Engineered Escherichia coli Nissle 1917 for the High Level Biosynthesis of γ-Aminobutyric Acid

Abstract

γ-Aminobutyric acid (GABA), a vital bioactive component, is biosynthesized via the decarboxylation of L-glutamate (L-Glu) catalyzed by glutamate decarboxylase (GAD). However, the GADs from various sources commonly suffer from low thermal stability, which hampers their industrial applications. In this work, four ancestral sequences of GAD (Anc19, Anc20, Anc28, and Anc30) were designed via an ancestral sequence reconstruction (ASR) approach. Thereafter, the genes were synthesized and heterologously expressed in the probiotic Escherichia coli strain Nissle 1917 (EcN). Among all variants tested, Anc28 exhibited the highest catalytic performance. The K_m and k_cat values were determined to be 26.80 mM and 57.41 s⁻¹, respectively, yielding a catalytic efficiency (k_cat/K_m) of 2.14 s⁻¹mM⁻¹, which was 2.71-fold higher than that of the wild-type enzyme. Meanwhile, compared with the wild-type GAD, Anc28 exhibited a 6.74 °C increase in T₅₀¹⁵ and a 4.1-fold extension in t_1/2 at 60 °C. Furthermore, the GABA synthesis system using dormant Escherichia coli Nissle (T7)/pET28a-gadB_Anc28 cells as the biocatalyst and pure water as a sole medium was also constructed. Upon completion of the 4 h reaction, the GABA titer reached 307.53 g/L with a conversion ratio of 99.36%. The resulting engineered strains were successfully employed for the efficient biosynthesis of GABA.

1. Introduction

γ-Aminobutyric acid (GABA), a primary inhibitory neurotransmitter in the mammalian brain, is recognized as a potent bioactive compound due to its diverse physiological roles, including its hypotensive, diuretic, sedative, immunomodulatory, and anticonvulsant effects [1,2]. In light of these health benefits, various methods for GABA production have been established, such as chemical synthesis, plant enrichment, and biosynthesis [3,4,5]. Among these, biosynthesis stands out as particularly advantageous and appealing, owing to its exceptional catalytic efficiency, mild reaction conditions, and favorable environmental compatibility.

The major pathway for the biosynthesis of GABA is mediated by the intracellular pyridoxal-5′-phosphate (PLP)-dependent glutamate decarboxylase (GAD; EC 4.1.1.15), which catalyzes the α-decarboxylation of L-glutamic acid (L-Glu) to GABA and release of CO₂ as a byproduct [6]. During the past decade, considerable efforts have been devoted to screening or constructing potential strains with high GAD activity and then optimizing the main physicochemical parameters involved in GABA synthesis. In those cases, engineered Bacillus subtilis, Corynebacterium glutamicum, Levilactobacillus brevis, Lactococcus lactis, Lactiplantibacillus plantarum, and Escherichia coli strains were the common whole-cell biocatalysts [7,8,9,10,11]. Among these microorganisms, E. coli strains have been regarded as robust hosts for the heterologous expression of various GADs and have achieved high GABA titers [12,13]. However, compared with the generally recognized as safe (GRAS) microorganisms, the challenges associated with the safety concerns involved in the manufacturing of GABA by these engineered E. coli strains need to be further overcome.

Aside from the appropriate microbial chassis cells, improvement in the enzymatic property of GAD has been regarded as another key determinant for the low-cost biosynthesis of GABA. During the past few years, a large number of microbial GADs have been cloned and characterized. The GADs from different sources exhibit diverse activities, biochemical properties, and molecular weights; however, their limited stability is a common issue that hampers their industrial application [14]. Thus, the design of stable and sophisticated GADs exhibiting catalytic efficiency has attracted more and more attention.

During the past decade, various protein engineering strategies, including site-directed mutagenesis, Ramachandran plot analysis, proline residue incorporation, sequence consensus analysis, and protein surface charge optimization, have been widely used to enhance the thermal stability of GAD enzymes. Typically, researchers have utilized the Consensus Finder server to design consensus mutations aimed at enhancing the thermostability of LbGAD [15]. This strategy resulted in the mutant GadB^T383K, which showed a 2.99 °C increase in T₅₀¹⁵ and a 17.72 min extension of its half-life (t_1/2) at 55 °C. Similarly, researchers have utilized FoldX to identify key residues affecting the unfolding free energy (ΔG_unfold). The resulting engineered variants, GadB^D203E and GadB^S325A, exhibited increases in T₅₀¹⁵ of 2.3 °C and 1.4 °C, respectively [16]. Despite these advances, overcoming the “activity–thermostability trade-off” remains a pivotal challenge in the development of high-performance GAD enzymes.

In recent years, ancestral sequence reconstruction (ASR) has emerged as a novel computational approach that employs maximum likelihood or Bayesian methods for statistical phylogenetic analysis to infer plausible sequences of extinct ancestral proteins within modern protein families. Currently, ASR is increasingly used in protein engineering to discover and design superior industrial enzymes with enhanced thermal stability [17]. For instance, ASR was used to obtain P450 enzymes and ketol-acid reductoisomerase (KARI) capable of withstanding high temperatures and prolonged reaction durations [18]. Among them, the engineered CYP3_N1 variant demonstrated a T₅₀¹⁵ of 66 °C, approximately 30 °C higher than that of existing CYP₃ enzymes. Its t_1/2 values were determined to be 595.4 min at 50 °C and 54.7 min at 60 °C, corresponding to 283.52-fold and 136.75-fold enhancements over CYP3A4, respectively. Meanwhile, the reconstructed ancestral KARI exhibited a T₅₀¹⁵ of 55 °C, which is nearly 8.5 °C higher than that of the E. coli-derived enzyme. Similarly, researchers have applied ASR to engineer a more robust ancestral GshF (Anc427) with a thermal denaturation temperature of 56.2 ± 0.2 °C, representing an increase of 10.8 ± 0.2 °C over the probe enzyme St-GshF [19]. Additionally, Anc427 exhibited significantly improved operational stability, with a t_1/2 of 3465.7 min at 40 °C, a 20-fold extension compared to St-GshF.

As mentioned above, obtaining high-performance GAD enzymes and selecting suitable host cells are prerequisites for the biosynthesis of GABA. To address the above issue, four ancestral sequences of GAD were designed via an ASR approach based on a phylogenetic analysis of extant homologous amino acid sequences. Then, genes encoding the designed sequences were artificially synthesized and expressed in the probiotic Escherichia coli strain Nissle 1917 (EcN) under the control of T7 promoter. Finally, the whole-cell biocatalyst harboring the reconstructed GAD with superior thermal stability was employed for GABA production. A whole-cell catalytic system was established in this study using pure water as the sole medium. The selection of L-Glu over monosodium glutamate (MSG) as the substrate provides a major advantage in simplifying downstream purification. Owing to its low solubility at the isoelectric point (pI ≈ 3.2), unconverted L-Glu can be rapidly removed by precipitation following pH adjustment.

Taken together, this work positions EcN as a promising host for metabolic engineering, opening new avenues and informing strategies to achieve higher yields of bioactive compounds.

2. Materials and Methods

2.1. Bacterial Strains, Plasmids and Growth Conditions

The bacterial strains and plasmids utilized in this study are listed in

Table 1. Bacterial strains and plasmids used in this work.

Strains/Plasmids	Characteristics	Source
Strains
DH5α	F− endA1 glnV44 thi-1 recA1 relA1 gyrA96 deoR nupG purB20 φ80dlacZΔM15 Δ(lacZYA-argF)U169, hsdR17(rK−mK+), λ−	Takara
EcN (T7)	Serotype O6:K5:H1 derivate, insertion of T7-RNA polymerase gene with lacUV5 promoter by deletion of malEFG operon	Lab stock
EcN (T7)/pET28a-gadB	EcN (T7) strain harboring pET28a-gadB	This work
EcN (T7)/pET28a-gadBAnc19	EcN (T7) strain harboring pET28a-gadBAnc19	This work
EcN (T7)/pET28a-gadBAnc20	EcN (T7) strain harboring pET28a-gadBAnc20	This work
EcN (T7)/pET28a-gadBAnc28	EcN (T7) strain harboring pET28a-gadBAnc28	This work
EcN (T7)/pET28a-gadBAnc30	EcN (T7) strain harboring pET28a-gadBAnc30	This work
Plasmids
pET-28a (+)	Expression vector, Kan+	Novagen
pET28a-gadB	The gadB segment in vector pET-28a	Lab stock
pET28a-gadBAnc19	The gadBAnc19 segment in vector pET-28a	This work
pET28a-gadBAnc20	The gadBAnc20 segment in vector pET-28a	This work
pET28a-gadBAnc28	The gadBAnc28 segment in vector pET-28a	This work
pET28a-gadBAnc30	The gadBAnc30 segment in vector pET-28a	This work

. E. coli DH5α (Takara, Beijing, China) and EcN (T7) were cultured aerobically at 37 °C in Luria-Bertani (LB) broth with shaking at 200 rpm. The corresponding solid medium was prepared by supplementing the liquid broth with 1.5% (w/v) agar. For plasmid maintenance, kanamycin (Sangon Biotech, Shanghai, China) was added to the medium at a final concentration of 50 μg/mL when necessary.

2.2. Analysis of Conserved Amino Acid Residue Sites in GAD

The Protein Contacts Atlas platform, a comprehensive and interactive resource for multiscale investigation of non-covalent biomolecular interactions, was applied to elucidate structure–function links [20]. The analytical procedure includes obtaining crystallographic data of the target protein from the Protein Data Bank (PDB), followed by secondary structure determination and calculation of interatomic distances and solvent-accessible surface area using the DSSP algorithm. The processed data were then exported as JSON-format files to enable interactive analysis and support the identification of conserved amino acid residues. In this investigation, the GAD from Levilactobacillus brevis (PDB ID: 5GP4) served as the structural template. Automated analysis was performed utilizing the Protein Contacts Atlas technology to discover conserved amino acid residues linked with GAD catalytic activity.

2.3. Construction of Phylogenetic Tree for GAD Ancestral Enzymes

A phylogenetic tree of ancestral GAD was built using the FireProt^ASR ancestral sequence computing method (https://loschmidt.chemi.muni.cz/fireprotasr/, accessed on 13 May 2025), which is based on a sequence reconstruction strategy that incorporates the freezing of active-site residues [21]. The query sequence used for the initial homology search was the GAD from Levilactobacillus brevis CGMCC 1306 (PDB ID: 5GP4, NCBI NR: WP_011668532). The approach was accomplished as follows: initial homologous sequences were gathered from the NCBI non-redundant protein database (NCBI NR) using the Enzyme Miner tool v2.0, followed by two rounds of Position-Specific Iterated BLAST (PSI-BLAST, v2.16.0) to retrieve sequences having conserved catalytic residues. Abnormal sequences were filtered out using a length-based filter, and gaps in the matched sequences were treated with the ancestral gap reconstruction technique, which is based on locally weighted regression. Finally, the phylogenetic tree was created utilizing PAML, with 1000 bootstrap replicates to assess node support. Bootstrap values ≥ 70% are shown on the branches of the phylogenetic tree. Internal nodes represent the inferred ancestral enzyme sequences.

2.4. Engineering of the EcN (T7) Strain

The designed AncGAD gene segments were synthesized by Anhui General Biology Company (Anhui, China), cloned into the pET28a(+) vector (Novagen, Nanjing, China) at the Nde I/Sal I (TransGen Biotech, Beijing, China) cleavage sites, and then converted into EcN (T7) competent cells by the electroporation technique. The recombinant plasmid pET28a-gadB was transformed into EcN (T7) competent cells using the same method, thereby generating the strain designated as the control strain.

2.5. Expression and Purification of AncGADs

The engineered EcN (T7) cells were cultured at 37 °C in LB broth containing 50 μg/mL kanamycin with shaking at 200 rpm. After the culture reached an OD₆₀₀ of approximately 0.6, 0.5 mM of IPTG (Sangon Biotech, Shanghai, China) was added to induce GAD overexpression at 30 °C and 180 rpm for 7–8 h. After induction, cells were extracted by centrifugation at 8000× g for 15 min, and the pellets were resuspended in 50 mM sodium phosphate buffer (pH 7.4). Recombinant AncGAD proteins carrying an N-terminal hexahistidine tag were liberated by high-pressure homogenization (AH100B, ATS Industrial Systems Ltd., Shanghai, China) and subsequently purified via Ni²⁺-affinity chromatography.

2.6. Characterization of the Biochemical Properties of AncGADs

To determine the optimum pH of AncGAD, 20 μL of the purified enzyme was mixed with 480 μL of a substrate solution containing 10 μmol/L PLP and 100 mmol/L L-Glu in either 0.2 mol/L sodium acetate–acetic acid buffer (pH 3.0–5.6) or 0.1 mol/L sodium phosphate buffer (pH 5.9–7.0). The reaction mixture was incubated at 37 °C for 5 min and then quenched by the addition of 0.2 mol/L NaHCO₃ solution (pH 9.8). The influence of temperature on enzyme activity was examined in a pH 5.0 reaction system (0.2 mol/L sodium acetate–acetic acid buffer containing 10 μmol/L PLP and 100 mmol/L L-Glu) by incubating the mixture at temperatures ranging from 20 °C to 80 °C for 5 min.

The kinetic constants of WT and AncGADs were determined at pH 5.0 and 37 °C by measuring the initial reaction rates at different L-Glu concentrations ranging from 10 to 200 mM. The purified enzyme concentrations were 0.5–1.5 mg/mL. The kinetic parameters $K_m$ and $V_{max}$ were determined by fitting the initial-rate data to the Michaelis–Menten equation. The $k_{cat}$ values were calculated from $V_{max}$ using the molar concentration of the purified enzyme and the predicted molecular weight of each GAD variant. All measurements were conducted in triplicate.

All reagents used in this Section 2.6 were purchased from Sangon Biotech Co., Ltd. (Shanghai, China).

2.7. Molecular Dynamics (MD) Simulation

Three-dimensional structures of wild-type LbGAD (WT) and AncGADs were predicted via AlphaFold2, followed by energy minimization using Discovery Studio 2020. Molecular docking was subsequently performed using AutoDock 4.2.6. Molecular dynamics (MD) simulations were then conducted for 100 ns at 310 K under the Amber 22 force field. The hydrogen-added structure was embedded within a 10 Å × 10 Å × 10 Å cubic box, solvated with water molecules at a density of 0.98 g/L, and charge-neutralized by introducing suitable counterions (Na⁺/Cl⁻). To compare the energy fluctuations and local flexibility between WT and AncGADs, the root mean square deviation (RMSD), radius of gyration (Rg), and root mean square fluctuation (RMSF) were calculated from the MD simulation trajectories. Structural visualization, residue mapping, and MD trajectory analysis were performed using PyMOL v2.5.4 and OriginPro 2024.

2.8. Measurement of Thermostability of AncGADs

The semi-inactivation temperature (T₅₀¹⁵) is defined as the temperature causing a 50% loss of initial GAD activity after a 15 min incubation. This parameter was determined by incubating purified AncGAD at temperatures ranging from 20 to 75 °C for 15 min, followed by activity measurement. The activity of a non-incubated sample was defined as 100%. Meanwhile, the half-life (t_1/2) of AncGAD was also assessed by incubating the enzyme at 60 °C, and aliquots were taken at different time points (10–180 min). Following immediate cooling on ice, the residual activity of each aliquot was measured and expressed as a percentage relative to the activity at 37 °C. The purified enzyme was used at a concentration of approximately 1 mg/mL. All thermostability measurements were performed in triplicate, and residual activities are expressed as the mean ± standard deviation (SD).

2.9. Fed-Batch Cultivation of Engineered EcN Strain

The engineered EcN (T7) strain was initially inoculated into 200 mL of LB medium containing 50 μg/mL kanamycin and cultured overnight at 37 °C in a shaking incubator. This seed culture was then transferred into a 5 L bioreactor (F3G-BGG, Bsniss, Shanghai, China) containing 2.5 L of fermentation medium composed of 10 g/L glucose, 20 g/L yeast extract, 20 g/L peptone, 10 g/L KH₂PO₄, 2.5/L (NH₄)₂SO₄, and 2.5 g/L MgSO₄, 0.2 g/L ZnSO₄, 0.2 g/L ferric ammonium citrate, and 1 g/L defoamer, and was purchased from Sangon Biotech Co., Ltd. (Shanghai, China). The initial temperature was maintained at 37 °C, and the pH was kept at approximately 7.0 through the addition of 25% (v/v) NH₃·H₂O. Dissolved oxygen (DO) was controlled at around 25% saturation by adjusting the agitation speed between 200 and 800 rpm, with a constant air flow rate of 2.0 vvm. When the cell density reached an OD₆₀₀ of about 25, protein expression was induced by adding 0.5 mM IPTG. Upon depletion of the initial carbon source, an 80% (w/v) glucose solution was fed into the bioreactor to maintain the glucose concentration between 0.5 and 5 g/L.

2.10. High Level Synthesis of GABA by Using Dormant Engineered EcN Cells

After IPTG induction, the culture broth was centrifuged at 6000× g for 20 min at 4 °C, and the supernatant was discarded. The cell pellet was washed twice with PBS (10 mM, pH 7.5). For scalable production of GABA, the bioconversion was conducted in a 5 L reactor using cells (OD₆₀₀ = 20) suspended in 1.5 L of deionized water. The reaction mixture was initially supplemented with 0.813 kg of L-Glu, 10 μM PLP, and 1.5 mL of defoamer and proceeded at 40 °C and 200 rpm under uncontrolled pH conditions. The time variations of GABA synthesis in the above reaction system were analyzed by HPLC [22].

3. Results and Discussion

3.1. Reconstruction of GAD Ancestral Enzymes

A popular view holds that ancestral enzymes likely possessed higher thermostability compared to their extant ones. Guided by this perspective, the use of ASR to build enzymes with better thermal stability has emerged as a prominent research strategy. Positive consequences from such efforts were demonstrated as early as the 2010s, with the thermostability of enzymes such as 3-isopropylmalate dehydrogenase [23], certain transcription factor proteins [24], and glycyl-tRNA synthetase [25] being successfully improved by this technique [26].

To preserve the functional integrity of the GAD ancestral enzymes, key active-site residues were identified based on our previous research [27] and analysis with the Protein Contacts Atlas system. As shown in Figure 1, 17 conserved amino acid sites that may be involved in catalytic activity were selected: F65, K89, C130, G164, Q166, W169, I178, M185, I211, T215, D246, A248, T254, L267, H278, K279, and P285. Subsequently, the phylogenetic tree of ancestral GADs was built using the FireProt^ASR system with the 17 conserved residues fixed (Figure S1) [28]. The ancestral nodes situated at the earliest branches of the phylogenetic tree-Anc19, Anc20, Anc28, and Anc30-were chosen as the target ancestral enzymes. Using their corresponding amino acid sequences as templates, codon optimization was performed for expression in E. coli using the JCAT codon optimization tool (http://www.jcat.de/) [29], resulting in the generation of optimized gene sequences that encode the corresponding ancestral GAD enzymes. The corresponding amino acid and gene sequences are presented in Supplementary Table S1. A multiple sequence alignment was performed among four AncGAD sequences (Anc19, Anc20, Anc28, Anc30) and a query sequence. The 17 conserved active-site residues (F65, K89, C130, G164, Q166, W169, I178, M185, I211, T215, D246, A248, T254, L267, H278, K279, P285) are also marked (Figure S2).

3.2. Expression and Purification of AncGADs in Engineered EcN (T7)

EcN is a well-established probiotic strain that lacks cytotoxins and enterotoxins and has been widely applied in the prevention and treatment of various gastrointestinal disorders [30,31]. These intrinsic properties, combined with its proven biosafety profile, render EcN an attractive chassis for the biosynthesis of compounds. In previous work, the T7-promoter-compatible strain EcN (T7) was constructed by chromosomal integration of the T7-RNA polymerase gene, driven by the lacUV5 promoter into the malEFG operon [32]. Based on this, the recombinant AncGADs were generated in the modified host strain EcN (T7) by IPTG-induced expression and purified via Ni²⁺-affinity chromatography. SDS-PAGE analysis (Figure 2) demonstrated that the purified AncGADs migrated as a single band with a molecular mass of approximately 53–55 kDa, which corresponds to their predicted molecular weights. The high purity of the obtained sample provides a solid foundation for further investigation of the enzymatic properties of AncGADs.

3.3. The Enzymatic Properties of AncGADs

The wild-type (WT) GAD used as a reference in this work was the GadB from Levilactobacillus brevis CGMCC 1306 (LbGAD). To clarify the function of AncGADs, temperature and pH dependencies of decarboxylation activity were investigated. As shown in Figure 3, the AncGADs variants exhibited a markedly higher optimum temperature than the WT, while their optimal pH remained similar. The Anc19 showed the highest optimum temperature at 65 °C, suggesting an increase of nearly 30 °C relative to WT. Notably, all variants retained more than 50% of their maximal activity across a wide temperature range of 45–70 °C, indicating markedly enhanced thermal stability. However, as for the pH dependence of activity, a rapid decline was observed above pH 5.5, with activity becoming virtually undetectable at pH 6.0, as reported in previous studies [33].

Routinely, the kinetic parameters of WT and AncGADs were also determined at 37 °C and pH 5.0. The K_m and catalytic efficiency (k_cat/K_m) of WT were 41.01 mM and 0.79 s⁻¹mM⁻¹, respectively. As shown in

Table 2. Kinetic analysis of the AncGADs with L-Glu as substrate.

GADs	Km (mM)	kcat (s−1)	kcat/Km (s−1mM−1)
WT	41.01	32.60	0.79
Anc19	54.78	48.23	0.88
Anc20	59.63	41.47	0.69
Anc28	26.80	57.41	2.14
Anc30	30.44	30.16	0.99

, both Anc19 and Anc30 exhibited small improvements in catalytic efficiency compared with the WT. Particularly, Anc28 displayed a lower K_m value, indicative of enhanced substrate affinity relative to the WT. Meanwhile, the k_cat value of Anc28 was markedly enhanced. This modification consequently enhanced its catalytic efficiency (k_cat/K_m) to 2.14 s⁻¹mM⁻¹, which was 2.71-fold higher than that of the WT, representing a substantial catalytic advantage. Collectively, the ASR strategy successfully endowed GAD with a higher optimum temperature and enhanced catalytic efficiency, highlighting its promising prospects in industrial applications.

3.4. AncGADs Exhibited Enhanced Thermal Stability

Owing to its superior catalytic efficiency and increased optimal reaction temperature, Anc28 was selected for further investigation into its improved thermal stability relative to the WT enzyme. As shown in Figure 4A, the T₅₀¹⁵ value of Anc28 was 62.94 °C, which was 6.74 °C higher than that of the WT enzyme (56.2 °C). Similarly, the t_1/2 of Anc28 was 65.43 min, which was a 4.1-fold improvement over that for the WT enzyme (15.95 min) (Figure 4B). Over the past decade, several GAD mutants with improved thermal stability have been obtained via different protein engineering strategies [15,34,35,36,37,38]. Particularly, compared with these previously reported GAD mutants (Table 3), Anc28 exhibits a remarkable improvement in thermal stability. Taken together, these results demonstrate that Anc28 combines high catalytic efficiency with significantly enhanced thermal stability, positioning it as a promising candidate for GABA synthesis.

Table 3. Summary of the reported GAD engineering strategies and effects.

Source of GAD	Engineering Strategy	Obtained GAD Mutants	Effects	Reference
E. coli MG1655	Site-directed mutagenesis based on the N-terminal structure analysis	GadBQ5D/V6I/T7E	T5015 increased by 7.7 °C	[34]
E. coli K12	Site-directed mutagenesis based on a homologous comparison of its isoform and the catalytic–substrate interactions	GadBT62S, GadBQ309A	Residual activity increased by 19% and 27% after 12 h at 45 °C, pH 4.3	[35]
L. brevis CGMCC 1306	Site-directed mutagenesis at consensus site followed by saturation mutation	GadBT383V	T5015 increased by 3.0 °C; t1/2 increased 1.2-fold at 37 °C	[15]
L. plantarum GM 1403	C-terminal truncation guided by homology modeling	GadBΔC11	T5015 increased by 1.82 °C	[36]
L. brevis CGMCC 1306	Proline residues were introduced at sites corresponding to those present in the thermophilic T. kodakarensis GAD	GadBG364P	t1/2 increased by 19.4 min and T5015 by 5.3 °C at 55 °C	[37]
L. brevis CGMCC 1306	Site-directed mutagenesis selected by Ramachandran plot analysis	GadBK413A	t1/2 increased 2.1-fold at 50 °C	[38]

Table 3. Summary of the reported GAD engineering strategies and effects.

Source of GAD	Engineering Strategy	Obtained GAD Mutants	Effects	Reference
E. coli MG1655	Site-directed mutagenesis based on the N-terminal structure analysis	GadBQ5D/V6I/T7E	T5015 increased by 7.7 °C	[34]
E. coli K12	Site-directed mutagenesis based on a homologous comparison of its isoform and the catalytic–substrate interactions	GadBT62S, GadBQ309A	Residual activity increased by 19% and 27% after 12 h at 45 °C, pH 4.3	[35]
L. brevis CGMCC 1306	Site-directed mutagenesis at consensus site followed by saturation mutation	GadBT383V	T5015 increased by 3.0 °C; t1/2 increased 1.2-fold at 37 °C	[15]
L. plantarum GM 1403	C-terminal truncation guided by homology modeling	GadBΔC11	T5015 increased by 1.82 °C	[36]
L. brevis CGMCC 1306	Proline residues were introduced at sites corresponding to those present in the thermophilic T. kodakarensis GAD	GadBG364P	t1/2 increased by 19.4 min and T5015 by 5.3 °C at 55 °C	[37]
L. brevis CGMCC 1306	Site-directed mutagenesis selected by Ramachandran plot analysis	GadBK413A	t1/2 increased 2.1-fold at 50 °C	[38]

To further investigate the differences in structural stability between the ancestral enzyme Anc28 and the WT, molecular dynamics (MD) simulations were performed for both variants. As illustrated in Figure 5A, the RMSD value of Anc28 rose sharply at the early simulation stage and stabilized after around 20 ns, with subsequent fluctuations primarily ranging from 1.1 to 1.3 nm. Conversely, WT showed larger RMSD fluctuations throughout the simulation, with a rising tendency in the later phase, reflecting weaker conformational stability. Further, Rg analysis demonstrated Anc28 reached stability at 20 ns, with its radius of gyration persistently smaller than WT (Figure 5B), suggesting a tighter structural arrangement. Collectively, these findings reveal that Anc28 achieved structural stability at around 20 ns, accompanied by diminished conformational fluctuation and tighter overall packing. Such structural features likely underpin its superior thermal stability.

Further RMSF analysis revealed that Anc28 exhibited lower values across multiple regions relative to the wild type, notably the N-terminal 45–98 residue segment, the 150–220 residue region, and partial C-terminal sequences beyond residue 330. This finding suggests that the motions of residues in these regions were partially restricted, leading to decreased local conformational fluctuations and improved structural stability. This observation implies restrained residue mobility within these domains, which curbs local conformational fluctuations and strengthens structural stability. Overall, the generally lower RMSF profile of Anc28 reflects diminished residue flexibility and elevated local structural rigidity, favoring the preservation of protein conformational stability (Figure 6).

Gibbs free energy (GFE) analysis showed the WT possessed an extensive low-energy region, with the energy minimum localized at an RMSD of 0.40–0.85 nm and Rg of 2.35–2.50 nm (Figure 7A,C). This suggests the wild type can adopt diverse low-energy conformations, consistent with its high structural flexibility. By contrast, Anc28 displayed a deeper and more concentrated free energy minimum corresponding to RMSD 0.35–0.70 nm and Rg 2.25–2.30 nm (Figure 7B,D). It adopts fewer low-energy conformations and presents a more compact stable structural state.

The combined RMSD, Rg, RMSF and free energy analyses confirm Anc28 achieves better overall and local structural stability than WT, featuring compact, rigid and energetically optimal conformations. The simulation results match the improved experimental T₅₀¹⁵ and t_1/2, elucidating the molecular mechanism behind enhanced thermostability. Relevant simulation data of the remaining three AncGAD variants can be found in the Supplementary Files.

3.5. High-Efficiency GABA Synthesis Employing Engineered EcN (T7)

To assess the feasibility of GadB with improved catalytic efficiency and enhanced thermal stability for GABA production, the GABA-producing capacity of the engineered E. coli strains EcN (T7)/pET28a-gadB and EcN (T7)/pET28a-gadB_Anc28 was further compared in a 5 L bioreactor. Owing to the low solubility of L-Glu, a large quantity of solid powder remained at the bottom of the 5 L bioreactor during the initial reaction stage. As the reaction proceeded, the solid L-Glu gradually dissolved, accompanied by a steady increase in the pH of the reaction system. The time course of GABA synthesis using engineered EcN (T7)/pET28a-gadB is shown in Figure 8A. At 4 h, the GABA concentration reached 283.38 g/L, with the pH increasing from 3.5 to 6.3. The reaction system approached equilibrium after 5 h, yielding 299.08 g/L GABA (with an average space–time yield of 59.82 g/L/h). Furthermore, HPLC analysis confirmed that the conversion rate of L-Glu approached 96.67%. As for EcN (T7)/pET28a-gadB_Anc28, a similar variation tendency was observed for GABA synthesis and the pH (Figure 8B). Remarkably, GABA rapidly accumulated, and the pH increased sharply, as the reaction proceeded. The GABA titer reached 301.37 g/L with a conversion rate of 97.37% at 0.5 h. Upon completion of the 4 h reaction, the GABA concentration reached 307.53 g/L with a conversion rate of 99.36%. Compared with previously reported systems [39,40,41,42,43], EcN (T7)/pET28a-gadB_Anc28 exhibits distinct advantages in both final product titer and average space–time yield (

Table 4. The GABA production performance of engineered strains.

Strains	Gene and Source	Concentration (g/L)	Productivity (g/L/h)	Conversion Ratio	References
E. coli XBT	GadB and GadC (E. coli)	5.46	0.114	89.5%	[39]
E. coli BL21(DE3)	GadB (L. lactis)	204.12	34	99%	[40]
E. coli XL1-Blue	GadB (L. lactis)	94.8	1.98	77.7%	[41]
E. coli BL21(DE3)	GadB (Saccharomyces cerevisiae)	252	N/S	99%	[42]
E. coli BL21(DE3)	GadBΔC11K17T/D294G/E312S/Q346H (L. brevis)	270.42	36.06	99.9	[43]

). Notably, trace residual L-Glu and the use of pure water as the sole reaction medium strongly facilitate the development of environmentally benign processes for GABA production. On the basis of these results, engineered E. coli EcN (T7) cells expressing the Anc28 are applicable for the preparative-scale biosynthesis of GABA from L-Glu.

4. Conclusions

As the key rate-limiting enzyme in GABA biosynthesis, GAD demands superior thermal stability and high catalytic efficiency to fulfill industrial requirements. In this study, ancestral sequence reconstruction (ASR) was employed to generate GAD variants with drastically enhanced catalytic activity and thermal stability, and their enzymatic properties were systematically characterized. Beyond enzyme engineering, a whole-cell GABA bioproduction system using pure water as the sole medium and engineered E. coli Nissle (T7) cells as biocatalysts was established. This system enabled efficient GABA production, reaching a high titer of 307.53 g/L with 99.36% conversion within 4 h at 40 °C. Collectively, this work not only significantly improved the thermostability of GAD but also provided a promising strategy for the green and efficient production of high-purity GABA, offering a scalable and environmentally friendly platform for industrial GABA production and a promising strategy for the green and efficient production of high-purity GABA under pure-water bioconversion conditions. Nevertheless, because the current production strain is plasmid-based and contains a kanamycin-resistance marker, future development of antibiotic-marker-free strains will be necessary before this approach can be considered a fully food-grade or GRAS-compliant process.