Enhancement of γ-aminobutyric acid production in recombinant Corynebacterium glutamicum by co-expressing two glutamate decarboxylase genes from Lactobacillus brevis

Abstract

γ-Aminobutyric acid (GABA), a non-protein amino acid, is a bioactive component in the food, feed and pharmaceutical fields. To establish an effective single-step production system for GABA, a recombinant Corynebacterium glutamicum strain co-expressing two glutamate decarboxylase (GAD) genes (gadB1 and gadB2) derived from Lactobacillus brevis Lb85 was constructed. Compared with the GABA production of the gadB1 or gadB2 single-expressing strains, GABA production by the gadB1–gadB2 co-expressing strain increased more than twofold. By optimising urea supplementation, the total production of l-glutamate and GABA increased from 22.57 ± 1.24 to 30.18 ± 1.33 g L⁻¹, and GABA production increased from 4.02 ± 0.95 to 18.66 ± 2.11 g L⁻¹ after 84-h cultivation. Under optimal urea supplementation, l-glutamate continued to be consumed, GABA continued to accumulate after 36 h of fermentation, and the pH level fluctuated. GABA production increased to a maximum level of 27.13 ± 0.54 g L⁻¹ after 120-h flask cultivation and 26.32 g L⁻¹ after 60-h fed-batch fermentation. The conversion ratio of l-glutamate to GABA reached 0.60–0.74 mol mol⁻¹. By co-expressing gadB1 and gadB2 and optimising the urea addition method, C. glutamicum was genetically improved for de novo biosynthesis of GABA from its own accumulated l-glutamate.

F. Shi and J. Jiang contributed equally to this work.

Introduction

γ-Aminobutyric acid (GABA), a four-carbon non-protein amino acid, is widely distributed in nature from microorganisms to plants and animals [14]. This amino acid is a well-characterised inhibitory neurotransmitter in animals that has various physiological functions, including hypotensive, anti-anxiety, tranquilising, analgesic and diuretic functions [4, 17], and can be applied to treat various neurological disorders [31]. Recent research has indicated that GABA may improve the protein synthesis rate in the brain [29] and suppress carcinomas, such as small airway-derived lung adenocarcinomas [22]. GABA has been considered a bioactive component in the food, feed and pharmaceutical fields. In addition to its bioactivities beneficial to animals, GABA functions as an important component in the acid resistance systems of certain bacteria, such as Escherichia coli [28], Lactococcus lactis [21] and Listeria monocytogenes [9], and is required for the normal oxidative stress tolerance in Saccharomyces cerevisiae [2]. GABA is synthesised by glutamate decarboxylase (GAD), a pyridoxal 5′-phosphate-dependent enzyme that catalyses irreversible α-decarboxylation of l-glutamate to GABA [30]. GAD is widely distributed in eukaryotes and prokaryotes [14]. Because many lactic acid bacteria (LAB) are probiotic to humans and animals, LAB would be the most suitable microorganisms for GAD gene cloning and GABA production [15, 18, 25].

Although GABA production using LAB with higher GAD activity is acceptable for food applications, GABA production requires addition of l-glutamate or monosodium glutamate as the precursor [16, 27] and expensive nitrogen sources for cultivating LAB [15, 35]. This type of production system is not cost-effective for industrial applications. It is important to find a cost-effective and convenient approach for synthesising GABA. Considering that Corynebacterium glutamicum, a non-pathogenic Gram-positive bacterium, overproduces the precursor of GABA, i.e. l-glutamate [34], GABA production by engineered C. glutamicum over-expressing the GAD gene from LAB warrants investigation. C. glutamicum is an important industrial microorganism because of its high productivity of vitamins, l-glutamate and other amino acids [5, 13]. The complete genome of C. glutamicum ATCC 13032 was sequenced by Kalinowski et al. [8] and Ikeda and Nakagawa [7]. The completed deciphering contributed to significant improvements in bioprocessing and downstream technology, as well as in molecular biology [11]. In C. glutamicum ATCC 13032, neither gene encoding GAD was identified, and GABA production was not reported. Recently, by expressing the GAD gene from Lactobacillus brevis Lb85, we succeeded in engineering C. glutamicum ATCC 13032 to produce GABA spontaneously, but production was low (2.15 ± 0.16 g L⁻¹) [23]. A later study indicated that, by expressing the GAD gene from E. coli W3110, C. glutamicum ATCC 13032 was able to produce 12.37 ± 0.88 g L⁻¹ of GABA [26].

Two genes encoding GAD, namely gadB1 and gadB2, were identified in L. brevis Lb85. The transcriptional regulator gene gadR and the l-glutamate/GABA antiporter gene gadC are upstream of gadB2. In our previous research, gadB1, gadB2, gadCB2 and gadRCB2 were introduced into C. glutamicum ATCC 13032 separately by an inducible expression vector (pDXW-8) [23]. The four recombinant strains were able to synthesise GABA using endogenous l-glutamate, with the strain ATCC 13032/pDXW-8-gadB2 yielding a slightly higher production level than the strain ATCC 13032/pDXW-8-gadB1, and the strain ATCC 13032/pDXW-8-gadRCB2 generating the highest production level (2.15 g L⁻¹). Such GABA production was not high, and the production should be induced by isopropyl β-d-1-thiogalactopyranoside (IPTG) because pDXW-8 is an inducible expression vector with a weaker tac promoter and a lacI repressor gene. In this study, to improve GABA production in recombinant C. glutamicum, a constitutive expression strain co-expressing gadB1 and gadB2 was constructed and compared with the gadB1 and gadB2 single-expressing strains. The urea addition strategy during the l-glutamate fermentation stage was researched, and using the optimal urea supplementation method, GABA production reached 27.13 ± 0.54 g L⁻¹ after 120 h of fermentation.

Materials and methods

Strains, media and growth conditions

The bacterial strains and plasmids used in this study are summarised in Table 1. L. brevis Lb85 exhibiting GABA production capacity was used for amplification of gadB1 and gadB2 genes. L. brevis was grown under static conditions in de Man–Rogosa–Sharpe (MRS) medium (10 g tryptone, 5 g yeast extract, 5 g beef extract, 5 g glucose, 3 g sodium acetate, 2 g citric acid diamine, 2 g K₂HPO₄·3H₂O, 0.1 g MgSO₄, 0.04 g MnSO₄ L⁻¹) at 30 °C. E. coli DH5α was used as the host for constructing and propagating the plasmids. E. coli BL21(DE3) was used as the host for expressing the gadB1 and gadB2 genes derived from L. brevis Lb85. E. coli was grown in Luria–Bertani (LB) medium (10 g tryptone, 5 g yeast extract, 10 g NaCl L⁻¹, pH 7.2) at 37 °C and 200 rpm with 50 mg L⁻¹ kanamycin, as required. C. glutamicum ATCC 13032, an l-glutamate-producing strain, was used to express the GAD genes derived from L. brevis Lb85 for GABA production. C. glutamicum was grown in LBG medium (LB supplemented with 5 g L⁻¹ glucose) at 200 rpm and 30 °C with 30 mg L⁻¹ kanamycin, as required.

Bacterial strains and plasmids used in this study

Strain or plasmid	Description	Source
Strains
L. brevis Lb85	L. brevis capable of producing GABA	[23]
E. coli DH5α	Wild-type E. coli	Novagen
DH5α/pDXW-10-gadB1	DH5α harbouring pDXW-10-gadB1	This work
DH5α/pDXW-10-gadB2	DH5α harbouring pDXW-10-gadB2	This work
DH5α/pDXW-10-gadB1–gadB2	DH5α harbouring pDXW-10-gadB1–gadB2	This work
DH5α/pDXW-10-gadRCB2	DH5α harbouring pDXW-10-gadRCB2	This work
E. coli BL21(DE3)	E. coli strain for expressing target gene(s)	Novagen
BL21/pET28a-gadB1	BL21(DE3) harbouring pET28a-gadB1	This work
BL21/pET28a-gadB2	BL21(DE3) harbouring pET28a-gadB2	This work
C. glutamicum ATCC 13032	Wild-type C. glutamicum	ATCC
ATCC 13032/pDXW-10	ATCC 13032 harbouring pDXW-10	This study
ATCC 13032/pDXW-10-gadB1	ATCC 13032 harbouring pDXW-10-gadB1	This work
ATCC 13032/pDXW-10-gadB2	ATCC 13032 harbouring pDXW-10-gadB2	This work
ATCC 13032/pDXW-10-gadB1–gadB2	ATCC 13032 harbouring pDXW-10-gadB1–gadB2	This work
ATCC 13032/pDXW-10-gadRCB2	ATCC 13032 harbouring pDXW-10-gadRCB2	This work
Plasmids
pDXW-10	A shuttle vector between E. coli and Corynebacterium	[33]
pDXW-10-gadB1	pDXW-10 harbouring gadB1	This work
pDXW-10-gadB2	pDXW-10 harbouring gadB2	This work
pDXW-10-gadB1–gadB2	pDXW-10 harbouring gadB1–gadB2	This work
pDXW-10-gadRCB2	pDXW-10 harbouring gadRCB2	This work
pET-28a(+)	E. coli expression vector, Kanr	Novagen
pET28a-gadB1	pET-28a(+) harbouring gadB1	This work
pET28a-gadB2	pET-28a(+) harbouring gadB2	This work

Construction of gadB1- and gadB2-expressing plasmids and strains

For expression of gadB genes in C. glutamicum, gadB1, gadB2 and gadB2 plus the upstream l-glutamate/GABA antiporter gene gadC and the transcriptional regulator gene gadR (gadRCB2) were amplified from the genomic DNA of L. brevis Lb85. The primers were designed using the L. brevis Lb85 gadB1 and gadRCB2 gene sequence [23] and are listed in Table 2. The 1,407-bp gadB1 gene was amplified using the gadB1(+) and gadB1(−) primer pair and ligated into the EcoRI and NotI restriction sites of the vector pDXW-10 to produce the plasmid pDXW-10-gadB1. pDXW-10 is a shuttle expression vector carrying a stronger tac-M promoter for cloning between E. coli and Corynebacterium [33]. The 1,440-bp gadB2 gene was amplified using the gadB2(+) and gadB2(−) primer pair and ligated into the NheI and PstI restriction sites of pDXW-10 to produce the plasmid pDXW-10-gadB2. Subsequently, the gadB2 gene with the tac-M promoter sequence was amplified from the plasmid pDXW-10-gadB2 using the tac(+) and gadB2(−) primer pair. The tac-M-gadB2 amplification product was ligated into the NotI and PstI restriction sites of the plasmid pDXW-10-gadB1 to produce the co-expression plasmid pDXW-10-gadB1–gadB2. The 3,785-bp gadRCB2 gene sequence was amplified using the gadRCB2F and gadB2R primer pair and ligated into the NheI and HindIII restriction sites of pDXW-10 to produce the plasmid pDXW-10-gadRCB2. All the plasmids transferred into competent E. coli DH5α cells were confirmed by restriction enzyme digestion and target gene amplification.

Primers for amplifying GAD genes from L. brevis

The restriction sites are in boldface. The sequences corresponding to the genomic DNA of L. brevis are underlined, and the start and stop codons are boxed. The Shine–Dalgarno sequences are in italics

The constructed plasmids pDXW-10-gadB1, pDXW-10-gadB2, pDXW-10-gadB1–gadB2 and pDXW-10-gadRCB2 were isolated from E. coli and transformed into electrocompetent C. glutamicum ATCC 13032 according to the protocol of Xu et al. [32], yielding gadB1-expressing strain C. glutamicum ATCC 13032/pDXW-10-gadB1, gadB2-expressing strain C. glutamicum ATCC 13032/pDXW-10-gadB2, gadB1–gadB2 co-expressing strain C. glutamicum ATCC 13032/pDXW-10-gadB1–gadB2 and gadRCB2-expressing strain C. glutamicum ATCC 13032/pDXW-10-gadRCB2. All plasmids from the positive transformants were verified by target gene amplification.

For expression of gadB genes in E. coli, gadB1 was amplified from the genomic DNA of L. brevis Lb85 using the gadB1(E+) and gadB1(E−) primer pair and ligated into the NdeI and EcoRI restriction sites of the vector pET-28a(+) to produce the plasmid pET28a-gadB1. Similarly, gadB2 was amplified from the genomic DNA of L. brevis Lb85 using the gadB2(E+) and gadB2(E−) primer pair and ligated into the NheI and SacI restriction sites of the vector pET-28a(+) to produce the plasmid pET28a-gadB2. The plasmids pET28a-gadB1 and pET28a-gadB2 were transformed into E. coli BL21(DE3), yielding E. coli BL21/pET28a-gadB1 and BL21/pET28a-gadB2 strains. The use of pET-28a(+) enabled us to fuse a 6xHis tag to the N-terminal of the GadB1 and GadB2 proteins and allowed IPTG-inducible expression of gadB1 and gadB2 genes in E. coli BL21(DE3).

Expression of gadB1 and gadB2 in E. coli and purification of GadB1 and GadB2

The recombinant E. coli BL21/pET28a-gadB1 and BL21/pET28a-gadB2 were cultivated for 12 h, transferred to 400 mL LB broth, and grown at 37 °C until the optical density (OD₆₀₀) reached 0.6. IPTG was added to final concentration of 1 mM, and cultivation was continued at 30 °C and 200 rpm for 5 h. The cells were collected, washed, and resuspended in lytic buffer [300 mM NaCl, 10 mM imidazole, 50 mM phosphate buffer saline (pH 8.0), 1 mM phenylmethylsulphonyl fluoride (PMSF)] to wet cell concentration of 0.1 g mL⁻¹ and disrupted using sonication. A total of 10 mg mL⁻¹ Triton X-100 was added to the sonicated cell suspension. The GadB1 protein was extracted, purified by nickel-chelating affinity chromatography, and dissolved in 20 mM sodium acetate buffer (pH 4.6).

GABA fermentation by recombinant C. glutamicum in shake flask

For GABA production in shake flask, recombinant C. glutamicum cells were precultured in seed medium (25 g glucose, 30 g corn steep liquor, 8 g urea, 1 g K₂HPO₄·3H₂O, 0.2 g MgSO₄ L⁻¹, pH 7.0–7.2) supplemented with 30 mg L⁻¹ kanamycin at 30 °C and 110 rpm for 7 h. The preculture broth was inoculated into 20 mL fermentation medium containing 30 mg L⁻¹ kanamycin in a 500-mL Erlenmeyer flask to final optical density (OD₅₆₂) of 2.0 and shaken by a reciprocating shaker at 30 °C and 110 rpm for 84 h. The fermentation medium contained 100 g glucose, 4 g corn steep liquor, 4 g urea, 2 g K₂HPO₄·3H₂O, 0.4 g MgSO₄, 0.2 g MnSO₄ and 0.29 g FeSO₄·7H₂O L⁻¹, and was initially adjusted to pH 7.0–7.2. Generally, at 10, 17 and 24 h of fermentation, 4 g L⁻¹ urea was added to the culture to maintain the neutral condition; this fermentation method is defined as the general fermentation method.

To determine the effect of the urea supplementation method on GABA production, four methods of urea addition were carried out during the first 24 h of fermentation, retaining the initial urea concentration at 4 g L⁻¹ and the total urea amount at 16 g L⁻¹. In three methods, urea was added during 10–24 h of fermentation three times (4 g L⁻¹ every 7 h, denoted as 4 + 3 × 4), four times (3 g L⁻¹ every 4.5 h, denoted as 4 + 4 × 3) or five times (2.4 g L⁻¹ every 3.5 h, denoted as 4 + 5 × 2.4). In the fourth method, urea was added to the broth six times during 6–24 h of fermentation (2 g L⁻¹ every 3.5 h, denoted as 4 + 6 × 2). One-time addition of urea (12 g L⁻¹) at 10 h of cultivation (denoted as 4 + 1 × 12) and addition of the total amount of urea (16 g L⁻¹) at the starting point of cultivation (denoted as 16 + 0) were conducted as control methods. Fermentation was performed on a reciprocating shaker at 110 rpm and 30 °C.

GABA fermentation by recombinant C. glutamicum ATCC 13032/pDXW-10-gadB1–gadB2 in fermentor

For GABA production in the fermentor, 100 mL seed culture was prepared in flasks at 30 °C for 8 h and transferred to a 3-L fermentor (BioFlo 110; New Brunswick Scientific, USA) containing 1.2 L fermentation medium. The temperature and aeration rate were kept at 31 °C and 1.5 vvm, respectively. The dissolved oxygen level was controlled at 20 % in the first 10 h and 30 % thereafter by coupling with the agitation speed. During the 8–22 h of fermentation, the pH was controlled at 7.5–8.0 by automatically adding 40 % urea solution, whereas after 36 h, the pH was adjusted to 4.8–5.3 by adding 6 M HCl. Samples were taken every 4 h to determine the OD₅₆₂ value and residual glucose. From 28 h of fermentation, glucose was fed by a peristaltic pump when the residual glucose in the medium was lower than 20 g L⁻¹.

Determination of pH level as well as extracellular and intracellular l-glutamate and GABA concentrations during fermentation

During fermentation, an appropriate volume of culture broth was harvested and centrifuged at 12,000 rpm and 4 °C for 5 min. The supernatant was used for measuring the pH as well as the l-glutamate and GABA concentrations in the fermentation broth. The precipitate was washed twice and used to determine the intracellular l-glutamate and GABA concentrations. pH was measured directly by a pH electrode (Mettler-Toledo, Germany). The l-glutamate and GABA concentrations were assayed using reversed-phase high-pressure liquid chromatography (HPLC, 1200 series; Agilent Technologies, USA). The cell concentration was determined by measuring the OD₅₆₂ value with a UV-1800 spectrophotometer (Shimadzu, Japan). To determine the intracellular l-glutamate and GABA concentrations, the washed precipitate was resuspended in an identical volume of phosphate buffer (20 mM, pH 7.2) and boiled for 15 min; the cell debris was discarded by centrifugation at 12,000 rpm for 10 min [20], and the supernatant was used for HPLC analysis. Before analysis, the supernatant sample was treated with trichloroacetic acid to precipitate proteins. The clear filtrate was used directly for HPLC assay. HPLC separation and quantisation were performed on a Thermo ODS C₁₈ column (250 mm × 4.0 mm, USA) by the o-phthalaldehyde pre-column derivatisation method [12]. The l-glutamate and GABA were detected at 338 nm, and the spectra were recorded online. The intracellular concentration was calculated with the intracellular volume of 1.6 mL g⁻¹ dry weight [10]. The conversion ratio of l-glutamate to GABA was calculated as the mol of GABA divided by the total mol of l-glutamate and GABA.

Extraction of crude enzyme from recombinant C. glutamicum and determination of GAD activity

The cells in the fermentation broth were harvested, washed once with chilled distilled water, and suspended in phosphate-buffered saline (20 mM, pH 7.2). After incubation with 1 g L⁻¹ lysozyme at 37 °C for 2 h, the cells were disrupted by ultrasonication at 4 °C. The supernatant (cell extract) was recovered and designated as crude enzyme after lysed cell debris was removed by centrifugation. The protein concentration of the enzyme solution was determined according to the Bradford method, with bovine serum albumin as standard.

Enzyme solution (<0.1 mL) was added to a reaction mixture containing 0.4 mL substrate buffer (0.2 M Na₂HPO₄–citric acid buffer, pH 4.8 and 10 mM sodium glutamate) and incubated at 37 °C for 60 min. The reaction was terminated by immediately boiling for 10 min. After centrifugation at 12,000 rpm for 10 min, the amount of GABA in the supernatant was determined by HPLC. One unit (U) of enzyme activity was defined as 1.0 μmol GABA produced in 1 min at 37 °C in the initial reaction mixture (0.5 mL), and the specific activity was expressed as U mg⁻¹ or U g⁻¹ of protein. The optimum pH for purified GadB1 was assayed over a pH range of 3.6–6.2, and 0.2 M sodium acetate buffer was used instead of 0.2 M Na₂HPO₄–citric acid buffer, pH 4.8.

Results

To improve GABA production in recombinant C. glutamicum, the constitutive expression vector pDXW-10, with a stronger tac-M promoter, was used for the expression of L. brevis Lb85 gadB genes in C. glutamicum, and the production of GABA was examined.

GABA production by recombinant C. glutamicum strains

First, GABA production was compared between C. glutamicum strains expressing only gadB1, only gadB2, and gadRCB2, and co-expressing gadB1–gadB2. These recombinant strains were fermented by the general fermentation method. After 84 h of fermentation, the strain ATCC 13032/pDXW-10-gadB1–gadB2 could accumulate 4.02 ± 0.95 g L⁻¹ of GABA in the fermented broth, higher than the strains ATCC 13032/pDXW-10-gadB1 (0.69 ± 0.40 g L⁻¹), ATCC 13032/pDXW-10-gadB2 (0.74 ± 0.16 g L⁻¹) and ATCC 13032/pDXW-10-gadRCB2 (2.41 ± 1.33 g L⁻¹); no GABA was detected in the control strain, ATCC 13032/pDXW-10. This result indicated that L. brevis Lb85 gadB genes were expressed and exhibited glutamate decarboxylase activity in recombinant C. glutamicum by pDXW-10, and the gadB1–gadB2 co-expression exhibited better activity. The ATCC 13032/pDXW-10-gadB1–gadB2 strain was used as the predominant strain for GABA production.

Effect of urea supplementation method on GABA production

l-Glutamate is the precursor of GABA. Urea supplementation has a great influence on the biosynthesis of l-glutamate because urea is not only the nitrogen source but also the basic reagent required for maintaining the neutral condition for l-glutamate biosynthesis in C. glutamicum. The l-glutamate dehydrogenase, which is critical for l-glutamate biosynthesis, exhibits the highest activity at pH 7.5 [24], whereas at pH 6.5, half of its activity is lost. Urea was supplemented during 10–24 h of fermentation because l-glutamate is synthesised within 36 h of fermentation. The accumulated l-glutamate can then be converted into GABA by the expressed GAD protein. To understand the possible influence of urea supplementation on pH variation, GAD expression and GABA production, four types of urea addition method were compared for recombinant C. glutamicum ATCC 13032/pDXW-10-gadB1–gadB2. The total addition of urea was maintained at 12 g L⁻¹ in all four methods. One-time addition of urea (12 g L⁻¹) at 10 h of cultivation and addition of the total amount of urea (16 g L⁻¹) at the cultivation starting point were conducted as control methods.

Initially, when urea was supplemented three times, with 4 g L⁻¹ each time (denoted as 4 + 3 × 4), the fermentation broth pH level decreased to 5.67 at 24 h just before the last supplementation, the GAD activity of the crude cell extract was 114.3 U g⁻¹ at 60 h of fermentation, and after 84 h, the l-glutamate and GABA concentration in the fermentation broth reached 18.55 ± 2.19 and 4.02 ± 0.95 g L⁻¹, respectively (Fig. 1). When urea was added four times, with 3 g L⁻¹ each time (denoted as 4 + 4 × 3), the fermentation broth pH level remained at 7.33 at 24 h, the GAD activity of the crude cell extract increased to 142.0 U g⁻¹ at 60 h, and after 84 h, the l-glutamate and GABA production increased to 21.55 ± 2.57 and 7.70 ± 0.95 g L⁻¹, respectively. When urea was supplemented five times (2.4 g L⁻¹ each time, denoted as 4 + 5 × 2.4) and six times (2 g L⁻¹ each time, denoted as 4 + 6 × 2), the fermentation broth pH level was 7.13 and 6.26, respectively, at 24 h, the GAD activity of the crude cell extract increased to 164.6 and 340.7 U g⁻¹, respectively, at 60 h, and after 84 h, the l-glutamate concentration decreased to 13.10 ± 1.99 and 11.51 ± 1.23 g L⁻¹, respectively, whereas the GABA concentration increased to 15.76 ± 1.55 and 18.66 ± 2.41 g L⁻¹, respectively. In the control methods, where 12 g L⁻¹ of urea was added in one time at 10 h of cultivation (denoted as 4 + 1 × 12) and all the 16 g L⁻¹ of urea was added at the cultivation starting point (denoted as 16 + 0), the fermentation broth pH level decreased to 5.37 and 5.30, respectively, at 24 h, the GAD activity of the crude cell extract decreased to 95.6 and 42.4 U g⁻¹, respectively, at 60 h, and after 84 h, the l-glutamate production was 10.74 ± 1.23 and 5.81 ± 0.30 g L⁻¹, respectively, and the GABA production was 1.76 ± 0.18 and 0.62 ± 0.05 g L⁻¹, respectively (Fig. 1), being much lower than the values produced using the urea supplementation methods with three to six additions. The more frequently urea was supplemented, the higher the GAD activity and GABA conversion ratio obtained, and the more GABA was produced. The conversion ratio of l-glutamate to GABA increased to 0.70 mol mol⁻¹, and the GABA productivity increased by 3.6-fold compared with the initial fermentation method (4 + 3 × 4). Addition of urea six times with 2 g L⁻¹ each time was adopted in subsequent fermentation.

Fig. 1

Effect of urea supplementation method on l-glutamate and GABA production by recombinant C. glutamicum ATCC 13032/pDXW-10-gadB1–gadB2. Cells were precultured in seed medium for 7 h at 30 °C and 110 rpm, inoculated into fermentation medium, and cultivated at 30 °C and 110 rpm for 84 h. By three methods, urea was supplemented during 10–24 h of fermentation in three times (4 g L⁻¹ every 7 h, denoted as 4 + 3 × 4), four times (3 g L⁻¹ every 4.5 h, denoted as 4 + 4 × 3) or five times (2.4 g L⁻¹ every 3.5 h, denoted as 4 + 5 × 2.4). In the fourth method, urea was supplemented during 6–24 h of fermentation in six times (2 g L⁻¹ every 3.5 h, denoted as 4 + 6 × 2). In the control methods, one-time addition of urea at 10 h of cultivation (12 g L⁻¹, denoted as 4 + 1 × 12) and addition of the total amount of urea at the starting point of cultivation (16 g L⁻¹, denoted as 16 + 0) were conducted. After cultivation, l-glutamate (white bars) and GABA (black bars) concentrations in the fermented broth were analysed. GAD activities from crude cell extracts of 60-h cultivations (grey bars) were assayed. Averages of three independent experiments are provided

Time course of GABA fermentation in recombinant C. glutamicum strains

In our previous study, we found that l-glutamate was generally synthesised within 36 h of fermentation, whereas GABA began to accumulate after 36 h and increased continuously thereafter [23]; thus 0–36 h can be regarded as the l-glutamate fermentation stage, and after 36 h can be regarded as the GABA fermentation stage. At the GABA fermentation stage, GAD must be activated. GAD can only exhibit activity in a relatively acid condition because the purified GadB1 protein after being expressed in E. coli BL21(DE3) exhibited optimum activity at pH 4.6, whereas at pH 5.8, approximately 85 % of its activity was lost (data not shown). GadB2 was expressed as an inclusion body in E. coli BL21(DE3). An additional GadB2 protein from L. brevis IFO12005 was reported to exhibit optimum activity at pH 4.5–5.0 [6]; at pH 5.8, approximately 80 % of its activity was lost. Apart from the l-glutamate fermentation stage, the pH level in the GABA fermentation stage will influence GABA biosynthesis. At the same time, the accumulation of l-glutamate and the conversion of l-glutamate to GABA will change the pH level. To elucidate the relationship between GABA biosynthesis and pH fluctuation, the time course of l-glutamate concentration, GABA concentration and extracellular pH level at the GABA fermentation stage were investigated.

During the fermentation of recombinant C. glutamicum ATCC 13032/pDXW-10-gadB1–gadB2, the l-glutamate concentration reached the highest values (27.55–28.17 g L⁻¹) at 36–48 h and continuously reduced thereafter, whereas GABA began to be synthesised before 36 h, and its concentration continuously increased (Fig. 2a). The GAD activity of the crude cell extracts at pH 4.8 remained at 172–341 U g⁻¹ during 24–84 h of fermentation (Table 3), with the activity at 60 h being the highest (340.7 ± 4.0 U g⁻¹). The expressed GadB1 and GadB2 proteins were not visible in the crude cell extract of fermentation broth by sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE). After 84 h of fermentation, the l-glutamate concentration decreased to 15.19 ± 0.71 g L⁻¹, whereas the GABA concentration increased to 19.38 ± 0.17 g L⁻¹. The conversion ratio of l-glutamate to GABA reached 0.65 mol mol⁻¹, and the total amount of l-glutamate and GABA reached 0.29 M. The strain grew rapidly and consumed much more glucose (75.5 g L⁻¹) before 36 h of fermentation, indicating that the carbon source glucose was mainly required at the l-glutamate fermentation stage. After 36 h, the strain grew slowly and consumed little glucose (Fig. 2b). During this GABA fermentation stage, 84 mM of extracellular l-glutamate was reduced, but 159 mM of GABA was formed, indicating that biosynthesis of l-glutamate did not cease after 36 h. With the conversion of l-glutamate to GABA, the pH level of the fermentation broth changed. It was lowest at 36–48 h (pH 5.56–5.53), in accordance with the highest l-glutamate and lowest GABA concentration at this period, then increased quickly until 60 h (pH 6.24) and slowly thereafter, in accordance with the decrease of the l-glutamate and increase of the GABA concentration at this period, because l-glutamate is an acidic [isoelectric point (pI) 3.22] molecule and GABA is a neutral (pI 7.19) molecule. Because the time course of the extracellular pH level during the GABA fermentation stage was in accordance with the conversion of l-glutamate to GABA, the pH level could be used as an indicator to reflect GABA synthesis in recombinant C. glutamicum.

Fig. 2

Time courses of GABA production during the GABA fermentation stage by recombinant C. glutamicum ATCC 13032/pDXW-10-gadB1–gadB2: a extracellular l-glutamate concentration (solid line with squares), GABA concentration (solid line with triangles) and pH level (dotted line with diamonds); b cell growth (squares) and glucose consumption (triangles); c intracellular l-glutamate concentration (squares) and GABA concentration (triangles). Cells were cultivated as described in Fig. 1 by supplementing urea six times. At each time point, the fermentation broth was collected and assayed for l-glutamate, GABA, pH, OD₅₆₂ and glucose. Each point represents the average of three independent experiments

 

GAD activity from crude cell extract of recombinant C. glutamicum

GAD activity (U g−1)	24 h	36 h	48 h	60 h	72 h	84 h
ATCC 13032/pDXW-10-gadB1–gadB2	171.6 ± 3.5	188.3 ± 2.9	259.1 ± 6.4	340.7 ± 4.0	256.6 ± 7.1	210.7 ± 2.9
ATCC 13032/pDXW-10-gadB2	151.7 ± 2.6	172.1 ± 4.1	186.3 ± 3.7	194.2 ± 5.6	136.7 ± 2.3	130.9 ± 3.1

Cells cultivated in fermentation medium at 30 °C for a certain time were collected, and GAD activities from the crude cell extracts were assayed as described in “Materials and methods” section. Averages of three independent experiments are shown

Continuous accumulation of extracellular GABA might be observed because of delayed export of intracellular GABA synthesised previously. The intracellular concentrations of l-glutamate and GABA during fermentation were therefore determined. In C. glutamicum ATCC 13032/pDXW-10-gadB1–gadB2, the intracellular concentration of l-glutamate decreased during 36–60 h, similar to the extracellular concentration, but increased obviously thereafter (Fig. 2c). The relatively higher concentration of intracellular l-glutamate than extracellular l-glutamate indicates that such retarded export of l-glutamate would occur. The intracellular concentration of GABA continuously increased during 36–60 h, suddenly declined at 72 h, and later increased to 15.14 ± 0.25 g L⁻¹. The nearly identical level of intracellular and extracellular GABA concentrations indicated that the synthesised GABA could be transported out of cells almost simultaneously.

During the fermentation of the control strain C. glutamicum ATCC 13032/pDXW-10, the l-glutamate concentration in the fermentation broth reached a similar level (27.49 ± 0.40 g L⁻¹) at 36 h, but increased slightly thereafter to 33.05 ± 0.10 g L⁻¹, whereas the GABA concentration could not be detected at any time (Fig. 3a). The pH level of the fermentation broth decreased slightly from 5.41 to 5.29 over 36–72 h and remained at 5.29–5.31. This strain exhibited a similar growth rate and glucose consumption pattern (Fig. 3b) to ATCC 13032/pDXW-10-gadB1–gadB2. The intracellular l-glutamate reached the highest concentration (84.57 ± 5.01 g L⁻¹) at 36 h and decreased continuously to approximately 50 g L⁻¹ (Fig. 3c), which was consistently much higher than the intracellular l-glutamate concentration of the gadB1–gadB2 co-expressing strain (Fig. 2c), most likely because of the continuous conversion of l-glutamate to GABA in the gadB1–gadB2 co-expressing strain. This concentration was consistently much higher than the extracellular l-glutamate concentration, especially early on (36 h), demonstrating the retarded export of l-glutamate. Simultaneously, the intracellular GABA concentration could not be detected at any time, as was the case for the extracellular GABA concentration (Fig. 3a), indicating that the control strain could not produce GABA.

Fig. 3

Time courses of l-glutamate production during the later period of fermentation by the control strain C. glutamicum ATCC 13032/pDXW-10: a extracellular l-glutamate concentration (solid line with squares), GABA concentration (solid line with triangles) and pH level (dotted line with diamonds); b cell growth (squares) and glucose consumption (triangles); c intracellular l-glutamate concentration (squares) and GABA concentration (triangles). Cells were cultivated, and fermentation broth was collected and assayed as described in Fig. 2. Each point represents the average of three independent experiments

The progression of the l-glutamate and GABA concentrations and the fermentation broth pH level of recombinant C. glutamicum ATCC 13032/pDXW-10-gadB2 was similar to that of ATCC 13032/pDXW-10-gadB1–gadB2, but the variation range was much smaller (Fig. 4a). Its growth pattern and glucose consumption pattern (Fig. 4b) were similar to those of ATCC 13032/pDXW-10-gadB1–gadB2. The l-glutamate concentration decreased to 18.57 ± 1.54 g L⁻¹, and the GABA concentration increased to 11.16 ± 0.68 g L⁻¹. The conversion ratio of l-glutamate to GABA reached 0.46 mol mol⁻¹, a much lower level than that in the gadB1–gadB2 co-expressing strain, demonstrating that the gadB1–gadB2 co-expressing strain is more efficient for single-step production of GABA. The GAD activity from the crude cell extracts remained at 131–194 U g⁻¹ during 24–84 h of fermentation, which was lower than that of the ATCC 13032/pDXW-10-gadB1–gadB2 strain (Table 3).

Fig. 4

Time courses of GABA production during the GABA fermentation stage by recombinant C. glutamicum ATCC 13032/pDXW-10-gadB2: a extracellular l-glutamate concentration (solid line with squares), GABA concentration (solid line with triangles) and pH level (dotted line with diamonds); b cell growth (squares) and glucose consumption (triangles). Cells were cultivated, and fermentation broth was collected and assayed as described in Fig. 2. Each point represents the average of three independent experiments

Prolongation of GABA fermentation time in ATCC 13032/pDXW-10-gadB1–gadB2

After fermentation for 84 h of recombinant C. glutamicum ATCC 13032/pDXW-10-gadB1–gadB2, the GABA concentration reached nearly 20 g L⁻¹ and continued to increase. Prolonged fermentation was performed, and the GABA accumulation was investigated.

During fermentation, this strain completed growth at 36 h but continued to consume 15 g L⁻¹ of glucose thereafter (Fig. 5a). At 120 h, the optical density of the fermentation broth was 34.1 ± 1.3, and approximately 9.5 ± 0.7 g L⁻¹ of glucose was not consumed. The GABA concentration in the fermentation broth increased continuously after 84 h, despite a relatively lower yield (0.216 g L⁻¹ h⁻¹) than before 84 h (0.342 g L⁻¹ h⁻¹) (Fig. 5b). At 120 h, the GABA concentration reached 27.13 ± 0.54 g L⁻¹, and the l-glutamate concentration decreased slightly to 13.27 ± 2.10 g L⁻¹. The conversion ratio of l-glutamate to GABA reached 0.74 mol mol⁻¹, and the total amount of l-glutamate and GABA reached 0.35 M. With extension of the fermentation time, the GABA production increased further.

Fig. 5

GABA production by C. glutamicum ATCC 13032/pDXW-10-gadB1–gadB2 under optimised condition: a cell growth (squares) and glucose consumption (triangles); b extracellular l-glutamate (squares) and GABA (triangles) production. Cells were cultivated, and fermentation broth was collected and assayed as described in Fig. 2. Averages of three independent experiments are provided

GABA production of ATCC 13032/pDXW-10-gadB1–gadB2 in fermentor

C. glutamicum ATCC 13032/pDXW-10-gadB1–gadB2 was then evaluated in the fermentor. During 8–22 h of fermentation, the pH was controlled at 7.5–8.0 by adding 40 % urea solution, whereas after 36 h, the pH was adjusted to 4.8–5.3 by adding 6 M HCl. The cells grew quickly in the first 12 h, and after 24 h, the cell concentration became stable (Fig. 6a). Glucose was consumed quickly in the first 24 h and was fed after 28 h when the residual glucose in the medium was lower than 20 g L⁻¹. There was no GABA production in the first 24 h, which was similar to the shake flask cultivation. The l-glutamate concentration increased to the highest value (43.64 g L⁻¹) at 36 h and reduced continuously thereafter, whereas the GABA concentration increased continuously during 24–60 h, and after 60 h, it became stable (Fig. 6b). At 60 h of fermentation, the GABA concentration in the fermentation broth reached 26.32 g L⁻¹; the total amount of l-glutamate and GABA reached 0.42 M, and the conversion ratio of l-glutamate to GABA reached 0.60 mol mol⁻¹.

Fig. 6

Fed-batch production of GABA by C. glutamicum ATCC 13032/pDXW-10-gadB1–gadB2 in fermentor: a cell growth (squares) and residual glucose (triangles), b  l-glutamate production (solid line with squares), GABA production (solid line with triangles) and pH level (dotted line with diamonds). Cells were cultivated, and fermentation broth was collected and assayed as described in Fig. 2

Discussion

In this study, an enhanced single-step GABA production system using a recombinant C. glutamicum strain co-expressing two glutamate decarboxylase genes (gadB1 and gadB2) was established. Through optimisation of the urea addition method, the GABA concentration increased from 4.02 ± 0.95 to 19.38 ± 0.17 g L⁻¹ after 84 h of fermentation in a shake flask, and after 120 h of fermentation, the production reached 27.13 ± 0.54 g L⁻¹. With cultivation in a fermentor, the GABA concentration reached 26.32 g L⁻¹ after 60 h of fermentation. In a recent study, a recombinant C. glutamicum strain expressing gadB of E. coli W3110 accumulated 12.37 ± 0.88 g L⁻¹ of GABA after 72 h of fermentation but decreased sharply thereafter [26]. Compared with the sole expression of the gadB2 gene of L. brevis Lb85 or the gadB gene of E. coli W3110, co-expression of the gadB1 and gadB2 genes of L. brevis Lb85 contributed significantly to the accumulation of GABA. This accumulation does not require addition of the GABA precursor l-glutamate.

In recombinant C. glutamicum, the fermentation conditions for l-glutamate biosynthesis and GABA conversion are notably different, with the former occurring in a neutral condition [1], and the latter occurring in an acidic condition. Generally, l-glutamate is accumulated first and GABA conversion occurs subsequently. The pH level of the fermentation broth at the first fermentation stage is maintained at approximately 7.0 to ensure l-glutamate biosynthesis. With the accumulation of acidic l-glutamate, the pH level decreases accordingly, thus the urea supplementation strategy is adopted during the first 24 h of fermentation because urea can be hydrolysed to ammonia in C. glutamicum by the abundant intracellular urease. The released ammonia can neutralise the generated acidic l-glutamate and maintain the pH level. After 24 h, with the l-glutamate concentration increasing to nearly its highest level, the pH level is left to decline automatically and GABA biosynthesis can begin.

Considering that urea supplementation has a great influence on the production of l-glutamate, as well as on the variation of the pH level and the resulting production of GABA, various urea supplementation methods were investigated in this study. By adding urea 3–6 times during 6–24 h of fermentation, for a total amount of 12 g L⁻¹, the pH level during the l-glutamate fermentation stage was maintained in a more suitable range. At the same time, the gadB expression was enhanced. At 60 h of fermentation, the in vitro GAD activity of the crude cell extract at pH 4.8 increased to 340.7 ± 4.0 U g⁻¹. After 84 h of fermentation, the total amount of l-glutamate and GABA increased to 30.18 ± 1.33 g L⁻¹, the GABA production increased to 18.66 ± 2.41 g L⁻¹ (Fig. 1), and the conversion ratio of l-glutamate to GABA increased to 0.70 mol mol⁻¹, which was much higher than for the general fermentation method or control methods. This result illustrates that frequent supplementation of urea during the l-glutamate fermentation stage results in an appropriate fluctuation of the pH level and efficient l-glutamate accumulation and subsequent GABA conversion.

At the GABA fermentation stage, the pH level varied regularly, being lowest at the beginning, markedly elevated thereafter, and weakly changed at the end (Figs. 2a, 4a). The variation range in the gadB1–gadB2 co-expressing strain was larger than that in the gadB2 single-expressing strain, in accordance with its higher GAD expression (Table 3) and higher conversion ratio of l-glutamate to GABA (0.65 mol mol⁻¹) and GABA yield (0.342 g L⁻¹ h⁻¹). The pH variation during the GABA fermentation stage can be used as an indicator to reflect the progress of GABA biosynthesis. The varied pH would influence the l-glutamate biosynthesis and GABA conversion. After 108 h, with the increased pH level, l-glutamate began to accumulate again and thus improved the GABA biosynthesis in the gadB1–gadB2 co-expressing strain (Fig. 5a). The co-expression of gadB1 and gadB2 and the sole expression of gadB2 in C. glutamicum indicate the potential for conferring acid resistance to this microorganism because the pH level in the recombinant strains was not as acidic as for the control strain, and after 36 h of fermentation, the pH level increased in the recombinant strains (Figs. 2a, 4a) but decreased in the control strain (Fig. 3a). A novel l-glutamate producer, Pantoea ananatis, which is able to grow in acidic pH, was isolated and sequenced recently [3]. It might be suitable for single-step GABA production following appropriate construction.

In C. glutamicum, the l-glutamate secretion or transport system has a significant effect on the l-glutamate production [19]. The control strain ATCC 13032/pDXW-10 showed retarded export of l-glutamate (Fig. 3). In recombinant C. glutamicum, l-glutamate does not need to be transported from cells so rapidly, because it is converted to GABA by the intracellular enzymes GadB1 and GadB2, and the GABA is secreted. Compared with the extracellular concentrations, the constantly higher level of intracellular l-glutamate concentration and nearly identical level of intracellular GABA concentration (Fig. 2) indicate the retarded export of l-glutamate and almost simultaneous export of GABA in ATCC 13032/pDXW-10-gadB1–gadB2, although a GABA uptake system GabP_Cg should exist in C. glutamicum, as discovered recently by Zhao et al. [36].

In this study, by modifying the urea supplementation strategy, an improved single-step GABA production system was obtained using a recombinant C. glutamicum strain co-expressing gadB1 and gadB2 genes from L. brevis Lb85. This system could effectively and continuously convert its own accumulated l-glutamate into GABA. Considering that the bioprocessing and downstream technology of l-glutamate production by C. glutamicum have matured sufficiently, the recombinant strain has great potential and superiority for industrial application if an l-glutamate-overproducing strain is used as the host bacterium. Because C. glutamicum is a microorganism that is generally recognised to be safe, the de novo GABA fermentation system could be applied for production of GABA as a component of functional foods and pharmaceuticals.

Acknowledgments

The authors thank the Program of State Key Laboratory of Food Science and Technology (contract no. SKLF-TS-201103) and the Fundamental Research Funds for the Central Universities (contract no. JUSRP 21109) for financial support.