Enhancing vitamin B₁₂ content in co-fermented soy-milk via a Lotka Volterra model

Microorganisms and culture conditions

Lactobacillus reuteri ZJ03 and P. shermanii ZJ01 were taken from the culture collection of Key Laboratory for Food Safety of Zhejiang Province. The stock of cells was maintained in glycerol 50% (v/v) at −70°C. The bacteria were propagated in vitamin B₁₂ test broth (Luqiao, Beijing, China) in standing cultures overnight at 37°C.

One milliliter of fermented soy-milk was mixed with 9 mL of saline solution. The sample was diluted to corresponding concentrations and spread onto solid agars such as MRS (pH of 5.0) and NaLa agar [20]. MRS agar was incubated at 37°C for 72 h and NaLa agar was incubated at 30°C for 7 days. Lactobacillus reuteri ZJ03 was counted from MRS agar of the white shiny smooth colonies. Propionibacterium shermanii ZJ01 was identified and counted from NaLa agar by the morphology of 1.0–2.5 mm, dull brown, lighter margin colonies. A subtraction method, as a control, could also be used to calculate counts of Propionibacteria by subtracting the number of L. reuteri from the total count in NaLa agar.

Fermentation

A basic medium in a glass flask was prepared using 100 mL soy-milk, 5 g glucose, 1 mL of 10⁸ CFU/mL of Lactobacillus reuteri inoculum and 1 mL of 10⁸ CFU of P. shermanii inoculum. The medium were adjusted pH to 6.5 and conducted at 30°C, if it was not mentioned separately.

To study the effect of pH on fermentations, initial pH values were adjusted to 6.0, 6.5, 7.0, 7.5, and 8.0 for this set of experiments. To study the effect of temperature on fermentations, experiments were conducted under 28, 30, 35, and 37°C. To study the combination effects of aerobic/anaerobic fermentation, experiments were conducted respectively under 6-day anaerobic and 1-day aerobic, 5-day anaerobic and 2-day aerobic, and 4-day anaerobic and 3-day aerobic fermentation conditions, respectively. All other trials, unless specified, were carried out at 30°C under 5-day anaerobic followed by 2-day aerobic fermentation conditions.

Determination of vitamin B₁₂ and metabolites

Vitamin B₁₂ was analyzed using a microbiological assay modified by Denter [21] and HPLC method [22]. Vitamin B₁₂ was extracted from 1 mL of soy-milk with 10 mL sodium acetate buffer (pH 6.0) in present of KCN and heated in a water bath for 30 min at 70°C. Then, the pH of the solution was adjusted to 7.0 and mixed with 10 mL hexane (Extra pure N-hexane, Merck, Darmstadt, Germany) then centrifuged for 15 min at 4010 g (Varifuge 3.0, Heraeus centrifuge, Heraeus Instruments, Hanau, Germany). The aqueous phase was collected and passed through a solid phase extraction column (SPE) (CEC181M6 United Chemical Technologies, Bristol, PA, USA), which had been washed by 3 mL methanol (Merck, Darmstadt, Germany) and 3 mL double distilled water (DDW) [from Reversed osmosis Mill-Q water (18 Ω) (Millipore, Billerica, MA, USA)], with the aid of a pump (AL 15, Knf Neuberger, Hamburg, Germany) to control the speed of drops at one drop per second. After 3 times washing by ultra purified water, 3 mL methanol was used as the eluate. After the solvent was evaporated to dry, the residue was dissolved by 1 mL ultra purified water. The solvent was filtered through a membrane filter (0.2 μm) (Macherey-Nagel, Düren, Germany) and the filtrate was analyzed by HPLC (Waters E2695, USA) using a RP-18 column (250*4 mm I.D., 5 μm, Merck, Darmstadt, Germany).

All chromatographic separations were carried out at room temperature. A flow of 0.5 mL/min, methanol with 0.1% formic acid (A) (Merck, Darmstadt, Germany) and ultra purified water with 0.1% formic acid (B), which were degassed by an ultrasonic water bath (Sonorex TK 52, ultrasonic waterbath, Bandelin electronics, Berlin, Germany), were used as mobile phases and the gradient elution was programmed as follows; 0–2 min 20% A; 2–3 min 20–25% A; 3–11 min 25–35% A; 11–19 min 35–20% A; 20–22 min 100–100% A; 22–26 min 100–20% A; 26–36 min 20% A. The injection volume was 100 μL and the column eluate was monitored by Diode Array Detector (Waters, USA) at 361 nm wavelength.

Metabolites were detected using HPLC (Waters E2695, USA) with an organic acid column (850 BP-OA H+, 300*7.8 mm, Benson Polymeric, Sparks, USA) as a solid phase. All chromatographic separations were carried out at 60°C. A flow of 0.6 mL/min with 26 mM sulfuric acid was used as the mobile phase. The injection volume was 10 μL and the column eluate was monitored by Lachrom RI Dectector. One milliliter fermentation liquid was centrifuged for 10 min at 17,000 g (Biofuge pico centrifuge, Heraeus Instruments, Hanau, Germany). Ten microliter supernatant was injected into HPLC for analysis.

Lotka Volterra model development/implementation

An assumption was made that both microorganisms grew naturally without any inhibition from themselves. L and P stand for CFU of L. reuteri and P. shermanii at time t. Q₁ and Q₂ respectively represent the physiological state of both microorganisms. μ_maxL and μ_maxP separately show the maximum species growth rate. L_max and P_max stand for the maximum CFU during fermentation. The coefficients of a and b means the interspecific competition parameters of P. shermanii on L. reuteri, vice versa.

According to the assumption made, Q_i/(1+Q_i) was set as 1. The integration of equation was made from t_i−1 to t_i. The differential equations (Eq. 1, Eq. 2) can not be solved manually. Hence, least squares method was introduced to estimate the coefficients of a and b using Matlab (Version 5.3.0.10183, Mathworks Inc., Natick, MA, USA). The transpose of A is A^T.

Aki=∫ti−1tiL(t)dt

i=1, 2, 3…m

A=[t1−t0−A11−A21:::tm−tm−1−A1m−A2m]

X=[μmaxLμmaxLLmaxμmaxLaLmax]

Y=[μmaxPμmaxPPmaxμmaxLbPmax]

X=(ATA)−1ATBL

Y=(ATA)−1ATBP

Temperature effects on co-fermentation

Temperature displayed a mild influence on interactions between these bacteria. With the increase of temperature, the inhibition effects of both bacteria increased (Table 1). In this study (Figure 1), 30°C is an optimal temperature for co-fermentation to produce vitamin B₁₂. An unexpected drop in cell numbers for L. reuteri and P. shermanii was observed from 28 to 37°C (Figure 2). The decrease of final concentrations of ethanol (Table 2) and increase of final concentrations of propionic acid (Table 2) from 28 to 37°C can be recognized as a reason for low cobalamin production and low cell densities of L. reuteri, confirmed by increasing values of interaction coefficient a (Table 1). Our data suggested that dramatic changes of metabolism were observed at 37°C. It is may due to the suitable temperature for L. reuteri to produce more lactate. Meanwhile, P. shermanii prefers to utilize lactate as primary carbon resources, leading to increasing of propionic acid. There is no report about the relationship between temperature and cobalamin production before. But some researchers [14, 23, 24, 25] reported a correlation between temperature and 3-HPA production or propionic acid, in which a cobalamin dependent enzyme was involved. Doleyres et al. [23] reported no significant difference in 3-HPA production at temperatures between 15 and 37°C. Another contradictory result suggested that 3-HPA production at 37°C was significantly higher than at other temperatures in any kind of media [25]. The optimal growth temperature for Propionibacterium spp. is however, almost 30°C [24].

Figure 1:

Vitamin B₁₂ production under various conditions. Each experiment was repeated for 5 times.

Figure 2:

Microorganisms content on 7^th day under various conditions. Each experiment was repeated for 5 times.

Initial pH effects on co-fermentation

With the increase of initial pH values, both bacteria had a synergistic effect on each other. But these phenomena did not enhance the production of vitamin B₁₂. pH is essentially important in influencing metabolites and cobalamin production. Both microorganisms have their own optimal pH and the corresponding suitable ranges. The optimal pH for 3-HPA production is 6.0 [24], whereas the best pH for propionic acid is between 7.0 and 7.2 [14]. The question of the optimal initial pH for cobalamin production in co-fermentation was solved by our work (Figure 1). At pH 6.5–7.0 the highest values of cobalamin were reached. Interestingly, as is shown in Table 1, the ratio of a/b at pH 6.5 and pH 7.0 were 13.54/16.64 and −10.19/−10.31, respectively. Except for final concentrations of L. reuteri at pH 6.5, no huge difference was observed between both microorganisms content among various pH conditions (Figure 2). At pH 6.5 they had an inhibitory effect, but at pH 7.0 they changed to a synergistic effect. Moreover, a very strong influence on production of ethanol, propionic acid, and acetate was observed (Table 2). A possible interpretation of high cobalamin production is that a higher production of propionic acid needs more cobalamin under acidic condition (pH 6.5–7.0), as reported by Hsu and Yang [26]. Another explanation could be the low activity of 3-HPA generation above pH 7.0 [24], whereas more ethanol was generated to balance the redox reaction in L. reuteri.

Oxygen supplementation effects on co-fermentation

Oxygen, as mentioned above, is involved in the DMBI generation in P. freudenreichii. In the presence of oxygen, growth is slower due to the inhibition of propionic acid, acetate, and succinate formation, but pyruvate is accumulated [27]. However, propionic acid, an inhibitory factor of both microorganisms, can be decomposed in the presence of oxygen. Some researchers [28, 29] have conducted an oxygen cycle to improve cobalamin production by mediating catabolism of glucose to propionic acid and acetate in the presence and absence of oxygen [14]. Oxygen also affects L. reuteri to synthesize more heme against toxic forms of oxygen [30]. In this study (Figure 1), 2-day aerobic fermentation showed a higher productivity of cobalamin than 1-day or 3-day aerobic fermentation. But an obvious increase inhibitory effect of bacteria can be observed from an increase of interaction coefficients during the increase of aerobic fermentation duration. The CFU of L. reuteri increased during the increase of oxygen supplied days (Figure 2). Propionibacterium shermanii demonstrated the opposite trend (Figure 2). Some researchers [29, 31] found that low dissolved oxygen was advantageous for cell growth, propionate decomposition and acetate production by P. shermanii decrement, which was also confirmed in this study (Table 2). The dissolved oxygen obviously led to a reduction of final ethanol concentrations (Table 2).

Interaction and co-fermentation

A Lotka Volterra model, known as an ecological predator-prey model, was applied to describe the competition between microorganisms. The interaction coefficients, which describe the antagonistic activities, were obtained by fitting the Lotka Volterra model with least square methods. The coefficients of a and b can be interpreted as the interspecific competition parameters of P. shermanii on L. reuteri, and vice versa. The average values of interaction coefficients of a (−4.66) and b (−3.60) from this study represented a less negative effect from P. shermanii on L. reuteri and a negative effect from L. reuteri on P. shermanii. However, some researchers [24] reported that the maximal cell numbers of L. acidophilus and P. shermanii were higher than in single culture fermentation.

With the exception of pH 6.5, all fermentations with high productions of cobalamin did not show strong antagonistic effects between the two microorganisms (Table 1). Experiments of oxygen supply for 1 day acquired huge negative values of interaction coefficients. It can be explained that oxygen to some extent became the main inhibitor for the growth of both microorganisms. Interaction coefficients were changing from positive values to negative values during fermentations with an increase of initial pH from 6.0 to 8.0. This means a high initial pH value is beneficial for growth of both bacteria. No significant difference of interactions was found in fermentation under different temperatures.

A synergistic effect in the co-fermentation of L. acidophilus and P. shermanii was described by Liu and Moon [24]. They reported no lactate accumulation in the medium. Acetic acid production rates per generation were lower in mixed cultures and growth rate was faster than before. Results from this study agreed with previous results [24] and partly confirmed them as well. The robust growth of mixed cultures was also indicated in this study. Fast increase of propionic acid, ethanol production, and OD values were observed (data not shown), along with an accumulation of lactate observed, particularly in fermentation at 37°C (Table 2). Propionibacterium sp. can reduce the lactate stress on L. reuteri and retard the decrease of pH. Moreover, at the same time, L. reuteri can decompose proteins from soymilk relying on full proteolytic system to meet the nitrogen requirement of itself and P. freudenreichii with a low ability of protease. Another hypothetical assumption is about the synthesis of Dmbi. 5,6-dimethylbenzimidazole, an important precursor of cobalamin, can only be formed in the presence of oxygen by P. freudenreichii and P. shermanii [16]. Because of this, after several days of fermentation under anaerobic conditions with P. freudenreichii or P. shermanii, the fermentation should be switched to aerobic conditions. However, Santos et al. [17] found that the gene of cobT of L. reuteri is 59% similar with Salmonella typhimurium, which could mean that L. reuteri has the ability to form Dmbi without oxygen. Furthermore, some analogs can improve production of cobalamin by protecting an inhibitory riboswitch [32].

Some researchers demonstrated that spent media cultured with LAB strains led to a low cell concentration of P. shermanii, which produced more cobalamin than before [33]. Nevertheless, mixed cultures in this study produced 1.6–2.4 fold more cobalamin than single fermentation (data not shown). For further work, a doubtless conclusion can be made that a co-fermentation with P. shermanii and L. reuteri can lead to a high production of cobalamin in soy-milk. Correlations between the ratio of a/b and vitamin B₁₂ production were also found. According to Figure 3, an obvious conclusion can be made that more vitamin B₁₂ production (Figure 1) can be produced when values of a/b approaches to 1. This means that both bacteria contribute to vitamin B₁₂ production. The model is is therefore demonstrated its feasibility in describing the response of vitamin B₁₂ production and predicting a response value within the appropriate ranges.

Figure 3:

Concentration of cobalamin production under various a/b.