Introduction

Acetoin (3-hydroxy-2-butanone) is a volatile compound widely existing in fruits, vegetables, dairy products, and some fermentative foods, and is mainly applied in food industries as a flavor enhancer at present [1]. As one of the 30 platform chemicals classified by the US Department of Energy, acetoin could be used as a precursor to produce liquid hydrocarbon fuels, diacetyl, acetylbutanediol, pyrazines, amino nitriles, etc. [2, 3]. Acetoin has two optical isomers: (3R)- and (3S)-acetoin. Currently, commercial acetoin is largely synthesized from chemical methods, which is a racemic mixture and is sufficient for these applications. However, optically pure acetoin has some specific applications in the synthesis of novel optically active materials such as α-hydroxyketone derivatives and liquid crystal composites [2]. Furthermore, (3R)-acetoin is a female sex pheromone of Amphimallon solstitiale which could attract swarming males, whereas neither racemic acetoin nor 2,3-butanediol (2,3-BD) showed activity [4]. Therefore, it is essential to develop an effective process to produce optically pure acetoin.

In recent years, three methods have been developed to produce (3R)-acetoin:enzymatic catalysis of 2,3-BD, whole-cell catalysis of 2,3-BD, and fermentative production from carbohydrates. Under the catalysis of alcohol dehydrogenase meso-2,3-BD is converted to (3R)-acetoin, but the final concentration of (3R)-acetoin was relatively low due to the limitation of substrate concentration and enzyme stability [5]. By overexpressing 2,3-BD dehydrogenase and NADH oxidase simultaneously, a whole-cell biocatalyst of recombinant Escherichia coli was successfully constructed, which could convert meso- or (2R, 3R)-2,3-BD to (3R)-acetoin [6, 7]. A titer of 86.74 g/L (3R)-acetoin with optical purity of 97.89% was achieved when hemoglobin protein was co-expressed [6]. In general, the production cost is relatively high by whole-cell catalysis due to the high cost of 2,3-BD.

Most of the microorganisms can produce a mixture of both isomers of acetoin or acetoin with unreported optical purity [2]. Part of the wild type or mutant strains can accumulate high concentration of acetoin with acceptable yields, such as Bacillus [8,9,10], Lactococcus lactis [11], Serratia marcescens [12]. However, the titer and optical purity of (3R)-acetoin were not satisfactory in all the discovered native strains. The reported highest optical purity of (3R)-acetoin obtained from native strains was 95.2% [8]. By metabolic engineering of strains, the titer and optical purity of (3R)-acetoin could be increased [13, 14]. Until now, a native microorganism which could produce high concentration of (3R)- or (3S)-acetoin with high optical purity has not been reported.

In this research, a newly isolated strain of Bacillus subtilis was reported for the production of (3R)-acetoin with glucose as carbon source. The effects of nutrients in fermentation medium on (3R)-acetoin production was evaluated by Plackett–Burman (P–B) design, and (3R)-acetoin could be accumulated to 83.7 g/L with optical purity of 99.4% in a 5-L batch fermenter. As one of the generally recognized as safe (GRAS) species, this strain demonstrates great potential for industrial-level production of optically pure (3R)-acetoin.

Materials and methods

Materials and reagents

The standard racemic mixture of acetoin was purchased from Shanghai Nuotai Chemical Co., Ltd. (China), and meso-2,3-BD was from Sigma-Aldrich. Glucose was purchased from Xiwang Sugar Co., Ltd. (China), corn steep liquor powder (CSLP) was from Jinzhou Yuancheng Biochemical Technology Co., Ltd. (China), and other chemicals were purchased from Sinopharm Chemical Reagent Co., Ltd (China).

The seed-culture medium and screening medium were prepared according to the seed medium and fermentation medium in the literature, respectively [15]. The initial pH of the media was adjusted to 7.0.

Microorganism isolation and strain identification

Five sea sediment samples from Dalian, China, were separately soaked in sterile deionized water and agitated vigorously before settling for 10 min. The supernatant of 1 mL was inoculated into 99 mL seed-culture medium and cultured at 200 rpm, 37 °C for 72 h. The obtained 72-h culture was properly diluted and spread onto solidified seed-culture medium. Bigger colonies were picked to test the acetoin production in screening medium (100 mL) by culturing at 200 rpm, 37 °C for 48 h. During culture, 0.5 mL broth was taken at 24-h and mixed with 0.5 mL 40% glycerol solution which was stored at − 20 °C for following study. Then, the isolate with highest (3R)-acetoin (1 mL stored sample) was inoculated into 99 mL screening medium containing 10 g/L acetoin and cultured at 200 rpm, 37 °C for 24 h. The 24-h culture was diluted and spread onto solidified seed-culture medium for re-screening. The obtained strain was purified by the streaking technique, and identified by 16S rRNA sequence analysis carried out at TaKaRa Biotechnology (Dalian) Co., Ltd.

Selection of key medium components using P–B design

The experiments including P–B design and medium optimization were carried out in triplicate by shaking-flask cultivation at 37 °C and 200 rpm. The primary experiment showed that the strain grew better at 200 rpm than 180 rpm, so the rotation speed of the shaker incubator was set at 200 rpm.


The supplements including glucose, CSLP, phosphate, metal ions and some trace elements were used to compose the P–B design (Tables 1, 2). A 20-run P–B design was used to screen 15 factors in fermentation medium by strain CGMCC 13141, and the experimental responses were analyzed by the method of least squares to fit the following first-order model:

$$Y={\alpha _0}+\sum {\alpha _i}{x_i}$$
(1)

where Y was the predicted response, αi was the regression coefficient, and xi was the coded level of the variable.

Table 1 Nutrient supplements for screening in the Plackett–Burman design
Full size table
Table 2 Experimental design and results of Plackett–Burman design
Full size table

Medium optimization

First, the experiments were carried out in the direction of the steepest ascent based on the first-order model obtained from P–B design. Then, the 5-level-2-factor CCD (central composite design) was employed to optimize the medium composition. With the conversion of glucose to (3R)-acetoin as response, a second-order polynomial was fitted to correlate the relationship between independent variables and response:

$$Y={\beta _0}+\sum {\beta _i}{x_i}+{\beta _{12}}{x_1}{x_2}+\sum {\beta _{ii}}{x_i}^{2}$$
(2)

where Y was the predicted response, β0 was the offset term, βi was the linear effect, βii was the squared effect, β12 is the interaction effect, xi represented the concentration of Mn2+ (X1) and Co2+ (X2), respectively. The fit of the quadratic polynomial model was evaluated by the determination coefficient (R2) and analysis of variance (ANOVA).

Fed-batch fermentation of (3R)-acetoin

With optimized medium as fermentation medium, the fed-batch fermentation was carried out in a 5 L stirred bioreactor (BIOTEC-5JG-7000A, BXBIO, China) with work volume of 1.5 L. The seed was inoculated (10%, v/v) into the fermentation medium at an initial natural pH of 7.3. The cultivation was carried out at 170 rpm, 37 °C with different aeration rate, and the pH of fermentation broth was decreased automatically to pH 5.9, then maintained by automatic addition of 5 mol/L NaOH. When the glucose in broth was less than 10 g/L, about 140 g solid glucose was added into the fermentor one time.

Analytical method

The concentration of glucose was determined by a glucose analyzer (Biosensor SBA-50, Shandong Academy of Sciences, China). The concentration of acetoin and 2,3-BD was analyzed by a GC system (Agilent Technologies 7890 A GC systems) equipped with a capillary column BGB-174 (30 m × 0.25 mm I D 0.25 µm df) and a FID detector. Before GC analysis, the sample (1 mL) was centrifuged at 12,000 rpm, 4 °C for 10 min. The supernatant (100 µL) was mixed with 900 µL absolute ethanol and centrifuged at 12,000 rpm, 4 °C for 10 min. The obtained ethanol supernatant was diluted to suitable concentration with absolute ethanol for GC analysis. The optical purity of (3R)-acetoin was the concentration ratio of (3R)-acetoin over the total of (3R)- and (3S)-acetoin. The biomass was represented by the optical density at 620 nm (OD620) detected by a spectrophotometer. All the samples and standards were analyzed in triplicate to obtain the average value.

Results and discussion

Strain isolation and identification

As a primary metabolite, acetoin production is usually related with bacterial growth. According to this experience, bigger colonies (diameter ≥ 2 mm) were picked during screening process. Among the 60 picked colonies, two strains (No. 11 and 13) exhibited higher production of (3R)-acetoin and little production of (3S)-acetoin, and two strains (No. 21 and 22) showed higher total concentration of acetoin and 2,3-BD. Strain No. 11 was selected for acclimation, and further screening was carried out. The strain (DL 11–11) showed highest titer of (3R)-acetoin, which was purified and reserved for the following studies.

The sequence of 16S rRNA of DL 11–11 was determined and submitted to GenBank (accession number: MG727662), and the phylogenetic tree was shown in Fig. 1. Sequence analysis showed strain DL 11–11 shared 100% identity with Bacillus sp. DL 01 which was later classified as B. subtilis [8], and 99% identity with Bacillus sp., B. subtilis, B. methylotrophicus, B. pumilus, and B. amyloliquefaciens. Therefore, the strain DL 11–11 was classified as B. subtilis and reserved in China General Microbiological Culture Collection Center (CGMCC, No. 13141).

Fig. 1
figure 1

Phylogenetic tree based on 16S rRNA sequences showing the position of strain DL 11–11 (accession number MG727662) among its closely related organisms. Numbers in parentheses are accession numbers of published sequences. The tree was constructed by the neighbour-joining method. Bootstrap values are shown as percentages of 1000 replicates. Bar: 0.01 estimated sequence divergence

Full size image

Effect of nutrients on the production of (3R)-acetoin

The medium composition and process control are the most important factors to enhance the production of a target compound for a given strain via fermentative production [2]. The method of P–B design is an efficient technique to eliminate insignificant factors and pick out factors which significantly influence the production of acetoin [10, 16], thus a smaller and more manageable set of factors is obtained.

During screening experiment, little (3S)-acetoin was detected in fermentation broth and the conversion rate was about 0.35 g (3R)-acetoin/g glucose. Therefore, the optical purity of (3R)-acetoin was not selected as response. The effect of nutrients on (3R)-acetoin production was evaluated using (3R)-acetoin concentration and conversion of glucose to (3R)-acetoin as responses. The experimental design and results were shown in Table 2. The highest (3R)-acetoin production (35.5 g/L) and glucose conversion to (3R)-acetoin (0.454 g/g) were obtained at the condition of Run 6, while the lowest (3R)-acetoin production (6.9 g/L) and glucose conversion (0.281 g/g) were obtained at Run 13 and 16, respectively. The (3R)-acetoin production at Run 3 was similar to that obtained at Run 6 while the glucose conversion was a little lower. The difference of factor levels among Run 3, 6, 13, and 16 indicated that carbon source (X1), phosphorous source (X3), and divalent metal ions such as Mg2+ (X6), Ca2+ (X9), Mn2+ (X10) and Co2+ (X15) played important role in (3R)-acetoin production and glucose conversion.

With the production of (3R)-acetoin as response, the fitted first-order model equation could be written as: Y = 21.54 − 2.96X1 + 2.44X2 + 3.18X3 + 0.55X4 − 0.61X5 + 1.70X6 + 1.61X7 − 0.035X8 − 0.15X9 + 3.95X10 − 1.82X11 − 0.26X12 − 0.055X13 − 0.67X14 + 3.24X15. The coefficient of each variable displayed the strength of the effect of this variable on (3R)-acetoin accumulation. The quality of fit of the polynomial model equation was expressed by the coefficient of determination (R2), which was 0.9658. The statistical analysis showed that factors having greatest impact on the accumulation of (3R)-acetoin were X10 (Mn2+, P = 0.006), X15 (Co2+, P = 0.01), X3 (HPO42−, P = 0.01), X1 (glucose, P = 0.02) and X2 (CSLP, P = 0.03). Among them, Mn2+, Co2+, HPO42−, and CSLP were positive factors, while glucose was negative in the tested range.

With the conversion of glucose to (3R)-acetoin as response, the fitted first-order model equation was: Y = 0.37 + 0.014X1 + 0.019X2 + 0.00625X3 + 0.017X4 − 0.00545X5 + 0.00975X6 + 0.00605X7 + 0.00155X8 − 0.011X9 + 0.020X10 − 0.031X11 − 0.00145X12 + 0.01X13 − 0.016X14 + 0.018X15. The value 0.9814 for R2 indicated that 98% of the variability in the response could be explained by the model. Based on statistical analysis, the factors with greatest impact on conversion were X11 (H3BO3, P = 0.001), X10 (Mn2+, P = 0.007), X2 (CSLP, P = 0.009) and X15 (Co2+, P = 0.01), in which Mn2+, CSLP, and Co2+ were positive while H3BO3 was negative.

CSLP is a type of cheap organic nitrogen source which can provide nitrogen source and some trace elements such as Mg, Fe, Ca, and Zn [17]. The study of B. subtilis 168 showed that the addition of CSLP (5–50 g/L) could stimulate the cell growth and acetoin reductase (AR) activity, increase the consumption rate of glucose and glucose conversion to acetoin and 2,3-BD by lowering the production of lactate and succinate [17]. With the increasing concentration of CSLP, the accumulation of 2,3-BD increased while acetoin concentration decreased. Therefore, the addition of CSLP at low concentration was favorable for acetoin production. In this study, CSLP showed positive effect on acetoin accumulation and glucose conversion under a range of 0.75– 2 g/L.

The divalent metal ions are essential in the fermentation of acetoin and played important roles in the catalysis and structural stability of the key enzymes in acetoin pathway (2,3-BD pathway) [10, 16, 18,19,20,21]. However, the preference of metal ions and the detailed effects were dependent on the strain used. For example, Mg2+ was one of the cofactors of α-acetolactate synthase from B. subtilis [19]. The ions of Ba2+, Ca2+, Mg2+, and Zn2+ all could activate and increase the activity of α-acetolactate decarboxylase (ALDC) from B. subtilis 168 and Lactococcus lactis DX [20, 21]. Mn2+ had a slightly active effect (< 10%) on ALDC from L. lactis DX but inhibited and decreased the activity of ALDC from B. subtilis 168 by almost 30%. The third key enzyme of acetoin pathway is AR/2,3-BD dehydrogenase (BDH) which catalyzes the conversion of acetoin and 2,3-BD. The presence of 3 mmol/L Mn2+ increased the relative activity of BDH from B. subtilis JNA 3–10 by 252.3% while only increased its AR activity by 112.6% [22]. In contrast, Mn2+ inhibited the activity of BDH from Serratia marcescens H30 by 35% but slightly increased its AR activity [23]. In this work, part of the phenomenon was similar to these published works.

To evaluate the production of a target product, the conversion rate and product accumulation are two important elements. The former is the key factor which influences the cost of raw materials, while product concentration would affect the cost of downstream processing. Although H3BO3 demonstrated greatest impact on conversion rate, its negative effect indicated that it should be set at low level. Thus, H3BO3 was excluded from the fermentation medium. The factors of Mn2+, Co2+, and CSLP displayed positive effect on the accumulation of (3R)-acetoin and the conversion of glucose to (3R)-acetoin, in which CSLP was less important according to the P values. The concentration of product could be increased by fed-batch fermentation, so the conversion of glucose to (3R)-acetoin was selected as the response, and Mn2+ and Co2+ were the factors for medium optimization.

Medium optimization using central composite design

A five-run steepest ascent path experiment was carried out (Table 3) to select the center point for CCD experiment. The conversion of glucose to (3R)-acetoin was increased along the path, and highest conversion was obtained from the combination of Mn2+ (0.05 mmol/L) and Co2+ (0.007 mmol/L). This combination was used as the center point.

Table 3 Design and results of the steepest ascent path experiment
Full size table

A 13-run CCD experiment was performed and the results were shown in Table 4. The corresponding second-order model equation could be written as: Y = 0.355 + 4.21X1 + 0.446X2 + 74.1X1X2 − 42.3X12 − 519.5X22, where R2 was 0.9001. Among the terms of the model, the linear coefficient of Co2+ (P = 0.005) was more significant than other factors. Therefore, the concentration of Co2+ displayed a great influence on the glucose conversion, which indicated that a little variation of its concentration would alter glucose conversion. The concentration of Mn2+ was almost significant in the linear and quadratic level (P < 0.05), while the interaction between the variables had no significant influence on conversion.

Table 4 Experimental design and results of central composition design
Full size table

According to the equation, the optimized fermentation medium for (3R)-acetoin production was as follows: glucose 100 g/L, CSLP 2 g/L, HPO42− 0.15 mol/L, NH4+ 0.2 mol/L, K+ 0.1 mol/L, Mg2+ 2.5 mmol/L, SO42− 2.5 mmol/L, Fe2+ 0.01 mmol/L, Zn2+ 0.001 mmol/L, Cu2+ 0.001 mmol/L, Mn2+ 0.053 mmol/L and Co2+ 0.004 mmol/L, and the predicted conversion of glucose to (3R)-acetoin was 0.468 g/g. Under the optimal medium, a conversion of 0.475 g/g was obtained by shaking-flask cultivation, indicating that the experimental and predicted values were in good agreement.

Effect of aeration rate on the production of (3R)-acetoin

To check the effect of aeration rate on the production of (3R)-acetoin, fed-batch fermentation was carried out in a 5-L fermenter. The initial pH of the fermentation medium was 7.3, which was favorable for the fast growth of Bacillus species [2, 8, 10, 22]. The pH was automatically decreased to pH 5.9 due to the production of organic acids, then maintained at pH 5.9 which was beneficial for the conversion of pyruvate to α-acetolactate [2, 24].

Dissolved oxygen is one of the key parameters in the fermentation of acetoin, which influences the cell growth and distribution of metabolites [10, 25]. The level of dissolved oxygen depends on the agitation speed and aeration rate. Agitation is usually the main power consumer of an industrial fermentation process. It was found that large amount of foam was produced when agitation was set at 200 rpm. Lowering the agitation speed could reduce the foam formation. Therefore, a low agitation speed of 170 rpm was applied in this study, while the aeration rate was carefully optimized (Fig. 2). The results showed that aeration rate displayed important impact on cell growth and (3R)-acetoin formation. With the elongation of incubation time, the cell density and (3R)-acetoin accumulation gradually increased. With enhanced oxygen supply, cell density was also increased, just like most of Bacillus strains [8, 16, 25]. The highest cell density was obtained under the aeration rate of 0.25 vvm. However, the stationary phase at 0.25 vvm was largely shortened compared with those at 0.15 vvm and 0.2 vvm, and the cell growth entered decline phase rapidly. After 80-h culture at 0.25 vvm, the pH value in fermentation broth began to increase, so the fermentation process was stopped.

Fig. 2
figure 2

(3R)-Acetoin production via fed-batch fermentation under different aeration rates. a 0.15 vvm; b 0.2 vvm; c 0.25 vvm. Filled square: OD620; open square: residual glucose; filled circle: (3R)-acetoin; open circle: (3S)-acetoin; filled triangle: meso-2,3-butanediol; ☆ dissolved oxygen

Full size image

When the environment was switched from aerobic to microaerobic and anaerobic, the expression of genes involved in acetoin pathway can be activated under low oxygen tension [26, 27]. Therefore, rapid accumulation of (3R)-acetoin was observed after 10-h culture where dissolved oxygen was decreased to around 0.1 ppm. Under higher oxygen supply, more carbon flux was distributed to biomass which resulted in lower acetoin production and glucose conversion [8, 25], while lower oxygen supply was not favorable for cell growth. In this study, the process under an aeration rate of 0.2 vvm led to better glucose conversion to (3R)-acetoin (0.448 g/g), and higher (3R)-acetoin production (83.7 g/L). The (3R)-acetoin accumulation under the aeration rate of 0.15 vvm and 0.25 vvm was 78.1 g/L and 69.8 g/L, respectively, and glucose conversion was 0.436 g/g and 0.403 g/g glucose, respectively.

During the fermentation, (3S)-acetoin was accumulated gradually. However, its concentration was less than 1 g/L during the fermentation process. The optical purity of (3R)-acetoin surpassed 99% in the fermentation broth, indicating the stable property in high stereoisomeric selectivity of the strain. This is important for practical application, for some strains showed decreased optical purity of (3R)-acetoin by fed-batch fermentation compared with shaking-flask cultivation [28]. Under an aeration rate of 0.2 vvm (3R)-acetoin with an optical purity of 99.4% was obtained at the end of fermentation.

In most of the natural microorganisms, acetoin is not the end-product, which could be converted to 2,3-BD by AR. When Bacillus strains were used to produce acetoin, a titer of 2,3-BD was usually detected in the midterm of fermentation which was finally converted to acetoin by BDH [9, 25]. In this study, the production of 2,3-BD was gradually increased with the elongation of incubation and kept at low concentration (Fig. 2). At the end of the fermentation, the concentration of meso-2,3-BD obtained at 0.15 vvm, 0.2 vvm and 0.25 vvm was 6.97 g/L, 6.44 g/L and 4.64 g/L, respectively. This result indicates the low activity of AR/BDH in strain CGMCC 13141, which was also found in another marine strain of B. subtilis DL01 [29].

Compared with other Bacillus strains which were cultured at 1 vvm and 350 ~ 600 rpm [9, 10, 30, 31], the strain CGMCC 13141 showed a low oxygen requirement. Such an operation condition was beneficial for the investment of equipment and energy consumption in industrialization. As shown in Table 5, fermentative production of (3R)-acetoin is more cost-effective in industrial-level production because sugar is much cheaper than 2,3-BD. Compared with other strains, strain CGMCC 13141 is competitive for industrial production of (3R)-acetoin. Although the concentration and productivity in this study were a little lower than those of the highest production by an engineered strain of Corynebacterium glutamicum [28], the yield and optical purity were much higher which were important factors for the cost and stereoisomeric selectivity of (3R)-acetoin. Moreover, the microorganism is a native strain and belongs to GRAS species. Therefore, the strain in this study showed excellent characteristics in (3R)-formation, stereoisomeric selectivity, glucose conversion, and oxygen requirement, which indicated the potential in industrial production.

Table 5 Comparison of biological production of (3R)-acetoin
Full size table

Conclusions

Bacillus subtilis CGMCC 13141 was isolated for the production of (3R)-acetoin by fermentation. Plackett–Burman design experiment showed that Mn2+, Co2+, and CSLP were the factors with great impacts on (3R)-acetoin accumulation and glucose conversion. By medium optimization using central composition design, the optimal concentration of Mn2+ and Co2+ was 0.053 mmol/L and 0.004 mmol/L, respectively. Fed-batch fermentation was carried out under different aeration rate, and a titer of 83.7 g/L (3R)-acetoin with an optical purity of 99.4% was achieved after 82-h culture under an aeration rate of 0.2 vvm. The conversion yield of glucose to (3R)-acetoin was 0.448 g/g, 91.5% of the theoretical value.