Introduction

As a third class of redox cofactors following pyridine nucleotide- and flavin-dependent cofactors, pyrroloquinoline quinine (PQQ) is an important quinone cofactor (Sun et al. 2014a, b). It is widely recognized as an essential nutritional component due to its ubiquitous presence in foods and also has noticeable antioxidant properties (Wei et al. 2010; Zhang et al. 2012). In addition, it acts like a growth-promoting factor, particularly as vitamins (Kasahara and Kato 2003). Recently, PQQ has also been shown to act as a bio-control agent for plant fungal pathogens (Han et al. 2008), as an inducer for protein kinases (Khairnar et al. 2007), and as a redox sensor leading to development of biosensors (Kurtinaitiene et al. 2010). Most of recent research work has shown that PQQ is more likely an anti-neurodegenerative, anticancer, and pharmacological agent and can work as a signal molecule (Misra et al. 2012; Yu et al. 2012). The current source of PQQ is mainly by chemical synthesis (Fouchard et al. 2004). However, the complicated multi-step organic synthesis leads to a very low production efficiency of PQQ, and the subsequent purification of PQQ from many isomers and various by-products becomes a big challenge. Comparably, the biological synthesis of PQQ has attracted increasing attention because of its environmental friendliness, cost-effectiveness, and high yield.

To date, the species used for microbial production of PQQ include Klebsiella pneumonia, Acinetobacter calcoaceticus, Methylobacterium extorquens, Gluconobacter oxydans, Enterobacter intermedium, Pseudomonas fluorescens, and Pantoea ananatis (Andreeva et al. 2011). Among them, the methanol-utilizing bacteria, such as Hyphomicrobium sp. strain TK 0441 (Urakami et al. 1992) and Methylovorus sp. MP688 (Xiong et al. 2011), gave a relatively high PQQ productivity. Further cloning and sequence analyses of PQQ biosynthetic gene clusters have been made using several PQQ producers, and 2 mg/L of PQQ was achieved in the culture of recombinant E. coli integrated with pqq ABCDE gene cluster from G. oxydans (Yang et al. 2010). However, few efforts have been paid to optimize culture medium and develop an efficient bioprocess to produce PQQ at high yield. It was reported that the primary effect of initial pH in the medium on PQQ production was investigated, and one constant-pH control strategy was adopted to improve the PQQ accumulation (up to 125 mg/L after 6 days of fermentation) (Xiong et al. 2011). However, the optimal pH values for cell growth and metabolite production were different during the entire fermentation, which has been exemplified by the fermentative production of some chemicals, such as arachidonic acid, propionic acid, lipids, and carotenoids (Li et al. 2015; Zhang et al. 2010). Recently, one high-yield strain (Methylobacillus sp. CCTCC M2016079) was screened out to produce 33.4 mg/L PQQ in the shake flask by developing one novel high-throughput protocol in our laboratory (Si et al. 2016). In the present work, systematic investigations on the effects of pH on biomass and PQQ production were carried out with our isolated strain, and the discrepancy in optimal pH values between cell growth and PQQ biosynthesis stimulated us to develop a novel two-stage pH fermentation process. In order to further improve the productivity and increase biomass during the pH-shifted stage, the combined feeding of carbon and nitrogen sources was examined in benchtop fermenters.

Materials and methods

Microorganism and culture medium

Methylobacillus sp. CCTCC M2016079 was derived from soil and maintained in our lab. Unless otherwise noted, the seed medium contained 3.0 g/L (NH4)2SO4, 1.4 g/L KH2PO4, 3.0 g/L Na2HPO4, 0.5 g/L MgSO4·7H2O, and 8 mL/L CH3OH; the fermentation medium contained 35 g/L CH3OH, 3.0 g/L (NH4)2SO4, 1.4 g/L KH2PO4, 3.0 g/L Na2HPO4, 1.53 g/L MgSO4·7H2O, 30 mg/L FeC6H5O2·6H2O, 30 mg/L CaCl2·2H2O, 5.0 mg/L MnCl2·4H2O, 3.15 mg/L CoCl2·6H2O, 418.7 μg/L para-aminobenzoic acid (PABA), 5.0 mg/L ZnSO4·7H2O, and 0.5 mg/L CuSO4·5H2O. All media were autoclaved at 115 °C for 20 min.

Production of PQQ with varied initial pH in the shake flasks

The strain was maintained on agar slants and incubated at 31 °C for 3–5 days. After growth and colony formation, one loop of colonies of Methylobacillus sp. CCTCC M2016079 was inoculated into 50 mL of seed medium in 250-mL flasks and cultivated for 36 h at 31 °C on a rotary shaker at 180 r/min. This culture was used as seed in the subsequent experiments. In the shake flasks, the initial pHs (4.8, 5.3, 5.8, 6.3, 6.8, 7.3, 7.8) of fermentation medium were adjusted by 1 M NaOH or 1 M H2SO4. Two hundred fifty-milliliter flasks containing 50 mL fermentation medium were inoculated with 10% (v/v) seeds and incubated on a shaker at 31 °C and 200 rpm for 48 h.

Batch production of PQQ with different pH controls in benchtop fermenters

The effects of different pH controls on PQQ production were further tested in the 10 L benchtop bioreactor (Shanghai Baoxing Bio-Engineering Equipment Co., Ltd) with a working volume of 7.0 L. The inoculum size was 10% (v/v) to give a final optical density of 0.4–0.5 at 600 nm. Dissolved oxygen level was adjusted to over 10% by adapting the stirrer speed, the air pressure, and the aeration rate. The temperature was kept at 31 °C throughout the fermentation process. The initial pH was either naturally changed or controlled at 5.3–7.3 with the addition of 30% NH4OH or 2 M H2SO4. For the two-stage pH control strategy experiment, the pH was controlled at 6.8 for the first 48 h and then changed to 5.8 during the rest of fermentation. Other conditions were similar to those from the experiment with the constant pH controls. Samples were periodically collected and analyzed for biomass, methanol, PQQ, and acetic acid.

Fed-batch production of PQQ with two-stage pH control in benchtop fermenters

When a two-stage pH control strategy was applied during the fed-batch fermentation, four varied feeding modes of methanol were further studied. For the pulse feeding fermentation, 140 mL of anhydrous methanol was added at 54 and 72 h respectively (mode A). In the constant feeding fermentation, anhydrous methanol was either fed at a fixed rate of 9.9 mL/h (mode B) from 66 to 108 h, or 11.5 mL/h (mode C) from 54 to 90 h, or 14.3 mL/h (mode D) from 54 to 84 h. Combined with the suitable methanol feeding, yeast extract was supplemented to further improve the productivity of PQQ at a low rate of 0.5 g/h from 90 to 114 h.

Analytical methods

The biomass was determined by the cell dry weight (DCW), which was collected by centrifugation at 5000×g and then dried at 105 °C overnight. The PQQ concentration in the broth was determined with HPLC according to our previous work (Si et al. 2016). The concentrations of methanol and acetic acid in the broth were analyzed by gas chromatography (Agilent 6820 system, USA) with HP-INNOWax column (30 m × 0.25 mm × 0.25um). Nitrogen was used as the carrier gas at a constant voltage of 15.0 psi (1 psi = 6894.76 Pa), and the initial temperature of the column was set at 50 °C. After holding for 3 min, the gas was heated to 150 °C at a rate of 15 °C/min. Then, the heating rate was changed to 3 °C/min up to 250 °C. The holding time was 3 min, and the injection volume was 1.0 μL.

Specific cell growth rate and specific PQQ production rate

The specific cell growth rate and specific PQQ production rate were estimated from experimental or fitted data of cell growth and PQQ production. The values of specific cell growth rate (μ x ) and specific PQQ production rate (μ p ) were obtained from the following equations:

$$ {\mu}_x=\frac{1}{x}\cdot \frac{dx}{dt} $$
(1)
$$ {\mu}_p=\frac{1}{x}\cdot \frac{dp}{dt} $$
(2)

μ x was determined from the slope of the semi-logarithmic plot of DCW versus fermentation time. The line of cell growth was interposed normally, and the cell density at a definite time (Dt = 0.1 h) was computed with OriginPro 8 SR3 (Version 8.0932; OriginLab Corporation, Northampton, MA, USA). Then, μ x was obtained at a definite time by computing cell DCW using Microsoft Excel. The value for μ p was obtained with the same method used to calculate μ x .

Results

Effect of initial pH on PQQ production in shake flasks

The effects of initial pH ranging from 4.8 to 7.8 on cell growth, PQQ accumulation, and acetic acid production were investigated in shake flasks (Fig. 1). When the initial pH of culture was 6.8, the biomass reached a maximum value of 8.1 g/L, and further increase of pH could lead to the decrease of biomass. However, the highest PQQ production (about 38.00 mg/L) was obtained with an initial pH of 5.8. These results suggested that the optimal pH conditions for cell growth and PQQ production were different. The acidic condition was beneficial for PQQ production while cell growth increased more rapidly with an initial pH close to neutral. In addition, the accumulation of by-product acetic acid was increased when pH was increased from 4.8 to 6.8, which indicated that lower initial pH could reduce the synthesis of the by-product (acetic acid). Therefore, it is essential to investigate the effects of pH on cell growth, methanol consumption, PQQ production, and acetic acid accumulation in a fermenter under a pH control strategy.

Fig. 1
figure 1

Effects of initial pH on cell growth, PQQ, and acetic acid concentration after 48 h cultivation at shake flask scale

Full size image

Effects of different constant pH on PQQ production in benchtop fermenters

The batch cultivation of Methylobacillus sp. CCTCC M2016079 was further performed with varied pH levels ranging from 5.3 to 7.3 (5.3, 5.8, 6.3, 6.8, and 7.3) in the 10-L benchtop fermenters. The time courses of cell growth, PQQ production, methanol consumption, and acetic acid accumulation were shown in Fig. 2. As pH increased, the lag time was evidently shortened and the cell growth rate was improved in the logarithmic phase of growth. When the pH was maintained at 6.8, the biomass reached a maximum value of 11.5 g/L at 72 h, while the lowest biomass was observed with the pH controlled at 7.3 (Fig. 2a); hence, a relatively high pH close to neutral was favorable to cell growth. Similarly, methanol consumption was increased when the pH was raised from 5.3 to 6.8 until 66 h, but a pH of 5.8 seemed more conducive for methanol consumption after 66 h (Fig. 2b). The comparison of PQQ production at different pH values indicated that PQQ was rapidly accumulated after 36 h of fermentation (Fig. 2c). The maximum PQQ production of 109.44 mg/L was obtained at a pH of 5.8, and the lowest PQQ concentration of 19.74 mg/L was observed at a pH of 7.3. This result illustrated that a proper acidic condition was more suitable for the biosynthesis of PQQ. As shown in Fig. 2d, the total acetic acid by-production was enhanced when the pH was increased from 5.3 to 6.8 and then decreased with further increase of pH up to 7.3, probably due to low biomass at that condition. The maximum acetic acid concentration of 4.38 g/L was obtained at a pH of 6.8, which suggested that the low pH was more conducive to reduce by-production of acetic acid.

Fig. 2
figure 2

Profiles of biomass (a), residual methanol (b), PQQ (c), and acetic acid (d) in the culture of Methylobacillus sp. CCTCC M2016079 under different pH controls in batch fermentation. All the experiment were carried out in 10-L benchtop fermenters

Full size image

Specific cell growth rate (μ x ) and specific PQQ formation rate (μ p ) were applied to describe the PQQ fermentation process. As shown in Fig. 3, the profiles of μ x and μ p had similar tendencies and the maximum of μ x and μ p were obtained at about 26 and 56 h, respectively. It was also concluded that pH 6.8 was better for cell growth at the beginning of fermentation to shorten the lag phase (Fig. 3a) but pH 5.8 was more beneficial for PQQ synthesis after 48 h, presenting a higher value of μ p (Fig. 3b). Based on the analyses of μ x and μ p , a two-stage pH control strategy was proposed, in which pH 6.8 was used to reach the optimal biomass before 48 h, while pH 5.8 was applied for better PQQ production after 48 h.

Fig. 3
figure 3

Effects of pH on specific cell growth rate (μ x ) (a) and specific PQQ formation rate (μ p ) (b). Nonlinear curve fits of μ x and μ p were carried out on account of the curves of cell growth and PQQ formation by origin 8.0 software according to logistic model

Full size image

Improvement of PQQ production by a two-stage pH control strategy in benchtop fermenters

Based on the above kinetic analysis, a two-stage pH control strategy was evaluated in a 10-L benchtop fermenter. Compared to the pH-constant (6.8, 5.8) or pH-uncontrolled fermentations, the pH two-stage process showed greater ability to produce PQQ. The maximum concentration of biomass (16.2 g/L) and PQQ (158.61 mg/L) with this two-stage pH process was distinctively higher than those from uncontrolled pH process (Fig. 4a) and constant pH processes (Fig. 2) and a very low level of by-product (2.71 g/L acetic acid) was accumulated (Fig. 4b). The obvious increase of PQQ concentration in this two-stage pH process was contributed by the following three factors: the suitable pH (5.8) environment for PQQ biosynthesis in the second pH-shifted phase, high biomass, and longer biosynthesis duration which might be explained by the reduced role of the lowest acetic acid concentration in the broth.

Fig. 4
figure 4

Improvement of PQQ biosynthesis in Methylobacillus sp. CCTCC M2016079 with two-stage pH or combined fed-batch strategies. a Uncontrolled pH, batch cultivation. b Two-stage pH control (pH 6.8 from the beginning to 48 h, pH 5.8 from 48 h to the fermentation end), batch cultivation. c Two-stage pH control (pH 6.8 from the beginning to 48 h, pH 5.8 from 48 h to the fermentation end), combined fed-batch fermentation (feeding anhydrous methanol at a constant rate of 11.5 mL/h from 54 to 90 h and yeast extract at a constant rate of 0.5 g/h from 90 to 114 h)

Full size image

Moreover, the batch fermentation of Methylobacillus sp. CCTCC M2016079 with natural pH was also carried out in a 10-L fermenter. The remarkable decrease of pH at the early stage was related to the rapid cell growth and methanol consumption. The biomass (11.91 g/L) and PQQ production (about 80.95 mg/L) obtained from the fermenter without pH control were much lower than those from the two-stage fermentation. This result suggested that although approximating changes of pH existed in this natural process, precise control of two-stage pH was superior for PQQ accumulation which was achieved by a relatively high cell density in the methanol-based fermentation.

Further improvement of PQQ production by a combined fed-batch strategy

To explore the potential of PQQ production, the two-stage pH control strategy was further investigated to improve PQQ productivity by combining different methanol feeding modes. Mode A was intermittent feeding of 140 mL methanol at 54 and 72 h respectively; modes B, C, and D were carried out by methanol feeding rate of 9.9, 11.5, and 14.3 mL/h, respectively. As shown in Table 1, mode A brought about a relatively high biomass (30.62 g/L) but low PQQ accumulation (205.54 mg/L); mode B gave the minimal biomass (26.53 g/L) and the lowest acetic acid by-production (2.86 g/L); mode C presented the highest PQQ concentration (272.21 mg/L) and productivity (2.84 mg/L/h) comparing with other feeding modes. Further increased feeding rate of methanol (mode D) would lead to the highest acetic acid by-production (4.17 g/L) and the minimum PQQ concentration (183.38 mg/L) after 90 h of cultivation. This result showed that, by integrating suitable constant methanol feeding process with the two-stage pH control strategy, the PQQ concentration and productivity (mode C) were further increased by 48.7 and 42.0% (mode C), respectively. Inspired by these findings, additional supplementation of yeast extract via mode C was also carried out (Fig. 4c). Notably, both biomass and PQQ production were significantly improved by feeding yeast extract from 90 to 114 h although acetic acid concentration was slightly increased up to 4.32 g/L simultaneously. The maximum biomass reached up to 39.32 g/L at 114 h, and the highest PQQ concentration (353.28 mg/L) and productivity (3.27 mg/L/h) were obtained at the end of fermentation.

Table 1 Comparison of PQQ production using different feeding strategies under two-stage pH control
Full size table

Discussion

Due to its increasing application in healthy food and pharmaceutical industries, some methanol-utilizing bacteria were screened out to accumulate a quantity of PQQ in the culture. However, the concentration and productivity of PQQ in the culture were relatively low and the highest productivity was 2.08 mg/L/h with Hyphomicrobium sp. strain TK 0441. Owing to low productivity, long fermentation time (480 h) was always needed to achieve a desired PQQ concentration (approximately 1.0 mg/mL) in the broth (Urakami et al. 1992). Although the PQQ biosynthetic gene clusters in some micro-organisms have been identified, the biosynthetic pathway of PQQ was very complex and remained unclear (Shen et al. 2012). Usually, a variety of environmental and nutritional factors such as cell morphology, culture parameters, and nutritional status can greatly affect PQQ biosynthesis (Misra et al. 2012). Therefore, besides genetic evolution of the isolated strains, systematic investigations on culture conditions and fermentation mode were very necessary to improve PQQ production in the laboratory.

Despite the stable cytoplasmic pH of microorganisms even within the growth media having varied pH values between 5.5 and 8.5, the pH of periplasmic space was most flexible, directly reflecting the pH of the surrounding medium. This was because the outer membrane was permeable to ions and low-molecular solutes that existed in the extracellular medium (Baneyx et al. 1991). In our experiments, it was found that the optimal pHs of culture for biomass accumulation and PQQ productivity were 6.8 and 5.8, respectively. The reason for enhanced PQQ biosynthesis under low pH circumstance (5.8) might be contributed by the fact that methanol dehydrogenase essential for PQQ biosynthesis was located within the periplasmic space and still exerted its high activity under low pH even at pH 5.6 (Kim et al. 2012). And it was also observed that the pqqA promoter could be induced when the host was transferred from pH 5.5 to pH 7.0, resulting in the upregulation of pqqA expression which then enhanced the biosynthesis of PQQ significantly (Ge et al. 2015). It was also known that the final step of PQQ formation was involved in a ring closure and eight-electron oxidation of AHQQ, which was catalyzed by PqqC. It was reported that this enzymatic reaction could be effectively carried out at a low pH value (pH 2.0) (Bonnot et al. 2013), which might be another reason for the enhancement of PQQ production under a relatively lower pH.

This discrepancy of optimal pH values between cell growth and PQQ biosynthesis laid a solid base for the development of a novel two-stage pH fermentation process. In our two-stage pH fermentation, cell growth rate was adjusted at a high level close to neutral pH during the first stage and then PQQ biosynthesis would start at a relatively high biomass level under low pH control. The results showed that PQQ concentration and productivity were improved by 44.9 and 45.5%, respectively, with regard to that under constant pH 5.8 control, and the highest concentration (158.61 mg/L) and productivity (1.47 mg/L/h) were attained. Meanwhile, the yield of PQQ increased from 2.93 mg/g methanol to 3.26 mg/g methanol. It was reported that two-stage pH control strategy could improve the productivity of several organic acids significantly by reducing the inhibitory effects on growth by the products, such as ketogluconic acid (Sun et al. 2014a), arachidonic acid (Li et al. 2015), and propionic acid (Zhuge et al. 2014). The present work demonstrated new application of this pH-shifted strategy for PQQ production by upgrading the expression of key biosynthetic genes with low pH control.

The feeding of different substrates was often employed to increase the production of end metabolites in the benchtop fermenters. By adjusting feeding rate of methanol, PQQ concentration and productivity were improved by 48.7 and 42.0%, respectively, compared with the pH-shifted batch fermentation, and the PQQ concentration and productivity were improved up to 272.21 and 2.84 mg/L/h, respectively. Accordingly, the yield of PQQ was also increased from 3.01 mg/g methanol to 4.02 mg/g methanol. However, in the above two-stage fed-batch process, the depletion of nitrogen sources might cause cell death or reduce cell viability during the second pH-shifted stage due to the only feeding of carbon source (methanol). To avoid this negative effect, the additional feeding of yeast extract in the second-stage pH phase was proposed. The results showed that the feeding of yeast extract could improve cell growth greatly, and the PQQ accumulation and yield were improved by 29.8 and 18.4%, respectively. Corresponding with this, the productivity was also improved up to 3.27 mg/L/h which was the highest reported in all the literature. Obviously, the combined feeding of carbon source (methanol) and nitrogen source (yeast extract) in the second pH phase could support the enhanced biosynthesis of PQQ based on a relatively high cell density, which was difficult to maintain without the combined feeding of yeast extract under the second neutral pH phase which was not optimal for cell growth.

In conclusion, the present work showed that the pHs suitable for cell growth and PQQ production could not be synchronized with Methylobacillus sp. CCTCC M2016079. A novel pH control strategy was adopted to greatly improve the productivity of PQQ by separating cell growth and product formation with two pH-control phases. Then, with the integration with the combined feeding of methanol and yeast extract, this two-stage pH fermentation provided further enhanced improvement of PQQ production. The present work is not only helpful for the industrial production of PQQ in the future but also beneficial to develop such similar efficient processes to improve the productivity when the pHs suitable for cell growth and product formation cannot be synchronized.