Physicochemical Properties and Bacterial Community Profiling of Optimal Mahewu (A Fermented Food Product) Prepared Using White and Yellow Maize with Different Inocula

Mahewu is a fermented food product from maize, commonly consumed in Southern Africa. This study investigated the effect of optimizing fermentation (time and temperature) and boiling time of white maize (WM) and yellow maize (YM) mahewu, with the use of the Box–Behnken-response surface methodology (RSM). Fermentation time and temperature as well as boiling time were optimized and pH, total titratable acidity (TTA) and total soluble solids (TSS) determined. Results obtained showed that the processing conditions significantly (p ≤ 0.05) influenced the physicochemical properties. pH values of the mahewu samples ranged between 3.48–5.28 and 3.50–4.20 for YM mahewu and WM mahewu samples, respectively. Reduction in pH values after fermentation coincided with an increase in TTA as well as changes in the TSS values. Using the numerical multi-response optimisation of three investigated responses the optimal fermentation conditions were observed to be 25 °C for 54 h and a boiling time of 19 min for white maize mahewu and 29 °C for 72 h and a boiling time of 13 min for yellow maize mahewu. Thereafter white and yellow maize mahewu were prepared with the optimized conditions using different inocula (sorghum malt flour, wheat flour, millet malt flour or maize malt flour) and the pH, TTA and TSS of the derived mahewu samples determined. Additionally, amplicon sequencing of the 16S rRNA gene was used to characterise the relative abundance of bacterial genera in optimized mahewu samples, malted grains as well as flour samples. Major bacterial genera observed in the mahewu samples included Paenibacillus, Stenotrophomonas, Weissella, Pseudomonas, Lactococcus, Enterococcus, Lactobacillus, Bacillus, Massilia, Clostridium sensu stricto 1, Streptococcus, Staphylococcus, Sanguibacter, Roseococcus, Leuconostoc, Cutibacterium, Brevibacterium, Blastococcus, Sphingomonas and Pediococcus, with variations noted for YM mahewu and WM mahewu. As a result, the variations in physicochemical properties are due to differences in maize type and modification in processing conditions. This study also discovered the existence of variety of bacterial that can be isolated for controlled fermentation of mahewu.


Introduction
Maize is one of the topmost cereal crops identified as a major source of food in many countries [1]. Maize is transformed into different food products using fermentation and such fermented foods including kenkey, kwete, munkoyo, uji and ogi, form part of the culture and nutrient sources of inhabitants of under-developed countries [2]. Fermented foods include beverages produced through microbial growth and the conversion of food components through enzymatic action. These beverages often possess considerably high nutritive Oven, New Delhi, India,) and dried at 60 • C for 24 h. The sprouts were removed manually and winnowed. The seeds were then milled (Perten Laboratory Mill 3310 Instruments AB, Helsinki, Finland) into fine flour and packaged into zip lock bags for further use. Unmalted white, yellow and wheat grains were sorted, cleaned, milled and packaged into Ziplock bags for further use.

The Optimisation of Parameters to Produce Mahewu
Different parameters such as boiling time (X 1 ), fermentation temperature (X 2 ) and fermentation time (X 3 ), with ranges of 10-20 min, 25-45 • C and of 16-72 h, respectively, were investigated. The choice of the parameters investigated was based on previous studies on the production of mahewu [12,15,16]. The three-factor Box-Behnken design gave a total of 17 experimental runs each for white and yellow maize, respectively (Table 1). Three responses, including pH (Y 1 ), TTA (Y 2 ) and TSS (Y 3 ) were investigated.
The Box-Behnken general second-order polynomial model in Equation (1) was used to depict the mathematical expression of the relationship between the process variables, including linear, quadratic and interactive effects.
where Y is the response, X 1 , X 2 , X 3 are factors, β 0 . . . β 33 are the coefficients of linear, quadratic and interaction terms. The response surfaces were represented with model equations and respective coefficients obtained using the Box-Behnken-response surface methodology.

pH, TTA and TSS
The pH of each mahewu sample was measured after the fermentation process using a pH meter (Hand-Held EcoSense pH10An pen Tester, Beijing, China). The titratable acidity (TTA) analysis was carried out by titrating 2 g of mahewu sample mixed in 20 mL distilled water against 0.1 N NaOH, with the use of phenolphthalein as an indicator, the result obtained was expressed in percentage (% lactic acid). The total soluble solids (TSS) determination was carried out using a refractometer (Hanna H196801, Woonsocket, RI, USA), and the results were expressed as • Brix.

Multi-Response Numerical Optimisation and Processing of Optimal Mahewu Samples
Based on the experimental results obtained from Table 1, a numerical multi-response optimisation of parameters (pH, TTA, TSS) was carried out on Minitab 16 (Minitab Lt. Coventry, UK). The optimal processing conditions obtained were a fermentation temperature of 25 • C for 54 h and a boiling time of 19 min for white maize mahewu while that of yellow maize mahewu were fermentation conditions of 29 • C for 72 h and a boiling time of 13 min. For both white and yellow maize, a total of eight (8)  The bacterial DNA extraction was performed using a ZymoBIOMICS TM DNA miniprep kit (Inqaba Biotech, Pretoria, South Africa) according to the manufacturer's instructions. DNA samples from raw maize flours, optimal mahewu samples and inocula used were extracted. The extracted DNA was stored at −80 • C for further processing. Subsequently, the quantity and quality of the eluted DNA was verified using nano-drop equipment (Implen Nano-photometer N60-Touch, Cape Town, South Africa). The extracted DNA was subsequently sent to Inqaba Biotech Pty (Ltd), Johannesburg, South Africa, for 16s rRNA sequencing.
16s rRNA Amplicon Sequencing The total DNA extracted earlier from samples was amplified at the 16S region using universal primers 5 -CCTACGGGNGGCWGCAG-3 and 5 -GACTACHVGGGTATCTAATCC-3. The hypervariable regions targeted were V3-V4. The reaction was carried out in 50 µL volumes containing 0.3 mg/mL BSA (Bovine Serum Albumin), 250 µM dTNPs, 0.5 µM of each primer, 0.02 U Phusion High-Fidelity DNA Polymerase (Finnzymes OY, Espoo, Finland) and 5× Phusion HF buffer containing 1.5 mM MgCl 2 on a GeneAmp PCR system using the following PCR conditions: initial denaturation at 95 • C for 5 min, 25 cycles of denaturation at 95 • C for 40 s followed by annealing for 2 min, extension at 72 • C for 1 min and a final step of extension at 72 • C for 7 min. Subsequently, PCR products were purified in an ExoSAP, Affymetrix, Inc. Santa Clara, CA, USA. The amplicon library was normalised and prepared for sequencing following the 16S rRNA gene library preparation guide (Amplicon, 2013). Sequencing was performed using an Illumina MiSeq Sequencer (Illumina, San Diego, CA, USA) with a MiSeq Reagent Kit v3 to generate 2 × 300 paired end reads at Inqaba Biotech Pty (Ltd), Johannesburg, South Africa.

Bioinformatics Analysis
The 16S sequencing data included the forward and reverse paired-end reads. The demultiplexed forward and reverse reads were imported with the CASAVA 1.8 pipeline (paired-end) using the Quantitative Insights into Microbial Ecology package (Qiime2) The demux.qzv visualisation provided the length of nucleotides to trim and truncate for the subsequent qiime denoise analysis. This further displayed the data quality scores, allowing the removal of low-quality reads <Phred33 with the deblur plugin. Sequence variant calling of the illumina-amplicon sequences and chimeric sequences was removed using deblur plugin.
The compositional and taxonomic analyses were conducted by using feature-classifier plugins, i.e., composition and taxa (https://github.com/qiime2/q2) (accessed on 24 May 2021) using a pretrained Naive-Bayes classifier SILVA 138. Subsequently, sequences were clustered according to similarities into an operational taxonomic unit (OTU), followed by the generation of a representative sequence for each OTU. Taxonomy of the resulting OTU was used for downstream taxonomic assignment. A phylogenetic tree with the phylogeny fasttree command was generated with the use of the feature table. Graphical representations were conducted using the phyloseq package with the Bioconductor version 3.0 and R version 3.5.1.

Statistical Analysis
Except for the profiling of the microbial community done with R software (R-4.1.2 for Windows), all other analyses were done in triplicate using the analysis of variance (ANOVA) in the Minitab 16 software (Minitab Lt. Coventry, UK) and differences were considered statistically significant if p ≤ 0.05.

pH, TTA and TSS
The role and importance of maize-based diets in the provision of health benefits is key to achieving food security, hence, the selection of a suitable model to generate the optimum conditions of factors for desirable mahewu products is necessary. Experimental results of the seventeen runs generated by the BBD (Table 2) for each maize-based product (YM and WM) revealed that pH values decreased with an equivalent increase in TTA. The pH values of the maize-based products (YM and WM) ranges from 3.48 to 5.28 for YM and 3.50 to 4.20 for WM (Table 2). It was observed that mahewu samples prepared for longer times (44-72 h) at 25 and 35 • C had lower pH values compared to those with the shorter fermentation time (16 h). It is also expected that at a low pH value, the growth of most spoilage bacteria can be inhibited [19]. Besides, at 45 • C the highest pH values were recorded for both maize types. As expected in all samples, as the pH values decreased, there was a corresponding increase in TTA, which is indicative of the acid fermentation that is expected to occur. Additionally, at the end of fermentation, the TSS values decreased from approximately 5.90 • Brix to 4.43 • Brix for white maize mahewu and 5.87 • Brix to 4.10 • Brix for yellow maize mahewu.
The differences in pH, TTA and TSS values for the experimental runs of the maizebased products (WM and YM mahewu) could be attributed to differences in cooking time, fermentation conditions as well as sample source and inoculum. It is important to note that in fermented beverages, pH, TTA and TSS are significant in determining microbial stability against food-borne pathogens and correlate with the taste of the products. Subsequently, the seventeen experimental results of the different types of mahewu products (Table 2) were used to generate an optimal condition for both WM and YM for the preparation of mahewu as described in Section 2.5.

Statistical Models and Validation
This study investigated the effects of independent process variables (fermentation temperature (X 1 ), fermentation time (X 2 ) and boiling time (X 3 )) of WM and YM on the production of mahewu. For numerical optimisation, the parameters determined were pH (Y 1 ), TTA (Y 2 ) and TSS (Y 3 ) and the Box-Behnken model equations representing each are provided in Equations (2)-(4) for white maize and Equations (5)-(7) for yellow maize.
All calculated R 2 values in this study were above 80% (Table 3) which signifies a good model fit as reported in the literature [13,20]. Lower R 2 values in this study could be explained by the reduction in TSS after fermentation, which could be due to the loss of nutrients during the germination process of the inoculum (sorghum malt). This makes sense because sprouting, as a metabolic process, involves nutrient breakdown and is also accompanied with the growth of cotyledons and the release of energy [21]. This finding is in line with the findings in the literature [21]. β is the coefficient of various models and β 0 represents the constant term, β 1 and β 2 are the linear effects of fermentation and cooking conditions. β 11 , β 22 and β 33 connote the quadratic effects and β 12 , β 13 and β 23 are the interactions. TTA = titratable acidity; TSS = total soluble solids. * Significant at p ≤ 0.05.
To obtain the optimum conditions in each case, the degree of importance of each of the responses (pH, TTA and TSS) were defined in the Minitab software to obtain the desired result in the resulting mahewu samples. Hence, the TSS was minimised while targeted values of 3.5 for pH and 0.5% for TTA were defined. The optimal processing conditions obtained were a fermentation temperature of 25 • C for 54 h and a boiling time of 19 min for white maize mahewu with desirability factor (D f ) of 0.95, while the fermentation conditions These optimal conditions were further used to investigate the pH, TTA and TSS on the eight different types of mahewu samples produced. Experimental results of the eight samples generated for each maize-based product revealed that pH values decreased (increased acidity) with an equivalent increase in TTA ( Table 4). The inverse proportional trend between pH and titratable acidity has been previously reported by several scientists [13,22].  Therefore, these results suggest an accumulation of organic acids with an increase in the activity of microbes and metabolism of the fermenting organisms. The pH of the optimal maize-based products (WM and YM mahewu) ranged between 3.41 to 4.51 for WM (Table 4) and 3.44 to 4.65 for YM (Table 4), whereas their TTA ranged between 0.47 to 0.68% for WM and 0.38 to 0.62% for YM. With significant differences (p ≤ 0.05) in the values obtained. These differences could be due to their difference in fermentation time and temperature which made the mahewu samples ferment differently. Interestingly, the values reported herein are within the acceptable range reported for mahewu and other fermented cereal beverages [11,15,[23][24][25].
The TSS is an approximate measurement of sugar content. Boiling the mahewu for an adequate measure of time is essential for starch gelatinisation and release of lockedup nutrients in yeast cells [26]. Therefore, the proliferation of fermentative microbes is driven by the hydrolysis of cooked starch to fermentable sugars by endogenous amylolytic enzymes [25]. As observed, the general pattern for all mahewu samples herein ( Table 4), showed that TSS decreased from approximately 5.90 • Brix to the experimental values obtained for both white and yellow maize mahewu at the end of the fermentation process. A similar trend has been documented [24,27]. This decrease in TSS at the end of fermentation could be because of the high bacterial load in the mahewu samples, which meant rapid utilisation of accessible solids [28]. Furthermore, fermentation of sugars might have caused lactic acid bacteria to multiply as acidity increased [29].

Profiling of Bacterial Community in Optimized Mahewu Samples
16S amplicon analysis showed the first report on the bacterial community present in optimised mahewu samples ( Figure 1) and they included Paenibacillus, Stenotrophomonas, Weissella, Pseudomonas, Lactococcus, Enterococcus, Lactobacillus, Bacillus, Massilia, Clostridium sensu stricto 1, Streptococcus, Staphylococcus, Sanguibacter, Roseococcus, Leuconostoc, Cutibacterium, Brevibacterium, Blastococcus, Sphingomonas and Pediococcus. Nevertheless, the difference in bacterial composition between yellow maize mahewu and white maize mahewu could be due to the differences in cooking time, fermentation conditions as well as sample source and inoculum.

Profiling of Bacterial Community in Malt and Flour Samples
To determine their relative abundance, samples were characterised by the analysis of different bacterial taxa at the phylum and genus level. Proteobacteria was the most abundant and prevalent phylum in the flour sample whereas Firmicutes was most prevalent in the malt samples followed by Proteobacteria (Figure 2) with Pseudomonas as the most abundant genus in the flour samples while Lactococcus was the most prevalent in the malt samples followed by Paenibacillus (Figure 3). The members of the phylum Proteobacteria and genus Pseudomonas have been reported in wholemeal wheat flours and barley grain [46,48]. Members of the phyla Firmicutes and Proteobacteria, as well as the genera Lactococcus and Paenibacillus found in malt samples in this study, have previously been reported in the literature [49,50]. The taxonomic classification of the lactic acid bacteria such as Weissella, Streptococcus, Pediococcus, Leuconostoc, Lactococcus, Lactobacillus and Enterococcus as revealed herein were the most dominant in the malt samples. In the same line, the interactions of lactic acid bacteria in food are typically involved in spontaneous fermentation and plays key role in food safety.
Nevertheless, the significant decrease and increase in pH and TTA, respectively, after the fermentation of mahewu prepared with the malt samples as compared to mahewu prepared with wheat flour could be because of the hydrolysis of some complex organic molecules such as phytin, protein and lipids to acetic acid, lactic acid, fatty acids, phosphate and amino acids during malting and fermentation. Similar observations have been reported in the literature by Muhammed and his co-workers on their work on wheat and millet [51].
Hence, the decrease in pH and increase in acidity accompanying fermentation could be attributed to the extent of these complex molecules as well as the digestibility of the Few studies have reported members of the genera Weissella, Leuconostoc, Lactobacillus, Streptococcus, Enterococcus, Lactococcus and Pediococcus as dominant lactic acid bacteria in fermented foods [30][31][32][33]. This was expected since all products exhibited a pH of around 3.5 to 4.6. Weissella and Leuconostoc species are heterofermenters that produce ethanol and CO 2 in addition to organic acids [33]. Lactobacillus and Streptococcus species are homofermenters producing lactic acid as a major product [33]. Members of the Leuconostoc genera have been considered as a starter to produce commercially fermented kimchi with good quality. It was reported that Enterococcus is active in the natural fermentation of meju, indicating that the bacteria may exist extensively in some fermentation steps of meju [30]. In another study, Lactococcus and Pediococcus were reported to be present in mahewu. Consistent with other studies, the bacterial community detected in this study has been previously reported in fermented products [10,[34][35][36][37][38][39][40][41].
The presence of Clostridium sensu stricto 1 enhances the production of butyrate, which induces the production of regulatory T cells and play an anti-inflammatory role [42]. Additionally, Tao and his colleagues reported that during fermentation, Clostridium sensu stricto 1 stimulated the proliferation of fibrolytic bacteria, which in turn degraded fibres to produce organic acid and monosaccharides [43]. Interestingly, it has been reported that the presence of lactic acid bacteria in fermented food may drastically reduce the duration and severe effects of gastrointestinal disorders [44]. It has also been established that some LAB-fermented foods have antimutagenic and anticarcinogenic activities [45][46][47].

Profiling of Bacterial Community in Malt and Flour Samples
To determine their relative abundance, samples were characterised by the analysis of different bacterial taxa at the phylum and genus level. Proteobacteria was the most abundant and prevalent phylum in the flour sample whereas Firmicutes was most prevalent in the malt samples followed by Proteobacteria (Figure 2) with Pseudomonas as the most abundant genus in the flour samples while Lactococcus was the most prevalent in the malt samples followed by Paenibacillus (Figure 3). The members of the phylum Proteobacteria and genus Pseudomonas have been reported in wholemeal wheat flours and barley grain [46,48]. Members of the phyla Firmicutes and Proteobacteria, as well as the genera Lactococcus and Paenibacillus found in malt samples in this study, have previously been reported in the literature [49,50]. The taxonomic classification of the lactic acid bacteria such as Weissella, Streptococcus, Pediococcus, Leuconostoc, Lactococcus, Lactobacillus and Enterococcus as revealed herein were the most dominant in the malt samples. In the same line, the interactions of lactic acid bacteria in food are typically involved in spontaneous fermentation and plays key role in food safety. malted and fermented wheat and millet. This is consistent with the work documented by Muhammed and his co-workers on their work on maize [51]. Additionally, the higher bacteria load in this samples led to reduced TSS values.
Bacterial composition was low in raw white maize flour (RWMF), raw yellow maize flour (RYMF) and wheat flour (WF) but a higher bacterial composition in the malted inocula may be the result of the malting process. With respect to microbial activity and safety, steeping is the most critical step in malting [52]. Microbes, particularly lactic acid bacteria rapidly increase due to the limited accessibility of oxygen during steeping [53]. During germination, the ability of bacteria constituents to form a complex matrix in the form of a biofilm make the bacteria taxa a dominant microflora. Additionally, bacteria constituents have been reported to be the dominant microflora during germination [54]. Additionally, metabolic changes which include the conversion of residual carbohydrates to fermentable sugars occur during germination. In this phase, Lactobacilli was found as the predominant LAB, whereas Leuconostoc species were the major LAB during steeping. Furthermore, kilning which happens to be the final phase during malting is very important to the quality and stability of microbes [54]. However, the microbes that present at this phase are certain heat resistant bacteria which can form spores and they include bacillus species [51]. Similarly, as reported in the literature, malting can significantly improve the grain nutritional profile such as proteins, phenolic compounds as well as functional microorganisms with health benefits [55].  Nevertheless, the significant decrease and increase in pH and TTA, respectively, after the fermentation of mahewu prepared with the malt samples as compared to mahewu prepared with wheat flour could be because of the hydrolysis of some complex organic molecules such as phytin, protein and lipids to acetic acid, lactic acid, fatty acids, phosphate and amino acids during malting and fermentation. Similar observations have been reported in the literature by Muhammed and his co-workers on their work on wheat and millet [51].
Hence, the decrease in pH and increase in acidity accompanying fermentation could be attributed to the extent of these complex molecules as well as the digestibility of the malted and fermented wheat and millet. This is consistent with the work documented by Muhammed and his co-workers on their work on maize [51]. Additionally, the higher bacteria load in this samples led to reduced TSS values.

Conclusions
It could be concluded that the optimum processing conditions for white and yellow maize mahewu are 25 °C for 54 h with a boiling time of 19 min for white maize mahewu and 29 °C for 72 h with a boiling time of 13 min for yellow maize mahewu. Isolation of bacterial nucleic acids from the natural environment has become a useful tool to detect bacteria that may be difficult to culture. However, the present study reveals the shift in the bacterial community of optimised mahewu produced with different inocula using a high-throughput microbiota determination approach. This research will serve as the foundation for standardised, high-quality production of white and yellow maize mahewu. Therefore, further studies are needed to identify organisms such as fungi and virus with the aim of developing novel healthy optimal mahewu products. Additionally, in order to establish the quality of optimal mahewu products, studies can be developed to determine the nutritional and phytochemical quality of these mahewu products. Additionally, future research should also focus on consumers acceptance of these products.   Bacterial composition was low in raw white maize flour (RWMF), raw yellow maize flour (RYMF) and wheat flour (WF) but a higher bacterial composition in the malted inocula may be the result of the malting process. With respect to microbial activity and safety, steeping is the most critical step in malting [52]. Microbes, particularly lactic acid bacteria rapidly increase due to the limited accessibility of oxygen during steeping [53]. During germination, the ability of bacteria constituents to form a complex matrix in the form of a biofilm make the bacteria taxa a dominant microflora. Additionally, bacteria constituents have been reported to be the dominant microflora during germination [54]. Additionally, metabolic changes which include the conversion of residual carbohydrates to fermentable sugars occur during germination. In this phase, Lactobacilli was found as the predominant LAB, whereas Leuconostoc species were the major LAB during steeping. Furthermore, kilning which happens to be the final phase during malting is very important to the quality and stability of microbes [54]. However, the microbes that present at this phase are certain heat resistant bacteria which can form spores and they include bacillus species [51]. Similarly, as reported in the literature, malting can significantly improve the grain nutritional profile such as proteins, phenolic compounds as well as functional microorganisms with health benefits [55].

Conclusions
It could be concluded that the optimum processing conditions for white and yellow maize mahewu are 25 • C for 54 h with a boiling time of 19 min for white maize mahewu and 29 • C for 72 h with a boiling time of 13 min for yellow maize mahewu. Isolation of bacterial nucleic acids from the natural environment has become a useful tool to detect bacteria that may be difficult to culture. However, the present study reveals the shift in the bacterial community of optimised mahewu produced with different inocula using a high-throughput microbiota determination approach. This research will serve as the foundation for standardised, high-quality production of white and yellow maize mahewu. Therefore, further studies are needed to identify organisms such as fungi and virus with the aim of developing novel healthy optimal mahewu products. Additionally, in order to establish the quality of optimal mahewu products, studies can be developed to determine