Efficient Whole-Cell Biocatalyst for Acetoin Production with NAD+ Regeneration System through Homologous Co-Expression of 2,3-Butanediol Dehydrogenase and NADH Oxidase in Engineered Bacillus subtilis

Acetoin (3-hydroxy-2-butanone), an extensively-used food spice and bio-based platform chemical, is usually produced by chemical synthesis methods. With increasingly requirement of food security and environmental protection, bio-fermentation of acetoin by microorganisms has a great promising market. However, through metabolic engineering strategies, the mixed acid-butanediol fermentation metabolizes a certain portion of substrate to the by-products of organic acids such as lactic acid and acetic acid, which causes energy cost and increases the difficulty of product purification in downstream processes. In this work, due to the high efficiency of enzymatic reaction and excellent selectivity, a strategy for efficiently converting 2,3-butandiol to acetoin using whole-cell biocatalyst by engineered Bacillus subtilis is proposed. In this process, NAD+ plays a significant role on 2,3-butanediol and acetoin distribution, so the NADH oxidase and 2,3-butanediol dehydrogenase both from B. subtilis are co-expressed in B. subtilis 168 to construct an NAD+ regeneration system, which forces dramatic decrease of the intracellular NADH concentration (1.6 fold) and NADH/NAD+ ratio (2.2 fold). By optimization of the enzymatic reaction and applying repeated batch conversion, the whole-cell biocatalyst efficiently produced 91.8 g/L acetoin with a productivity of 2.30 g/(L·h), which was the highest record ever reported by biocatalysis. This work indicated that manipulation of the intracellular cofactor levels was more effective than the strategy of enhancing enzyme activity, and the bioprocess for NAD+ regeneration may also be a useful way for improving the productivity of NAD+-dependent chemistry-based products.


Introduction
Acetoin (3-hydroxy-2-butanone, AC) is an extensively-used spice that naturally exists in corn, grape, cocoa, apple, butter, coffee, etc. Widely used in food and beverage industry, AC also serves as a platform compound in many other industries [1]. It is one of the 30 platform chemicals that are given priority to their development and utilization by the US Department of Energy [2]. Although there are many chemical synthetic methods for AC preparation [3], its market is limited by the disadvantages of traditional chemical synthesis. On the other hand, with the further development of green chemical technology and the constant improvement of the environmental protection consciousness, non-toxic and non-pollution biological technology inevitably become the main direction of industrial development and consumers prefer security natural products even though they are generally more expensive than the corresponding chemical compounds.
Nowadays, a lot of efforts have been made to develop natural AC production using fermentative [4], enzymatic [5] or biocatalytic technologies [6]. A number of bacteria have abilities to produce AC, including the genera Klebsiella, Paenibacillus, Gluconobacter, Bacillus, Serratia, etc. [7][8][9][10][11]. However, in most of these species, AC just plays a part as by-product of 2,3-butanediol (2,3-BD), which is another important bio-based platform chemical [12]. Bacillus species, which can produce various of industrial products [13], have been proved with AC as its major fermentation product under specific conditions [14]. Many efforts have been made to improve the production of AC from Bacillus strains. Liu et al. isolated a B. licheniformis strain that could produce 41.3 g/L of AC [4]. Zhang et al. isolated the B. subtilis JNA-3-10 and produced 42.2 g/L of AC [15]. Fermentation optimization strategies have been used to improve AC production, such as optimizing the medium components [16], controlling the level of dissolved oxygen and controlling the fermentation pH [17]. Metabolic engineering strategies were also applied to improve AC production through modifying metabolic branchpoints in the network [14,18,19]. However, so much long fermentation duration lead to a low AC productivity. To our knowledge, the highest productivity of AC by Bacillus strains is just 1.42 g/(L?h) [4]. In addition, the mixed acid-butanediol fermentation of Bacillus strains will metabolize a certain portion of sugars to the by-products of organic acids such as lactic acid and acetic acid, which causes energy cost and increases the difficulty of product purification in downstream processes [20].
Recently, the introduction of NAD + regeneration system could dramatically improve AC production and decrease the yield of NADH-dependent by-products [6,11]. Sun et al. obtained 75.2 g/ L AC with a productivity of 1.88 g/(L?h) by S. marcescens H32 with over-expression of a water-forming NADH oxidase [11]. Xiao et al. developed a co-expression system with 2,3-butanediol dehydrogenase and NADH oxidase in E. coli produced AC at a high productivity of 3.06 g/(L?h) [6]. Although have relative high AC productivities, AC yield of this biocatalyst was still far away from the highest report of 89.2 g/L achieved by Wang et al. using fermentation method with 2,3-BD as substrate by Gluconobacter oxydans DSM 2003 [10]. Therefore, combining both advantages of fermentation and cofactor regeneration, a potential strategy of introducing a biocatalytic process with NAD + regeneration system for efficient natural AC production in B. subtilis is proposed by us. Whole-cell biocatalyst has been intensively explored for the production of valuable compounds because excellent selectivity and NAD + reserves that provides a continuous source of cofactors [21]. Trough NAD + regeneration system, the cellular cofactor level, redox state and the corresponding enzymatic activity are expected to have major effects on the performance of the wholecell biocatalysts. In this whole-cell biocatalyst, 2,3-BD is used as substrate and only AC can be obtained in the short biocatalyst period. This is also a good solution to develop derivative process for industrially produced 2,3-BD utilization, which can not be commercially utilized so far.
In previous work of our lab, when B. subtilis was fermented with glucose as substrate, 2,3-BD was the major product in prophase of fermentation, whereas in the anaphase of fermentation, 2,3-BD was reversely transformed to AC [15]. The reversible conversion between AC and 2,3-BD was responsible by the enzyme AC reductase/2,3-BD dehydrogenase (AR/BDH EC 1.1.1.4). Generally, AR/BDHs have the similar optimum pH-values for oxidation and reduction [22][23][24]. However, very special property of this enzyme that has different optimum pH-values for oxidation and reduction was proved [25,26]. Intrigued by this point, our group experimentally demonstrated that AR/BDH from B. subtilis also has very different optimum pH-values [27]. The results indicated that this enzyme preferentially catalyzes the reduction/oxidation reaction in the acidic/alkaline condition. Similar to other reductases and dehydrogenases, AR/BDH requires NAD + and NADH as cofactors [28,29]. As one pair of key cofactors, NADH and NAD + play a critical role in over 300 biochemical reactions including oxidation and reduction [30,31], which helps to maintain the elementary requirement for metabolism and growth in cells [32]. During the catalytic process of BDH, NAD + is reduced to NADH accompanied by the oxidization of 2,3-BD [6]. Therefore, an driving force performed by NADH oxidase (NOX EC 1.6.99.3) was required for sweeping away the obstruction proposed on NAD + regeneration. As a flavoprotein, NOX uses oxygen to produce either water or hydrogen peroxide [33], which can partially or completely inhibit cell growth. In this work, the homologous NOX from B. subtilis, similar to the NOX from Lactobacillus brevis [33], generates NAD + from NADH by reducing O 2 to H 2 O was used as the core enzyme for NAD + regeneration [34], because construction of a homologous NAD + regeneration system can be more efficient and safe for AC production. The constructed biocatalyst of B. subtilis, in which AR/BDH and NOX were co-expressed (shown in Figure 1), adjusted the intracellular NADH/NAD + ratio and NAD(H) level and strongly pumped the catalytic reaction. After optimization of the biocatalytic conditions, AC production and productivity was highly improved.
B. subtilis and E. coli were cultivated in Luria-Broth (LB) medium routinely and were used as the cloning and expression hosts, respectively. When necessary, antibiotics (100 mg/mL ampicillin or 50 mg/mL kanamycin) were supplemented to the medium to maintain the plasmids.

Strains, plasmids and primers
The strains, plasmids and primers used in this study were given in Table 1.

Construction of recombinant plasmids
The bdhA and yodC genes were amplified using forward primer P1/P3 and reverse primer P2/P4 and B. subtilis 168 as template. The plasmid pMA5-bdhA/pMA5-yodC were constructed by inserting bdhA/yodC between the BamH I and Mlu I sites of pMA5. The yodC gene with with the upstream region including HpaII promoter and RBS was then amplified using forward primer P5 and reverse primer P6 and pMA5-yodC as template. The plasmid pMA5-bdhA-yodC was constructed by inserting HpaII-yodC between the Sph I and Hind III sites of pMA5-bdhA.

Transformation and selection of recombinant strains
The ligated DNAs were used to transform E. coli JM109. Positive colonies were selected on agar plates containing ampicillin, and the plasmids were confirmed using restriction enzyme analysis and DNA sequencing. The recombinant plasmids were then used to transform B. subtilis 168 [35]. The transformants were screened for their ability to grow on LB agar plates containing kanamycin.

Assays of AR/BDH and NOX activities and SDS-PAGE analysis
The cells were harvested by centrifugation at 10000 rpm for 10 min at 4uC and washed twice with 50 mM sodium phosphate buffer (pH 6.5). Then the harvested cells were resuspended in  50 mM sodium phosphate buffer (pH 6.5) containing 0.1 mM PMSF, and 1 mM DTT. After ultrasonic disruption, cell debris was removed by centrifugation at 15000 rpm at 4uC for 30 min to obtain a crude enzyme solution for the enzyme assay. The AR/BDH and NOX activities were determined spectrophotometrically by measuring the change of absorbance at 340 nm and 25uC corresponding to the oxidation of NADH (e 340 = 6220/M?cm) or the reduction of NAD + . AR activity was determined using 50 mM sodium phosphate buffer (pH 6.5) containing 50 mM acetoin and 5 mM NAD + . BDH activity was determined using 50 mM glycine-NaOH buffer (pH 8.5) containing 100 mM 2,3-butanediol and 5 mM NADH. One unit of activity (U) corresponds to 1 mmol NAD(H) formed per minute [23].
NOX activity was determined using 50 mM sodium phosphate (pH 7.0) containing 0.3 mM EDTA, 50 mM FAD and 0.3 mM NADH. One unit of activity (U) corresponds to 1 mmol NAD + formed per minute [36].
The enzymes were assayed by sodium dodecyl sulfatepolyacrylamide gel electrophoresis (SDS-PAGE) according to Laemmli method [37]. AR/BDH and NOX were also detected by Western blotting by using a mouse monoclonal anti-His 6 antibody. The extracts were loaded in a 12% SDS-PAGE gel that was blotted onto a polyvinylidene difluoride (PVDF) membrane (Millipore) and treated with the primary anti-His 6 antibody. To visualize the relevant bands, the blot was treated with a secondary goat anti-mouse antibody (Bio-Rad) conjugated to horseradish peroxidase (HRP), and the bands were detected by chemiluminescence with luminol and peroxide with the aid of a Bio-Rad Chemidoc XRS. The molecular weight of the subunit of AR/ BDH and NOX were determined by comparing the relative mobility of perfect protein Marker 14.3-97.2 kDa (Takara). The protein concentration was determined by Bradford method [38] using BSA as the standard protein.    The whole-cell biocatalytic activity was assayed by measuring the formation of AC, and one unit was defined as 1.0 mM AC produced per hour per OD 600 of culture at 37uC.

Determination of NADH and NAD + concentrations
The intracellular concentrations of NADH and NAD + were determined using the Amplite Fluorimetric NAD/NADH Ratio Assay Kit (15263) from AAT Bioquest (Sunnyvale, CA, USA) according to the manufacturers' instructions. The assay kit provides a convenient method for sensitive detection of NAD + , NADH and their ratio. The signal can be easily read by either a fluorescence microplate reader at Ex/Em = 530-570/590-600 nm (maximum Ex/Em = 540/590 nm) or an absorbance microplate reader at ,576 nm [39].

Analysis methods
The bio-profiles were monitored periodically over the experiment period. Samples were centrifuged at 10000 rpm for 10 min, and the supernatant was used for further analysis. Concentrations of AC and 2,3-BD were monitored by gas chromatography (Jie Dao GC1600, China, FID detector, N 2 flow rate of 50 ml/min, detector temperature of 250uC, and a column temperature of 160uC). Biomass was measured at OD 600 (UNICO UV-2000) of the culture broth after appropriate dilution with water. All assays were performed by triplicate cultures. The samples were determined at least twice for three extracts derived from independent cultures, and standard deviations of the biological replicates were represented by error bars.

Construction of the whole-cell biocatalyst
To construct the whole-cell biocatalyst, the following strains of B. subtilis 168, B. subtilis 168/pMA5, B. subtilis 168/pMA5-bdhA, B. subtilis 168/pMA5-yodC and B. subtilis 168/pMA5-bdhA-yodC were detected to their enzyme activities and whole-cell biocatalytic activities. The results were shown in Table 2. B. subtilis 168/pMA5-bdhA and B. subtilis 168/pMA5-yodC showed the BDH and NOX activity of 124.5 mU/mg and 570.3 mU/mg, respectively. As expected, the co-expression system of B. subtilis 168/pMA5-bdhA-yodC showed a BDH activity of 152.8 mU/mg and a NOX activity of 346.4 mU/mg. SDS-PAGE analyze of the recombinant protein expression was shown in Figure 2. AR/BDH and NOX have the specific molecular weights of 37.3 kDa and 25.7 kDa, respectively. The results of SDS-PAGE and enzyme activity assay indicated that bdhA were successfully co-expressed with yodC.
In comparison with free enzyme activities, the whole-cell biocatalytic activities prepared from the recombinants were defined as the formation of 1.0 mmol AC per hour per OD 600 of culture at 37uC, which were much more direct to prove the biocatalytic abilities. As shown in Table 1, B. subtilis 168 could not effectively catalyze 2,3-BD to AC because of low AR/BDH activity and the restriction of NAD + pool. By over-expression of AR/BDH or NOX, the whole-cell biocatalytic activities of B. subtilis 168/pMA5-bdhA and B. subtilis 168/pMA5-yodC were both increased, but no more than 30 % compared to B. subtilis 168. However, the whole-cell biocatalytic activity of B. subtilis 168/pMA5-bdhA-yodC was nearly twice that of B. subtilis 168, indicating the NAD + regeneration system worked successfully.

Effects of intracellular NADH and NAD + concentrations on AC biosynthesis
The recombinant strains and the parent strain were cultured in the fermentation medium. Homologous over-expression of NOX in B. subtilis was expected to regenerate the overall intracellular NAD + pool for continuously converting AC to 2,3-BD. As shown in Figure 3A, B. subtilis 168/pMA5-bdhA-yodC achieved the highest conversion rate of 2,3-BD to AC comparing to the other biocatalysts. Both B. subtilis 168/pMA5-yodC and B. subtilis 168/ pMA5-bdhA-yodC could maintain efficient productivities to catalyze 2,3-BD to AC, indicating NAD + was regenerated by NOX.
The significant role of cofactors played in these biocatalysts was further proved by comparing the intracellular concentrations of NAD + and NADH in the recombinants, in which the overall amount of intracellular NAD + and NADH were almost constant within 6 hours. The intracellular NAD + pools in recombinant B. subtilis 168/pMA5-bdhA-yodC and B. subtilis 168/pMA5-yodC were greatly improved by over-expression of NOX ( Figure 3B and 3C), and NAD + was continuously regenerated to keep a persistent AC productivity. Although over-expression of AR/BDH by B. subtilis 168/pMA5-bdhA effectively converted 2,3-BD to AC in the first 2 hours, the reduced NAD + pool then restricted its process for the bioconversion.
By comparing of the intracellular NAD + and NADH levels and the ratio of NADH to NAD + (Figure 3D), NAD + was proved be regenerated by over-expression of NOX. After 6 h, the intracellular NAD + concentration of B. subtilis 168/pMA5-bdhA-yodC  increased to 2.75 mmol/(L?OD 600 ), higher than B. subtilis 168 and B. subtilis 168/pMA5-bdhA. Corresponding to above results, the intracellular NADH concentration of B. subtilis 168/pMA5-bdhA-yodC decreased to 1.29 mmol/(L?OD 600 ), lower than B. subtilis 168 and B. subtilis 168/pMA5-bdhA. Although B. subtilis 168/ pMA5-yodC achieved the highest NOX activity, which resulted in the lowest NADH/NAD + ratio, it did not showed the highest conversion rate from 2,3-BD to AC. The results suggested that a relatively stable dynamic redox balance of low NADH/NAD + ratio as well as a high AR/BDH activity were necessary for efficiently and persistently converting 2,3-BD to AC. So that, the biocatalyst of B. subtilis 168/pMA5-bdhA-yodC, which coexpressed AR/BDH and NOX was chosen for later experiments.

Optimization of the biocatalytic reaction
To achieve higher productivity, the whole-cell biocatalytic reaction was optimized. As an important parameter that often limit enzyme activity and stability in technical applications [40], the effect of pH on the conversion rate of the biocatalyst was investigated. As shown in Figure 4A, AR/BDH had different optimum pH-values of 8.5 and 6.5 for oxidation and reduction, respectively. Meanwhile, the optimum pH value for NOX was 9.0, and this enzyme could maintain 80 % relative activity under pH 8.0-9.0. So the efficiency of the whole-cell biocatalyst could be affected by adjusting the pH-values of the conversion solution. As shown in Figure 4B, reactions with 40.0 g/L of 2,3-BD as substrate in the biocatalyst B. subtilis 168/pMA5-bdhA-yodC were conducted under different pH-values for 6 h. The conversion rate was measured by monitoring the yield of AC. The results indicated that the conversion rate of this biocatalyst was gradually increased by improving the pH levers in the alkaline condition and the highest conversion rate was acquired at pH 8.5. Thus, we use pH 8.5 as the optimum conversion pH-value for the following experiments.
Temperature can also affect the efficiency of the whole-cell catalytic processes, including enzyme activity and cellular maintenance [10]. Thus, we studied the effect of temperature on AC conversion rate using this biocatalyst. As shown in Figure 5A, AR/ BDH and NOX had different optimum temperature. The optimum oxidation temperature for AR/BDH activity was 50uC (similar to the optimum reduction temperature, data was not shown), but the optimum temperature for NOX activity was 35uC. However, both enzymes could maintain 80 % relative activities under the temperature 40-45uC. So the whole-cell bioconversion was assayed under the following temperatures of 20uC, 30uC, 37uC (the best growing temperature), 40uC, 45uC and 50uC, respectively ( Figure 5B). The results indicated that the highest conversion rate of this biocatalyst was obtained at a relatively high temperature 40uC, closed to the best growing temperature. Finally, a temperature of 40uC was chosen for the optimum temperature.
Chemical stimulators such as Mg 2+ , Mn 2+ and Ca 2+ can improve the activity of AR/BDH [27]. However, another enzyme in this biocatalyst, NOX, had been proved that mental ions have neither stimulation effect nor inhibition effect on its activity [34]. Thus, the following chemicals of MgCl 2 , MnCl 2 , CaCl 2 , FeCl 2 and FeCl 3 were studied of their effect on the biocatalyst, and they were added to the conversion solution at the final concentrations of 0.5 mM, 2.5 mM, 5.0 mM, 7.5 mM and 10.0 mM, respectively. Of all the chemicals listed in Figure 6A, 2.5-5.0 mM Mn 2+ showed remarkable stimulation effect on AC conversion rate. Other metal ions just slightly stimulated the convention rate of 2,3-BD to AC. To find the optimum Mn 2+ concentration on the efficiency of this biocatalyst, the final concentration of Mn 2+ was adjusted from 2.0 mM to 8.0 mM ( Figure 6B). The results indicated that 5.0 mM Mn 2+ was favorable for this whole-cell biocatalyst.
Substrate concentration is another important factor in wholecell catalytic processes, which may result in substrate inhibition. The effect of 2,3-BD concentration on the conversion rate using the biocatalyst was tested (Figure 7). The results showed that the biocatalyst achieved the highest conversion rate with 40 g/L 2,3-BD as substrate. While 2,3-BD concentration kept on increasing, the conversion rate could be inhibited by high concentration of substrate [6]. AR/BDH can catalyze the stereospecific oxidation of (2R,3R)-2,3-BD and meso-2,3-BD to (3R)-AC and (3S)-AC, respectively [41]. However, it cannot catalyze the conversion of (2S,3S)-2,3-BD. Thus, while using 40.0 g/L of mixed stereospecific 2,3-BD as substrate, the highest AC yield was 31.5 g/L after 12 h. Although 2,3-BD could not be totally converted into AC, it can be easily separated from AC which has a low boiling point. Therefore, 40.0 g/L of 2,3-BD was chosen as the optimum substrate concentration for this biocatalyst.

Whole-cell biocatalyst using repeated batch strategy under optimized conditions
The whole-cell biocatalyst was conducted under optimized conditions as described above in a 5-L fermentor. As shown in Figure 8A, 31.5 g/L AC was acquired from 40.0 g/L 2,3-BD after 12 h with a productivity 2.62 g/(L?h). No other products were detected during the process. Since AR/BDH was a reversible enzyme, the whole-cell biocatalyst could not completely transform 2,3-BD to AC, high concentration of AC would restrain the conversion. As shown in Figure 8B, The initial 2,3-BD concentration was 40.0 g/L, and 30.2 g/L of 2,3-BD was added at 8 h. However, due to the limitation of high concentration of product, the yield of AC could not be effectively increased.
To achieve a higher product yield, repeated batch strategies could efficiently enhance the concentrations of the aim products. Due to the NAD + regeneration system, high AC yield could be obtained through more than once whole-cell biocatalyst by using the same batch bacterium without any other additional NADH/ NAD + . The initial 2,3-BD concentration was 40.0 g/L, after each conversion cycle of 8 h, the cells were harvested and then added by 40.0 g/L of fresh 2,3-BD solution to continue the enzymatic reaction. As shown in Figure 9, there was a stable and efficient increase on AC yield in the first two batch conversion. However, AC productivity gradually decreased in the third batch. Thus, after total 40 h of bioconversion, 91.8 g/L of AC was produced from 120.0 g/L 2,3-BD with the productivity of 2.30 g/(L?h), which was the highest production by biocatalyst (Table 3). This is the first report of applying NAD + regeneration system to produce acetoin in B. subtilis. By repeated batch strategy, the whole-cell biocatalyst achieves the purpose of efficient and sustainable producing of AC, and finally reaches the highest AC production record by biocatalysis. In addition, compared to the difficulty of the downstream purification processes of mixed acidbutanediol fermentation, this bioprocess is relatively simple and security, and residual 2,3-BD can be easily separated from AC.

Conclusions
A biocatalyst for AC production was successfully constructed by introducing the NAD + regeneration system into B. subtilis, in which AR/BDH and NOX were co-expressed. After optimization    Table 3. Comparing of microbial production of acetoin using different fermentative strains or biocatalysts. of this converting reaction, repeated batch strategy was further applied on this biocatalyst, and 120.0 g/L 2,3-BD was converted into 91.8 g/L AC with the productivity of 2.30 g/(L?h). To our knowledge, this is the highest report of AC production by biocatalyst. However, AR/BDH stability and the inhibitory effect of acetoin/2,3-butanediol to this enzyme should be further studied, and modeling of this enzyme and site directed mutation is now undergoing. This work proposed an efficient approach for nature AC production and microbial-based biofuel utilization of 2,3-BD.