Dias et al. (2012) discussed the fact that, although color is an important sensory characteristic, product flavor may have a greater impact on its acceptance.
The acidity and pH could be the parameters responsible for the decrease in beverage acceptance, since strong correlations between acceptance and acidity (−0.98 for sucrose and −0.85 for stevia, p < .05) and pH (0.99, p < .05 for both Represents the stevia sweetened beverage and represents the sucrose sweetened beverage. Figura 4. Análisis de L* (a), a* (b), b* (c) y ΔE (d) durante los 30 días de vida útil. Las barras de error representan las desviaciones estándar. Las mismas letras minúsculas señalan que no hay diferencias significativas para la misma muestra en distintos periodos de vida útil a p ≤ 0.05. La misma letra mayúscula indica que no hay diferencias significativas para las distintas bebidas durante el mismo periodo de vida útil analizado a p ≤ 0.05.
Representa la bebida endulzada con stevia y representa la bebida endulzada con sacarosa. samples) could indicate that the tasters who took part in the sensory tests preferred less acidic beverages. Salmerón et al. (2015) discuss that the production of acetic and lactic acids by the kefir microorganisms potentially affect both physicochemical parameters and the overall sensory characteristics of the beverages. The stevia-sweetened kefir beverage showed global acceptance above 5.00 throughout the whole shelf-life and buying intention above 2.50 for the first three periods analyzed, which could show a willingness to accept stevia as a substitute for sucrose. Similar scores were found by Randazzo et al. (2016) when testing different fruit juices as substrates for kefir fermentation. Pimentel et al. (2015) also found similar values when working with probiotics sweetened with sucralose, however sucralose is derived from substances considered toxic, so the use of stevia may benefit from being a natural substance with less risk of toxicity (Tandel, 2011).
Functional foods are part of a new conception of food, since, in addition to their nutritional compositions, they beneficially affect one or more functions of the organism. Within this context, the relevance of the present study was to respond to an increasing demand for functional products on the food market (Markowiak & Slizewska, 2017). The interaction between the in vivo probiotic and prebiotic functions in a synergistic way is of interest, since prebiotics are known to increase the viability of some microbial strains (Pandey et al., 2015) and promote even greater beneficial effects, other than those intrinsic to them separately.

Conclusion
The beverages differed in their total calories, due to the noncaloric quality of the stevia extract, giving it commercial potential. Both beverages could be considered potentially probiotic due to their lactic acid bacteria and yeast count during the storage period and had good sensory acceptance. Although the instrumental color analyses showed significative differences, these differences were only perceived in the sensory evaluations after 20 days of storage. The results showed that both beverages functioned as viable matrices for kefir and could be implemented on the food market as new synbiotic products, with the stevia beverage having the advantage of its low caloric content and that it is a natural sweetener with low toxicity. The beverages could serve as promoters of a healthier diet and a balanced gut microbiome, especially to those suffering from dairy intolerance since they contain beneficial microorganisms and nondigestible carbohydrates functioning in a synergic way. The use of a fruit-based matrix can increment the daily consumption of vitamin-rich products and promote an adequate fruit ingestion.