White maize-Zea mays, White Sorghum-Sorghum bicolor, Pearl millet (Pennisetum glaucum L.) were purchased from Igbona market in Osogbo, Nigeria, while lemon (Citrus limon) fruits were purchased from Durban, South Africa. The γPGA sodium salt, media, chemical, and reagent were Oxoid products and Sigma-Aldrich grade.

The fermentation process of OSL and LOSL from different cereals followed the modified method described by Ajayeoba et al. (2016). Liquor samples of OSL and LOSL were taken from the unsieved ogi supernatant at every 6 h during a 72 h fermentation period to test for the presence of Bacillus species.

Isolation of Bacillus spp. was carried out according to the modified method described by Chettri et al. (2016). About 10 ml of each liquor was homogenized with 90 ml sterile saline water in a stomacher (Fisher Scientific, United Kingdom) for 1 min. The mixture was serially diluted and heated at 100°C for 2 min, incubated for 24 h at 37°C and enumerated on sterile nutrient agar. Mucoid colonies were purified and preserved in 15% (v/v) glycerol at -80°C for further analysis.

The morphological and physiological characterization of the purified isolates were based on microscopic shape, Gram staining reaction, catalase, motility test, endospore formation, production of carbon dioxide from glucose, ammonia from arginine, growth at different temperatures (30 and 45°C), NaCl concentrations (1.5 and 2.5%) and pH in nutrient broth following the method of Schillinger and Lücke (1987). Furthermore, Voges– Proskauer test, nitrate reduction, starch hydrolysis, casein hydrolysis, citrate utilization test, bile salt tolerance, anaerobic growth, and sugar fermentations were determined following the method of Duc et al. (2004). Taxonomic key of Slepecky and Hemphill (2006) was followed for the identification of Bacillus spp.

The stickiness of prospective Bacillus isolates was measured and screened for potentials of presumably γPGA production according to the modified described by Chettri et al. (2016). The precipitated γPGA was purified, dialyzed overnight against distilled water, freeze-dried and used to evaluate some functional properties of γPGA.

The yield of each γPGA polymer was detected using spectrophotometric ultraviolet assay while polysaccharide and protein contents were determined using phenol-sulphuric acid and bicinchoninic acid as described by Zeng et al. (2012). Glucose and bovine serum albumin were used as standards. Furthermore, the molecular weights of γPGA were detected by measuring the diffusion distance of the concentric zone on plates according to Zeng et al. (2013). After incubating the plate at 37°C for 5 h, the distance observed was measured by a Vernier caliper and the molecular weight was calculated with the equation; (y = a + bx, where a = 10.607, b = -0.000677013, y = diffusion distance, x = molecular weight). Fourier Transform Infrared Spectroscopy (FT-IR) (PerkinElmer) was also used to determine organic groups in the range of 4000–400 cm^-1 (Konglom et al., 2012).

16s rRNA sequences of the selected isolates were identified by PCR at Inqaba Biotech, South Africa using 27F (5′-AGAGTTTGATCCTGGCTCAG-3′) and 1492R (5′-CGGTTACCTTGTTACGACTT-3′) primers (Altschul et al., 1997). PCR products were sequenced, assembled and edited with software BioEdit 7.2.5. Basic Local Alignment Search Tool (BLAST)¹ was used to compare the consensus sequences with those deposited in the GenBank DNA database. A phylogenetic tree based on 16s rRNA genes was constructed to determine the closest bacterial species by the neighbor-joining method (Saitou and Nei, 1987), using MEGA7 (Kumar et al., 2016). Sequence divergence among LAB isolates were quantified using the Maximum Composite Likelihood method (Tamura et al., 2004).

Bacillus species have been identified in various fermented food products. The growth of any microorganism in a fermenting medium may be influenced by initial inoculum, pH, type of fermentation, food substrate, temperature and nature fermenting microorganism (Karovičová and Kohajdova, 2007; Todorov and Holzapfel, 2015). In this study, the colony forming units of Bacillus spp. gradually decreased with an increase in fermentation time and varied with the type of cereal (Table 1). Although the number of Bacillus species was higher in OSL than LOSL, it was not significantly different (p > 0.05). After 18 h of fermentation, the growth of Bacillus spp. was not observed in LJOSL. However, Bacillus spp. was isolated in OSL (millet) at 48 h of fermentation. The varied colony forming units (CFU/ml) of Bacillus in fermented liquor suggest differences in the mechanisms of fermentation in each cereal. Findings are in agreement with the report of Okeke et al. (2015) that Bacillus species are dominant in all forms of ogi liquor at the early stage of fermenting but may gradually phase out as other competing microorganisms emerge. The gradual decrease of Bacillus spp. may be due to the progressive acidification of OSL and LOSL, growth of other LAB and yeast (Ijabadeniyi, 2007; Obi, 2014; Ajayeoba et al., 2016). Isolates were randomly selected from OSL and LJOSL (4 isolates from each category and at each fermenting intervals) and phenotypically characterized as B. lichenformis, Bacillus subtilis, Bacillus cereus, Bacillus thuringienesis, and Bacillus amyloliquefaciens respectively. A total of 40 isolates exhibited varying degrees of stickiness were observed at the early hours of fermentation (0–12 h). PGA-producing Bacillus spp. are potentially characterized with shiny, milk white and mucoid morphology (Li et al., 2008; Kolton et al., 2017) as shown in Figure 1A.

The PGA production of the isolates was tested at 25, 30, and 45°C (Table 2). The isolates formed an insoluble fibrous precipitate after addition of twice the volume of 95% ethanol into the PGA medium (Figure 1B). On the basis of high fibrous precipitate at 30°C and pH 7.5, and stickiness of >15 cm, 4 isolates of Bacillus spp. (TA004, TA006, TA011, TA012) were used for further analysis (Table 2).

The yield, protein and polysaccharide contents of the γPGA varied with the type of isolate. However, the molecular weights of γPGA polymers (Table 5) showed that TA004 had the highest (261.44 kDa) while TA011 had the least (194.97 kDa). The addition of ethanol to the broth supernatant used in growing γPGA-producing Bacillus is known to precipitate polysaccharides and proteins (Buescher and Margaritis, 2007). The polysaccharide and protein components are the major constituent of γPGA matrix that influences its functionality. The protein nature of γPGA determines the economic and environmental application of the polymer (Ho et al., 2006). The molecular weight of γPGA was determined by diffusion distances (TA 004–10.43 mm; TA006-10.47 mm; TA011-10.48 mm; TA012-10.44 mm) in agar plate as shown in Table 5.

Microbial production of γPGA can have molecular weights ranging from 100 to over 2,000 kDa (Ho et al., 2006; Ogunleye et al., 2015; Luo et al., 2016). The yield and molecular weight of PGA are dependent on nature and pH the substrate, microbial species involved, composition, cultivation conditions, and time of fermentation (Markovsky et al., 2012; Lee et al., 2018). These influences the specificity (Ogunleye et al., 2015) and explains the importance of the substrate to isolate PGA-producing Bacillus spp. Although potent ACE-inhibitory compounds reported were released from fermented protein-based products (Chettri et al., 2016), this research shows that other fermented products may have ACE inhibitory compounds in some Bacillus spp., depending on the fermenting conditions. Furthermore, the physicochemical characteristics of the polymers, such as molecular weight, charge, chemical composition and polydispersity influences the properties and applications of a polymer (Markovsky et al., 2012). Most of the ACE inhibitors known for drug delivery have low molecular-weight-compounds, usually between 30 and 50 kDa (Lee et al., 2018) because it does not readily diffuse through normal capillaries and glomerular endothelium, therefore additional downstream purification steps will be necessary to isolate specific molecular weight fractions from γPGA used in this study (Richard and Margaritis, 2001). Antimicrobial properties are desired attribute of a substance intended for food and biomedical applications. There are few reports on the antimicrobial functions of pure γPGA. Although it is reported that higher molecular weight of polymer may diffuse slowly, the factors that are considered for antimicrobial activity are surface charge, hydrophobicity, hydrophilicity, and intramolecular hydrogen bonding (Khalil et al., 2017; Wang et al., 2017), and these differ between producing strains.

The accession numbers of Bacillus spp. identified in this study are TA004 (B. subtilis-GenBank:MH782061), TA006 (B. amyloliquefaciens- GenBank:MH782075), TA011 (B. subtilis- GenBank:MH782088), TA012 (B. subtilis- GenBank:MH782083). The confidence degree, E = 0.0 and homology similarity are between 99 and 100% for all the isolates. The phylogenetic tree constructed with neighbor-joining method was based on the evolutionary relationships and three distinct clusters were formed (Figure 4) Bacillus subtilis (TA004, TA011, TA012) showed 85% similiarity with B. subtilis JCM 1465, B. tequilensis 10b, B. subtilis NBRC 13719 and B. subtilis BCRC 10255 as shown in cluster 1. TA006 (Bacillus amyloliqueficiens) was observed in same clade showing 61% similarities with B. siamensis PD-A10 and B. amyloliquefaciens NBRC 15535 among others. B. subtilis and B. amyloliquefaciens are known γPGA producers (Chettri et al., 2016; Luo et al., 2016). These relationships suggest that some Bacillus spp. identified in Ogi steep liquor could have possible application in industrial settings.