Ten cassava tubers were harvested from two different markets located in the Bacongo and Makélékélé districts of the Brazzaville, Republic of Congo. Each sample was processed to test the presence or name of biosurfactants.

To control the fermentation of cassava tubers, we peeled and washed 500 g of cassava tubers, which we then left to ferment for 4 days. After fermentation, we proceeded to defibration to recover the retted dough. Once drained, this dough was traditionally kneaded to begin the transformation process into chikwangue, which was heated to boiling for 2 to 4 hours. Chikwangue was left at room temperature to cool. The emulsification test was carried out considering all key stages of fermentation (Batch 1: Solution containing the liquid of freshly harvested and crushed cassava tubers. A 5 ml solution was obtained after filtration. Batch 2: Solution whose fermented and retted tubers were mixed with physiological saline. A 5 ml solution was obtained after filtration. Batch 3: 10 g of Chikwangue were mixed with physiological saline. The mixture was vortexed. All batches of samples were subjected to a protein assay using the Bradford method. A 5 mL solution was obtained after filtration. Each batch was carried out three times in a row to optimise reproducibility. In this experiment, a physiological saline control was used. Each batch was mixed with 5 ml of essence (v/v). The mixture was shaken vigorously for 5 minutes with a vortex mixer (VELP Scientifica, Italy). The cassava tubers were then incubated at room temperature at room temperature for 24 hours. Subsequently, the height of the emulsion layer and the total height of the mixture were measured (all experiments were carried out in triplicate), and the emulsification index (E24%) was calculated using the standard formula E24% = (He/Ht) × 100, where He represents the height of the emulsion, Ht the total height of the mixture and E24% the percentage of emulsification after 24 hours.

To explain the presence of biosurfactant in batches 1 and 2, 10 g were aseptically collected in a sterile tube at different stages of chikwangue preparation. Samples were homogenized with physiological sterile water and distributed in sterile microbiological tubes. Then successive dilutions were performed, and the bacterial suspensions were inoculated in a Mossel agar medium (10.0 g of peptone, 1.0 g of meat extract, 10.0 g of mannitol, 10% of egg yolk, 0.01 g of polymyxin B sulphate, 0.025 g of phenol red, 10.0 g of sodium chloride, 14.0 g of agar and pH 7.2) to promote the growth of Bacillus. The plates were incubated at 37˚C for 24 hours and the colony count was carried out in triplicate. The purification of the isolates was carried out rigorously by successive subcultures using Luria-Bertani (10 g of peptone, 5 g of yeast extract, 10 g of NaCl). The morphological characteristics of the colonies, such as shape, size, and color, were recorded. Morphological characterisation was performed using a light microscope (OPTIKA, Italy), and Gram-staining was performed with a solution of 3% potassium hydroxide (KOH). For future experiments, all purified isolated cultures were stored at −20˚C in Luria-Bertani (LB) broth containing 20% glycerol (v/v).

Isolates with a good percentage of the emulsification index were subjected to genomic DNA extraction and purification. This experiment was carried out using the NucleoSpin Microbial DNA Kit (Macherey-NAGEL, Germany). Briefly, isolates were cultured in 5 ml of LB broth for 24 hours at 37˚C with shaking. DNA purity was assessed by the UV absorbance ratio (260/280 nm). 1 μL of template DNA with concentrations equalling 10 - 20 ng/μL. Universal 16S rRNA primers fD1 (5'-AGAGTTTGATCCTGGCTCAG-3') and rP2 (5'-ACGGCTACCTTGTTACGACTT-3') were used to amplify the 16S RNA gene (Eurogentec). 5 µl of each amplification product was mixed with 2 μL of loading buffer (BIOKE, The Netherlands). The mixtures were then electrophoresed on a 1% (w/v) agarose gel. The molecular weight marker used was the 10-kb 2-Log DNA sample (BIOKE, The Netherlands).

The confirmation of the identification of Bacillus species was carried out according to the method proposed by Kaya Ongoto et al. [18] . The species B. amyloliquefaciens, B. subtilis, B. pumilus, B. licheniformis, B. altitudinis, B. mojavensis, B. safensis and B. atrophaeus were identified through the amplification of the fibE gene, responsible to produce the fibrinolytic enzyme. PCR products were purified using the Gel Extraction Kit (Omega Biotek), the purified products were subjected to Sanger sequencing (3130xl Genetic Analyser, Applied Biosystems). The sequences obtained were aligned with Bio Numerics 7.5 software (Applied Maths, Belgium) and manually corrected to resolve the discrepancies between the sense and antisense strands. The sequences were compared with homologous sequences contained in sequence databases through the NCBI portal using the BLASTn programme ( https://www.ncbi.nlm.nih.gov/ ). A multiplex endpoint PCR assay was used to identify B. cereus enterotoxins: hemolysin BL (hblD) with Eurogentec primers (hblD-F (5’-ATGAAAAAATTTCCATTCAAAGTACTAAC-3’) and hblD-R (5’-GAATATCATTCCAACTTTCTTTAGCGGC-3’), non-haemolytic enterotoxin (nhe) with primers nheB-F (5’-ATGGCTCTATCAGCACTTATGGCAG-3’) and nheB-R (5’-TTAAGCTTTTTTCGTATCTACTACTTTAATAC-3’) and enterotoxin FM toxin genes (entFM) with primers entFM-F (5’-GTTGCAGTTCCAGGTATGGATTCTGC-3’) and entFM-R (5’TTCTGCACTAATGAACTGACCGTTTCC-3’). Amplification of the cytotoxin K gene (CytK) was detected in isolates classified as B. mycoides CytK-F1 (5′-CAAAACTCA(T/C)CTATGCAATTATGCAT-3′), CytK-F2 (5′-AAAATGTTTAGCATTATCCGCTGT-3′) and CytK-R (5′-ACCAGTTGTATTAATAACGGCAATC-3′). The phylogenic tree has been designed to facilitate the execution of phylogenetic workflows. The new generation phylogenetic services of NGPhylogeny.fr have been used to perform iTOL (interactive of live) [19] .

Isolates identified with a higher percentage of biosurfactant production were extracted as previously done [20] . Briefly, after 24 hours of incubation, the bacterial suspension of each isolate was centrifuged at 11,000 rpm for 15 minutes and the obtained supernatant was treated with HCl until a pH of 2.0 was obtained. The mixture obtained was then incubated at 4˚C for 24 hours. After incubation, a deposit was observed at the bottom of the tube, indicating the presence of the biosurfactant extract. This mixture was centrifuged again at 11,000 rpm for 15 minutes and once the supernatant was removed, the precipitate was collected by elution in 500 µL of 1X phosphate buffer saline (137 mM NaCl, 2.7 mM KCl, 10 mM Na2HPO4 and 1.76 mM KH₂PO₄) and kept cool to perform the inhibition test.

The antibacterial activity of the biosurfactant extract was evaluated against seven laboratory pathogenic strains (E. coli, S. aureus, S. flexneri, S. enterica, B. cereus, P. aeruginosa and Klebsiella sp), according to the method described previously [21] . The wells were prepared aseptically in plates containing counting agar. The microorganism to be tested was inoculated in the gel, then a volume of 75 µl of the biosurfactant extract was deposited in the wells. After an incubation period of 24 hours at 37˚C, the diameter of the inhibition zones was measured. The average of the three measurements was taken to ensure the reproducibility of the results.

To explain the increase in total crude secreted protein concentration in fermented and retted cassava paste (RCP) and diluted chikwangue (DC), we first resuspended 10 g of RCP and 10 g of chikwangue in 1X phosphate buffer saline (137 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, and 1.76 mM KH₂PO₄). The solutions (RCP and DC) were vortexed and filtered using filters with a diameter of 0.22 µm to exclude bacteria and fibres. After filtration, the solutions were centrifuged at 10,000 xg and concentrated 10 times using concentrators (PierceTM Protein Concentrators PES, 5K MWCO, 0.5 to 100 ml). The protein assay had been performed using the Bradford method (Bio-rad, USA). The proteolytic, amylolytic, and cellulolytic activities of the solution had been tested to degrade casein in skim milk, cassava starch, and cellulose, on the one hand. On the other hand, the ability of B. amyloliquefaciens, B. subtilis, B. pumilus, B. licheniformis, B. altitudinis, B. mojavensis, B. safensis and B. atrophaeus, B. mycoides, B. cereus, B. thuringensis, B. mycoides, Paenibacillus polymixa and B. tequilensis strains to produce proteases, amylases, cellulases was based on their ability to degrade casein in skim milk, cassava starch, and cellulose as substrates. The methodology carried out was previously described [7] [20] . In terms of proteolytic activities, an overnight culture of Bacillus strains was prepared in Luria broth (LB) at 37˚C. The 10 ml of culture was then centrifuged at 12,000 × g for 10 min and the supernatant was concentrated 10 times and 75 µL of each culture supernatant was transferred to a well containing 1% agarose and skim milk (10%). Regarding the amylolytic and cellulolytic activities, 1 g of starch and 0.5 g of cellulose were added separately in 100 ml of LB agar. The Petri dishes were incubated at 37˚C for 24 h. Positive activity was revealed by the presence of a halo around the well; the diameters of the hydrolysed halo were measured. A Lugol solution (Merck, Germany) was used to reveal positive amylolytic and cellulolytic activities. The percentage of lysis for each isolated colony was evaluated using the formula% L = ((DT DC)/DT), where% L is the percentage of lysis, DT is the total diameter (colony and halo) and DC is the colony diameter. The average of the three measurements was determined.

GraphPad Prism version 8.0 software was used to determine significance thresholds. Principal component analysis (PCA) was used to investigate the correlation between pathogenic bacteria in antagonistic activities. Before ordination, strains abundance data was transformed to better meet the assumptions of normality [22] using ln (x + 1). All analyses were performed using CANOCO (Canonical Community Ordination, version 4.5) [23] .

To demonstrate the presence of biosurfactants in fresh cassava, fermented tubers, and chikwangue, we tested the emulsifying activity of samples from batches number 1, 2 and 3. The results show that batch number 1, the emulsification index was zero (E24: 0%), which shows that fresh cassava does not contain any biosurfactants. However, batches 2 and 3 allowed to obtain emulsification indices (E24) ranged from 50% to 100% ( Figure 1(A) , Figure 1(E) ). Bacillus spp. is a remarkably diverse bacterial species that is capable of growth within many environments. In parallel, the protein concentrations of the samples were measured. The concentration of the sample of tubers without fermentation was 450 mg/mL. On the other hand, the sample of fermented cassava paste was 4 g/ml and the chikwangue was 2.5 g/mL. The extraction of biosurfactants in the presence of HCl allowed one to show that the total biosurfactants retained their emulsifying activity ( Figure 1(B) , Figure 1(F) ). The total crude extract of biosurfactants was able to inhibit the growing of S. flexneri, Klebsiella sp., Salmonella sp., E. coli, Pseudomonas, Staphylococcus aureus and Bacillus cereus ( Figure 1(C) , Figure 1(D) ). The crude biosurfactant sample directly from fermented cassava tuber paste had diameters ranging from 2.4 ± 0.4 (cm) to the chikwangue 3.8 ± 0.4 (cm) compared to biosurfactant samples whose diameters varied between 1.9 ± 0.7 (cm) to 3.4 ± 0.4 (cm) except for antagonism with the strain of Staphylococcus strain, which had a large diameter of the order of 4.7 ± 0.4 (cm) ( Figure 1(C) , Figure 1(D) ).

After serial dilutions, isolation, and culture of the unfamiliar bacteria, the culture characteristics, morphology observation, biochemical test, and precursory antibacterial test had been done. A total of 100 isolates were obtained. An emulsification test was performed on the 100 isolates. 16% of isolates including Mon33, Mon7liche, Mon7alt, Mon10sub, Mon12moj, Mon25, Mon31, Mon55, Mon77, Mon26, Mon30, Mon24, Mon36, Mon40, mon67sub, and mon67amy were the subject of molecular identification using 16S RNA, cytK, hblD, nheB and entFM genes. The sequences were submitted to the NCBI portal with the following accession numbers B. cereus (Genbank: PQ432941.1), B. licheniformis (Genbank: PQ432758.1), B. altitudinis (Genbank: PQ432754.1), B. subtilis (Genbank: PQ432759.1), B. mojavensis (Genbank: PQ432755.1), B. tequilensis (Genbank: MT994788.1), B. subtilis (Genbank: MT994789.1)

Paenibacillus polymyxa (Genbank: PQ452544.1), B. velezensis (Genbank: PQ452545.1), B. thuringiensis (Genbank: PQ432763.1), B. pumilus (Genbank: PQ432762.1)B. subtilis, (Genbank: MT994787.1), B.mycoides (Genbank: PQ432890.1), B.thuringiensis (Genbank: PQ432766.1), B. subtilis (Genbank: PQ432757.1) and B. amyloliquefaciens (Genbank: PQ432756.1). B. species belonging to group II were amplified with the fibE gene encoding subtilisin. This concern B. subtilis, B. amyloliquefaciens, B. subtilis, B. pumilus, B. licheniformis, B. altitudinis, B. mojavensis, B. safensis, B. atrophaeus and B. mojavensis. The detected DNA fragments at around 800 bp were sequenced, which confirmed the bacterial species. The B. cereus species had haemolytic activity. PCR amplification from the nicked genes of strain B. cereus with primers hblD-F/hblD-R, nheB-F/nheB-R, entFM-F/entFM-R allowed us to make the difference between B. thuringensis and B. cereus. PCR amplification of the cytK gene from strain B. mycoides with the set CytK_F1 and CytK_R allowed the production of a 369-bp amplicon, while the usage of the primer pair F2-CytK and R-CytK resulted in the formation of a 238-bp amplicon.

We established the phylogenetic tree of the strains that were obtained using live iTOL interactive of live ( Figure 2 ). As a result, Bacillus spp. Respect their phylogenic relationship.

The amylolytic, proteolytic and cellulase activities of fresh tubers, fermented cassava tuber paste and chikwangue were compared with the capacity of the different strains identified. The percentages of activities range between 60 and 100%. However, it should be noted that chikwangue presents a percentage of 0% in its proteolytic, amylolytoic and cellulosic activity. Chikwangue could emulsify up to 47% ( Figure 3 ). The suspension was concentrated 10 times and emulsified to 100%.

Biosurfactant extracts of Bacillus strains including B. subtilis, B. amylolechefasciens, B. licheniformis, B. mycoides, B. altitudinis, B. cereus, B. thuringensis, B. mojavensis, B. safensis Pb. polymyxa, B. velezensis, and B. subtilis strain 48 (used as a positive control) were obtained. Antibacterial tests on pathogenic bacteria (E. coli, S. flexneri, S. enterica, Klebsiella sp., P. aeruginosa, S. aureus, and B. cereus) were performed. This work revealed that all biosurfactant extracts of Bacillus strains could inhibit the growth of pathogenic bacteria. The diameters of the Gram-negative pathogen bacteria (S. enterica, S. flexneri, E. coli, Klebsiella sp., P. aeruginosa) ranged from 2.3 ± 01 cm to 5.5 ± 0.4 cm. On the other hand, the diameters of Gram-positive pathogenic bacteria (B. cereus and S. aureus) varied between 1.5 ± 0.5 cm and 4.0 ± 0.2 cm. These values appear to be lower than those obtained with Gram-negative bacteria. The diameters of S. enterica ranged from 2.6 ± 0.3 cm to 4.8 ± 0.2 cm. The biosurfactant of B. subtillis was the most significant inhibition compared to Salmonella. The diameters of E. coli ranged from 2.3 ± 0.1 cm to 5.5 ± 0.4 cm. The biosurfactant of B. licheniformis was the most significant inhibition compared to E. coli. ( Table 1 ). The bacterium of the genus Shigella were more sensitive to biosurfactants isolated from Bacillus whose diameters varied between 3.1 ± 0.1 cm and 4.7 ± 0.4 cm. The bacteria of the genus Klebsiella were also sensitive to biosurfactants isolated from Bacillus whose diameters varied between 3.1 ± 0.1 cm and 4.1 ± 0.5 cm ( Table 1 ).

Distances and correlations were determined based on PCA analyses and dindogram construction based on inhibition diameters ( Figure 4(A)-(C) ). Synergy of the antibacterial effect was determined. Based on the Euclidean distances, the antibacterial effect was grouped. Note the groups whose antibacterial effect is correlated with group 1 consisting of B. cereus and S. aureus ( Figure 4(B) ). Both bacteria are Gram-positive bacteria. However, group 2 is larger. It consists of two subgroups. The inhibition values of group 2a, including S. flexneri and E. coli are close and those of group 2b correlate with S. enterica, P. aeruginosa and Klebsiella bacteria ( Figure 4(B) ).

In the Euclidean distance analysis, the longest distance was observed between B. cereus and S. flexneri whose value was around 7.08. The lowest value was observed between B. cereus and S. aureus whose distance was 2.36 ( Figure 4(A)-(C) ).