Optimization of nutritional requirements and ammonium feeding strategies for improving vitamin B 12 production by Pseudomonas denitrificans

Statistical experiment design and data analysis were used to establish the major factors in a chemically defined medium and to develop an ammonium control strategy to optimize the specific vitamin B12 production rate (Yp) of Pseudomonas denitrificans. Through Plackett-Burman design, the major factors of glucose, ammonium sulfate and KCl were selected as the significant factors affecting vitamin B12 biosynthesis and these were further optimized by central composite design with response surface methodology. The maximum Yp of 34.2 μg/gDCW/h was obtained in batch cultivation under the estimated optimal initial composition of glucose (93.6 g/l), (NH4)2SO4 (7.93 g/l) and KCl (1.24 g/l). Ammonium control strategies in fed-batch fermentation showed that when ammonium concentration was maintained at 40 mmol/l, the maximum Yp reached 36.0 ± 1.31 μg/gDCW/h, which was 57.2% higher than that of the control (22.9 ± 0.83 μg/gDCW/h). This ammonium control strategy successfully enhanced the industrial production, resulting in a stable high vitamin B12 production of 212.02 ± 3.03 mg/l and Yp of 37.1 μg/gDCW/h.


INTRODUCTION
Two aspects are of critical importance in the successful industrial optimization of industrial vitamin B 12 fermentation processes.Firstly, a full understanding is needed regarding the influence of nutritional composition on microbial fermentation.Secondly, the metabolism and physiological characteristics of the industrial microbial strains must be known in detail.A study on changes in intracellular metabolic flux distributions under different environments, combined with the real-time process parameters, would provide important information for fermentation optimization.Analysis of the correlation between primary metabolism and biosynthesis of the product of interest would also direct rational strain improvement through genetics and modification by molecular biology.
To carry out 13 C-based metabolic flux analysis, a chemically defined medium is usually necessary in order to eliminate interference from unknown carbon sources on the isotopmer balance and metabolic flux ratio calculations.However, the productivity obtained under 13 Cbased chemically defined media is always dramatically lower than that obtained using the complex media commonly used in industrial production.As well, the physiological characteristics of the microorganisms can also dramatically differ when grown under these two diverse conditions.Therefore, an appropriate medium is necessary for the effective 13 C metabolic network analysis in industrial microbial strains.
A clearly defined compositional medium could also provide an effective platform for investigating the effects of various factors on metabolism and productivity.In industrial vitamin B 12 fermentation, the by-products like beet molasses and corn steep liquor are extensively used as carbon and nitrogen sources.However, the uncertain component and different quality of materials frequently results in large fluctuations of industrial fermentation.These have been also reported in fermentations of cephalosporin C (Li et al., 2008b), erythromycin (Warren et al., 2002) and avermectins.Therefore, determining the significant factors seriously affect the industrial productivities and developing an optimal controlling strategy would prove useful for improving these fermentation techniques.
This study presented experiments designed to screen and optimize a chemically defined medium for vitamin B 12 production by P. denitrificans, using Plackett-Burman design and central composite design.The effects of ammonium concentrations on vitamin B 12 fermentation, based on this medium, were also investigated in fed-batch fermentation.An optimal ammonium control strategy was also successfully applied to enhance industrial vitamin B 12 production.

Shake flask culture
Pre-culture was carried out in a 500 ml Erlenmeyer flask containing 100 ml of seed medium inoculated with cells from a fresh slant and cultured at 28°C on a rotary shaker at 260 rpm for 18 h.The seed culture was then transferred to a 500 ml Erlenmeyer flask containing 50 ml fermentation medium with 10% inoculum and incubated at 32°C on a rotary shaker at 260 rpm for 120 h.

L bioreactor fed-batch culture
Fed-batch fermentations were performed in a 15 L turbine-agitated bioreactor (Shanghai Guoqiang Inc., China) with a 10 L working volume.The inoculum was 10% (v/v) of the whole volume of the culture broth.The fermentation temperature and agitation speed were adjusted to 32°C and 400 rpm, respectively.The aeration rate was 0.8 vvm and the pH was controlled automatically at 7.0 to 7.2 with 3 M NaOH.The fed-batch processed combined pulse feeding of ammonium sulfate solution (400 g/l) to control the ammonium concentration.Each condition was tested twice and Wang et al. 10553 the data shown represent the mean values.Previous research found that dissolved oxygen concentration was sustained at a zero level constantly during fermentation because of the high oxygen consumption rate of P. denitrificans and an uptake rate controlling strategy was also applied in this research (Scott, 2003).When the total sugar in the broth dropped to 20 g/l, feeding of glucose solution (500 g/l) began and the residual glucose concentration was maintained at 1.5 to 2.0% throughout the entire fermentation process.

10,000 L bioreactor fed-batch culture
In a 10,000 L fermenter, the first seed culture was prepared in a 150 L seed fermenter containing 100 L sterile medium with inoculums from eight fresh slants.Cultivation was carried out at 28°C for 40 h with an aeration rate at 0.5 vvm and agitation speed at 180 rpm.Then the primary seed culture (70 to 80 L) was transferred into a 1,000 L secondary seed fermenter with 700 L of seed medium.It was cultivated at 28°C with stirrer speed at 130 rpm and aeration rate at 0.4 vvm for 30 h.Fermentation was performed in a 10,000 L fermenter equipped with three 3-bladed propeller impellers, a temperature probe, pH probe (Mettler Toledo) and dissolved oxygen probe (Mettler Toledo).The pH of the fermentation medium was adjusted to 7.0 and 7.2 before inoculation.The second seed (about 5% of the final working volume) was transferred into 10,000 L fermenter with 7,500 L working volume.The fermentation process was carried out at 32°C for 166 h.The glucose feeding strategies were similar to that mentioned earlier in the 15 L bioreactor.

Determination of dry cell weight (DCW) and vitamin B 12
Samples were centrifuged at 8000 g for 10 min and washed twice with distilled water; the cell precipitate was dried to a constant weight at 105°C.The vitamin B 12 concentration was determined by HPLC.Broth samples (25 ml), to which 2.5 ml of 8% NaNO 2 and 2.5 ml of glacial acetic acid were added, were boiled for 30 min, then the upper aqueous phase was measured using a HPLC system equipped with a C 18 column, (4.6 mm, 25 cm and 5 μm) with an UV detector (254 nm).85% of 0.25 M sodium acetate anhydrous (pH was adjusted to 3.6 with acetic acid) and 15% of methanol in water was used as the mobile phase with a flow of 1.0 ml/min at 40°C (Kamarthapu et al., 2008).

Determination of ammonium and δ-aminolevulinic acid (δ-ALA) concentrations
The concentrations of ammonium-nitrogen (NH 4 + -N) in broths were quantified by the method described by Kanda, which uses Ophenylphenol as a substitute for liquid phenol (Hofmann et al., 2008).δ-ALA concentration in the fermentation broth was determined according to the method reported by Choi et al. ( 1999).

Experimental design and data analysis
For selection of significant factors for vitamin B 12 production, eight independent variables, including glucose, KCl, Na 2 HPO 4 , MgSO 4 , betaine, (NH 4 ) 2 HPO 4 , (NH 4 ) 2 SO 4 and TES, were tested and identified via the Plackett-Burman design (Christensen and Nielsen, 2000).All trials were carried out in triplicate and the average vitamin B 12 yields were treated as responses.The main effect of each variable was simply calculated as the difference Table 1.Plackett-Burman design matrix for eight variables with the observed vitamin B 12 production.4) 0( 5) 0( 7 between the average of measurements made at the high setting (+1) and low setting (-1) for that factor.The experimental design is shown in Table 1.The Plackett-Burman experimental design is based on the first order model (Equation 1):

Run
Where, Y is the response (vitamin B 12 yield), β 0 is the models intercept, β i is the linear coefficient and X i is the level of the independent variable.This model does not describe the interactions among factors; it is used to screen and evaluate the important factors that influence the response.The significance of each variable was determined via Student's t-test using SAS software.
The next step in the formulation of the medium was to determine the optimum levels of significant variables for vitamin B 12 production.For this purpose, the response surface methodology (RSM) was used for the augmentation of total vitamin B 12 production (Li et al., 2008a).The significant variables utilized were as follows: glucose, KCl and (NH 4 ) 2 SO 4 , each of which was assessed at five coded levels (-1.68, -1, 0, +1 and +1.68) and a total of 20 experiments were carried out, as shown in Table 3.All variables were taken at a central coded value, which was designated as zero.The minimum and maximum ranges of the variables were used and the full experimental plan with regard to their values in actual and coded form is provided in Table 3.The statistical software package Design-Expert-7.0 was used to generate the model to predict the effect of combined variables on vitamin B 12 production.For a three-variable system, the responses can be predicted by the following second-order regression equation: Where, Y is the dependent variable (vitamin B 12 yield or specific vitamin B 12 production rate); X 1 , X 2 and X 3 are the independent variables as mentioned earlier; β 0 is the regression coefficient at center point; β 1 , β 2 and β 3 are linear coefficients; β 12 , β 13 and β 23 are second order interaction coefficients; β 11 , β 22 and β 33 are quadratic coefficients.The values of the coefficients, the optimum levels and R 2 were determined as mentioned earlier.

RESULTS AND DISCUSSION
Evaluation of significant variables by Plackett-Burman Design.
The design matrix selected for the screening of significant variables for vitamin B 12 production and the corresponding responses are shown in Table 1.The adequacy of the model was calculated and the variables showing statistically significant effects were screened via Student's t-test for ANOVA with SAS statistical software (Table 2).Factors showing p-values of less than 0.05 were considered to have significant effects on the response and were therefore, selected for further optimization studies.Glucose, with a probability value of 0.0018 was determined to be the most significant factor, followed by (NH 4 ) 2 SO 4 (0.0115) and KCl (0.0298).
Glucose had a positive effect on the vitamin B 12 production.Vitamin B 12 is a highly complicated molecule as it contains a tetrapyrrole-derived macrocycle (Martens et al., 2002;Wang et al., 2010) and somewhere around 30 genes are necessary for its complete de novo biosynthesis (Wang et al., 2010).In these processes, large amounts of energy and cofactors were required for maintenance and vitamin B 12 production in the oxygen-dependent fermentation by P. denitrificans (Sano, 2009).Therefore, a high glucose concentration is necessary for greater vitamin B 12 production.Nitrogen as (NH 4 ) 2 SO 4 also plays an important role in cell  R 2 (coefficient of determination) = 0.99; X 1, glucose (g/l); X 2 , KCl (g/l); X 3, (NH 4 ) 2 SO 4 (g/l).
growth and product fermentation and higher vitamin B 12 production was achieved as N source concentrations were increased.Statistical analysis also showed that KCl had a significant effect on vitamin B 12 production.As is well known, K + has a significant function in the regulation of intracellular osmosis.The K + concentrations in the cytoplasm of growing cells increases greatly with increases in osmolarity of the defined growth medium to allow expulsion of redundant putrescine (Borodina et al., 2005).Vitamin B 12 biosynthesis requires eight methylation reactions, with S-adenosylmethionine (SAM) serving as the methylating agent.Kamarthapu et al. (2008) proposed that, SAM synthetase was stimulated by the monovalent cation K + at 100 mM and the further research demonstrated the presence of conserved amino acids at positions D286 for the K + binding site, confirming the need for K + for SAM synthetase activity.Betaine is an important methyl-group donor for vitamin B 12 biosynthesis in P. denitrificans (Lofty et al., 2007).In this study, betaine concentrations from 2.5 to 5.5 g/l had little effect on vitamin B 12 production, so betaine concentration was selected at 2.5 g/l.All other non significant variables were neglected and the optimum

Optimization of fermentation conditions by central composite design
The experimental design, the levels of vitamin B 12 production and the specific productivity obtained from different concentrations of additives are summarized in Table 3.
The experimental levels of vitamin B 12 production and the specific productivity were used to generate the second-order regression models.A second order polynomial function was fitted to the experimental results of vitamin B 12 production and the specific productivity with statistical software.The response of vitamin B 12 production (Y) and the specific vitamin B 12 production rate (Y p ) can be expressed in terms of the following regression equations (3, 4), respectively: The quality of fit of the second-order regression equations was checked using the coefficient of determination R 2 (multiple correlation coefficient) and was calculated to be 0.9902 and 0.995 with vitamin B 12 production and the Y p as the response, respectively, indicating that 99.02 and 99.5% of the variability in the response could be explained by the model.The ANOVA table is presented in Table 4.The model F-values of 112.11 and 21.19 showed the models were significant and the adequacy of second-order regression models.
According to the linear coefficients of second-order equations, statistical analysis demonstrated that glucose and (NH 4 ) 2 SO 4 had significant effects on vitamin B 12 production, while KCl and (NH 4 ) 2 SO 4 both had considerable effects on the Y p within the tested ranges, at the probability level of 95%.
For predicting these optimal points mathematically, second-order regression models were used to develop response surface curves relating vitamin B 12 productivity to various levels of additives.Figures 2 and 3 showed the response surface plot for vitamin B 12 production and Y p as a function of KCl and (NH 4 ) 2 SO 4 and glucose, respectively.By moving along the X and Y axes, it can be demonstrated that increasing the levels of KCl and (NH 4 ) 2 SO 4 had a conspicuous effect on vitamin B 12 production (Figure 2).As a result, the stationary ridge shape was observed on the surface plot of the two fermentation parameters and the maximum response inside the design boundary.Based on the medium optimization evaluated from the model of Equation 3, the predicted largest vitamin B 12 yield was 67.2 mg/l when the glucose, (NH 4 ) 2 SO 4 and KCl concentration were 72.2, 7.85 and 0.98 g/l, respectively.Figure 3 showed the interaction of glucose and (NH 4 ) 2 SO 4 had a significant effect on Y p , the plot exhibited that higher concentrations of glucose and (NH 4 ) 2 SO 4 were favorable for the Y p .The maximum predicted Y p of 33.3 μg/gDCW/h was attained when the glucose, (NH 4 ) 2 SO 4 and KCl concentration were 93.6, 7.93 and 1.24 g/l, respectively.The results showed that higher initial concentrations of glucose and KCl were important for accelerating the vitamin B 12 biosynthesis rate of P. denitrificans, while lower glucose concentration was favorable for higher vitamin B 12 production.
The verification experiments carried out at two optimal conditions showed the observed experimental vitamin B 12 production and the Y p was 65.8 ± 1.5 mg/l and 34.2 ± 1.0 μg/gDCW/h, respectively, which were approximately 103.3 and 87.5% higher than that of the control (32.4 ± 1.3 mg/l and 18.2 ± 0.7 μg/gDCW/h with basal medium before applying P-B optimization).The great similarity between the predicted and observed results demonstrated the accuracy and applicability of  X max , Maximum cell concentration (g/l); growth rate: maximum cell density divided by the bath culture time (g/L/h); Yp, specific vitamin B 12 production rate (μg/gDCW/h).Ammonium controlling strategies: control, with the basal fermentation medium and not ammonium controlled; Run-a, Run-b and Run-c: the ammonium concentration were controlled at 22 ± 1.8, 40 ± 2.0 and 65 ± 2.5 mmol/l by feeding ammonium sulfate in vitamin B 12 fermentation with optimal medium, respectively.
the central composite design as an extremely powerful method for optimizing the vitamin B 12 medium (Li et al., 2000).

Ammonium controlling strategies in fed-batch fermentation (15 L) with chemically defined media
Vitamin B 12 , a secondary metabolite of P. denitrificans, was mainly synthesized during the stationary growth period, and the fed-batch cultivation was applied in vitamin B 12 fermentation of P. denitrificans.Except for the initial nutrients, the components of the medium in the stationary state of cell growth also exerted significant effects on vitamin B 12 biosynthesis.The earlier mentioned results had proved that (NH 4 ) 2 SO 4 was the significant factor for both vitamin B 12 production and the specific B 12 production production rate.Therefore, the effects of ammonium concentration on vitamin B 12 biosynthesis were further investigated in fed-batch culture with a chemically defined medium.With the rapid cell growth, the residual ammonium concentration drastically dropped from 248 ± 2.9 to 23.5 ± 3.2 mmol/l at 47 h (Figure 4) and then the ammonium concentrations were controlled at 22 ± 1.8, 40 ± 2.0 and 65 ± 2.5 mmol/l by feeding ammonium sulfate solution (Figure 4).The fed-batch fermentation with the basal fermentation medium and no ammonium control was used as the control.
As shown in Table 5, the period of batch phase in optimal chemically defined medium was only 37.5 ± 0.8 h, less than that of the control (48 ± 1.5 h) and the cell growth rate was 41.7 to 61.2% higher than that of the control.Cell growth rate was enhanced by the increased ammonium concentration.A pronounced enhancement of vitamin B 12 synthesis was achieved under an ammonium concentration of 40 ± 2.0 mmol/l and the final vitamin B 12 production was 165 ± 2.2 mg/l, which was 84.4% greater than that of the control.The specific vitamin B 12 production rate (Yp) also increased when the ammonium concentration increased to 40 ± 2.0 mmol/l.The maximum Yp reached 36.0 ± 1.31 μg/gDCW/h; 57.2% higher than that of the control (22.9 ± 0.83 μg/gDCW/h) in fed-batch fermentation (Table 5).However, when ammonium concentration increased to 65 ± 2.5 mmol/l, vitamin B 12 biosynthesis was dramatically inhibited, the vitamin B 12 production decreased to 93.5 ± 2.2 mg/l and the Yp dropped to 19.1 ± 2.07 μg/gDCW/h.Determination of the intermediates showed that higher specific secretion rates of glycine, glutamate and succinate were obtained when the ammonium concentration was increased to 40 ± 2.0 mmol/l.These intermediates were the critical precursors of δ-ALA biosynthesis and thus, promoted vitamin B 12 production.Glutamate joint the transaminase for many amino acids biosynthesis and ammonium was necessary for the biosynthesis of glutamate from α-ketoglutarate by reductive ammonia fixation via glutamate dehydrogenase (Muller et al., 1989).Therefore, maintaining the appropriate ammonium concentration at 40 ± 2.0 mmol/l was favorable for higher glutamate production.The other amino acids such as aspartate, serine, alanine and organic acids α-ketoglutarate and acetate showed little difference under all the four conditions.However, under the highest ammonium concentration of 65 ± 2.5 mmol/l, the specific secretion rate of glycine, glutamate and succinate were significantly decreased.This might be caused by the significant inhibition of high ammonium concentration on the activity of pyruvate dehydrogenase and then led to the highest specific puruvate accumulation rate of 121.1 ± 13.8 μg/gDCW/h, which were 9 folds higher than that of the other conditions.Overall, the control of the ammonium concentration at 40 mmol/l led to higher accumulation of relevant intermediates and therefore, greater vitamin B 12 biosynthesis by P. denitrificans.

Scale-up of the optimal ammonium control strategy in a 10 m 3 fermenter
The results from the fed-batch fermentation with a chemically defined medium demonstrated that the nitrate source levels had significant effects on vitamin B 12 fermentation.In vitamin B 12 fermentation, the byproducts like beet molasses and corn steep liquor are extensively used as nitrogen sources due to their low cost.The vitamin B 12 productivities in large-scale industrial fermentation were always tremendously influenced by the uncertainty of the components and changing quality of raw materials, especially the large differences in the nitrogen content of the raw materials.Therefore, in order to eliminate negative effects of nitrogen content fluctuation in the medium on vitamin B 12 production, the optimal ammonium controlling strategy was implemented on the industrial fermentation in a 10 m 3 fermenter with complex medium.Figure 5 shows the kinetics of ammonium concentration, biomass growth, vitamin B 12 production and δ-ALA concentration in largescale fermentations with and without ammonium controlling strategy.As cell growth increased, the residual ammonium concentrations decreased sharply to 15.0 ± 3.0 mmol/l at 37 ± 2.8 h at the two kinds of cultivation and then ammonium concentrations were maintained at 35 to 43 mmol/l by pulse feeding of ammonium sulfate solution, compared with that of the control)(Figure 5a).The cell growth, δ-ALA concentration and vitamin B 12 production had no different before the ammonium concentration was controlled.Although, the cell growth with the ammonium controlled cultivation had little higher than that of the control (Figure 5b), a remarkable enhancement of vitamin B 12 biosynthesis was achieved under ammonium controlled cultivation (Figure 5c).A stably vitamin B 12 production reached 212.0 ± 3.0 mg/l, 20.5% higher than that of the control (176.2 ± 2.9 mg/l).The maximum specific vitamin B 12 production rate reached 37.1 μg/gDCW/h under ammonium controlling strategy, which was 14.6% higher than that of without ammonium control (31.6 μg/gDCW/h).Analysis of the δ-ALA concentrations showed that maintaining ammonium concentration at 35 to 43 mmol/l would greatly accelerate δ-ALA biosynthesis and the high flux of δ-ALA generation thus improved the vitamin B 12 production (Figure 5d).This ammonium control strategy was successfully applied to industrial production in a complex medium and a stable and higher vitamin B 12 production was achieved.
Previous research shown that glutamate and methionine are necessary for growth of P. denitrificans in a chemically defined medium (Christiansen et al., 2002), while our results demonstrated that the industrial strain of P. denitrificans can grow well with glucose as the sole carbon source.The specific vitamin B 12 production rate reached 36.0 ± 1.31 μg/gDCW/h under the optimal chemically defined medium in fed-batch fermentation, which was similar to that with a complex medium (37.1 μg/gDCW/h) in large-scale industrial production (Li et al., 2000)..The physiological characteristics of P. denitrificans in industrial vitamin B 12 production can be well implemented in this chemically defined medium.Therefore, the obtained optimal chemically defined medium and ammonium controlling strategy could be used for further 13 C-metabolic flux analysis of P. denitrificans, so as to reveal the metabolic network distributions and implement a rational optimization on industrial vitamin B 12 fermentation.Furthermore, a stable high productivity and more econo-mical industrial vitamin B 12 production were successfully achieved.

Conclusion
In this research, glucose, ammonium sulfate and KCl were selected as the major factors affecting vitamin B 12 productivity of P. denitrificans from eight varieties through P-B design and the optimal concentration and the interaction between these factors were estimated with RSM design and a second-order polynomial model.Further experiments showed that maintaining the ammonium concentration at 35 to 43 mmol/l was necessary for enhanced vitamin B 12 biosynthesis.One advantage of optimizing a chemically defined medium and establishing a control strategy was that a higher specific vitamin B 12 production rate was achieved, which was similar to that of an industrial complex-medium condition.In addition, the ability to use glucose and inorganic ammonium as the sole carbon and nitrogen sources should allow further metabolic flux analysis on vitamin B 12 fermentation of P. denitrificans by 13 C labeling experiments, as the influence of unlabeled carbon sources on labeling abundance is eliminated.The ammonium controlling strategy was able to prevent the fluctuation in total-N concentrations of the raw materials in industrial production, allowing a stable 20.5% augmentation of B 12 production.

Figure 2 .Figure 3 .
Figure 2. The response of vitamin B 12 yield as a function of (NH 4 ) 2 SO 4 and KCl concentration based on the central composite experimental results.

Figure 4 .
Figure 4. Profiles of ammonium and glucose concentration with different ammonium controlling strategies in the fed-batch vitamin B 12 fermentation in 15 L bioreactors with chemically defined medium.Ammonium concentration were controlled by feeding ammonium sulfate solution at Run-a, 22 ± 1.8 mmol/l (▲); Run-b, 40 ± 2.0 mmol/l (•); Run-c, 65 ± 2.5 mmol/l (■); the control (with the basal fermentation medium and not ammonium controlled, ).

Table 2 .
Model coefficient estimated by multiplies linear regression.

Table 3 .
Central composite design matrix for the three variables along with the observed and predicted vitamin B 12 production and specific vitamin B 12 production rate (μg/gDCW/h).

Table 4 .
Model coefficients estimated by multiple linear regression (ANOVA).

Table 5 .
The physiological parameters of the fed-batch vitamin B 12 fermentation by P. denitrificans under various ammonium controlling strategies after 168 h cultivation.