A novel strain of acetic acid bacteria Gluconobacter oxydans FBFS97 involved in riboflavin production

A novel bacterial strain of acetic acid bacteria capable of producing riboflavin was isolated from the soil sample collected in Wuhan, China. The isolated strain was identified as Gluconobacter oxydans FBFS97 based on several phenotype characteristics, biochemicals tests, and 16S rRNA gene sequence conducted. Furthermore, the complete genome sequencing of the isolated strain has showed that it contains a complete operon for the biosynthesis of riboflavin. In order to obtain the maximum concentration of riboflavin production, Gluconobacter oxydans FBFS97 was optimized in shake flask cultures through response surface methodology employing Plackett–Burman design (PBD), and Central composite design (CCD). The results of the pre-experiments displayed that fructose and tryptone were found to be the most suitable sources of carbon and nitrogen for riboflavin production. Then, PBD was conducted for initial screening of eleven minerals (FeSO4, FeCl3, KH2PO4, K2HPO4, MgSO4, ZnSO4, NaCl, CaCl2, KCl, ZnCl2, and AlCl3.6H2O) for their significances on riboflavin production by Gluconobacter oxydans strain FBFS97. The most significant variables affecting on riboflavin production are K2HPO4 and CaCl2, the interaction affects and levels of these variables were optimized by CCD. After optimization of the medium compositions for riboflavin production were determined as follows: fructose 25 g/L, tryptone 12.5 g/L, K2HPO4 9 g/L, and CaCl2 0.06 g/L with maximum riboflavin production 23.24 mg/L.

Morphological, biochemical, and physiological characteristics. The isolated strain FBFS97 was examined for morphological properties according to the methods described previously 27 . The ability to produce 2-keto-D-gulonate, 2,5-Diketo-d-gluconate, and 5-keto-D-gulonate was performed through detection using the HPLC technique as reported before by Blake et al. 28 . Catalase and oxidase activities, utilization of different carbon sources, acetic acid production, as well as the isolated bacterial strain growth at different temperatures and pH values were investigated as reported by Yamada and others 29-31 . Genome sequencing and assembly. Total genomic DNA of FBFS97 was extracted using the GenElute Bacterial Genomic DNA kit (Sigma) according to the manufacturer's recommendations. FBFS97 whole genome was sequenced using the PacBio RSII platform utilizing 2 single-molecule real-time (SMRT) cells, was used to acquire the raw sequence reads at the Genome Technology Facility (GTF) in Lausanne, Switzerland. The highquality reads were then assembled de novo into a single contig using Hierarchical Genome Assembly Process version 3.0 (HGAP 3.0) in SMRT Analysis version 2.3.0. Automatic annotation and gene prediction were carried out using Prokka version 1.1.0. Circos was used to create the genome circle (https ://circo s.ca/tutor ials/lesso ns/). Molecular identification by 16S rRNA and phylogenetic analysis. The 16S rRNA gene fragment was amplified using universal primers (27F 5′-AGA GTT TGATCMTGG CTC AG-3′ and 1492R 5′-TAC GGY TAC CTT GTT ACG ACTT-3′). The amplification profile included an initial denaturation at 94 °C for 4 min, followed by 30 amplification cycles of 94 °C for 30 s, 55 °C for 30 s and 72 °C for 1.5 min. The PCR product was sent to Shanghai Sangon Biological Engineering Technology & Services Co., Ltd., for sequence determination. The phylogenetic tree was constructed through the neighbor joining (NJ) method of MEGA 7.0 software 32 .
Chromatography conditions. A high-resolution UHPLC Dionex Ultimate 3,000 coupled with a Q-Exactive hybrid quadrupole-orbitrap mass spectrometer was applied. In brief, 5 ml of culture suspension was centrifuged and filtrated with a 0.22 µm membrane. 20 µl of the filtrated sample was injected into an Accucore aQ C18 Polar Endcapped LC column (150 mm × 2.1 mm, 2.6 µm) at 25 °C and the flow rate was 0.2 mL/min. The analysis was achieved at 270 nm with 30% (v/v) methanol, 70% (v/v) water and 0.1% (v/v) formic acid as a mobile phase. The detection of riboflavin was made in positive electrospray ionization mode (ESI +). Riboflavin was identified based on the retention time and fragmentation pattern of a known standard. www.nature.com/scientificreports/ Riboflavin production conditions. The fermentative production of riboflavin by FBFS97 was carried out in 250 mL Erlenmeyer flasks containing 50 mL of fermentation media (according to the experiment). The inoculated flasks were incubated at 28 °C on a rotatory incubator shaker at 180 rpm for 72 h in aerobic conditions. After incubation time, the samples were collected by centrifugation at 8,000 rpm for 5 min. A 0.8 mL of the culture supernatant was mixed with 0.2 mL of 1 M NaOH, and then 0.4 ml of the resulting solution was neutralized with 1 mL of 0.1 M potassium phosphate buffer (pH 6.0). The concentration of riboflavin was determined by measuring the absorbance at 444 nm. The standard curve was constructed using pure riboflavin standard (Sigma).
Experimental designs and statistical analysis for medium optimization. Optimization of the fermentation medium was carried out using PBD followed by CCD.
Screening of the main minerals influencing on riboflavin production using PBD. To study the significant variables for riboflavin production, eleven selected minerals were screened out using Plackett-Burman design by Minitab 18. The PB experiment was carried out in 12 runs according to PB is K + 1 where K indicates to the number of variables, each variable was represented at two levels, maximum and minimum which are denoted by ( +) and (-), respectively. The levels of each variable are listed in (Table 6). All experiments were accomplished in triplicate and the average of riboflavin production was treated as response. The model was subjected to the analysis of variance (ANOVA). P-value was used to calculate the significance of the factors where, factors with a significance (P < 0.05) were considered to have a significant effect on riboflavin production. Plackett-Burman experimental design is based on a first order model: where Y is the response or dependent variable (riboflavin production), β 0 is the model intercept, β i is the regression coefficient, and X i is the level of the independent variable.
Central composite design (CCD). In this study, four significant minerals which exerted a positive effect on riboflavin production were used to optimize by employing CCD, each at five coded levels, very low level, low level, central level, high level and very high level indicated as − 2, − 1, 0, 1, 2 respectively. The CCD experiments consisted of 30 trails were designed using Design-Expert 12 (StatEase, Inc., USA). All experiments were done in duplicate and the average of riboflavin production obtained was taken as the response. The production of riboflavin was analyzed with statistical software package Design-Expert 12. After the responses were measured for each trail, each trail was fitted to an independent second order polynomial equation: Model validation experiment. In order to estimate the accuracy of the optimum medium composition which predicted by the response surface model. All validation experiments have been performed using the medium composition resulting from the model in triplicate.

Results and discussion
Morphological, cultural and physiological characterization. The isolated strain FBFS97 exhibited the general morphological, cultural and physiological characteristics of the AAB as was shown in (Table 1). FBFS97 cells were red shaped and Gram-negative. It was also a strictly aerobic, non-spore forming, motile strain, as well as catalase-positive, and oxidase-negative . Colonies of this strain were beige, circular, smooth, and raised to convex with an entire margin, glistening, and formed transparent zones through using CaCO 3 to produce acetic acid on GYC agar plates. It can produce acetic acid from d-glucose and ethanol but did not able to overoxidise acetic acid to CO 2 and H 2 O.Also, d-gluconic acid, 2,5-diketo-gluconic acid and a water-soluble brown pigment were produced from d-glucose. FBFS97 exhibited glucose tolerance up to 10% (w/v). It was able to grow at pH 3 and 37 °C but the optimum growth temperature was between 25 and 30 °C and optimal pH was 6. The results of acid production with different sugars displayed that FBFS97 produced acid from sucrose, D-mannitol, Dglucose, maltose, D-galactose, D-sorbitol, and D-fructose but not from α-lactose, β-cyclodextrixs, and lanolin (Tables 1,2).   Table 2. Differential characteristics of isolated strain FBFS97 and its phylogenetically nearest relatives in the genus Gluconobacter. 97, FBFS97; Go, G.oxydans NBRC14819 T ; Gr, G. roseus NBRC3990 T ; Gk, G. kanchanaburiensis LMG 26774 T ; Gz, G. ZW160-2 T . + , Positive; − , negative; b, some strains in the genus are positive; vw, very weakly positive; w, weakly positive;nd, not determined. Data cited from Spitaels and others 33,34 .

Genome features and identification of riboflavin in the G. oxydans FBFS97 fermentation media.
To investigate the riboflavin production related genes, the genome of Gluconobacter oxydans FBFS97 was sequenced, annotated and explored. The complete genome of FBFS97 is comprised of a single circular chromosome of 3,988,308 bp with an average coverage of 245 x, and a G + C content of 66.6% (Fig. 2). The prediction and annotation of FBFS97 genome resulted in 3,582 ORFs ,and 76.83% of the genome were assigned to the genes with predicted function. Also, the genome contains 12 ribosomal RNA operons (5S, 16S, 23S) , 59 tRNA and 11 other RNA genes. The complete genome of FBFS97 has a total of 3,500 putative coding sequences, among which 2,752 are assigned a putative function, and 748 encode hypothetical proteins. The genome properties and statistics are summarized in Table 3. Based on the Uniprot alignment 36 and KEGG database 37 in FBFS97 genome, the full set of riboflavin production related genes were annotated. The putative genes encoding of the riboflavin synthesis pathway on the genome of this strain is represented by RibBA, RibD1, RibD2, PYRP2, Rib4, Rib5, RibF, and bluB. In order to confirm the production of riboflavin in the fermentation broth of FBFS97 growth and riboflavin production. The production of riboflavin by G.oxydans FBFS97 was detected at different time points over the range of 0 -72 h. The growth started in the first hours of incubation, however, a slight riboflavin production was initiated at about 24 h. The production increased markedly at about 32 h when the growth was almost finished. The maximum riboflavin production was 0.33 mg/L at about 64 h, after this point, the production of riboflavin was started to decrease continuously until the end of the incubation period (Fig. 4). The medium pH quickly dropped from pH 6.0 on the first day to pH 2.0 at the end of incubation, which had a negative effect on riboflavin production. At pH 2.0 the riboflavin production starts to decrease until has become negligible or absent after 72 h of the incubation time 38 .
Screening of the main minerals affecting on riboflavin production using PBD. Pre-experiments were carried out to investigate the effect of four various carbon sources (glucose, fructose, maltose, and mannitol) and four different nitrogen sources (yeast extract, tryptone, peptone, and malt extract) on riboflavin production by G.oxydans FBFS97. The results obtained were analyzed by using a one way ANOVA test. The highest  (Tables 4, 5). Hence, fructose and tryptone were used as carbon and nitrogen sources for further experiments. The results showed that this improved medium was capable of enhancing riboflavin production from 0.33 mg/l to more than 5 mg/L in 64 h. Plackett-Burman design was used to screen and detect the effect of the 11 minerals (FeSO 4 , FeCl 3 , KH 2 PO 4 , K 2 HPO 4 , MgSO 4 , ZnSO 4 , NaCl, CaCl 2 , KCl, ZnCl 2 , and AlCl 3 .6H 2 O) on the riboflavin production using the improved medium. PB experimental design exhibited an obvious variation of riboflavin production from 0.32 to 3.72 mg/l. The maximum riboflavin production (3.72 mg/L) was observed in the run 9, whereas the minimum  www.nature.com/scientificreports/ www.nature.com/scientificreports/ riboflavin production (0.32 mg/L) was found to be in the run 1(      Table 6. Screened variables (g/L) and their two levels in the PBD, along with riboflavin production. FeSO 4 (X 1 ), FeCl 3 (X 2 ), KH 2 PO 4 (X 3 ), K 2 HPO 4 (X 4 ), MgSO 4 (X 5 ), ZnSO 4 (X 6 ), NaCl (X 7 ), CaCl 2 (X 8 ), KCl (X 9 ), ZnCl 2 (X 10 ), and AlCl 3 .6H 2 O (X 11 ). www.nature.com/scientificreports/ bar in this chart is proportional to the absolute value of its associated regression coefficient or estimated effect. In this study, Pareto chart showed that FeSO 4 , FeCl 3 , and KH 2 PO 4 were the most effective significant variables on riboflavin production but the effect of these three minerals were all nigative. Figures 5, 6 show the main effects plot of each variable on riboflavin production, we can notice that two variables from the eleven different independent variables called K 2 HPO 4 and CaCl 2 affect positively riboflavin production, where the six variables called FeSO 4 , FeCl 3 , KH 2 PO 4 , NaCl, KCl, and ZnCl 2 affect negatively riboflavin production. Therefore, only the significant variables with positive effect were used for further optimization experiments. The experimental results was obtained by using the following equation:

Optimization of significant variables by CCD.
From the Placket-Burman results, the most effective variables that significantly positive impact riboflavin production by FBFS97 were K 2 HPO 4 and CaCl 2 , consequently, they were selected for further optimization utilizing central composite design. K 2 HPO 4 renders as an important phosphorus source for substance vitality such as cell growth and product formation . Also, K 2 HPO 4 along with KH 2 PO 4 uses as a pH buffer for the medium 39 .Adding CaCl 2 to the medium can improve the meta-  www.nature.com/scientificreports/ bolic activity of bacterial strain. Calcium has an important role on the growth of microorganisms because of its ability to control the cell permeability 40 . In the present study, the CCD was employed to investigate the interactions between the significant variables and determine their optimum values for riboflavin production. A total of 30 experimental combinations for four significant variables (fructose, tryptone, K 2 HPO 4 , and CaCl 2 ) were achieved at five levels with the center points replicated six times in the experiment (run order : 4, 18, 20, 22, 23, 29). The range of variables concentrations at different coded levels shown in Table 7. The maximum riboflavin production 23.67 mg/L was observed in run number 16, while the minimum riboflavin production 0.095 mg/L was achieved in run number 28. The design matrix and the experimental and predicted values for riboflavin production noted in Table 7. Design Expert 12 software was used to analyze the experimental data, and the resultant second-order polynomial equation for riboflavin production was as the following: where Y is the predicted value of riboflavin production, X 1 is fructose, X 2 is tryptone, X 3 is K 2 HPO 4 , and X 4 is CaCl 2 .
(4) Y (riboflavin production) = 16.63 − 2.51X 1 + 3.26X 2 − 0.7775X 3 + 0.0475X4 − 0.8663X 1 X 2 − 3.26X 1 X 3 + 0.36X 1 X 4 + 2.27X 2 X 3 + 1.25X 2 X 4 + 0.7738X 3 X 4 −2.51X 2 1 −1.91 X 2 2 − 2.70 X 2 3 + 0.5205 X 2 4 Table 7. Design matrix of experimental runs for riboflavin production by G.oxydans FBFS97 using central composite design (CCD), representing the riboflavin production as affected by fructose (X1), tryptone (X2), K 2 HPO 4 (X3), and CaCl 2 (X4), together with the predicted riboflavin production and the levels of variables. www.nature.com/scientificreports/ Statistical analysis. The statistical significance of the fitted model was evaluated by multiple regression and the analysis of variance (ANOVA) which was tested using Fisher's test value Table 8. The model F-value of 18.80 with P < 0.0001 implies that the model is highly significant and there was only a 0.01% chance that this high F-value could occur due to noise. The lack of fit was not significant relative to the pure error (P ˃0.05). Furthermore, it can be noticed from the degree of significant that X 1 , X 2 , X 1 X 3 , X 2 X 3 , X 2 X 4 , X 1 2 , X 2 2 and X 3 2 are significant model terms. The coefficient of determination (R 2 ) of the model was 0.9461 which indicated that the model could be used reliably for the riboflavin production in this study. The Predicted R 2 of 0.7240 is in reasonable agreement with the Adjusted R 2 of 0.8957 which implied a good adjustment between the predicted and observed values. Adeq Precision ratio of 16.5602 indicates an adequate signal to noise ratio. The positive coefficient values indicate that individual effect (X 2 , X 4 ), interaction effects (X 1 X 4 , X 2 X 3 , X 2 X 4, X 3 X 4 ) and quadratic effect (X 4 2 ) increase the production of riboflavin, whereas other negative coefficient values indicate to decrease in riboflavin production.
Response surface plots. The three-dimensional response surface curves were plotted to explain the interaction of the variables and obtain the optimal level of each variable required for riboflavin production by FBFS97. 3D surface plots were created for the response (riboflavin production) at any two independent variables while keeping the other variables at their 0 level. Figure 7A shows the interaction effect of fructose and tryptone concentrations in riboflavin production. As can be seen in the plot, the increase of fructose concentration leads to a decrease in riboflavin production. While the increase of tryptone concentration enhanced the riboflavin production until the center point and then riboflavin production decreased gradually with a further increase in the concentration of this variable. The interaction effect between these two variables were not significant, pointing that out there is no significant correlation between them, thus they did not help much in the riboflavin production increasing. Figure 7B reveals the effect of fructose and K 2 HPO 4 concentrations on riboflavin production.Where a negative effect on riboflavin production was observed at the higher concentrations of both variables. This result was in agreement with the finding of Marjan et al 41 . who used fructose as the only carbon source along with K 2 HPO 4 to enhance the production of riboflavin by Bacillus subtilis ATCC 6,051, also they mentioned to that the excessive fructose and K 2 HPO 4 in the culture medium can be effected negatively on riboflavin production. The plot of Fig. 7C indicates the effect of the interaction of fructose and CaCl 2 on riboflavin production. The maximum production of riboflavin was obtained at the lowest level of CaCl 2 and level 0 of fructose and then riboflavin production was decreased from the central point along with further concentrations increase of both variables. Figure 7D represents the effect of tryptone and K 2 HPO 4 on the production of riboflavin by FBFS97. A gradual increase of riboflavin production was Table 8. Results of the analysis of variance (ANOVA) for the quadratic model used for optimizing riboflavin production by G.oxydans FBFS97. N non-significant, df degree of freedom, C.V coefficient of variation, P level of significant, F Fisher's function. www.nature.com/scientificreports/ found when the initial tryptone and K 2 HPO 4 concentrations were increased until the optimum value and then riboflavin production decreased with further increase in the concentration of both variables. Figure 7E depicts the interaction of tryptone and CaCl 2 . Riboflavin production increased as tryptone and CaCl 2 concentrations increased. The optimum riboflavin production was observed at the central point and then a slight decrease was observed with further increase in the concentration of both variables. Figure 7F indicates the effect of K 2 HPO 4 and CaCl 2 on riboflavin production in the FBFS97 fermentation medium. At a moderate level of K 2 HPO 4 the production riboflavin was high, as the CaCl 2 was at its low level. A slight decrease was noticed towards the central point as the concentration of both variables at their middle level, then a decline in the curve was observed with further increase of K 2 HPO 4 . In addition, the interaction terms between these variables were not significant, indicating that there is no significant correlation between the two variables, thus the interaction between them did not help much in increasing the production of riboflavin. www.nature.com/scientificreports/ Validation of the model results. In order to confirm the reliability of the statistical design and validate the results, an experiment with the new medium composition which predicted by the statistical model was carried out. It was obtained that the optimal levels of the variables for riboflavin production by G.oxydans FBFS97 were fructose 25 g/L, tryptone 12.5 g/L, K 2 HPO 4 9 g/L, and CaCl 2 0.06 g/L. The production of riboflavin by FBFS97 was 23.24 mg/L found from the validation experiment was close to the predicted value of 23.2 mg/L, which confirms the accuracy of the model with a great degree of desirability of 97.7% . Compared with some strains of lactic acid bacteria, the production of riboflavin by FBFS97 as a wild strain was higher than the riboflavin production by Lactobacillus fermentum GKJFE (3.49 mg/L), Lactobacillus plantarum NCDO1752 (0.6 mg/L), and Leuconostoc lactis NZ9000 (0.7 mg/L) as mutant strains 42 .

conclusion
A new riboflavin producer G.oxydans FBFS97 was isolated from the soil sample collected in Wuhan, China. Then, the effect of medium composition on riboflavin production by FBFS97 was investigated. Fructose and tryptone were chosen as suitable sources of carbon and nitrogen. PBD was applied to choose the most significant minerals effective on riboflavin production by FBFS97. The results of CCD coupled with response surface modeling (RSM) displayed that the optimum concentrations of the selected variables were fructose 25 g/L, tryptone 12.5 g/L, K 2 HPO 4 9 g/L, and CaCl 2 0.06 g/L with the maximum riboflavin production 23.24 mg/L. The results of this study revealed that G.oxydans FBFS97 could be a new potential candidate for the industrial application of riboflavin production.