Enhanced vitamin B₆ production in engineered Escherichia coli via restricted mix-carbon feeding and pressure-controlled fermentation

Abstract

Pyridoxine (PN), a major commercial form of vitamin B₆, is mainly produced via chemical synthesis, raising environmental concerns. Microbial fermentation offers a greener alternative, but low biomass and titer limit industrial application. In this study, we developed an integrated strategy combining DO-stat restricted mix-carbon feeding with two-stage pressure-controlled fermentation conditions, and medium optimization to enhance cell growth and PN production in Escherichia coli. DO-stat feeding enabled dynamic regulation of oxygen and carbon sources, while a two-stage pressure-controlled (0.1 bar for 0–24 h, 0.2 bar thereafter) improved both growth and PN biosynthesis. Meanwhile, a novel medium (CRS-67), optimized via Taguchi design, increased shake flask PN titer by 275 mg/L, 75.5 % over CS medium. Under systematic fermentation optimization in a 5-L bioreactor, OD₆₀₀ reached 142.8 and PN titer reached 3.33 g/L—the highest reported level to date. This study presents a robust and scalable process for microbial PN production and offers insights for industrial biomanufacturing of vitamin B₆.

1. Introduction

Vitamin B₆ is a vital compound in the pharmaceutical industry, as well as in the production of food and feed additives and cosmetics [[1], [2], [3]]. Vitamin B₆ denotes a class of chemically related compounds, comprising pyridoxal (PL), pyridoxine (PN), and pyridoxamine (PM), together with their respective phosphorylated derivatives: 5′-pyridoxal phosphate (PLP), 5′-pyridoxine phosphate (PNP), and 5′-pyridoxamine phosphate (PMP) [1,4,5]. Vitamin B₆ plays a crucial role as a cofactor for numerous proteins and enzymes, rendering it one of the most essential vitamins [6]. The majority of enzymes that utilize PLP as a cofactor are involved in diverse biochemical processes, including amino acid biosynthesis, decarboxylation, racemization, Cα–Cβ bond cleavage, elimination, and α-, β-, and γ-substitution reactions [7]. In the biosynthetic pathway of deoxysugars, specific enzymes utilize PMP as a cofactor [[8], [9], [10]]. Among the various forms of vitamin B₆, PN is the major commercial form. PN is primarily produced through chemical synthesis; however, this method is associated with high raw material consumption, complex processing steps, and significant environmental pollution. In recent years, the development of efficient, green, and sustainable microbial fermentation strategies has emerged as a research hotspot, aiming to provide an alternative to traditional chemical synthesis [11,12].

Microbial production of PN primarily focuses on the metabolic engineering modification of microbial strains. Various microorganisms, including E. coli [[13], [14], [15], [16]], Bacillus subtilis [17,18], and Ensifer meliloti [19,20], have shown appreciable advances in vitamin B6 biosynthesis. Some limitations prevent theirs industrial scalability: E. meliloti requires 168 h fermentations for 1.3 g/L PN (productivity: 7.74 mg/L/h) and suffers from inadequate genetic tools [19,20], whereas B. subtilis achieves only 65 mg/L titer despite dependency on toxic precursor 4-hydroxythreonine feeding [17,18]. E. coli has a clear genetic background, abundant genetic manipulation tools, short fermentation period, and contains a natural PN synthesis pathway, making it the preferred strain for PN production. Nevertheless, the fermentation titer and overall production efficiency of PN remain suboptimal. In our previous work, an engineered E. coli strain was constructed-through protein engineering and iterative multimodule optimization-to achieve a PN titer of 1.4 g/L in fed-batch fermentation [13]. To enhance strain stability and scalability, a single-plasmid high-yield E. coli strain was subsequently developed, which reached 1.95 g/L of PN under fed-batch fermentation conditions [14]. In another study, it achieved PN titer of 2120.7 ± 7.8 mg/L and productivity of 32.1 mg/L/h in 30 L batch fermentation by integrating directed evolution of key enzymes, pathway reconstruction, branch disruption, and precursor enhancement [15]. Although E. coli has improved the metabolic flux of PN synthesis pathway through metabolic engineering, including promoter optimization, enzyme fusion, and cofactor engineering, it has largely solved the intracellular limitations of PN synthesis. However, the extracellular barrier—optimal nutrient composition of the culture medium, optimal feeding regulation strategy, appropriate oxygen transfer efficiency—severely limits the cell growth and the high titer of PN in the bioreactor. In fed-batch fermentation, appropriate feeding control is critical for promoting cell growth and maximizing product yield, highlighting the need to explore optimized feeding strategies in the subsequent study.

Depending on the rate of nutrient assimilation or demand for the nutrients, different nutrient feeding techniques are used in fed-batch culture, such as DO-stat [[21], [22], [23], [24], [25], [26], [27]], constant-rate [21,24,[28], [29], [30], [31]], pH-stat [[31], [32], [33], [34]], exponential feeding [27,[35], [36], [37]]. Lv et al. demonstrated a 28 % increase in biomass and a 22.75-fold enhancement in β-carotene productivity in Yarrowia lipolytica using optimized fed-batch processes [21]. Nygaard et al. employed a combination of DO-stat and exponential feeding strategies, resulting in PHB titers of 17.5 g/L and 25.7 g/L, respectively-representing a 40 % improvement over the wild-type Cupriavidus necator [27]. Additionally, Saroj et al. established a pH-triggered feeding strategy in E. coli, where supplementation was initiated when pH > 7, achieving 22.6 g/L of N-methylanthranilate (NMA) [31].

During high-cell-density aerobic fermentation, the cellular oxygen demand can surpass the maximum oxygen transfer capacity, rendering oxygen availability a limiting factor that impairs microbial growth and metabolic activity [38]. Increasing pressure in the head space of the bioreactors is an effective approach to enhance the solubility of oxygen in the culture medium, thereby alleviating oxygen limitation and improving fermentation performance [39,40]. However, excessive pressure may exert stress on microbial cells, potentially compromising their physiological functions and productivity [41,42]. Consequently, pressure-controlled strategies require optimization tailored to the specific microbial strain and fermentation process characteristics for maximal production efficiency.

To address growth limitations of engineered E. coli strains in bioreactor systems and improve pyridoxine (PN) production, this study systematically evaluates multiple fed-batch regulation strategies, in combination with the rational optimization of fermentation medium composition. By precisely controlling the growth phases, significant improvements were achieved in both PN titer and cell density. This integrated approach provides a solid foundation for the development of scalable and efficient microbial processes for industrial PN manufacturing.

2. Materials and Methods

2.1. Strains and media

The PN producing strain used in this study was TZ03 from the previously published paper [14]. For shake-flask fermentation, activated colonies from the frozen stock were inoculated into tubes containing 5 mL of seed medium containing glycerol 12 g/L, glucose 4 g/L, yeast extract 8 g/L, peptone 7 g/L, succinate 2 g/L, α-KG 2 g/L, MgSO₄·7H₂O 0.2 g/L, MnSO₄·5H₂O 0.01 g/L, FeSO₄·7H₂O 0.01 g/L, and Na₂HPO₄·12H₂O 35.8 g/L. Fermentation media are listed in Table S1.

2.2. Fed-batch fermentation

The fed-batch fermentations were performed in 5 L bioreactors, with the initial glucose concentration precisely set to 10 g/L in the culture medium. Kanamycin was added to the medium to a final concentration of 50 μg/mL. Fed-batch fermentation was conducted as follows: The activated lawn from cryopreservation tube was inoculated into a 1 L shake flask containing 200 mL of seed medium (pH 6.5), and cultured at 37 °C and 220 rpm for 12 h. Subsequently, the culture was inoculated into a 5-L bioreactor with 2 L of fermentation medium (pH 6.5) at a dose of 10 % (v/v). Fermentation was performed at 37 °C with agitation from 200 to 800 rpm and an airflow of 2 L/min. DO was maintained at 30 % and pH at 7.0 using NH₃·H₂O/H₃PO₄. The feeding medium contains 450 g/L glycerol, 70 g/L glucose, 20 g/L Peptone, 2 g/L MgSO₄·7H₂O, 0.1 g/L FeSO₄·7H₂O, 0.1 g/L MnSO₄·5H₂O and 6 g/L succinate. Fed-batch fermentation was initiated automatically via a peristaltic pump upon depletion of the initial glucose in the medium. Throughout the fermentation process, feeding was halted whenever the combined residual concentrations of glycerol and glucose exceeded 2 g/L. Samples were taken every 2 h to determine the OD₆₀₀, residual glycerol and the PN titer.

The exponential feeding takes into account the specific growth rate of microorganisms to provide an appropriate feeding rate. The feed rate (F) uses the equation obtained from the laboratory.

$$\mathrm{F}={\text{Ke}}^{\left(\mathrm{t}-\Delta \mathrm{T}\right)}+\mathrm{b}$$

K: Base feeding constant (g/L/h), t: Elapsed fermentation time (h), ΔT: Time delay (h), (adjusts initiation point of exponential phase), b: Baseline feed rate (g/L/h) (maintains minimum nutrient supply), K and b as 20 g/L/h and 1.2 g/L/h respectively.

2.3. Analytical methods

Cell density (OD₆₀₀) was measured using a Hybrid Multi-Mode Reader (Synergy Neo2, BioTek, USA). Determination of residual glycerol and glucose: Samples were centrifuged at 12,000 rpm for 2 min, supernatant were diluted to an appropriate multiple with sterile water, the residual glycerol was determined using a SBA-40E biosensor and calibrated with 1 g/L glycerol and glucose. Bioreactor pressure was controlled by a Siemens S7-1200 Programmable Logic Controller (PLC). Real-time pressure data from sensors were continuously fed back to the PLC, which regulated a proportional valve through PID algorithms to maintain the setpoint within ±0.01 bar. PN titer was determined using a Thermo Fisher UltiMate™ 3000 high-performance liquid chromatography (HPLC) system equipped with an FLD-3400 fluorescence detector. The analysis was performed according to a modified gradient program based on a previously reported method [13]. Briefly, fermentation broth samples were centrifuged, and the resulting supernatants were analyzed by high-performance liquid chromatography (HPLC) with fluorescence detection. The excitation and emission wavelengths were set at 293 nm and 395 nm, respectively. PN was separated on an octadecylsilyl (ODS) column (Cosmosil AR-II, 250 × 4.6 mm, 5 μm; Nacalai Tesque) under gradient elution conditions. Mobile phase A consisted of 33 mM phosphoric acid and 8 mM l-octanesulfonic acid, adjusted to pH 2.4 with KOH. Mobile phase B consisted of 80 % (v/v) acetonitrile. The gradient program was as follows: 0–5 min, 0–1 % B; 5–10 min, 1–19 % B; 10–20 min, 19–28 % B; 20–25 min, 28–63 % B; 25–27 min, 63–0 % B; and 27–30 min, hold at 0 % B. The flow rate was maintained at 0.8 mL/min.

All data are presented as the mean ± standard deviation (SD) from three independent biological replicates. Statistical analysis was performed using Minitab Statistical Software 22 and OriginPro (version 9.1).

3. Results

3.1. Effects of different feeding strategies on vitamin B₆ production

Previous studies demonstrated that the integration of protein engineering with systematic multimodule optimization substantially improved the yield of PN [13]. Subsequently, a single-plasmid strain was constructed, and its performance was further optimized through integrative metabolomics and transcriptomics analyses, along with refinement of the culture medium composition. As a result, a PN titer of 1.95 g/L was achieved in a 5 L bioreactor [14]. Despite notable progress, low fermentation productivity and slow microbial growth remain key challenges in PN biosynthesis. Optimizing fermentation regulation strategies offers an effective means to overcome these limitations. Among various process strategies, fed-batch fermentation plays a pivotal role in optimizing microbial production by enabling precise control over nutrient supply, which in turn strongly influences both cell growth and metabolite accumulation. However, feeding strategies for PN production have not yet been systematically optimized or evaluated at the bioreactor scale.

To achieve precise regulation of mixed carbon source feeding and thereby coordinate the metabolic balance between cell growth and PN biosynthesis, this study systematically evaluated four fed-batch fermentation strategies using the CS medium (Fig. 1): low-rate constant feeding, high-rate constant feeding, exponential feeding, and DO-stat feeding. These feeding strategies encompass a range of nutrient delivery models, from fixed-rate to feedback-controlled approaches, reflecting representative paradigms for carbon source regulation in microbial fermentation. Constant-rate feeding is widely adopted in industrial applications due to its operational simplicity. In this framework, low-rate constant feeding can mitigate metabolic stress and maintain stable cell growth, whereas high-rate feeding facilitates rapid biomass accumulation and enhances productivity in the early fermentation phase. Exponential feeding mimics nutrient demands during the exponential growth phase and enables dynamic alignment of substrate supply with cellular proliferation, thereby optimizing overall fermentation efficiency. DO-stat feeding is a responsive, feedback-regulated strategy that adjusts substrate delivery in real-time based on DO levels, serving as a proxy for metabolic activity and oxygen demand. This method effectively prevents substrate accumulation and oxygen limitation by synchronizing nutrient supply with microbial respiration. To enhance PN production, the four aforementioned feeding strategies were compared in terms of their effects on PN titer under fed-batch fermentation conditions.

Fig. 1.

Effects of different feeding strategies on PN production after 48 h fed-batch fermentation in a 5-L bioreactor. Time-course profiles of OD₆₀₀ (blue), PN titer (black), Glu (purple) and Gly (red). A: Low-rate constant feeding (fixed rate: 10 g/L/h), B: Exponential feeding (details in Section 2.2 Materials and Methods), C: High-rate constant feeding (fixed rate: 30 g/L/h), D: DO-stat feeding (feeding triggered when DO > 50 %; halted when DO < 50 %).

In the 5 L bioreactor, the different strategies exhibited significant differences in mix-carbon source regulation, cell growth, and product accumulation (Fig. 1). Low-rate constant feeding refers to the continuous addition of a feeding solution at a fixed rate of 10 g/L/h. Under this strategy, fermentation achieved a maximum OD₆₀₀ of 22.68 and a peak PN titer of 0.978 g/L (Fig. 1A). Exponential feeding resulted in a maximum OD₆₀₀ of 44.9 and a peak PN titer of 0.962 g/L during the fermentation process (Fig. 1B). High-rate constant feeding conducted at a fixed rate of 30 g/L/h, yielded maximum OD₆₀₀ of 48.2 and a peak PN titer of 0.716 g/L during fermentation (Fig. 1C). DO-stat feeding involves coupling substrate addition to DO levels, with feeding initiated when DO exceeds 50 % and halted when it falls below 50 %. This strategy proved most effective for both cell growth and PN production, achieving a maximum OD₆₀₀ of 70.5 and a peak PN titer of 1.116 g/L during fermentation (Fig. 1D). These results demonstrate that DO-stat feeding not only effectively prevents excessive accumulation of mixed carbon sources but also dynamically responds to cellular metabolic demands, thereby enabling precise matching between carbon supply and metabolic activity. Therefore, a DO-stat control strategy was implemented for the subsequent fermentations to maintain DO at a constant set point.

3.2. Optimization of medium composition enhances cell growth and vitamin B₆ production

After establishing DO-stat feeding as the optimal strategy, we observed that the OD₆₀₀ of E. coli growth began to plateau after 34 h in the 5 L bioreactor, suggesting potential nutrient limitation in the later fermentation stage. This indicated that the existing medium composition might restrict further cell growth and product formation. As the fundamental basis for microbial growth and product synthesis, culture medium composition critically influences metabolic pathways and flux distribution. Therefore, follow-up experiments systematically investigated and optimized the synergistic effects of carbon sources, nitrogen sources, and trace elements to fully exploit the production potential of the engineered strain and enhance PN biosynthesis efficiency.

Medium screening and optimization strategies have been widely adopted to promote microbial growth and enhance the titer of target metabolites. To systematically evaluate the impact of different media on PN production and cell growth, this study used the CS medium as a starting point to design and test seven modified media formulations (CRS-1 to CRS-7). The media were strategically adjusted in carbon sources, nitrogen sources, and trace elements to enhance cellular adaptability to fermentation conditions, optimize nutrient supply, and redirect metabolic flux toward PN biosynthesis. In addition, certain formulations incorporated betaine to alleviate osmotic stress, employed inorganic buffering systems to enhance pH stability, replaced costly nitrogen sources to reduce production cost, and increased organic nitrogen levels to promote cell growth, thereby improving PN production efficiency. These formulations were evaluated via 48 h shake flask fermentations in 250 mL flasks (Fig. 2). Among these formulations, CRS-3, CRS-6, and CRS-7 demonstrated higher PN titers than the initial CS medium in 250 mL shake flask experiments. However, the CRS-6 medium resulted in lower OD₆₀₀ than CRS-3 and CRS-7, suggesting reduced biomass accumulation. To further investigate the performance of these media at a larger scale, CRS-3, CRS-6, and CRS-7 were selected—alongside the CS medium—for validation in 5 L bioreactor fermentations.

Fig. 2.

Effects of different medium compositions on PN titer and cell density after 48 h cultivation in 250 mL shake flasks. PN titer (blue bars) and OD₆₀₀ (black dots).

To further evaluate the effects of different medium formulations on the PN production performance of the engineered strain, fed-batch fermentation was carried out in a 5 L bioreactor over a 48-h cultivation period. The original CS medium was used as the control, whereas CRS-6 and CRS-7 served as the experimental groups (Fig. 3). Despite using the same PN-producing E. coli strain, significant differences were observed among the media. In the case of CRS-3, the highest OD₆₀₀ reached 119, with a maximum PN titer of 1.02 g/L (Fig. 3B). For CRS-6, the maximum OD₆₀₀ was only 40; however, it achieved the highest PN titer of 1.50 g/L (Fig. 3C). CRS-7 resulted in an OD₆₀₀ of 48.5 and a PN titer of 1.24 g/L (Fig. 3D). In contrast, the control CS medium yielded an OD₆₀₀ of 75 and a PN titer of 1.17 g/L (Fig. 3A). Among the tested formulations, CRS-6 showed the best performance in terms of PN production, achieving a 330 mg/L increase in titer compared with the CS medium, highlighting its superior capacity to promote PN biosynthesis. However, cell growth under the CRS-6 medium, as indicated by OD₆₀₀, was significantly lower than that observed with the CS medium. This suggests that although PN biosynthesis was enhanced, biomass accumulation was limited, potentially due to suboptimal concentrations or imbalances of certain essential nutrients in the medium. Therefore, to further improve the overall fermentation performance, a systematic optimization of the CRS-6 formulation is necessary to achieve a better balance between cell growth and PN production.

Fig. 3.

Effects of medium composition on PN production during fed-batch fermentation in a 5-L bioreactor. Panels A, B, C, and D represent media CS, CRS-3, CRS-6, and CRS-7, respectively. PN titer and OD₆₀₀ were measured to evaluate the effect of medium composition on fermentation efficiency. Time-course profiles of OD₆₀₀ (blue), PN titer (black) and DO (yellow).

Single-factor experiments were conducted using the CRS-6 medium (Table 1) as baseline, systematically varying individual components while holding others constant. The concentrations of yeast extract, peptone, MnSO₄, FeSO₄, and succinic acid were individually adjusted to evaluate their effects [14,43,44]. A classical one-variable-at-a-time (OVAT) approach was employed, where all other components were held constant while individually varying the concentration of each target component [[45], [46], [47], [48]]. This method is widely used in microbial fermentation studies and facilitates the identification of key factors influencing product biosynthesis. Detailed analysis of the experimental results revealed that the following components positively contributed to the improvement of PN titer: yeast extract (8 g/L), peptone (5–20 g/L), MnSO₄ (0.01 g/L), and FeSO₄ (0.1 g/L). These nutrients play essential roles in amino acid supply, redox balance, and cofactor availability, enhancing metabolite biosynthesis. In contrast, variations in succinic acid concentration showed no significant effect on PN accumulation and were therefore not subjected to further optimization in subsequent experiments.

Table 1.

One-variable-at-a-time optimization.

Num	Yeast Extract (g/L)	Peptone (g/L)	MnSO4 (g/L)	FeSO4 (g/L)	Succinic acid (g/L)	OD600	PN (mg/L)
1	2	12	0.3	0.01	2	16.08 ± 0.87	428.48 ± 23.00
2C	8	12	0.3	0.01	2	15.25 ± 1.00	527.07 ± 66.61
3	16	12	0.3	0.01	2	15.24 ± 0.24	384.64 ± 40.66
4	32	12	0.3	0.01	2	16.99 ± 1.69	250.22 ± 34.11
5	8	5	0.3	0.01	2	18.69 ± 2.01	564.13 ± 29.95
6	8	20	0.3	0.01	2	17.04 ± 1.26	406.91 ± 12.30
7	8	40	0.3	0.01	2	17.20 ± 0.56	386.18 ± 9.79
8	8	12	0.01	0.01	2	13.69 ± 5.11	556.62 ± 15.54
9	8	12	0.1	0.01	2	16.49 ± 1.08	343.67 ± 43.95
10	8	12	0.6	0.01	2	16.96 ± 0.84	264.00 ± 24.79
11	8	12	0.3	0.1	2	15.55 ± 1.00	532.06 ± 26.59
12	8	12	0.3	0.5	2	13.87 ± 2.08	355.92 ± 46.33
13	8	12	0.3	1	2	22.04 ± 1.52	301.03 ± 46.85
14	8	12	0.3	0.01	1	15.01 ± 0.74	439.62 ± 27.05
15	8	12	0.3	0.01	3	16.53 ± 0.92	291.49 ± 15.52
16	8	12	0.3	0.01	4	22.72 ± 0.55	307.25 ± 25.32

2^C is the control medium CRS-6.

Building on single-factor experiment findings, we employed a Taguchi design to systematically evaluate medium composition effects on PN biosynthesis and cell growth (Table 2). A three-level orthogonal array optimized four critical components: FeSO₄, MnSO₄, yeast extract, and peptone. These concentration ranges were strategically defined to span physiologically critical thresholds—from nutrient deficiency to metabolic saturation—as supported by preliminary data and industrial feasibility requirements. The results showed that variations in these factors significantly influenced both PN accumulation and biomass production. The Taguchi method, developed by Dr. Genichi Taguchi, is a robust parameter design technique widely applied in bioprocess optimization [[49], [50], [51]]. By enabling the identification of optimal parameter combinations with a minimal number of experimental trials, the proposed method effectively improves product quality and process stability while reducing overall costs.

Table 2.

A three-level Taguchi optimization.

Num.	FeSO4 (g/L)	MnSO4 (g/L)	Yeast Extract (g/L)	Peptone (g/L)	OD600	PN (mg/L)
1	0.06	0.006	4	3	13.87 ± 1.29	321.84 ± 19.01
2	0.06	0.01	8	5	14.46 ± 1.04	436.24 ± 5.88
3	0.06	0.015	10	8	23.94 ± 4.59	414.08 ± 1.37
4	0.1	0.006	8	8	24.13 ± 6.52	447.11 ± 10.44
5	0.1	0.01	10	3	16.61 ± 2.51	482.35 ± 13.47
6	0.1	0.015	4	5	16.93 ± 6.89	459.20 ± 9.04
7	0.15	0.006	10	5	29.71 ± 9.06	639.99 ± 21.48
8	0.15	0.01	4	8	26.87 ± 3.36	511.71 ± 30.46
9	0.15	0.015	8	3	10.11 ± 1.67	317.93 ± 3.15
0CR	0.01	0.01	5	5	20.69 ± 1.80	364.00 ± 7.89
0CRS−6	0.01	0.3	8	12	13.90 ± 0.20	519.21 ± 6.42

0^CR and 0^CRS−6 are the control medium CR and CRS-6.

The concentrations of medium components (Table 2) and their corresponding yields were analyzed using Minitab software, with results detailed in Table S1–S4. Analysis of the signal-to-noise (S/N) ratio response table for PN titer (Table S2) and the mean response table (Table S4), alongside the S/N ratio (Table S3) and mean response tables (Table S5) for OD₆₀₀, identified yeast extract and FeSO₄ as primary factors influencing PN production. Main effect plots for means and S/N ratios (Fig. S1 and S2), generated via Minitab, further confirmation of the optimization of medium composition to enhance PN yield.

Evaluation of S/N ratios and mean values revealed that MnSO₄ and yeast extract exerted the most significant effects on PN production: MnSO₄ contributed substantially to process stability, while yeast extract elevated overall yield. Conversely, peptone markedly promoted cell growth (OD₆₀₀) but exhibited minimal impact on PN stability.

Based on the combined performance of PN titer and cell density, formulation group 7 was identified as the optimal medium (Table 2) and designated as CRS-67. Under this condition, the PN titer reached 639.99 ± 21.48 mg/L, with an OD₆₀₀ of 29.71 ± 9.06, both of which were the highest values among all tested groups.

3.3. Synergistic enhancement of vitamin B₆ production through restricted mix carbons feeding and pressure regulated fermentation

The optimized medium CRS-67, developed through single-factor and Taguchi design experiments, enabled both robust cell growth and significantly enhanced PN production in fed-batch fermentation. Under conditions of low residual glucose, a maximum OD₆₀₀ of 66.9 and a PN titer of 2.79 g/L were achieved (Fig. 4), representing a 46.84 % increase compared with the previously obtained titer of 1.9 g/L. These findings highlight the positive impact of medium optimization and feeding strategies on PN production. However, DO level could not be effectively maintained within the target range (30 %–40 %) throughout the fermentation process, which was likely due to the absence of pressure-controlled in the bioreactor, resulting in oxygen limitation.

Fig. 4.

Effects of the optimized CRS-67 medium on PN production and OD₆₀₀ during fed-batch fermentation in a 5-L bioreactor. Time-course profiles of OD₆₀₀ (blue), PN titer (black) and DO (yellow).

To overcome the limitation of dissolved oxygen, pressure-controlled was introduced during the fermentation process by applying different levels of constant pressure to enhance oxygen solubility. Experiments were conducted under four pressure conditions: 0.1, 0.2, 0.3, and 0.4 bar (Fig. 5A–D). At 0.1 bar, the maximum OD₆₀₀ reached 108.2, and the highest PN titer was 2.88 g/L (Fig. 5A). At a pressure of 0.2 bar, the OD₆₀₀ was 106.8, and the PN titer was further increased to 3.09 g/L (Fig. 5B). However, when the pressure was increased to 0.3 bar, both values declined to 105.4 and 2.29 g/L, respectively (Fig. 5C). A further increase to 0.4 bar resulted in a more pronounced decrease, with OD₆₀₀ and PN titer, which decreased to 63.8 and 2.16 g/L, respectively (Fig. 5D). These results suggest that moderate increases in pressure can enhance PN production, whereas excessive pressure levels inhibit E. coli growth and PN accumulation.

Fig. 5.

Effect of different tank pressure-controlled on PN production. (A–D) Fermentations were carried out under constant tank pressures of 0.1, 0.2, 0.3, and 0.4 bar respectively. OD₆₀₀ and PN titer were continuously monitored throughout the fermentation. Time-course profiles of OD₆₀₀ (blue), PN titer (black), and DO (yellow).

To further optimize oxygen supply during PN fermentation, a two-stage pressure-controlled strategy was developed based on the phase-dependent oxygen demands of microbial metabolism. As shown in Fig. 6, pressure was maintained at 0.1 bar during the initial 24 h to match the minimal oxygen requirements of rapid cell division in the exponential growth phase, and increased to 0.2 bar thereafter to address the peak oxygen demand during PN biosynthesis (Fig. 6). This strategy significantly enhanced both OD₆₀₀ and PN titer, reaching 142.8 and 3.33 g/L, respectively—significantly higher than under constant-pressure conditions. Moreover, the glucose and glycerol concentrations remained consistently low, indicating that the DO-stat feeding strategy precisely regulated the carbon source supply, preventing substrate accumulation and metabolic repression. By integrating mixed carbon feeding with a two-stage pressure-controlled fermentation strategy, oxygen supply is dynamically matched to the cells' metabolic demand, significantly enhancing both PN productivity and cell growth. This combined approach offers a feasible and efficient solution for industrial-scale PN biosynthesis.

Fig. 6.

Dynamic profiles of PN fermentation under a two-stage pressure-controlled strategy (0.1 bar → 0.2 bar). Time-course variations are shown for OD₆₀₀ (blue, indicating cell growth), PN titer (black), glucose concentration (Glu, red), glycerol concentration (Gly, green), and DO (yellow).

4. Discussion

To our knowledge, this study is the first to integrate DO-stat to restrict mix-carbon feeding with pressure-regulated control to enhance vitamin B₆ (PN) production in E. coli. We systematically investigated the impact of various feeding strategies on PN biosynthesis, including low-rate constant feeding, exponential feeding, high-rate constant feeding, and DO-stat feeding. Among them, DO-stat feeding was identified as the most effective, as it enables a dynamic response to cellular oxygen demand and precise regulation of mixed carbon sources. When combined with stage-specific pressure-controlled, this strategy not only promotes robust cell growth but also significantly improves PN production efficiency, offering a promising and scalable solution for industrial fermentation processes.

To address the limiting factors of the initial medium for enhancing PN production, this study further combined single-factor optimization and Taguchi design methods to systematically optimize four key components: FeSO₄, MnSO₄, yeast extract, and peptone. As a result, a novel bioreactor-compatible fermentation medium, CRS-67, was developed. Validation experiments at the 250 mL shake flask level demonstrated that the optimized CRS-67 medium increased the PN titer to 639.99 ± 21.48 mg/L, representing an improvement of approximately 275 mg/L compared with the initial CS medium (364.00 ± 7.89 mg/L), thereby confirming the significant enhancement achieved through medium optimization.

At the 5 L bioreactor scale, we revealed a pressure-regulated promoting effect of pressure on PN biosynthesis: low pressure (0.1 bar) favored PN accumulation during the early phase, while moderate pressure (0.2 bar) facilitated sustained production in the mid-to-late stages. Moderate stress induced by stage-specific pressure control likely enhances PN production by influencing cellular physiology and metabolic flux. Early low pressure may improve substrate uptake and oxygen availability, while medium pressure in the later stage could redirect metabolic resources toward pyridoxine biosynthesis. Stress-induced changes in redox balance and energy demand may further stimulate PN-related pathways. Together, these effects suggest that controlled pressure acts as both a physiological and metabolic regulator to promote PN accumulation. Accordingly, a two-stage pressure-controlled strategy was established based on the fermentation phase characteristics, with the pressure maintained at 0.1 bar during the first 24 h and then switched to 0.2 bar. This strategy significantly improved overall PN production without compromising cell growth.

The final results demonstrated that, under this systematically optimized strategy, OD₆₀₀ reached 142.8, the final PN titer was 3.33 g/L, representing a 57 % increase over the previously reported highest level (2.12 g/L),the average productivity was 45.0 mg/L/h, surpassing the previously reported maximum of 32.1 mg/L/h, the maximum productivity peaked at 111.8 mg/L/h, and at lower cost,all of which significantly outperforms. In summary, this study achieved a remarkable enhancement in PN biosynthesis efficiency through comprehensive process optimization, encompassing feeding strategy, medium formulation, and fermentation parameter control. These findings provide a solid technological foundation and theoretical basis for the large-scale industrial production of PN.

CRediT authorship contribution statement

Long Chen: Writing – review & editing, Writing – original draft, Visualization, Validation, Software, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Miao-Miao Xia: Writing – review & editing, Resources, Project administration, Formal analysis, Conceptualization. Hui-Na Dong: Writing – review & editing, Supervision, Resources, Funding acquisition. Hao-Ran Ma: Visualization, Methodology. Xu-Yang Huang: Methodology, Formal analysis, Conceptualization. Gong-Jin Shen: Visualization, Data curation. Zhao-Xia Jin: Writing – review & editing, Supervision, Resources, Project administration, Funding acquisition, Formal analysis, Data curation, Conceptualization. Lin-Xia Liu: Writing – review & editing, Supervision, Resources, Project administration, Funding acquisition. Da-Wei Zhang: Writing – review & editing, Supervision, Resources, Project administration, Funding acquisition.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following are the Supplementary data to this article:

figs1.

figs2.