Production of organic flavor compounds by dominant lactic acid bacteria and yeasts from Obushera, a traditional sorghum malt fermented beverage

Abstract Single and mixed starter cultures of lactic acid bacteria (LAB): Weissella confusa MNC20, Lactobacillus plantarum MNC21, Lactococcus lactis MNC24 and Lactobacillus fermentum MNC34 and yeasts: Issatchenkia orientalis MNC20Y and Saccharomyces cerevisiae MNC21Y were used to produce Obushera, a fermented sorghum beverage. Microbial counts, pH, sugars, organic acids, and volatile compounds in starter culture and spontaneous fermentations were monitored during 48 hrs. Maximum counts of LAB (8.4–9.4 log cfu g−1) and yeasts (7.5 ± 0.1 cfu g−1) starter cultures were attained in 6–48 hrs. Weissella confusa, Lc. lactis, and Lb. fermentum showed possible acid sensitivity while I. orientalis produced surface films. LAB starter cultures and their combinations with S. cerevisiae lowered pH from 5.83 to <4.5 (3.50–4.13) in a shorter time (12 hrs) than spontaneous fermentations (24 hrs). Lactococcus lactis and W. confusa metabolized glucose the fastest (p < .05) during the first 6 hrs. Lactobacillus fermentum, Lb. plantarum, and S. cerevisiae utilized glucose and maltose concurrently. Lactobacillus plantarum and S. cerevisiae additionally utilized fructose. S. cerevisiae metabolized sugars the fastest (p < .05) during the first 12–24 hrs. Lactobacillus plantarum and W. confusa produced the highest (p < .05) amounts of lactate (5.43 g kg−1) and diacetyl (9.5 mg kg−1), respectively. LAB also produced acetate, ethanol, acetaldehyde, acetone, and acetoin. Coculturing LAB with S. cerevisiae reduced (p < .05) lactate and diacetyl yield. Yeasts produced high amounts of acetaldehyde and methyl alcohols. Issatchenkia orientalis produced higher (p < .05) amounts of 2‐methy‐1‐propanol and 3‐methyl‐1‐butanol than S. cerevisiae. Combinations of LAB with S. cerevisiae produced a profile flavor compounds close to that of spontaneously fermented Obushera. These combinations can be adopted for controlled fermentation of Obushera and related fermented cereal products.

Relying on natural fermentation is a major hindrance to large-scale commercial processing of traditional fermented products mainly because natural fermentation is associated with variations in product quality and safety (Holzapfel, 2002). Starter cultures with desirable properties can be identified and applied to ensure reduction in processing time, consistent product quality, and safety (Holzapfel, 2002).
However, selecting appropriate starter cultures requires understanding the microbial diversity of the products and the roles played by specific organisms.
The flavor profile of cereal fermented products is composed of sugars, organic acids, aldehydes, ketones, alcohols, and esters (Mugula et al., 2003;. In sourdough fermentation, LAB are mainly responsible for acidification while yeasts produce ethanol, methyl alcohols, methyl aldehydes, and acetaldehyde among other volatiles (Gobbetti et al., 1995;Hansen & Hansen, 1996). The contribution of LAB toward production of volatile organic compounds in traditional African cereal fermented products has not been given much attention. Although the interaction between LAB and yeasts is known to enhance growth of either group of microbes (Mugula et al., 2003;Omemu et al., 2007), the effect of this interaction on flavor production has not been reported. Furthermore, differences in production of volatile organic compounds by the yeasts have also not been evaluated.
This study evaluated the individual and interactive effects of the dominant yeasts: S. cerevisiae and I. orientalis and the different LAB (Lb. plantarum, Lb. fermentum, W. confusa, Lc. lactis) on the flavor profile of traditional fermented cereals. Obushera, of the Obutoko type, a popular Ugandan fermented beverage prepared from malted sorghum (Mukisa, 2012) was used as a model for comparison. Information generated from this study will enable the identification of potential starter cultures for traditional fermented cereal products. JF512471, JQ754464, JQ754435, and JQ754436, respectively) were used in this study. These LAB and yeasts were previously isolated from Obushera and identified by sequencing 16S rRNA and the ITS1-5.8S -ITS2 region of rRNA, respectively (Mukisa, Porcellato, et al., 2012). The above strains were selected for this study since they belong to the dominant genera and species involved in the fermentation of Obushera . The LAB and yeasts were initially used as single starter cultures and later as mixed starter cultures.

Mixed starter cultures included a primarily homolactic combination of
Lb. plantarum + Lc. lactis. Coculturing Lc. lactis with Lb. plantarum was previously observed to enhance acid production in a cereal fermentation . This starter culture therefore acted as a basis for two heterolactic combinations: Lb. plantarum + Lc. lactis + W. confusa and Lb. plantarum + Lc. lactis + Lb. fermentum.
The yeast strain S. cerevisiae was used in two LAB/yeast mixed starter cultures: Lb. plantarum + Lc. lactis + S. cerevisiae (with homolactic LAB) and Lb. plantarum + W. confusa + S. cerevisiae (with homolactic and heterolactic LAB). I. orientalis was not used in the mixed starter cultures because it formed undesirable white films in the product.
Pure LAB and yeast isolates were maintained at −80°C in broth containing 15% glycerol (v/v): MRS (Merck KGaA, Darmstadt, Germany) and tryptone glucose yeast extract broth (Oxoid Ltd, Hampshire, UK), respectively. Starter cultures for inoculation were prepared by growing strains in 250 ml of respective broths at 30°C for 24 hrs. Cells were recovered by centrifugation (7,500 × g for 10 min at 4°C) and subsequently rinsed with sterile distilled water. The cell pellet was resuspended in 25 ml of Ringer's solution containing 15% glycerol and stored in aliquots at −80°C. The resulting suspension contained approximately 10 9 cfu ml −1 (for LAB) or 10 7 cfu ml −1 (for yeasts) determined by taking plate counts on de Man, Rogosa and Sharpe (MRS) agar and Rose Bengal Chloramphenicol agar (RBCA), respectively (Merck KGaA, Darmstadt, Germany).

| Preparation of Obushera samples
Obushera of the Obutoko type (from malted sorghum) was prepared using flour from red sorghum (Sorghum bicolor (L.) Moench) of the Sekedo variety. One batch of sorghum malt was prepared as described earlier .
The formulation described by Mukisa et al. (2010) was used to prepare Obushera. Briefly, slurries were made by mixing 50 g of sorghum malt in 400 ml of sterile distilled water held in 1 L glass jars. The slurries were heated in a water bath to 90°C, held for 10 min and then cooled to 30°C prior to inoculation. The slurries were then inoculated with single or mixed starter cultures resulting in initial cell concentrations of 6.6 ± 0.2 log cfu g −1 and 4.5 ± 0.3 log cfu g −1 for individual LAB and yeasts strains, respectively (Mukisa, 2012). Natural fermentation was initiated by adding 9 g of sorghum malt (2% malt) to the slurries (Mukisa et al., 2010). After inoculation, samples were homogenized for 1 min in an Omni mixer (Omni International, Waterbury, USA). Approximately 100 ml of sample was aseptically transferred into 200 ml sterile glass jars and incubated at 30°C. Samples were taken at t = 0, 6, 12, 24, and 48 hrs for the analysis of pH, titratable acidity, sugars, organic acids, volatile organic compounds, and microbial counts. Each sampling time was allocated a separate container.
Three independent fermentations were made for each starter culture combination.

| Microbial counts
LAB and yeast counts were determined by plating appropriate serial dilutions of Obushera in ¼ strength Ringer's solution on MRS agar and RBCA, respectively. LAB counts were taken after incubation at 30°C for 2-3 days while yeasts were counted after incubation at 25°C for 2-5 days.

| Data analysis
Results were presented as arithmetic means ± standard deviations (Mean ± SD) of three independent fermentations. Data were subjected to one-way analysis of variance (ANOVA) to test for significant differences at p = .05. Mean comparisons were made, using the Fisher's least significant difference (LSD) test to determine which means were significantly different.

| Growth of starter cultures in Obushera
The study evaluated the ability of the individual LAB and yeast starter cultures to grow in Obushera to levels that have been reported in previous studies on naturally and starter culture fermented cereal beverages. Individual LAB and yeast starter cultures grew in Obushera increasing in numbers from 6.6 to 8.4-9.4 log cfu g −1 and 4.5 to 7.6 log cfu g −1 , respectively within 6-24 hrs ( Figure 1a).
In contrast LAB and yeast counts in naturally fermented Obushera increased from 5.0 log cfu g −1 to 8.9 log cfu g −1 after 24 hrs and 5.2 log cfu g −1 to 7.6 log cfu g −1 in 24-48 hrs, respectively (results not shown). Lactococcus lactis and W. confusa attained maximum counts the earliest (8.4-8.8 log cfu g −1 in 6 hrs). Counts of Lc. lactis, W. confusa, and Lb. fermentum decreased significantly (p < .05) by 1.3-2.0 log cfu g −1 between 6 and 48 hrs while those of Lb. plantarum were stable at 9.0 ± 0.1 log cfu g −1 after 12 hrs. Issatchenkia orientalis produced visually disagreeable surface films, a characteristic which was also observed in spontaneously fermented Obushera kept beyond 48 hrs.

| Sugar utilization by LAB and yeast starter cultures
The main sugars observed in Obushera were maltose, glucose, and fructose. Spontaneously fermented samples (inoculated with 2% malt) contained significantly higher (p < .05) starting amounts of maltose (16.5 g kg −1 ), glucose (5.7 g kg −1 ) and fructose (1.8 g kg −1 ) ( Lactococcus lactis and W. confusa metabolized glucose significantly (p < .05) faster than the other starter cultures during the first 6 hrs.
Saccharomyces cerevisiae had the highest rate (p < .05) of breakdown of glucose, maltose, and fructose (reduced these to a minimum in 12-24 hrs).

| Production of ethanol
Ethanol was the major alcohol produced in spontaneously fermented samples and ranged from 0.62 g kg −1 to 2.00 g kg −1 between 12 and 48 hrs (Table 2 and (Table 2). Starter cultures with S. cerevisiae showed the highest rate of ethanol production reaching peak concentrations (1.4-1.56 g kg −1 ) after 12 hrs (Figure 4). Ethanol production by S. cerevisiae was not affected by co-culturing the yeast with LAB.

| Production of diacetyl, acetoin, and acetone
Spontaneously fermented Obushera did not accumulate diacetyl but contained the highest amounts of acetoin after 12 hrs (Figure 3d).

| Growth of starter cultures
Since not all microbial genera are of equal importance in fermentation (Holzapfel, 1997), candidate isolates for starter culture development have to be evaluated for their contribution during fermentation.
Selecting appropriate starter cultures involves evaluating candidate strains for their viability and survival in the product, their ability to utilize the major carbohydrate substrates available, rate of acidification, and production of desirable flavor compounds among others (Holzapfel, 2002). It is also important to determine the effects of coculturing selected isolates on their activity such as acidification rate and production of flavor compounds.
With regard to growth of the starter cultures, individual LAB and yeast isolates used in the current study were able to grow in Obushera and attained maximum counts similar to those reported for spontaneously fermented Obushera (Muyanja, Narvhus, & Langsrud, 2004;. However, maximum counts for the starter cultures were attained in a much shorter time (12-24 hrs) than in spontaneous fermentation (24-48 hrs). Fast growth is desirable since it results in fast acidification and flavor development thus contributing toward shortening of processing time as well as ensuring product safety (Holzapfel, 2002). The decline in cell numbers of W. confusa, Lc. lactis, and Lb. fermentum during fermentation could imply that these strains are acid sensitive. The stability of Lb. plantarum counts can be attributed to the acid tolerance of the strain MNC 21, which has been observed in other studies (Byakika, 2015;Mukisa, 2012). Despite its viability and flavor production potential, I. orientalis was considered undesirable for production of fermented cereal beverages on the basis of surface film formation. Surface film formation by I. orientalis is commonly associated with spoilage of beverages (Stratford & James, 2003).
Indeed, the development of surface films is one of the spoilage characteristics of Obushera that consumers may use to reject the product (Mukisa, 2012

| Utilization of sugars
The ability of starter cultures to metabolize sugars available in a substrate is an important aspect for starter culture selection. The major sugars in sorghum malt fermentations are maltose, glucose, and fructose. All starter cultures were able to utilize glucose thus enabling them to grow in and ferment the product. Differences were observed in the patterns and rates of utilization of maltose, glucose, and fructose by the individual LAB and yeast isolates. The ability of, and rate at which, an individual isolate metabolizes one or more sugars do influence its productivity and competiveness in a mixed starter culture (Gobbetti, 1998). Rapid utilization of sugars by S. cerevisiae makes it more competitive than LAB, when in coculture, resulting in diminished lactate production while ethanol yield was unaffected. Similar effects on lactate production were observed during sourdough fermentations when S. cerevisiae was cocultured with Lb. plantarum (Gobbetti, Corsetti, & Rossi, 1994) or Lb. sanfranciscensis (Gobbetti, 1998). In contrast, coculturing Lb. brevis with S. cerevisiae (Meignen et al., 2001) or Lb. plantarum with I. orientalis (Mugula et al., 2003) did not affect lactate production. It is possible that in the latter cases, LAB were faster than the yeasts in metabolizing the sugars while the inability of I. orientalis to utilize maltose, as seen from this study, could have made it less competitive. The rapid utilization of glucose by W. confusa and Lc. lactis in the first 8-12 hrs enables them to rapidly acidify the product to a pH below 4.5 in a period of about 6 hrs.
One key observation during the spontaneous fermentation of Obushera was that the concentration of maltose increased with maltose remaining above its taste threshold concentration of 13 g kg −1 for at least 24 hrs (Laska, Schüll, & Scheuber, 1999). This increase in sugars results from hydrolysis of starch by cereal amylases and is responsible for the desirable sweet taste of products such as Obushera.
Therefore, low initial amounts of sugars in starter culture fermented samples, coupled by rapid depletion by the starter cultures could compromise the taste acceptability of some products for which sweetness is an important attribute. It may therefore be necessary to introduce a saccharification step prior to inoculation with starter culture in order to achieve desirable levels of sweetness in such products.

| Lactate and acetate production
Acid production is important for both taste and safety of fermented products. Lactate imparts a sour taste which is an important sensory attribute of LAB fermented products (Mukisa et al., 2010). To ensure safety of lactic fermented products, some studies recommend attaining a rapid pH drop to ≤4.5-4.0 and a titratable acidity of about 0.7% lactic acid (Steinkraus, 1996). Fast acidification to pH below 4.0 inhibits spoilage bacteria, enteropathogens, and Bacillus sp (Holzapfel, 1997;Steinkraus, 1996). The ability to achieve rapid acidification in a shorter time than observed in the traditional process should therefore  Values are means ± standard deviations of three independent fermentations. Values in the same column with similar superscripts (a -e) are not significantly different (p > .05). ND: not detected. Acetald (acetaldehyde); 2-ME-pro-al (2-methyl-1-propanal); 2-ME-but-al (2-methyl-1-butanal); 3-ME-but-al (3-methyl-1-butanal).
be one of the important criteria for selecting LAB starter cultures.
All the LAB starter cultures used in this study were able to drop the pH of Obushera to below 4.5 in 6-12, but were not able to raise the acidity to 0.7%. Accelerated acidification by LAB starter cultures has also been reported in related studies (Mugula et al., 2003;Muyanja, Narvhus, & Langsrud, 2004;. Spontaneously fermented Obushera attained a pH below 4.5 in a much longer time (24 hrs) and only attained acidity of about 0.35%.
Lactobacillus fermentum, Lc. lactis, and W. confusa only reached a maximum acidity of 0.2% in 12-24 hrs while Lb. plantarum was able to raise the acidity to 0.53% in 24 hrs. In this respect, Lb. plantarum would be a better candidate as a starter culture for such fermentations because of its superior acid production. This starter culture has been reported to produce up to 1.7% lactic acid from sorghum malt extracts (Byakika, 2015). High lactate production in Lb. plantarum strains is attributed to versatility in metabolism of carbohydrates coupled with tolerance to acid conditions ( Van der Meulen et al., 2007). In some instances, excessive lactate production by Lb. plantarum could result in an undesirable sour taste (Hansen & Hansen, 1996). Therefore, the reduction in lactate production observed when S. cerevisiae was cocultured with the LAB-mixed starter cultures can help prevent excessive souring during fermentation, as long as the desired level of acidity is attained. This interaction is reported to have moderated the souring effect of Lb. plantarum in sourdough bread (Hansen & Hansen, 1996).

| Ethanol production
Ethanol production in Obushera is responsible for the characteristic alcoholic flavor (Mukisa et al., 2010). Obushera of the Obutoko type is acceptable with an alcohol content of up to 10 g kg −1 , but levels up to 45 g kg −1 result in a decline in acceptability (Mukisa, 2012).

Concentrations of ethanol in sorghum based fermentations such as
Obushera and Togwa vary with type of product and duration of fermentation ranging from 0.1 g kg −1 to 60.0 g kg −1 (Mugula et al., 2003;.
Ethanol in spontaneously fermented Obushera (1. 29-2.00 g kg −1 between 24 and 48 hrs) was within expected limits which do not negatively impact on acceptability. Saccharomyces cerevisiae produced the highest amounts of ethanol in 12 hrs (1.55 g kg −1 ), while ethanol production by Lb. fermentum, I. orientalis, and W. confusa was comparable (0.79-0.88 g kg −1 ). Ethanol is a product of carbohydrate metabolism by obligate heterofermentative LAB and yeasts, but the latter are generally recognized as high producers (Hansen & Schieberle, 2005

| Production of acetaldehyde
Variations in acetaldehyde concentrations during fermentation followed a similar pattern previously observed by . In the current study, yeasts produced the highest concentrations of acetaldehyde (8.99-10.69 mg kg −1 ) during the first 24 hrs and were followed by Lb. plantarum and Lc. lactis (1.76-2.14 mg kg). Acetaldehyde is produced by LAB and yeasts, mainly as an intermediate in ethanol production (Liu & Pilone, 2000).
However, yeasts and in particular S. cerevisiae, are known to produce higher amounts compared to LAB (Liu & Pilone, 2000). Coculturing Lb. fermentum or W. confusa with the Lb. plantarum + Lc. lactis starter culture prevented accumulation of acetaldehyde. The acetaldehyde was probably reduced to acetate/ethanol by action of acetaldehyde/ethanol dehydrogenase from W. confusa or Lb. fermentum (Zaunmüller, Eichert, Richter, & Unden, 2006). Acetaldehyde imparts a pleasant fruity aroma when present in low concentrations, but results in pungent odors at high concentrations (Liu & Pilone, 2000).

| CONCLUSION
This study reveals that besides producing lactic acid, LAB in cereal fermentations may also contribute to production of other flavor compounds notably acetate, ethanol, acetaldehyde, diacetyl, acetone, and acetoin. Yeasts mainly contribute toward production of ethanol, acetaldehyde, methyl aldehydes, and methyl alcohols. With regard to acidification, Lb. plantarum, Lc. lactis, W. confusa, and Lb. fermentum, or their selected combinations can be used to achieve accelerated fermentation of cereal products. However, use, or inclusion, of Lb. plantarum is encouraged because of its high acid production potential and stability to acid. Despite the acid sensitivity of Lc. lactis and Wc. confusa, their fast growth and glucose metabolism may contribute toward rapid acidification. Addition of S. cerevisiae to LAB starter cultures may be crucial for the attainment of a flavor profiles typical of traditional fermented cereal products. S. cerevisiae also prevents excessive lactate and diacetyl production, a phenomenon which could negatively or positively affect the flavor profile. The starter culture combinations: Lb. plantarum + Lc. lactis + S. cerevisiae, or Lb. plantarum + Lc. lactis +W. confusa + S. cerevisiae can be adopted for the fermentation of Obushera.