High Cell Density Cultivation of A Recombinant Bacillus Subtilis for Nattokinase Production

Background: Nattokinase (NK), a brinolytic enzyme, can be produced by culturing recombinant Bacillus subtilis in Luria-Bertani broth in a shaking ask. For use as a nutraceutical, however, a large-scale preparation and a simple purication process are required. Results: The present study utilized a fed-batch process to cultivate a B. subtilis strain carrying a pHT01 plasmid with an NK-encoding gene (B. subtilis/pHT01-aprN1). For batch A (FB A), with a pH-stat two-stage fermentation strategy, we achieved an activity of 2910.5 ± 21.6 U mL -1 and a specic activity of 30.32 U ml -1 OD 600-1 . Then, we changed the strategy with a later induction and lower feeding rate to pursue higher cell density and thus higher enzyme activity, a 11.9-fold activity of 4521.8 ± 23.8 U mL -1 was acquired, however, the specic activity was lower than FB A. For the third batch, low-glycerol-level-maintain feeding strategy was followed, and nally, a NK activity of 7778 ±17.28 U mL -1 was obtained, according to our knowledge, it was the highest activity assayed by the brin plate method ever reported. Furthermore, fermentation supernatant was successively puried by ammonium sulfate precipitation and nickel column anity chromatography with a total NK recovery rate of 65.2%. Conclusions: Our results indicate that there is a balance between the cell growth rate and NK expression when recombinant Bacillus subtilis is cultured with a fed-batch process. The equilibrium state can be attained by optimizing the induction and feeding strategy, and thus a high cell density and enzyme activity can be achieved.


Introduction
Thrombosis, which is responsible for high morbidity and mortality in humans [1], can be effectively treated by thrombolytic drugs, but these are associated with adverse effects [2,3]. Therefore, it is necessary to develop new biological substances, especially prophylactic food-source thrombolytic agents with low immunogenicity and preventative, long-term effects that are convenient for oral administration and stable in the gastrointestinal tract. Nattokinase (NK), which decreases the ability of blood to clot, is traditionally taken from natto, a Japanese solid-state fermented soybean food [4,5]. Currently, NK is used as a dietary supplement as well as a prophylactic or a curative thrombosis medicine [5,6]; however, the process used to purify it from natto is complicated and accompanied by signi cant loss of bioactivity. Accordingly, fed-batch fermentation by adopting genetically modi ed bacteria may improve enzyme yield [7][8][9][10].
Several reports have shown that fed-batch culture led to increases in NK activity of 2.1-25-fold relative to batch culture and that the addition of glycerol during the cell growth phase increased NK production signi cantly [10][11][12][13][14][15][16]. Moreover, various protein puri cation methods have been utilized for NK puri cation [6,16]. Taken together, these ndings suggest that the expression of NK by genetically modi ed bacteria may reach a much higher level if e cient protocols for high cell density fermentation and subsequent puri cation are obtained.
We previously constructed a B. subtilis 168 strain containing a pHT01-aprN plasmid [17]. In this study, we further investigated the expression of NK by culturing this strain under different induction and feeding strategies. In addition, we investigated the e ciency of puri cation with ammonium sulfate precipitation and Ni-NTA a nity chromatography.

Results And Discussion
NK expression by ask culture and fed-batch experiment At least ve different NK activity assay methods have been reported to date [6,11,12,18,19]. Among these, the brin plate method and chromogenic method are the most widely used, but the caseindegradation and JBSL (Japan Bio Science Laboratory Co., Ltd) methods may show much higher values of enzyme activity [11,12,14].
The NK activity obtained at 24 h in flask culture was 380.14 ± 5.71 U mL -1 , while the OD 600 reached 13.45 Many studies have demonstrated the signi cant contribution of media ingredients [20,21] and nutrient feeding strategy to NK production [6,15]. Berenjian con rmed that glycerol was a noteworthy carbon source in uencing cell density during the fermentation of B. subtilis natto, and that the highest activity of NK was obtained by adding 3% glycerol as a carbon source [15]. It has also been reported that 2,6pyridine dicarboxylic acid (PDCA) and metal ions such as Ca 2+ and Mg 2+ could improve osmotic pressure and help to maintain enzyme conformation, thereby improving the NK activity [17,22]. Moreover, Wang demonstrated that glucose, K 2 HPO 4· 3H 2 O and MgSO 4 ·7H 2 O played key roles in the production of NK, and they obtained an activity of 12.34 FU mL -1 [23]. Taken together, the results of these studies showed that NK activity could be improved dozens of times by media optimization.
In addition to media, feeding solutions are critical factors that in uence NK activity, which should support cell growth and recombinant protein production while avoiding substrate inhibition and other related problems [12,15].
Based on this information, we selected a mixture of glycerol, yeast extract, PDCA and a concentrated inorganic mixture solution as the fermentation broth, and a mixture of glycerol and yeast extract as the feed broth.
The strategy of induction, including cell density at the time of induction, inducer concentration, preinduction growth and post-induction incubation time, can also affect the e ciency of protein expression.
The aim of this study was to investigate the effects of using pH-stat and low-glycerol-level-maintaining strategies on NK expression by B. subtilis 168/pHT01-aprN1.
Three experiments were performed to examine the effects of induction time, feeding time and feeding rate on the NK activity of fermentation broth. The results are presented in Figure 1 and Table 1.
Two-stage fermentation strategy has been used for production of recombinant protein and other metabolites [24][25][26][27]. For FB A, we proposed a two-stage fermentation strategy and expected to get a high production. In the rst stage (cell growth stage), we fed 600 mL of media into the fermenter with a high ow rate at the eighth hour. Because the nutrition was not su cient for cell growth, the glycerol concentration decreasing rapidly to about 70 mmol L -1 at 12 h and the OD 600 not varying markedly after 13 h. However, the NK activity still increased signi cantly until 18 h. The nal NK activity was 2910.5 ± 21.6 U mL -1 and the speci c activity was 30.32 U ml -1 OD 600 -1 .
Although we achieved a 7.7-fold activity compare to ask culture, the cell density and the total activity were not as high as reported previously [6,12]. It suggested that later induction and lower feeding rate maybe to the bene t of higher cell density and thus higher enzyme activity [28,29]. According to this hypothesis, for FB B, the expression was induced at the fth hour, which was 1 h later than FB A, and the induced OD 600 was up to 27.3 ± 1.0, which was higher than that of FB A (17.6 ± 0.4). A three-stage feeding strategy was used, and the feeding of FB B started at 14 h, when the glycerol concentration was as low as 119.8 ± 1.3 mmol L -1 , which might have favored cell growth by reducing substrate inhibition.
The glycerol content was su cient to support the cell growth for 24 h. Although the OD 600 values from 15 h to 19 h were not varied signi cantly due to dilution and a new medium environment, it nally reached a high value of 208.8 ± 1.9 at 24 h, with a 11.9-fold activity of 4521.8 ± 23.8 U mL -1 . However, the speci c activity was 21.66 U ml -1 OD 600 -1 at 24 h, which was lower than that of FB A, and the NK activity did not synchronously increase following cell growth during the late fermentation stage, which implied that there should be a balance between the cell growth rate and enzyme expression.
The results of FB A and FB B suggested that maybe we should not unilaterally pursue high cell density, and instead the expression should be induced at an earlier time and kept a low glycerol content during the feeding period. Consequently, for FB C, continuous feeding was adapted, starting at 10 h when the OD 600 had reached 105.3 ± 1.1 and almost two-thirds of the initial glycerol had been consumed. The glycerol concentration was controlled to around 50 mmol L -1 by adjusting the feeding rate.
As expected, an activity of 7778.0 ± 17.3 U mL -1 and a speci c activity of 44.86 U ml -1 OD 600 -1 was achieved, and these values were 1.7-, 2.6-, and 26-fold higher than those of batch B, batch A and the flask culture, respectively. According to our knowledge, this was the highest ever reported activity value by the brin plate method [9,10,12,15].

Recovery rate of NK by puri cation
The recovery rate of NK obtained by ammonium sulfate precipitation was 89.1% (Table 2), which was consistent with a study by Garg [16]. The high recovery rate and simple operation showed that this method could be used to purify NK on a large scale.

Page 5/12
The puri cation was followed by Ni-NTA a nity chromatography, and the total recovery rate was 65.2% ( Table 2). The high imidazole concentration in wash buffer B led to a high loss of NK when samples were puri ed using Ni-NTA a nity chromatography. Accordingly, the imidazole concentrations were adjusted to zero in wash buffer A and incubated supernatant, and to 10 mmol L -1 in wash buffer B; thus, a recovery rate of 88% was obtained in this step. Different imidazole concentrations in elution buffer C were also investigated, but there were no signi cant differences in the range of 100 mmol L -1 to 500 mmol L -1 .
According to the NK expression level and the recovery rate, the process developed here may be applied for large scale production of NK.

SDS-PAGE and Western blotting analysis of NK
SDS-PAGE analysis demonstrated that a 28 kDa protein was a crucial component in the supernatant from induced B. subtilis 168/pHT01-aprN1, but that it was not present in the supernatant from B. subtilis 168/pHT01 and non-induced B. subtilis 168/pHT01-aprN1 fermentation broth ( Fig. 2 A/B), suggesting that recombinant NK could be expressed in a soluble form. Western blotting with a His-tag-speci c monoclonal antibody also showed a speci c signal at 28 kDa, whereas no cross-reaction occurred in the total soluble proteins from non-induced B. subtilis 168/pHT01-aprN1 broth, which con rmed that the 28 kDa protein was the recombinant NK, as expected (Fig. 2 C).
In view of these reports, NK was produced by fed-batch cultures of recombinant B. subtilis, and its production was improved to 7,778 U mL -1 from 380 U mL -1 of ask culture using pH-stat and low-glycerollevel strategies. Future studies will design a process and set a kinetic model for fermentation optimization.

Shake ask experiments
To prepare a stock culture, one colony of B. subtilis 168/pHT01-aprN1 was inoculated in 1 mL Luria-Bertani (LB) medium and then incubated overnight at 37°C in a shaking incubator at 200 revolutions per minute (rpm). A ask with 25 mL fed-batch culture media was then inoculated with 7.5% (v/v) of stock culture and incubated at 37°C until reaching mid-log phase (OD 600 of 0.8-1.0). Next, the temperature was adjusted to 28°C and held for 20 min, after which a 25-μL aliquot of 0.5 mmol L -1 isopropyl β-D-1thiogalactopyranoside (IPTG) was added and the culture was further incubated for 45 h at 28°C in a shaking incubator (200 rpm) for NK expression.

Fed-batch experiments
An in-situ sterilizable 15 L bioreactor (BIOSTAT ® B, Germany) equipped with pH and pO 2 probes was used for fed-batch experiments. One milliliter of glycerol stock bacteria was inoculated into 100 mL LB medium and cultured for 12 h at 37°C in a shaking incubator (200 rpm), after which the pre-culture solution was aseptically inoculated into 1 L LB medium and incubated under the same conditions for 12 h. Next, 700 ml of seed cultures were inoculated for 7-L fed-batch fermentation in the bioreactor. The value of oxygen dissolved was set at 20%, which was cascade controlled by agitation speed (300-1000 rpm) and air ow rate (3-20 L min -1 ). The pH was set as 7.0 and automatically controlled by adding 2 mol L -1 HCl or 2 mol L -1 NaOH.
In the rst fed-batch experiment (FB A), the temperature was adjusted to 28°C at 3.5 h, while IPTG was added 30 min later as the OD 600 value reached about 18. A total of 600 mL feed media was supplemented at 8 h with a ow rate of 30 mL min -1 . In the second fed-batch experiment (FB B), the expression was induced at 5 h when the OD 600 value had reached 27, and the feed media was constantly fed after 14 h of cultivation with a ow rate of 6 mL min -1 for 1 h, then 3.6 mL min -1 for 2 h, and nally 2.4 mL min -1 for another hour. In the third experiment (FB C), the expression was induced at 3 h when the OD 600 reached about 9 and the feed media was fed at 10 h. The feeding rate depended on the content of glycerol, which was controlled at a concentration of 50 mmol L -1 . Cell growth was monitored at various times by measuring the OD 600 values.

NK activity assay
Quantitative analysis of NK activity was conducted by the brin plate method, with slight modi cation [18]. bovine brinogen solution and 12 g L -1 agarose solution was then warmed in a 45°C bath, after which 10 μL of 500 U thrombin was added to 15 mL of the mixture solution in a 90 mm petri dish and kept at room temperature for 1 h to form brin. Holes with a 2 mm diameter were made on the brin plate and 10 µL of each diluted supernatant of fermentation broth was then added. The plates were subsequently incubated at 25°C for 16 h, after which the areas of the lysis zones on the brin plates were measured and the brinolytic activities were determined according to the standard curve of urokinase.

Glycerol concentration measurement
The concentration of glycerol was determined enzymatically using a free glycerol assay kit (Sigma, St. Louis, MO, USA) according to the manufacturer's procedures.
Puri cation and lyophilization of NK  Figure 1 B. subtilis 168/pHT01-aprN1 fed-batch fermentation. Panels A, B and C show the fed-batch fermentation pro les from FB A, FB B and FB C: bacterial growth (empty squares), enzyme activity (empty triangles) and glycerol concentration ( lled circles), respectively. Induction and feeding times are indicated by small and large arrows, respectively.