A brief review on glutamate decarboxylase from lactic acid bacteria

: Glutamate decarboxylase ( L -glutamate-1-carboxylase, GAD; EC 4.1.1.15) is a pyridoxal 5-phosphate-dependent enzyme, which catalyzes the irreversible α-decarboxylation of L -glutamic acid to γ-aminobutyric acid (GABA) and CO 2 . The enzyme is widely distributed in eukaryotes as well as prokaryotes, where it – together with its reaction product GABA - fulfils very different physiological functions. The occurrence of gad genes encoding GAD has been shown for many microorganisms, and GABA-producing lactic acid bacteria (LAB) have been a focus of research during recent years. A wide range of traditional foods produced by fermentation based on LAB offer the potential of providing new functional food products enriched with GABA that may offer certain health-benefits. Different GAD enzymes and genes from several strains of LAB have been isolated and characterized recently. GABA-producing LAB, biochemical properties of their GAD enzymes, and possible applications are reviewed here. the C-terminal region Ile454-Thr468 of this enzyme increased activity in the pH range of 5 to 7, with the  11 variant showing significantly better results, increasing the catalytic efficiency of the variant at pH 5.0 and 7.0 by a factor of 1.26 and 28.5, respectively. The authors concluded that the C-terminal region is involved in decreasing the activity of L. plantarum GAD at higher pH values by closing up the catalytic site as a result of pH-induced conformational changes [79]. In a similar way, a C-terminally truncated variant of L. brevis GAD, in which the terminal 14 amino acids had been removed by site-directed mutagenesis, showed improved activity at higher, around neutral pH values [85]. These studies point to the importance of the C-terminus of GAD for improved accessibility of the active site and this increased and this mesophilic GAD with homologous thermophilic enzymes to identify amino acid residues that might affect stability. Two mutant enzymes were obtained and showed higher thermostability with their half-inactivation temperature 2.3°C and 1.4°C higher than the wild-type enzyme. Furthermore, the activity of the variants was 1.67-fold increased during incubation at 60°C for 20 min. They suggested that this approach can be an efficient tool to improve the thermostability of GAD [101].


Introduction
Lactic acid bacteria (LAB) are Gram-positive, acid-tolerant, non-sporulating bacteria forming cocci as well as rods, and sharing common physiological and metabolic characteristics. Even though many genera of bacteria produce lactic acid as their primary or secondary metabolic end-product, the term 'lactic acid bacteria' is conventionally reserved for genera in the order Lactobacillales, which includes Lactobacillus, Lactococcus, Leuconostoc, Pediococcus and Streptococcus, and in addition Carnobacterium, Enterococcus, Oenococcus, Tetragenococcus, Vagococcus, and Weisella. LAB are important for a wide range of fermented foods, and are widely used as starter cultures in traditional and industrial food fermentations [1].
Lactic acid formed during the fermentation of carbohydrates as one of the main metabolic products can affect the physiological activities of LAB. Under acidic conditions, several LAB have developed different acid-resistance systems to maintain cell viability; these systems include for example the F0F1-ATPase system or cation/proton antiporter/symporter systems such as the K + -ATPase, which contribute to pH homeostasis in the cytosol by the translocation of protons [2]. In addition, glutamate or arginine-dependent systems, which require the presence of glutamate and arginine, respectively, as substrates, contribute to acid resistance of LAB. The first enzyme in the arginine-dependent system is arginine deiminase, which degrades arginine to citrulline and NH3. Citrulline is then further converted to ornithine and exported from the cell by an ornithine/arginine antiporter. While the arginine-dependent system is based on the production of an intracellular alkaline compound, the glutamate-dependent system consumes an intracellular proton by combining Table 1. Diversity of GABA-producing LAB, isolation sources and GABA production. GABA concentrations as found in food products fermented with this strain or from biocatalytic, while-cell transformations of glutamate are given plantarum, L. helveticus, or L. futsaii [2,13,31,32,33,47]. Among these, L. brevis, a heterofermentative LAB, is one of the best-studied organisms [43], known for forming high levels of GABA under appropriate conditions (Table 1). Traditionally, fermented food samples containing GABA are used to screen for and isolate GABA-producing LAB, and it is not surprising that food samples with high GABA content may result in the isolation of promising strains showing good GABA-forming properties. Typical fermented foods used for isolating GABA-producing LAB are kimchi, where in one study 68 out of 230 LAB isolates showed the ability to convert glutamate to GABA [44] or other fermented vegetable, cheese [16] or fermented milk products, as well as various fermented meat or fish products including sausages, Thai fermented fish plaa-som [45], or traditional fermented Cambodian food, mainly based on fish, where 6 out of 68 LAB isolates showed a significant GABAproducing ability [1]. These screening / isolation strategies often resulted in the identification of strains capable of efficiently converting glutamate or in the discovery of novel, not-yet-identified producers of GABA, which show promise as starter cultures for various fermented foods enriched in GABA. For example, the novel GABA producer Lactobacillus zymae, which can grow on up to 10% NaCl and is able to utilize D-arabitol as carbon source, was isolated from kimchi [46]. Recently, Sanchart et al. have isolated the novel GABA-forming strain Lactobacillus futsaii CS3 from fermented shrimp (Kung-som) [47]. This isolate was able to convert 25 mg/mL of monosodium glutamate to GABA with a yield of more than 99% within 72 h. These studies (Table 1) showed that genera Lactobacillus and Lactococcus are the predominant GABA-producing LAB, but also other genera such as Enterococcus were studied in this respect. A novel GABA-producing Enterococcus avium strain was isolated from the Korean traditional fermented anchovy and shrimp (jeotgal), and was shown to produce 18.47 mg/mL GABA within 48 h in a medium containing glutamate as substrate. A recent study looking at LAB isolated from traditional Japanese fermented fish products (kaburazushi, narezushi, konkazuke, and ishiru) showed that out of 53 randomly picked LAB isolates 10 showed the ability of transforming considerable amounts glutamate into GABA, and identified Weissella hellenica as a novel GABA producer [41]. Thus, these new genera expand the list of microorganisms as GABAproducing bacteria and can be open up new and different applications in the food industry. This may lead to a wider application and flexibility of starter cultures in the food industry [9].

Occurrence and organization of GAD genes
The conversion of glutamate to γ-aminobutyric acid is catalyzed by glutamate decarboxylase [glutamic acid decarboxylase, GAD, systematic name L-glutamate 1-carboxy-lyase (4aminobutanoate-forming), EC 4.1.1.15], which catalyzes the irreversible α-decarboxylation of glutamate [5,48]. GAD employs pyridoxal 5'-phosphate (PLP) as its cofactor, and is found in numerous microorganisms such as bacteria [3], fungi [49] and yeasts [50]; furthermore, GAD is found in plants [51], insects and vertebrates (52). GAD is an intracellular enzyme that is utilized by LAB to encounter acidic stress by decreasing the proton concentration in cytoplasm in the presence of Lglutamate [2,6,54]. This system, the so-called glutamate-dependent acid-resistant system (GDAR), provides protection under acidic condition, and therefore the ability of LAB to perceive and cope with acid stress is crucial for successful colonization of the gastrointestinal tract (GIT) and survival under acidic environments such as in fermented food. The GDAR system consists of two homologous inducible glutamate decarboxylases, GadA and GadB, and the glutamate/γ-aminobutyrate antiporter GadC [20,48]. The corresponding genes, i.e., gadA, gadB and gadC, are expressed upon entry into the stationary phase when cells are growing in rich media independently of pH, and are further induced upon hypoosmotic and hyperosmotic stress, or in the log-phase of growth in minimal medium containing glucose at a pH of 5.5 [53,55]. Siragusa et al. demonstrated that three strains with a GDAR system, L. bulgaricus PR1, L. lactis PU1 and L. plantarum C48, were able to survive and synthesize GABA under simulated gastrointestinal conditions [26]. Recently, the GABA-producing strain L. futsaii CS3 was shown to be only decreased by 1.5 log cycles under simulated gastrointestinal conditions, indicating that the GDAR system contributes to resistance to the conditions in the GIT and that GABA-producing LAB thus have a potential as functional probiotic starter cultures [47]. GAD systems and the organization of the gad operons among LAB species are highly variable [57]. Numerous studies reported that some LAB species such as S. thermophilus [5], L. brevis [6,7], L. lactis [42] have one or two gad gene (i.e., gadA, gadB) together with the antiporter (gadC). Interestingly, E. avium 352 carries 3 gad genes [58]. Typically, L. brevis contains two GAD-encoding genes, gadA and gadB, sharing approximately 50% amino acids sequence identity [6]. In contrast, the gadB genes is absent in strain L. brevis CD0817 [59], and the amino acid sequences identities of gadA and gadC from L. brevis CD0817 against other L. brevis strain are 91% and 90%, respectively. The transcriptional regulator gene gadR plays a crucial role in GABA production and acid resistance in L. brevis. Gong et al. reported that deletion of gadR in L. brevis ATCC 367 resulted in lower expression of both the gadB and gadC gene, a concurrent reduction in GABA synthesis and an increased sensitivity to acidic conditions [6]. Expression levels of gadR are varied among different LAB strains. The gadR gene was expressed 13-155-fold higher than gadCB in L. brevis NCL912 during the cultivation period [60]. In contrast, expression of gadR in L. brevis CGMCC1306 was observed to be much lower compared to gadCB. The role of gadA and gadB in L. brevis CGMCC1306 was investigated by disruption of the genes gadA, gadB and gadC resulting in complete elimination of GABA formation and increased sensitivity to acidic conditions, suggesting that both GAD proteins and the antiporter are essential for GABA production and acid resistance [61].
A genomic survey was conducted by Wu et al. to gain insight on the distribution of the gad operon and genes encoding glutamate decarboxylase in LAB [7]. Most strains of L. brevis (14 strains) as well as some strains of L. reuteri (6 strains), L. buchneri (2 strains), L. oris (3 strains), L. lactis (29 strains), and L. garvieae (5 strains) were shown to have an intact gad operon, the majority of these strains were shown to contain either gadA or gadB, whereas gadC is only present in the genomes of certain strains and noticeably lacking in L. plantarum, suggesting that the characteristic of GABA production is strain-dependent. Similar results were obtained by Yunes et al. who showed that L. fermentum (9 strains), L. plantarum (30 strains) and L. brevis (3 strains) typically contain gadB genes. In addition, no antiporter gene was observed next to gadB in L. plantarum 90sk and the expression of gadB was increased in early stationary phase and at low pH (3.5-5) [62]. The gadB gene from S. thermophilus encoding 459 amino acids has been investigated. The transposase genes Tn1216 (5' and 3') and Tn1546 are located downstream and upstream of hydrolase genes flanking the gadB/gadC operon as a result from horizontal gene transfer. This sequence implied that the order of gadB and gadC in S. thermophilus ST110 is similar to S. thermophilus Y2 [63], yet in different order from that reported for L. brevis [60], L. plantarum [62] and L. lactis [64].
The L. reuteri 100-23 genome was investigated by Su et al. for its gad operon [65]. This genome contains gadB and two genes for the antiporter (gadC1 and gadC2), as well as the glutaminaseencoding gene gls3, indicating that glutamine serves as a substrate for the synthesis of GABA. The organisation of the gad operon is in different order for other species of LAB (L. lactis and L. plantarum) as glutaminase (gls3) is in between the antiporters gadC1 and gadC2, while gadB is accompanied by gadC1 [65]. The full length of gad genes has been cloned and sequenced for several species and strains of LAB. Li et al. cloned gadA from L. brevis NCL912, and the whole gene fragment (4615 bp) including gadR, gadC, gadA and gts (glutamyl t-RNA synthetase) was successfully amplified. Their work suggested that the high GABA production capacity of L. brevis NCL912 may be linked to the gadA locus forming a gadCA operon complex that ensures the coordinated expression of GAD and the antiporter [60]. A core fragment of the gad gene from L. brevis OPK3 was cloned and successfully expressed in E. coli. The nucleotide sequence revealed that the open reading frame of the gad gene consisted of 1401 bases encoding 467 amino acid residues. The sequence showed 83%, 71% and 60% homology to GAD from L. plantarum, L. lactis and Listeria monocytogenes, respectively [66].
A phylogenetic tree constructed from available GAD sequences in the NCBI protein database shows that amino acids sequences of GAD are highly conserved within the same species (Figure 1), and that GAD is widely distributed in a number of LAB including L. brevis, L. buchneri, L. delbrueckii subsp. bulgaricus, L. fermentum, L. futsaii, L. paracasei, L. parakefiri, L. paraplantarum, L. plantarum, L. plantarum subsp. argentoratensis, L. reuteri, L. sakei, L. lactis, and S. thermophilus. All of these LAB are commonly found in fermented foods and with some of these are commonly used as starter cultures in food industries. In addition, GAD is also found in other lactobacilli including L. acidifarinae, L. aviaries, L. coleohominis, L. farraginis, L. japonicas, L. koreensis, L. nuruki, L. oris, L. rossiae, L. rennini, or L. suebicus ( Figure 1). These organisms have not been studied for their capacity to synthesize GABA nor have their GAD system been studied, and hence they could be of interest with respect to GABA production and GABA-enriched food. Figure 1. Phylogenetic analysis of glutamate decarboxylase from different species of LAB (maximumlikelihood method). The phylogenetic analysis was performed after the alignment of GAD sequences using MUSCLE in MEGA X software.

Glutamate decarboxylase
Glutamate decarboxylase is an intracellular enzyme that is found ubiquitously in eukaryotes and prokaryotes. GAD exhibits different physiological roles especially in vertebrates and plants, and its presence is highly variable among organisms [52]. GAD is a pyridoxal 5'-phosphate (PLP) dependent enzyme and as such belongs to the PLP-dependent enzyme superfamily, which contains at seven different folds [67], with GAD from LAB showing the type-I fold of PLP-dependent enzymes [68]. A number of important catalytic reactions including α-and -eliminations, decarboxylation, transamination, racemization and aldol cleavage are catalyzed by various members of this superfamily of enzymes [69]. GAD activity relies on the binding of its co-factor PLP, and belongs to group II of PLP-dependent decarboxylases [70]. In GAD from L. brevis GCMCC 1306, the active site entrance is located at the reface of the cofactor PLP, and PLP is covalently attached to a lysine (K279) via an imine linkage, referred to as an internal aldimine [85]. This lysine is strictly conserved in group II PLP-dependent decarboxylases. The corresponding lysine in E. coli GAD is at position 276, and when mutating this residue, the variant has less flexibility and affinity to both its substrate and the cofactor [71]. In addition to this covalent attachment, PLP is positioned in the active site via a number of H bonds between the phosphate group of PLP and surrounding amino acids, while the pyridine ring of PLP forms hydrophobic interactions with side chains of various amino acids in the active site [68]. Molecular docking of the substrate glutamate into the active-site of the holo form of L. brevis GAD showed several noncovalent interactions including hydrogen bonds between the O2, the O3 and the O4 atoms of the substrate L-Glu to various parts of the GAD polypeptide chain. Furthermore, electrostatic interactions between the negatively charged oxygen atom of the α-carboxyl and the γcarboxyl group of L-Glu and the positively charged nitrogen atom of residue R422 as well as H278 and K279, respectively, were proposed [72]. The flexible loop residues Tyr308-Glu312 in L. brevis GAD is located near the substrate-binding site, and is important for its catalytic reaction. Furthermore, the conserved residue Tyr308 play crucial role in decarboxylation of L-Glu. Thr 215 and Asp246 are the two catalytic residues in L. brevis GAD, which are also highly conserved and promote decarboxylation of L-Glu [71,73].
During catalysis a transamination reaction occurs, and PLP, which is covalently attached to a Lys in the active site of GAD in its resting state, now becomes covalently bonded to the substrate glutamate, forming a Schiff base or what is referred to as an external aldimine, which can then be transformed to a quinonoid intermediate [67,74]. In a small fraction of catalytic cycles when glutamate is decarboxylated, a subsequent alternative transamination of the quinonoid intermediate of the reaction can occur, and succinic semialdehyde (SSA) and pyridoxamine-5'-phosphate (PMP) are formed. The latter will immediately be released from the enzyme, resulting inactive apoGAD ( Figure 2), which can be regenerated to the active GAD-PLP complex when free pyridoxal 5'phosphate is present, thus completing a cycle of inactivation and activation. However, when free PLP is not present, GAD will not be reactive as a function of time and substrate concentration [62,[67][68][69]74,77].

Biochemical insights into glutamate decarboxylase from lactic acid bacteria
GAD from LAB typically consists of identical subunits with molecular masses ranging from 54 to 62 kD, and is formed in its mature holoform even when produced heterologously. The oligomerization, typically resulting in the formation of a homodimer, is crucial for activity of the Lactobacillus spp. enzymes. Some ambiguity about the active form of GAD isolated from different isolates of L. brevis and its quaternary structure exists in the scientific literature. Hiraga et al. reported that treatment with high concentrations of ammonium sulfate results in an active tetrameric form with the enzyme from L. brevis IFO12005 GAD [72]. The presences of ammonium sulfate apparently stabilized GAD from this source as the purified enzyme was found to be rather unstable, and the dimeric form showed no activity. Moreover, the presence of ammonium sulfate apparently did not affect the overall conformation but had effects on the active site of the protein. Studies by Yu et al. showed that GAD from L. brevis CGMCC 1306 is active as a monomer, while GAD from other LAB are generally active as dimers [85]. Subsequent structural studies on this enzyme revealed, however, that GAD from L. brevis CGMCC 1306 is active as a dimer, even though elucidation of the crystal structure resulted in a distorted asymmetric trimer. The authors concluded that this observed trimer is only the result from crystallographic packing and not the biological form [68].
As mentioned above, a number of LAB carry two GAD-encoding genes, gadA and gadB. Frequently, studies have focused on the purification and characterization of GadB, e.g., from E. raffinosus [75], L. plantarum [79], L. brevis [78], L. sakei [80], L. paracasei [18] since the expression levels of recombinant GadB are typically higher than those for GadA [55]. A recent study by Wu et al. showed that gadA transcript was highly upregulated (55-fold) in strain L. brevis NPS-QW-145 at the stationary phase of growth. Subsequently, both GadA and GadB were recombinantly produced and characterized. GadA showed a pH profile of activity near the neutral region, with the optimal activity found in the range of pH 5.5-6.6, in contrast to GadB, which is more active under acidic conditions (3.0-5.5). Presence of both of these two enzymes, GadA and GadB, in the L. brevis genome will give the organism a significant advantage to produce GABA over a broad range of pH (3.0-6.0) and thus to more efficiently maintain pH homeostasis. These findings suggest that extending the activity of GadA to the near-neutral pH region offer a novel genetic diversity of gad genes from LABs [7].
A number of GAD have been expressed and characterized from a variety of LABs. In general, the N-and C-terminal regions of GAD from different sources show significant differences, and this might affect recombinant GABA production. As shown in a sequence alignment (Figure 3), the sequence HVD(A/S)A(S/F)GG is highly conserved among LAB GAD, and a lysine residue (Lys279 in) plays crucial role in the PLP binding site. Table 2 summarizes biochemical properties of GAD from different strains [18,42,81,82]. Typically, the pH optima of GAD are found between 4.0 and 5.0. GAD from L. zymae, E. avium M5, S. salivarius subsp. thermophilus Y2 and L. paracasei NFRI 7415 have an optimum activity of above 40°C, which does not coincide with the optimal temperature for growth of these strains [46,72,82,83]. Different ions can affect the stability and activity of GAD from different sources ( Table 2). GAD from E. avium M5 is activated in the presence of CaCl2 and MnCl2 but the activity is decreased by CuSO4 and AgNO3 [82]; comparable results were also obtained for GAD from other LAB sources, L. zymae [46] and L. sakei A156 [80].
Since GAD is mainly active under acidic conditions, several engineering approaches were employed to broaden its activity, especially at the near-neutral pH region. To this end, Shi et al. applied both directed evolution and site-directed mutagenesis at the β-hairpin region and C-terminal end of L. brevis GAD [84]. By using a plate-based screening assay employing a pH indicator as assay principle, they could identify several variants and positions that improved activity at pH 6.0. Furthermore, they selected three residues (Tyr308, Glu312, Thr315) in the β-hairpin region for site directed mutagenesis based on homology modelling, since these residues exhibit different interaction with surrounding amino acids in the model at different pH values. By combining various positive mutations, they could increase the catalytic efficiency of GAD from L. brevis 13.1-and 43.2-fold at pH 4.6 and 6.0, respectively, as compared to the wild-type enzyme [84]. The role of the C-terminus for the pH dependence of catalysis of L. plantarum GAD was investigated by Shin et al. employing mutagenesis [79]. Deletions of three and eleven residues in the C-terminal region Ile454-Thr468 of this enzyme increased activity in the pH range of 5 to 7, with the 11 variant showing significantly better results, increasing the catalytic efficiency of the variant at pH 5.0 and 7.0 by a factor of 1.26 and 28.5, respectively. The authors concluded that the C-terminal region is involved in decreasing the activity of L. plantarum GAD at higher pH values by closing up the catalytic site as a result of pHinduced conformational changes [79]. In a similar way, a C-terminally truncated variant of L. brevis GAD, in which the terminal 14 amino acids had been removed by site-directed mutagenesis, showed improved activity at higher, around neutral pH values [85]. These studies point to the importance of the C-terminus of GAD for improved accessibility of the active site and this increased activity especially at higher pH values, and thus the C-terminal loop is an essential target for enzyme engineering for GABA production at fluctuated pH conditions [79,85].    6. Improvement of GAD activities and GABA production GABA biosynthesis can be employed using whole cell reaction, recombinant bacteria and purified GAD. GAD from various sources of LAB have been overexpressed in different hosts including E. coli [86], L. plantarum [87], L. sakei [88], Corynebacterium glutamicum [21] and B. subtilis [89]. Utilization of whole cells for the biocatalytic conversion of glutamate to GABA has some drawbacks including the conversion of GABA to succinic semialdehyde by the enzyme GABA transaminase (GABA-T), which is often found in bacteria and might decrease GABA yields during cultivation. To prolong and thereby increase GABA production, continuous cultivation [90], fedbatch fermentation [91], as well immobilized cell technology [92,93]   approaches effectively increased GABA productivity by improving cell viability resulting in longer periods of cultivation. GABA biosynthesis and production could be enhanced by optimizing fermentation conditions, with attention given to different factors including the carbon source, concentration of added glutamate, pH regulation, incubation temperatures, nitrogen sources, cofactor and feeding time [34,94]. A study by Lim et al. showed that under optimized conditions, strain L. brevis HYE1 produced 18.8 mM of GABA. Monosodium glutamate (MSG) or L-glutamate are the main substrate for the production of GABA using either appropriate GAD-containing cells or pure GAD [27]. LAB with GAD activity may furthermore require pyridoxal 5'-phosphate (PLP) as a coenzyme to enhance GABA production. The addition of 0.5% MSG increased GABA production by E. faecium JK29, which reached 14.9 mM after 48 h of cultivation [38]. A concentration of 6% MSG and the addition of 0.02 mM PLP were found to be optimal conditions for L. brevis K203 for GABA production [95]. This strategy of increasing glutamate supplementation could not be used for all strains though; when Lglutamate was added at concentrations of 10 to 20 g/L to a medium of S. thermophilus GABA production could not be enhanced. It was suggested that this strain is not able to tolerate high glutamate concentrations [35]. High glutamate concentrations increase the osmotic pressure in the cells, and this stress can disturb the bacterial metabolism [39]. Fermentation time and temperature are also a key factor for GABA production. Villegas et al. investigated GABA formation by L. brevis CRL 1942, and found that 48 h of fermentation at 30°C employing 270 mM of MSG resulted in a maximum GABA production of 255 mM in MRS medium, indicating that the GABA production is time-dependent manner [94].
Metabolic pathway engineering has been performed to achieve enhanced GABA production. The key points here are the direct modulation of GABA metabolic pathways. A whole-cell biocatalyst based on E. coli cells expressing the gadB gene from L. lactis was used as the starting point of this engineering approach. By introducing mutations into this GadB to shift its decarboxylation activity toward a neutral pH, by modifying the glutamate/GABA antiporter GadC to facilitate transport at a neutral pH, by enhanced the expression of soluble GadB by introducing the GroESL molecular chaperones, and by inhibited the degradation of GABA by inactivation of gadA and gadB from the E. coli genome an engineered strain was constructed that achieved a productivity of 44.04 g GABA per L and h with an almost quantitative conversion of 3 M glutamate [96].
Several mutational approaches such as directed evolution and site-specific mutagenesis are considered as powerful tools for optimizing or improving enzyme properties. Several researchers have applied these approaches to improve GAD activity [83,[97][98][99][100] that was applied in whole-cell biocatalysts. In order to improve GAD activity over an expanded pH range, recombinant C. glutamicum cells were obtained by expressing L. brevis Lb85 GadB variants. These variants were constructed by combining directed evolution and site-specific mutagenesis of GadB to improve activity at higher pH values (see above) since L. glutamicum grows best around neutral pH [83]. C. glutamicum is an industrial producer of glutamate, and by introducing these GadB variants into this organism, GABA could be produced without the need of exogenous glutamate on a simple glucosebased medium, with yields of up to 7.13 g/L [83].
Insufficient thermostability is often a common problem associated with industrial enzymes, and most GAD show low stability even at moderate temperatures. A rational strategy for improving thermostability is to identify critical regions or amino acid residues by sequence alignments. Alternatively, structural information can be used which indicates flexible regions, and to subsequently strengthen these regions [101]. Identification of the consensus sequences can also improve thermostability of proteins [102]. Recently, Zhang et al. developed a parallel strategy to engineer L. brevis CGMCC 1306 GAD. They compared the sequence and structure of this mesophilic GAD with homologous thermophilic enzymes to identify amino acid residues that might affect stability. Two mutant enzymes were obtained and showed higher thermostability with their halfinactivation temperature 2.3°C and 1.4°C higher than the wild-type enzyme. Furthermore, the activity of the variants was 1.67-fold increased during incubation at 60°C for 20 min. They suggested that this approach can be an efficient tool to improve the thermostability of GAD [101].
The use of purified GAD seems to be economically more feasible then whole-cell biocatalysis when aiming at producing pure GABA due to simplified downstream purification of this compound from less complex reaction mixtures. A numbers of immobilization technique have been performed for re-use purposes, such as GadB immobilized on calcium alginate beads that are then employed in a bioreactor [103], a GAD/cellulose-binding domain fusion protein immobilized onto cellulose [104], and GAD immobilized to metal affinity gels [105]. The performance of immobilized GAD in a fedbatch reactor was evaluated, which showed high productivity of GABA as the substrate concentration in the medium was kept constant by feeding solid glutamate. Moreover, no significant decrease of enzyme activities were observed during the reaction when the inactivation reaction of PLP to succinic semialdehyde and pyridoxamine-5'-phosphate during catalysis was avoided by adding addition of a small amount of α-ketoglutaric acid to the reactor, which regenerates PLP [100]. Sang-Jae Lee et al. performed immobilization of L. plantarum GAD using silica beads and showed high stability under acidic and alkaline conditions with improved thermostability [105]. In addition, the immobilized GAD converted 100% of glutamate to GABA [106]. The results suggest that immobilization gives an advantageous result for industrial application when using (partially) purified GAD for the GABA from glutamate.

The role of glutamate decarboxylase in the manufacturing of bio-based industrial chemical
Agricultural waste and waste stream from biofuel production are now being considered as a low-cost source of glutamate for biotechnological conversion into GABA, and production of biobased chemicals [106]. These protein-rich materials are mainly bioethanol by-product streams including dried distiller's grains with solubles (DDGS) from maize and wheat, or vinasse from sugarcane or sugar beet, but also plant leaves, oil or biodiesel by-products, and slaughterhouse waste. In the future, algae could also provide an additional source for biodiesel and thus become a natural source low-cost source of glutamic acid.
The protein-rich fraction of plants can be further split into more-and less-nutritious fractions, for example by hydrolyzing the proteins and separating the essential (nutritious) amino acids from the non-essential (less nutritious) ones. Non-essential amino acid such as glutamic acid and aspartic acid, which have no significant value in animal feed, can be utilized for preparing functionalized chemicals. Recently, a by-product from tuna canning industry, tuna condensate, was shown to be a useful material for the production of GABA. Tuna condensate contains significant amounts of glutamine but relatively little glutamate. Glutamine was first converted to glutamate by a glutaminase from Candida rugasa, and in a second step L. futsaii GAD converted glutamate to GABA. Both steps were catalyzed by immobilized whole cells [107]. Recently, it was shown that supplementation of arginine to media containing glutamate can enhance GABA production, and that the simultaneous addition of arginine, malate, and glutamate enabled GABA production already during exponential growth at relatively high pH (6.5) [108].
The structure of glutamic acid resembles many industrial intermediates, so it can be transformed into a variety of chemicals using a relatively limited number of steps. Decarboxylation of glutamic acid to GABA, enzymatically performed by GAD, is an important reaction of the pathway from glutamic acid to a range of molecules. GABA is for example an intermediate for the synthesis of pyrrolidones. Such an approach can be used to produce for example N-methyl-2-pyrrolidone (NMP), which is used as an industrial solvent. Combining the enzymatic decarboxylation of glutamate performed by GAD with the one-pot cyclization of GABA to 2-pyrrolidone and subsequent methylation will thus yield NMP [109]. Another interesting material synthesized by ring-opening polymerization of 2-pyrrolidone is Nylon 4 (110), a four-carbon polyamide suitable for application as an engineering plastic due to its superior thermal and mechanical properties [111]. Contrary to other nylon polymers, Nylon 4 is heat-resistant, biodegradable, biocompatible and compostable [111].

Future trends and conclusions
The demand of functional foods is increasing and marked by the awareness of consumers in maintaining health and prevention of degenerative diseases. Therefore, exploration of bioactive compounds such as GABA are important. The GAD system plays a crucial role in GABA biosynthesis. A number of studies on cloning, expression and characterization of both gadA and gadB lead to deciphering the role of the gad genes in the GABA metabolic pathway and its importance for LAB. Since the production of GABA is dependent on the biochemical properties of GAD, more study on the biochemical properties of GAD are important especially for those enzymes derived from LAB isolated from food fermentation processes as this will facilitate the optimization of the fermentation process and support the selection of suitable starter cultures for these processes that will bring more GABA-enriched food to the consumer. Recent structural information of GAD from LAB will facilitate enzyme-engineering approaches to improve GAD towards enhanced thermostability or improved activity over a broad range of pH. However, structural information is currently only limited to GAD from L. brevis, and thus structural studies on GAD from other GABA-producing LAB is needed in order to understand their catalytic and structural properties in more depth. The elucidation of molecular mechanisms and roles of GABA production, knowledge of the regulatory aspects of GABA production, and profound comprehension of GABA producing cell physiology will offer the basis and tools to increase GABA yield at genetic and metabolic levels.