Probiotic Enrichment and Reduction of Aflatoxins in a Traditional African Maize-Based Fermented Food

Fermentation of food products can be used for the delivery of probiotic bacteria and means of food detoxification, provided that probiotics are able to grow, and toxins are reduced in raw materials with minimal effects on consumer acceptability. This study evaluated probiotic enrichment and detoxification of kwete, a commonly consumed traditional fermented cereal beverage in Uganda, by the use of starter culture with the probiotic Lactobacillus rhamnosus yoba 2012 and Streptococcus thermophilus C106. Probiotic kwete was produced by fermenting a suspension of ground maize grain at 30 °C for a period of 24 h, leading to a decrease of the pH value to ≤ 4.0 and increase in titratable acidity of at least 0.2% (w/v). Probiotic kwete was acceptable to the consumers with a score of ≥6 on a 9-point hedonic scale. The products were stable over a month’s study period with a mean pH of 3.9, titratable acidity of 0.6% (w/v), and Lactobacillus rhamnosus counts >108 cfu g−1. HPLC analysis of aflatoxins of the water-soluble fraction of kwete indicated that fermentation led to an over 1000-fold reduction of aflatoxins B1, B2, G1, and G2 spiked in the raw ingredients. In vitro fluorescence spectroscopy confirmed binding of aflatoxin B1 to Lactobacillus rhamnosus with an efficiency of 83.5%. This study shows that fermentation is a means to enrich with probiotics and reduce widely occurring aflatoxin contamination of maize products that are consumed as staple foods in sub-Saharan Africa.


Introduction
The sustainable production of traditional foods in sub-Saharan Africa offers a viable opportunity to fight increasing hunger and malnutrition [1]. Cereals such as millet, sorghum, and maize are important sources of food in Africa [2], and are predominantly cultivated for human nutrition, in particular, for children [3][4][5]. The maize-based African fermented kwete is a fermented beverage, which is traditionally produced by Luo communities, but now commercially available in many 2012 strain has been previously applied for the preparation of African traditional products like uji (fermented maize), mutandabota (fermented pulp of the baobab fruit and milk), zomkom (fermented wheat), and the naturally fermented milk lait caillé [1,31,33]. Recently, Di Stefano et al. [34] reported the growth parameters, organoleptic characteristics, and acceptability of fermented millet by use of L. rhamnosus GR-1 and S. thermophilus C106, which provided a good reference point for this study. The inclusion of probiotic starters in a product may affect the products' sensory properties and, hence, acceptability [35][36][37]. It is therefore not only essential to ensure growth of the probiotic, but to also compare sensory characteristics and consumer acceptability of the probiotic-enriched, traditional fermented products.
In this paper, we evaluated the ability of L. rhamnosus yoba 2012 and S. thermophilus C106 to propagate in kwete, a traditional maize-based drink. In addition, we studied survival of the probiotic during storage and compared consumer acceptability with commercial traditional products previously introduced on the market. We also monitored the effect of the probiotic starter culture on the levels of aflatoxins during fermentation, and confirmed the ability of the probiotic to bind aflatoxin B 1 . The results of this study demonstrate the potential of utilizing widely consumed locally available traditional foods as carriers for probiotics, which adds health benefits and improves product safety.

Ingredients for Probiotic Kwete
The dried probiotic starter culture comprising L. rhamnosus yoba 2012 and S. thermophilus C106 [31] was obtained from the Yoba for Life Foundation (Amsterdam, The Netherlands) and stored at −40 • C prior to use. Hulled maize flour for preparation of probiotic kwete was purchased from Maganzu millers, Kampala, Uganda, and kept at room temperature in a dry place prior to use.

Preparation of Probiotic Kwete
Probiotic kwete was prepared using a modification of the traditional method described by Muyanja and Namugumya [8]. To prepare one liter of probiotic non-alcoholic kwete, 150 g of hulled maize flour was mixed with 100 mL of water to form dough. The dough was roasted to golden brown on a saucepan over a hotplate with continuous turning to prevent burning. The roasted dough was subsequently mixed in one liter of boiled water to form a slurry. The slurry was boiled for 15 minutes to form porridge and to inactivate all unfavorable microorganisms. The porridge was subsequently cooled down to 45 • C followed by inoculation with prefermented maize porridge of 2% (v/v). The porridge was left to ferment at 30 • C for 24 h, with the acidity and pH monitored at t = 0, 3, 6, 9, 12, and 24 h. Bacterial counts were taken at t = 0, 12, and 24 h. The experiment was carried out in triplicate on separate days. The shelf stability of probiotic kwete was determined by physicochemical, microbiological, and consumer acceptability tests. Samples were fermented for 24 h and stored in sterile bottles under refrigeration at 4 • C. Analyses were done at weekly intervals for a period of four weeks.

Inoculation Approaches
In this study, we evaluated the production of probiotic kwete by the use of three inoculation approaches for fermentation at 30 • C. These approaches included (i) prefermented milk (ii) direct inoculation with the dried starter culture, and (iii) prefermented maize porridge. For the first procedure with prefermented milk, one gram of probiotic dried starter culture was used for one liter of milk and incubated overnight at 37 • C, as described previously [38]. Subsequently, the kwete base was inoculated with 2% (v/v) of the prefermented milk. The second inoculation was performed by direct addition of one gram of the dried starter culture to one liter of kwete base. The third inoculation procedure includes 2% prefermented maize porridge. This porridge was obtained by 50 g of hulled maize flour mixed with one liter of water to a final concentration of 5% (w/v). The mixture was boiled to obtain a thick porridge followed by cooling down to 45 • C and inoculation with one gram of dried starter culture. Fermentation was carried out at 30 • C for 24 h. Samples during fermentation after all three inoculation procedures were taken at t = 0 and 24 h for analysis of pH, acidity, and colony forming units of L. rhamnosus yoba 2012 and S. thermophilus C106.

Enumeration of Colony Forming Units
Serial dilutions of probiotic kwete samples were prepared by using four-times-diluted Ringer's solution. Counts of lactic acid bacteria (LAB) were determined by pour plating of selected serial dilutions in sterile MRS agar for L. rhamnosus yoba 2012 and M17 agar for S. thermophilus C106, followed by incubation anaerobically at 30 • C for 48 h. Yeast counts were determined by surface spreading of selected dilutions in potato dextrose agar and incubating at 30 • C for 2-5 days. Coliform counts were determined by pour plating selected serial dilutions in violet red bile lactose agar and incubation at 30 • C for 48 h. All media were obtained from CONDA Laboratories (Madrid, Spain).

Determination of Titratable Acidity and pH
Titratable acidity was determined by weighing 10 mL of the kwete sample. The sample was subsequently filtered through Whatman (Whatman International Ltd, Maidstone, England) number 1 filter paper. The filtrate was titrated against a standardized solution of 0.1 N NaOH with phenolphthalein as the indicator [39]. The experiment was performed in triplicate. The pH was determined using a bench top FiveGo digital pH meter (Mettler Toledo, Greifensee, Switzerland) which was calibrated using DKD-certified buffers (Mettler Toledo) of pH 4.00, 7.00, and 9.20 prior to analysis. The pH of the samples was determined in duplicate.

Consumer Acceptability of Probiotic Kwete
A panel of 62 Ugandan students (equal ratio males and females) evaluated the acceptability of the probiotic kwete one day after fermentation in a double-blind study. The acceptability of the probiotic was compared with non-probiotic, traditionally prepared kwete [8]. Panelists ranked their acceptability of various attributes using a 9-point hedonic scale [40]. Water bottles were provided to rinse the palate in between tasting of samples. The group means differences between probiotic and traditionally made kwete (control) were analyzed using a t-test. All statistical analyses were performed using XLSTAT software (version 2012.4.03, Addinsoft, Paris, France).

Total Aflatoxin Concentration in Kwete
To evaluate the potential of L. rhamnosus yoba 2012 in mitigating the risk effect of aflatoxins in a maize-based traditional food, 20 mL of kwete base was spiked with 1.25 mL of 120 ng mL −1 total aflatoxins (40.0 ng mL −1 B 1 , 40.0 ng mL −1 G 1 , 20.0 ng mL −1 B 2 , and 20.0 ng mL −1 G 2 ). The aflatoxin standards were obtained from Bioo Scientific (Austin, Texas, USA). Control experiments were set as follows: unfermented kwete without aflatoxins, fermented kwete without aflatoxins, unfermented kwete base with aflatoxins, unfermented kwete base with 0.92% lactic acid (pH 4.4) with aflatoxins, and kwete spiked with aflatoxins after fermentation. Incubations were carried out for 24 h at 30 • C with samples taken at t = 0, 12, and 24 h for aflatoxin B 1 , B 2 , G 1 , and G 2 quantification. The water-insoluble phase of kwete was removed by centrifugation at 3000g for 20 minutes at room temperature. The supernatant was applied to an immunoaffinity column according to the instructions of the manufacturer (AFLASTAR TM R Romer Labs Inc, Union, Missouri, USA). Briefly, 0.5 mL of the extract containing aflatoxins (B 1 , B 2 , G 1 , and G 2 ) was diluted to 2.5 mL with deionized water prior to clean-up using the immunoaffinity column. The column was then washed with 4 mL of 16% methanol to remove any unbound aflatoxins, and the bound aflatoxins were eluted using 2 mL absolute methanol. Aliquots of 100 µL of the extract were injected into the HPLC column equilibrated with methanol. The aflatoxins were eluted using a methanol/acetonitrile/water (8:27:65 (v/v)) solution at a flow rate of 0.7 mL/min. Detection and quantification were performed by a fluorescence detector operated at excitation and emission wavelengths of 365 nm and 450 nm, respectively.

Aflatoxin B 1 Binding to Lactobacillus rhamnosus
In order to further substantiate the mechanism for the reduction of aflatoxins during the controlled fermentation of kwete, the binding affinity of L. rhamnosus yoba 2012 to aflatoxin B 1 was determined. Briefly, the probiotic L. rhamnosus yoba 2012 was cultured in de Man, Rogosa, Sharpe (MRS) broth with 0.1% (v/v) Tween 80 at 37 • C in an atmosphere of air containing 5% CO 2 for 24 h. The cell pellet was collected by centrifugation at 3200g for 10 minutes at room temperature, and washed twice with physiological saline to remove excess MRS broth. The cell pellet was resuspended and serially diluted with physiological saline to generate nine different concentrations in optical densities ranging from 0 to 1 using Ultrospec 2100 pro spectrophotometer (Amersham Biosciences, Piscataway, New Jersey, USA) set at 600 nm. These dilutions were prepared in a microtiter plate and centrifuged to remove supernatant prior to aflatoxin B 1 binding. An aflatoxin B 1 solution in physiological saline of 1.0 µg mL −1 was added to bacterial cell pellets in a microtiter plate, and the cell suspension was incubated at 37 • C for 30 minutes, followed by centrifugation at 3200g for 10 minutes at room temperature. The residual aflatoxin B 1 in the supernatant was analyzed by fluorescence spectroscopy using the Fluostar Omega microplate reader (BMG Labtech, Ortenberg, Germany) operating with a 390 nm excitation filter and a 480 nm emission filter. The fluorescence of the residual aflatoxin B 1 was plotted versus the cell concentrations. Curve fitting was carried out with GraphPad Prism version 7 (GraphPad Software, San Diego, CA, USA).

Fermentation of Kwete Using the Yoba Starter Culture
The L. rhamnosus yoba 2012 and S. thermophilus C106 bacteria propagated well in the kwete base with notable changes in pH and acidity ( Figure 1). Bacterial growth resulted in lactic acid production and an increase in titratable acidity from 1.8‰ to 7.0‰, and a decrease of the pH from 6.2 to 3.9 after 24 h of fermentation ( Figure 1). Fermentation of the probiotic cereal-based kwete with a pH of 4.4 and an acidity of 4.5‰ after twelve hours is relatively slow compared to milk fermentation with the same bacterial starter culture [38].
Nutrients 2018, 10, x FOR PEER REVIEW 5 of 15 equilibrated with methanol. The aflatoxins were eluted using a methanol/acetonitrile/water (8:27:65 (v/v)) solution at a flow rate of 0.7 mL/min. Detection and quantification were performed by a fluorescence detector operated at excitation and emission wavelengths of 365 nm and 450 nm, respectively.

Aflatoxin B1 Binding to Lactobacillus rhamnosus
In order to further substantiate the mechanism for the reduction of aflatoxins during the controlled fermentation of kwete, the binding affinity of L. rhamnosus yoba 2012 to aflatoxin B1 was determined. Briefly, the probiotic L. rhamnosus yoba 2012 was cultured in de Man, Rogosa, Sharpe (MRS) broth with 0.1% (v/v) Tween 80 at 37 °C in an atmosphere of air containing 5% CO2 for 24 hours. The cell pellet was collected by centrifugation at 3200g for 10 minutes at room temperature, and washed twice with physiological saline to remove excess MRS broth. The cell pellet was resuspended and serially diluted with physiological saline to generate nine different concentrations in optical densities ranging from 0 to 1 using Ultrospec 2100 pro spectrophotometer (Amersham Biosciences, Piscataway, New Jersey, USA) set at 600 nm. These dilutions were prepared in a microtiter plate and centrifuged to remove supernatant prior to aflatoxin B1 binding. An aflatoxin B1 solution in physiological saline of 1.0 µg mL −1 was added to bacterial cell pellets in a microtiter plate, and the cell suspension was incubated at 37 °C for 30 minutes, followed by centrifugation at 3200g for 10 minutes at room temperature. The residual aflatoxin B1 in the supernatant was analyzed by fluorescence spectroscopy using the Fluostar Omega microplate reader (BMG Labtech, Ortenberg, Germany) operating with a 390 nm excitation filter and a 480 nm emission filter. The fluorescence of the residual aflatoxin B1 was plotted versus the cell concentrations. Curve fitting was carried out with GraphPad Prism version 7(GraphPad Software, San Diego, California, USA).

Fermentation of Kwete Using the Yoba Starter Culture
The L. rhamnosus yoba 2012 and S. thermophilus C106 bacteria propagated well in the kwete base with notable changes in pH and acidity ( Figure 1). Bacterial growth resulted in lactic acid production and an increase in titratable acidity from 1.8 ‰ to 7.0 ‰, and a decrease of the pH from 6.2 to 3.9 after 24 hours of fermentation ( Figure 1). Fermentation of the probiotic cereal-based kwete with a pH of 4.4 and an acidity of 4.5 ‰ after twelve hours is relatively slow compared to milk fermentation with the same bacterial starter culture [38].

Acceptability of Probiotic Kwete
The consumer acceptability scores of probiotic kwete in comparison to traditionally made kwete assessed by a panel of 62 university students are shown in Table 1. The acceptability scores for color, aroma, and overall acceptability of probiotic kwete were generally comparable (p > 0.05) to the local (traditional) kwete on the market. Although the taste of probiotic kwete was highly acceptable, it was quite different from the locally made type, which had very sour and slightly alcoholic flavors with a pH of 3.4 ± 0.1. The latter pH value was similar to those reported by Muyanja and Namugumya [8]. The acceptability scores for probiotic kwete ranged from 3 to 9 ('dislike moderately' to 'like extremely'). Table 1. Comparison of the consumer acceptability scores of probiotic kwete produced by L. rhamnosus yoba 2012 and S. thermophilus C106 with a commercial brand on a 9-point hedonic scale: 9 = like extremely, 8 = like very much, 7 = like moderately, 6 = like slightly, 5 = neither like nor dislike, 4 = dislike slightly, 3 = dislike moderately, 2 = dislike very much, 1 = dislike extremely. Values are means and standard deviations (n = 62 respondents).

Sample
Acceptability Scores

Shelf Stability of Probiotic Kwete
Shelf stability for probiotic kwete was evaluated by monitoring changes in pH, acidity, viability of L. rhamnosus yoba 2012, S. thermophilus C106, coliforms, yeasts, and consumer acceptability during refrigerated storage for a period of four weeks ( Figure 2). The pH of the products ranged between 3.2 and 4.0, and the acidity ranged between 0.6% and 0.7% (p < 0.05) during four weeks of storage. Lactobacillus rhamnosus yoba 2012 cell counts remained above 10 8 cfu g −1 during the entire storage period, but S. thermophilus C106 dropped 3 log units after 2 weeks to 10 4 cfu g −1 in week 4. Coliforms, yeasts, and molds were not detected (<1 cfu g −1 ) in the samples. Probiotic kwete remained acceptable during the four weeks of storage with overall acceptability scores ranging, on average, between 6.0 to 7.7 (equivalent to 'like moderately' to 'like very much'). Overall acceptability scores did not vary significantly (p > 0.05) during storage.

Effect of Inoculation Method
The preparation of fermented milk with the freeze-dried starter yoba starter culture bacteria has been widely applied throughout Uganda in areas with access to milk. However, fermented milk as starter culture to ferment kwete base would not be helpful for people living in areas with limited milk availability. Therefore, two alternative approaches for initiating the fermentation of probiotic kwete were evaluated (Table 2). Both inoculation approaches (direct use of the dried starter culture and fermented maize porridge) have been used to produce probiotic kwete at room temperature for 24 hours. The changes during fermentation in pH, titratable acidity, L. rhamnosus yoba 2012, and S. thermophilus C106 counts are shown in Table 2. The probiotic bacterium L. rhamnosus yoba 2012 performed well for all three fermentations with an increase of two to three log units. The bacterium S. thermophilus C106 propagated well in kwete inoculated with fermented milk, showing an increase of more than two log units. However, in the absence of milk, in the case of the dried starter and prefermented kwete inoculation methods, the counts only increased slightly and remained at seven log units. For all fermentations, the pH dropped from 5.6-6.3 to 4.2-3.9, and the acidity increased from 0.2% to 0.3%-0.5%. The highest pH and acidity differences were observed for the kwete fermentation inoculated with prefermented maize porridge.

Effect of Inoculation Method
The preparation of fermented milk with the freeze-dried starter yoba starter culture bacteria has been widely applied throughout Uganda in areas with access to milk. However, fermented milk as starter culture to ferment kwete base would not be helpful for people living in areas with limited milk availability. Therefore, two alternative approaches for initiating the fermentation of probiotic kwete were evaluated ( Table 2). Both inoculation approaches (direct use of the dried starter culture and fermented maize porridge) have been used to produce probiotic kwete at room temperature for 24 h. The changes during fermentation in pH, titratable acidity, L. rhamnosus yoba 2012, and S. thermophilus C106 counts are shown in Table 2. The probiotic bacterium L. rhamnosus yoba 2012 performed well for all three fermentations with an increase of two to three log units. The bacterium S. thermophilus C106 propagated well in kwete inoculated with fermented milk, showing an increase of more than two log units. However, in the absence of milk, in the case of the dried starter and prefermented kwete inoculation methods, the counts only increased slightly and remained at seven log units. For all fermentations, the pH dropped from 5.6-6.3 to 4.2-3.9, and the acidity increased from 0.2% to 0.3%-0.5%. The highest pH and acidity differences were observed for the kwete fermentation inoculated with prefermented maize porridge.

Reduction Aflatoxins B 1 , B 2 , G 1 , and G 2 by Fermentation
The reduction of aflatoxins B 1 , B 2 , G 1 , and G 2 , spiked into the maize kwete base, was assessed during fermentation for 24 h with the yoba starter culture by HPLC analysis. The chromatogram in Figure 3 shows the concentration of aflatoxins from the immunoaffinity-purified water-soluble fraction at t = 0, 12, and 24 h. A notable decrease was recorded in the concentration of aflatoxins B 1 , G 1 , B 2 , and G 2 of 92% ± 0.1%, 91.4% ± 0.2%, 91.8% ± 0.2%, and 90.9% ± 0.2%, respectively, after a period of 12 h of fermentation. However, after 24 h of fermentation, no detectable levels of aflatoxins were left in the sample, showing that yoba starter culture bacteria efficiently removed the concentration of all of the four major types of aflatoxins ( Figure 3). In order to exclude that the reduction of aflatoxins was a result of other factors than the fermentation by the starter culture, we carried out a number of control experiments. These experiments indicated that no aflatoxins can be detected in case they are not added to the raw ingredients, that fermentation with the starter culture is required for the reduction, and that an incubation of 24 h in an acidic environment (lactic acid, pH 4.4) does not lead to a reduction of the four major aflatoxins (Table 3).

Reduction Aflatoxins B1, B2, G1, and G2 by Fermentation
The reduction of aflatoxins B1, B2, G1, and G2, spiked into the maize kwete base, was assessed during fermentation for 24 hours with the yoba starter culture by HPLC analysis. The chromatogram in Figure 3 shows the concentration of aflatoxins from the immunoaffinity-purified water-soluble fraction at t = 0, 12, and 24 hours. A notable decrease was recorded in the concentration of aflatoxins B1, G1, B2, and G2 of 92% ± 0.1%, 91.4% ± 0.2%, 91.8% ± 0.2%, and 90.9% ± 0.2%, respectively, after a period of 12 hours of fermentation. However, after 24 hours of fermentation, no detectable levels of aflatoxins were left in the sample, showing that yoba starter culture bacteria efficiently removed the concentration of all of the four major types of aflatoxins ( Figure 3). In order to exclude that the reduction of aflatoxins was a result of other factors than the fermentation by the starter culture, we carried out a number of control experiments. These experiments indicated that no aflatoxins can be detected in case they are not added to the raw ingredients, that fermentation with the starter culture is required for the reduction, and that an incubation of 24 hours in an acidic environment (lactic acid, pH 4.4) does not lead to a reduction of the four major aflatoxins (Table 3).

Binding of Aflatoxin B1 by Lactobacillus rhamnosus
The binding of the major aflatoxin B1 to the probiotic bacterium L. rhamnosus yoba 2012 was studied by monitoring the residual aflatoxin after incubation of a dilution series of cell suspensions with aflatoxin B1 at a concentration of 1.0 µg mL −1 . The fluorescence aflatoxin B1 was plotted as a function of OD600 (Figure 4)

Discussion
Introduction of bacterial probiotic strains in traditional fermented foods can be used as a means to convey their health benefits [41]. In this study, we used the probiotic model bacterium L. rhamnosus GG, since there is a wealth of scientific evidence showing its beneficial effects in the prevention and treatment of gastrointestinal diseases, including rotavirus and Clostridium difficile-associated diarrhea, and travelers' and antibiotic-associated diarrhea (AAD) [42][43][44][45]. In addition, this strain is readily accessible in its generic form, L. rhamnosus yoba 2012, packed in a lyophilized state together with S. thermophilus C106 in a sachet as the yoba starter culture [31].
The yoba starter culture bacteria successfully fermented the traditional maize-based food kwete, as evident from the production of lactic acid shown by a decrease in pH and simultaneous increase in titratable acid. As required for microbiological safety and stability of lactic acid-fermented beverages [46][47][48][49], the observed pH values of probiotic fermented kwete were ≤4.3, and the amount of titratable acid was at least 0.6% after 24 hours of fermentation at 30 °C. It should be noted that in case of natural kwete fermentations, it can take between 24 to 120 hours to attain these pH and acidity values, while-in line with our findings-with defined starter cultures, these values are reached within 12 to 24 hours of fermentation [6]. The maximum acidity levels observed during storage of probiotic kwete, of 0.7%, corresponded to the maximum levels of acid production previously observed with starters containing L. rhamnosus GG for fermentation of maize porridge with added barley [50].
In this study, kwete was used as a substrate to enhance growth of the probiotic L. rhamnosus yoba 2012, reaching a maximum of 1.0 × 10 9 cfu g −1 after 24 hours fermentation of kwete at a temperature of 30 °C. These counts of colony forming units were similar to those reported for other traditional products serving as a substrate for the same starter culture, including mutandabota (a dairy product containing baobab pulp), uji (fermented maize and sorghum beverage), and zomkom (a fermented sorghum beverage [1,31]. Maximum counts of L. rhamnosus yoba 2012 in kwete were also comparable

Discussion
Introduction of bacterial probiotic strains in traditional fermented foods can be used as a means to convey their health benefits [41]. In this study, we used the probiotic model bacterium L. rhamnosus GG, since there is a wealth of scientific evidence showing its beneficial effects in the prevention and treatment of gastrointestinal diseases, including rotavirus and Clostridium difficile-associated diarrhea, and travelers' and antibiotic-associated diarrhea (AAD) [42][43][44][45]. In addition, this strain is readily accessible in its generic form, L. rhamnosus yoba 2012, packed in a lyophilized state together with S. thermophilus C106 in a sachet as the yoba starter culture [31].
The yoba starter culture bacteria successfully fermented the traditional maize-based food kwete, as evident from the production of lactic acid shown by a decrease in pH and simultaneous increase in titratable acid. As required for microbiological safety and stability of lactic acid-fermented beverages [46][47][48][49], the observed pH values of probiotic fermented kwete were ≤4.3, and the amount of titratable acid was at least 0.6% after 24 h of fermentation at 30 • C. It should be noted that in case of natural kwete fermentations, it can take between 24 to 120 h to attain these pH and acidity values, while-in line with our findings-with defined starter cultures, these values are reached within 12 to 24 h of fermentation [6]. The maximum acidity levels observed during storage of probiotic kwete, of 0.7%, corresponded to the maximum levels of acid production previously observed with starters containing L. rhamnosus GG for fermentation of maize porridge with added barley [50].
In this study, kwete was used as a substrate to enhance growth of the probiotic L. rhamnosus yoba 2012, reaching a maximum of 1.0 × 10 9 cfu g −1 after 24 h fermentation of kwete at a temperature of 30 • C. These counts of colony forming units were similar to those reported for other traditional products serving as a substrate for the same starter culture, including mutandabota (a dairy product containing baobab pulp), uji (fermented maize and sorghum beverage), and zomkom (a fermented sorghum beverage [1,31]. Maximum counts of L. rhamnosus yoba 2012 in kwete were also comparable to those reported for other starter cultures with lactic acid bacteria, such as L. reuteri, L. acidophilus (LA5 and 1748), and L. rhamnosus GG in maize porridge [50]. The ability of L. rhamnosus yoba 2012 to grow in cereal bases, such as kwete, is attributed to the availability of sugars such as glucose, fructose, and maltose from maize and millet/sorghum malt for kwete [8], in addition to free amino nitrogen from cereal malt [51]. The traditional production of kwete with an undefined mixture of yeasts, Lactobacillus, and Lactococcus species often results in poor product quality and short shelf life, requiring a consumption of kwete within 24 h after production [52]. In probiotic kwete prepared by the use of the yoba starter culture, we did not identify any (harmful) coliforms, yeasts, and molds in the samples during 4 weeks of storage.
Different probiotic starters uniquely affect the flavor profile, sensorial properties and, ultimately, the acceptability of products in which they are introduced. For instance, mild acidity, relatively high amounts of acetaldehyde, and the presence of the human isolate L. plantarum NCIMB 8826 correlated with higher acceptability scores of barley-and oat-based probiotic beverages [53]. In another study, L. rhamnosus LRB and L. acidophilus PRO produced probiotic mageu (a fermented maize beverage), whose sensory properties and acceptability scores were closer to that of the control than the product produced by L. casei BGP1 and L. paracasei BGP93 [36]. Therefore, it is necessary to establish the effect of the addition of probiotics on the acceptability of traditional fermented foods. Benchmarking the new probiotic product with existing related traditional products helps in gauging the success of the probiotic product [49]. This study shows that use of the yoba starter did not significantly affect the acceptability of kwete. The acceptability of sensory characteristics was comparable to the commercial product on the market. The consumers took note of a difference in the taste, but appreciated the probiotic kwete for its sweet and sour taste with a mean score of 6.8 ± 1.4 compared to 6.5 ± 1.6 of commercial brand. Therefore, kwete produced using the yoba sachet culture can be readily accepted and frequently purchased by consumers, thus increasing accessibility of probiotics in Uganda.
For a product to be considered probiotic, it should contain a minimum of 10 6 cfu per mL or gram of the probiotic microorganisms at the time of consumption [54]. Consuming 100-1000 mL per day of such a product provides the recommended daily dose (10 8 -10 9 cfu), essential for conveying the health benefits of probiotics [54,55]. Probiotic kwete contained a minimum of 4.0 × 10 8 cfu g −1 of L. rhamnosus yoba 2012 during four weeks of storage at 4 • C. Thus, a minimum daily consumption of 10 mL of probiotic kwete per day would be more than sufficient to meet the recommended daily intake of probiotics. With respect to shelf stability, probiotic kwete generally remained stable and acceptable during the entire study period of four weeks. Traditional kwete is normally produced and consumed within 24 h [56]. However, the use of yoba starter culture made the product stable for four weeks, thus improving its shelf life.
Detoxification of aflatoxins in food prior to consumption is a novel approach to curb their toxic effects. Several technologies have been employed to eliminate or reduce levels of aflatoxins in food, but only a handful are accepted for use and, as of yet, none offer 100% efficiency [57]. The use of microorganisms to detoxify aflatoxins has been given more consideration [25,27]. In this study, yoba starter culture bacteria, which are being used extensively in Uganda, Kenya, and Tanzania to produce fermented milk, demonstrated an excellent ability to reduce aflatoxins, during fermentation of kwete base, to non-detectable levels. There was a notable reduction in total concentrations containing all four major aflatoxins (B 1 , B 2 , G 1 , and G 2 ) from 7.0 ng mL −1 to non-detectable levels ( Figure 3). The detoxification of aflatoxins in kwete could be the result of binding as well as of degradation, as binding alone would not reduce the toxin from the food substrate to non-detectable levels [58]. We speculate that aflatoxin degradation is a specific property of our starter culture, as other studies reported less than 100% removal by L. rhamnosus strains [59][60][61]. However, under the experimental conditions used so far, we have not been able to confirm degradation of aflatoxins by pure cultures of bacterial strains in the yoba starter culture.
Our in vitro fluorescence experiments did confirm binding of aflatoxin B 1 to a cell suspension of L. rhamnosus yoba 2012 at OD 600 of 0.5, which reduced the aflatoxin B 1 concentration of 1.0 µg/mL to 17%. Preliminary results indicated that the binding of aflatoxin B1 to S. thermophilus C106 was less efficient, with a reduction of aflatoxin B 1 to 86% at the same cell density. Aflatoxin binding to lactic acid bacteria was previously suggested as a safe means to reduce the bioavailability and enhance excretion of the toxin from the body [62,63]. Although the mechanism of binding is still poorly understood, cell surface polysaccharide, peptidoglycans, and teichoic acids have been suggested as the binding sites [59,64,65]. Here, we show that the yoba starter, including L. rhamnosus yoba 2012 and S. thermophilus C106, were able to remove 100% of 120 µg kg −1 total aflatoxins spiked in the water-soluble fraction of kwete, which is highly relevant considering the range of aflatoxin concentrations we previously found in maize flour in households in Uganda [22].

Conclusions
This study showed that yoba starter culture bacteria were able to produce kwete products with comparable acceptability to commercially available traditional products. The yoba starter culture bacteria were able to ferment kwete, reducing the pH to below 4.0 in 24 h at room temperature. The products remained stable during refrigerated storage for a month. This study demonstrated that yoba starter culture bacteria can reduce aflatoxins during fermentation to non-detectable levels. Accordingly, fermentation with this starter culture can positively contribute to reduction of the risk of aflatoxins in maize-based foods widely used in schools in Uganda.