Characterization and improved properties of Glutamine synthetase from Providencia vermicola by site-directed mutagenesis

In this study, a novel gene for Glutamine synthetase was cloned and characterized for its activities and stabilities from a marine bacterium Providencia vermicola (PveGS). A mutant S54A was generated by site directed mutagenesis, which showed significant increase in the activity and stabilities at a wide range of temperatures. The Km values of PveGS against hydroxylamine, ADP-Na2 and L-Glutamine were 15.7 ± 1.1, (25.2 ± 1.5) × 10−5 and 32.6 ± 1.7 mM, and the kcat were 17.0 ± 0.6, 9.14 ± 0.12 and 30.5 ± 1.0 s−1 respectively. In-silico-analysis revealed that the replacement of Ser at 54th position with Ala increased the catalytic activity of PveGS. Therefore, catalytic efficiency of mutant S54A had increased by 3.1, 0.89 and 2.9-folds towards hydroxylamine, ADP-Na2 and L-Glutamine respectively as compared to wild type. The structure prediction data indicated that the negatively charged pocket becomes enlarged and hydrogen bonding in Ser54 steadily promotes the product release. Interestingly, the residual activity of S54A mutant was increased by 10.7, 3.8 and 3.8 folds at 0, 10 and 50 °C as compared to WT. Structural analysis showed that S54A located on the loop near to the active site improved its flexibility due to the breaking of hydrogen bonds between product and enzyme. This also facilitated the enzyme to increase its cold adaptability as indicated by higher residual activity shown at 0 °C. Thus, replacement of Ala to Ser54 played a pivotal role to enhance the activities and stabilities at a wide range of temperatures.

significantly reduced by adenylylated form while the γ-glutamyl transfer activity exists in both forms 36 . The biosynthetic activity of GS is catalyze the reaction of glutamate and ammonia to form glutamine and the γ-glutamyl transfer activity catalyze the transfer of γ -glutamyl moieties to water or amino acids, or peptides 6 . GS are usually categorized as GSII, GSIII, and GSI 37 . GSI enzymes specially exist in prokaryotes, and their structures are dodecameric 38,39 but recently GSI enzymes have also been identified in mammals and plants 2,40 . GSI enzymes are classified into two subdivisions 37 , GSIα and GSIβ. GSIα genes are found in the thermophilic bacterium, Thermotoga maritima and the Euryarchaeota. GSIβ type genes exist in other bacteria such as E. coli, Synechocystis PCC6803, Aquifex aeolicus, Crenarcheon sp., and Sulfolobus acidocaldarius 37 .
In this study, a GSIβ glutamine synthetase protein structure from Salmonella typhimurium (PDB ID: 1F1H) was adapted as a homology model to build up PveGS structure 41 . This dodecameric structure was in complex with two ammonium analogues, thallous ion (Tl + 473, Tl + 474), two manganese ion and a vital substrate ADP. Tl + 473 coincides with the ammonium substrate binding site reported by Liaw et al. 4 , and the Tl + 474 is the binding site of the ammonium group of the substrate glutamate 41 . Eisenberg has reported that hydroxylamine in the transfer reaction, ammonium ion in the biosynthetic reaction, and water in the glutamine hydrolysis reaction presumably bind at the same site 38 .
Coupling the structural information with efficient site-directed mutagenesis, the microbial expression techniques enabled successful engineering of a protein with the desired characteristics 21,23,42,43 . The objective of the present study was to obtain a mutant of glutamine synthetase obtained from a marine bacterium P. vermicola. The gene has been successfully cloned expressed and characterized to study its biochemical properties. A mutant S54A with improved catalytic efficiency, cold adaptability and higher thermostability was obtained by site-directed mutagenesis based on homology modelling of PveGS. The mutant with improved catalytic features can be exploited for synthetic reactions at industrial scale.

Results
Gene cloning and sequence analysis. The glutamine synthetase was cloned successfully from P. vermicola genomic DNA with an ORF of 1410 base pairs encoding for 469 amino acids (primers presented in Table S1). The amino acid sequence of PveGS was aligned with the five-reported bacterial GS structures. Four highly conserved amino acids located in the loop region (Asp51, Tyr179, Asn24, Tyr398) and one amino acid Glu328 was located in the flap of active-site ( Fig S1). Soluble PveGS protein was expressed and purified from E. coli BL21 (DE3) harbouring a recombinant plasmids pGEX-6p-1-PveGS. The molecular mass of the purified recombinant GS was 52 kDa after the removal of GST-tag, while its molecular mass was observed larger than expected on SDS-PAGE (Fig. 1A) due to the mino acids composition. The purified protein then loaded onto gel filtration The optimum temperature of the wild type and S54A. The activity was determined at a gradient temperature ranging from 0 to 60 °C and activity at 35 °C was defined as 100%; (C) The optimum pH of the WT and S54A. The activity assay was carried out at 35 °C for 30 min and the specific activity under the optimum pH was defined as 100%. (D) and (E) Thermostability of WT and S54A was measured under different temperatures for 2 h, and samples were taken every 20 min for 2 h. The specific activity without incubation was defined as 100%. (F) pH stability of WT and S54A. The residual activity was measured at 35 °C for 30 min and the highest activity was defined as 100%.
Homology modelling. Homology modelling for PveGS was performed on the basis of target-template alignment, and its initial partial geometry. Heavy-atom coordinates were obtained from the conserved residues which were found between the template and PveGS. The backbone coordinates were collected when residue identity was variable. Backbone geometries were modelled from fragments of high-resolution chains from the PDB (Protein Data Bank) when there were no assigned backbone coordinates according to previous library 44 . Two appropriate structural templates were from PDB among a set of pre-alignment family, with an E value of 9.8 × 10 −231 and 7.3 × 10 −5 . The best hit template 1F1H (PDB ID) is the crystal structure of glutamine synthetase from S. typhimurium, which was adapted as modelling template after removing the first amino acid methionine of PveGS. The model quality was evaluated using Ramachandran plot by MOE. The result of Ramachandran plot suggested that 462 residues were in the maximum allowable area of the plot excluding six residues (Phe81, Met98, Lys386, Lys395, Thr406 and Ser468), indicating that the model was stable in stereochemistry.
The structure modelling and molecular weight analysis indicated that it is a dodecameric structure consists of two stacked hexamers as the typical GSI enzymes. Hydrogen bonding and hydrophobic interactions hold the two GS rings together. Each subunit possesses a C-terminus and N-terminus, in which C-terminus stabilizes the GS structure by insertion into the hydrophobic region of the subunit across the other ring. The N-terminus located on the surface and exposed to the solution environment. In addition, the central channel is formed via six four-stranded β-sheets composed of anti-parallel loops from the twelve subunits 38,45 . The PveGS monomer model (red chain) superimposed to a subunit (cyan chain) of 1F1H hexamer ring were showed in Fig. 2A, as the black arrow pointing at, and different colours represented the different monomers. The superimposition between the model of PveGS and the crystal structure of 1F1H from S. typhimurium (green) was presented in Fig. 2B. The homology model constructed by MOE2009 and the structure of α-helices and β-strands (yellow) was depicted in Fig. 2C.

Ligand interaction analysis. Every
PveGS monomer possessed an active site named 'bifunnel' (Fig. 2D), which is the binding site of three distinct substrates: nucleotide, ammonium ion, and amino acid 4,46,47 . The bifunnel top was the binding site for ATP, ADP as well as nucleic acids, and glutamine, glutamate, together with ammonium bind to the bottom region. As demonstrated in Fig. 2D, ADP located in the upside of bifunnel and two ammonium analogues, thallous ion, and ammonia group of glutamate binding sites, were in the lower end. Ligand interaction revealed that ADP molecules located in a pocket formed by Asp50′ (from neighbour chain), Cys90, Ile92, Glu94, Lys170, Tyr180, Pro182, Glu221, Val222 and Glu293 (Fig. 2E). Tl + 473 interacts with Met49, Ser54, Tyr179′, His210, Glu212 and Val213 (Fig. 3A) in which Ser54, Asp 51, Glu212 and Tyr179′ (Fig. 3B) form a negatively charged ammonium pocket. The ammonium ion could occupy this pocket and donate hydrogen bonds, similar to that of the GS model from S. typhimurium as described by Liaw et al. 4 . Tl + 474 located in the negatively charged pocket for binding the amino group of the glutamate in the biosynthetic reaction, interacts with Glu132, Glu213, Gln219, Gly266, Ser267, Gly268 and His270 (Fig. 3C).
Based on the homology model, interaction analysis and sequence alignment, site-directed mutagenesis was performed on V15A, A36T, S54A, L126F, E213L, T224A, A249V and N265A. Among the generated mutants, S54A showed increased activity than wild type as well other mutants and thus selected for further investigation. The homology model was constructed for mutant GS S54A using the same template and the predicted E value was 2.2 × 10 −230 . From the homology model, it suggested that about 10 residues are out of the maximum allowable area, but it is still stable in stereochemistry. After site-directed mutagenesis in site Ser54, it was changed into Ala54 and the conformational analysis and ligand interaction of Tl + 473 enclosed by amino acids was presented in Fig. 3D,E.
Effects of temperature and pH on enzyme activity and stability. The relative activity of the native (PveGS) and mutant (S54A) at various temperatures was presented in Fig. 1B. The optimal temperatures for the native and mutant enzymes were 35 °C and their activities were defined as 100%. PveGS activity declined sharply over 40 °C and showed no activity at 0 °C, but S54A retained ~100% activity at 40 °C, 64% at 10 °C and 35% at 0 °C. This indicated that the mutant has been adopted to show activity in the cold conditions i.e. 0 and 10 °C. In the thermo-stability test, the wild and mutant enzymes retained over 95% activity after incubation at 40 °C for 2 hours. The activity of WT was decreased sharply after 40 min of incubation at 50 °C, however, S54A retained 50% activity after 60 min. (Fig. 1D,E). The native and mutant enzymes had the optimum pHs of 8.0 and 7.0 respectively (Fig. 1C). In the pH stability test at pH 11 the mutant S54A and native can retains 60 and 40% activities respectively. Additionally, S54A exhibited a respectable stability at the acidic condition (pH 4.0), retaining 60% activity as compared to 40% for the native PveGS (Fig. 1F).

Effects of metal ions and chemical reagents.
These effects of metal ions and chemical reagents were determined to check their activation effects and stability of enzyme. The activity of wild type enzyme was sharply increased by Mn 2+ , NH 4 + (1 and 5 mM) and Na + (5 mM) and moderately increased by K + , Mg 2+ , Cu 2+ , Co 2+ , Li + , Ni 2+ and Na + at tested concentrations. The mutant S54A, has shown highest activity in the presence of Mn 2+ (1 and 5 mM) followed by the other metal ions. But the activity of wild type was inhibited in the presence of Ba 2+ , and Fe 3+ (Table S2). S54A was highly stable at 5 mM Ni 2+ , Li + , K + , Ba 2+ and Mg 2+ , retaining158.9, 142.3, 141.3, 141.6 and 131.7% activity respectively, followed by Na + , NH 4 + , Co 2+ , Cu 2+ and Zn 2+ , while the wild type enzyme showed about 100% activity only at 5 mM Ni 2+ , Li + , K + , Ba 2+ and Mg 2+ . The activity of both the enzymes was completely inhibited in the presence of 5 mM EDTA (Table S3).

Effects of detergents.
To determine the potential industrial scale application of PveGS, its tolerance towards various detergents was analysed (Table S4). These results indicated that the WT enzyme has better tolerance than S54A, indicating S54A was more sensitive to the solution environment. The wild type enzyme noted the highest relative activity of 160 and 154% in 5 mM Triton X-100 and Tween 80, respectively, higher than that in the presence of non-ionic detergents such as 5 mM Tween 20 and 0.5% CHAPS as well as 5 mM cationic surfactant CTAB. However, the anionic surfactant SDS inhibited the activity at the concentration of 0.1 and 0.5% when compared to the control. Meanwhile, the mutant activity decreased to about 18 to 40% relative to that of the control and the native. The activity of native enzyme was also inhibited in the presence of 0.1 and 0.5% SDS. Kinetic measurements. The enzyme kinetic parameters for the wild type and mutant enzymes were presented in Table 1. The kinetic analysis indicated that the K m of PveGS against hydroxylamine, ADP-Na 2 and L-Glutamine were (15.7 ± 1.1), (25.2 ± 1.5) × 10 −5 and (32.6 ± 1.7) mM, and the k cat were (17.0 ± 0.6), (9.14 ± 0.12) and (30.5 ± 1.0) s −1 respectively. The kinetic analysis of mutant S54A suggested that the K m of PveGS against hydroxylamine, ADP-Na 2 and L-Glutamine were (4.6 ± 0.3), (10.2 ± 5.6) × 10 −6 and (23.8 ± 1.2) mM, and the k cat were (19.6 ± 0.5), (15.2 ± 0.2) and (41.9 ± 1.1) s −1 respectively. The catalytic efficiency against hydroxylamine, ADP-Na 2 and L-Glutamine of the mutant S54A was increased by 3.1-, 0.89 and 2.9-folds respectively, as compared to wild type PveGS,

Discussion
In the present study, both the native and site-directed-mutagenesis-modified glutamine synthetases from marine P. vermicola were investigated for their catalytic properties under optimized conditions. The native (PveGS) and mutant (S54A) enzymes were characterized in terms of optimum temperature, pH, and thermal stability. The PveGS and S54A were found to be highly active at 40 °C for 2 h, which is similar to the previously reported GSI of P. ruminicola 48 . The mutant S54A showed a relatively high residual activity, about 64 and 35% at 10 °C and 0 °C respectively. These findings have been supported by a previous study similar to the wild esterase and lipase reported from Psychrobacter sp. Ant 300 and Pseudomonas sp.732 49,50 . As reported previously, the low ratio of Arginine/(Arginine + Lysine), low proportion of proline residues, small hydrophobic core, lesser salt bridges and aromatic-aromatic interactions are the common characteristics for a cold-adapted enzyme [51][52][53][54] . A smaller number of ion pairs and weak charge-dipole interactions in αhelices could contribute to the poor thermo-stability 55 . In the present investigation, S54A was located on the loop of the active site, which leads to more plasticity and flexibility to the structure. Meanwhile, the hydrogen bonds between the product and enzyme was destroyed, which is facilitating the release of the substrate at a lower temperature. The cold-adapted enzyme is a promising source to the pharmaceutical industries, agriculture and chemical industries 56 , for cold conditions are required for biosynthesis of fragile pharmaceutical compounds to avoid adverse side-effects and conserve energy 54 .
Glutamine synthetase is a metallo-protein possessing a metal binding site in its active site to accommodate the cofactors (Mg 2+ or Mn 2+ ) that are important for its activity. Cations binds to the metal binding site of the enzyme as previously reported 57 . A previous study revealed that there were mono-valent cation sites in GS so that these cations could stabilize the quaternary structure of GS and alkali ions could compete partially with NH 2 OH in the γ-glutamyl transfer reaction 4 . PveGS and S54A exhibited a relatively low residual activity (6.1 and 4.4%) in the presence of 5 mM EDTA, because the combination of Mn 2+ with EDTA resulted in the lack of activating metal ions 58 . The presence of cationic detergent (CTAB) and anion detergent (SDS) can break the non-covalent bond between the proteins, leading to the decrease of the activity. Non-ionic interaction between the protein and lipid separated, and the relative residual activity of PveGS and S54A was different at a higher concentration, which was like a previous report about the glutamine synthetase from Azospirillum brasilense 58 .
The ligands to the Tl + 473 ion are shown as atomic models in which Ser 54, Asp 51, Glu 212′, and Tyr 179′ (from the neighbouring chain) form a negatively charged ammonium pocket, and the ammonium ion could occupy the site and donate hydrogen bonds, which is similar to a previously described model of GS from S. typhimuriu 4 . Then, the negatively charged pocket forms the binding site of Tl + 474 to bind the amino group of the glutamate. Both the negatively charged pockets were situated on the same half side of the bottom of the bifunnel. The other half side of the bottom, which is positively charged, stabilizes the R-and γ-carboxylate groups of the glutamate in biosynthetic activity as well as the glutamine in γ-glutamyl transfer activity 41 .
After site-directed mutagenesis in site Ser54, was mutated into Ala54, and the conformational analysis indicated that the mutation destroys the side chain interaction between Gly56 and Ser54 as well as the receptor contact of Tl + 473, which was the ammonium group of the substrate-binding site of hydroxylamine and cofactor ADP 4 . Meanwhile, the exposure of the ligand atoms was also changed. The substrate hydroxylamine bound with the negatively charged ammonium pocket where the binding of S54A was not as strong as that of the wild type. The ammonium group could transfer to the glutamine with less effective hydrogen bonds, resulting in higher catalytic efficiency. On the other hand, the S54A has changed the hydrogen bonds to the L-Glutamic acid γ-monohydroxamate, thus improving the catalytic efficiency for different types of substrates.

Material and Methods
Bacterial strains and plasmid. The marine bacterium P. vermicola strain CGS6 was obtained from Marine Culture Collection of China (http://www.mccc.org.cn/). P. vermicola was grown in nutrient agar medium containing peptone (1%, w/v), yeast extract (0.5%, w/v), NaCl (2%, w/v), and agar (1.5%, w/v). Plasmid pGEX-6p-1 (GE Healthcare, USA) was used as a vector for protein expression. For cloning and expression, E. coli DH5α ADP-Na2 L-Glutamine Hydroxylamine Cloning and Site-directed mutagenesis. The PveGS gene was amplified using PveGS-F and PveGS-R primers (Table S1) containing BamHI and XhoI restriction sites respectively. The genomic DNA of P. vermicola was used as template. Then the target gene was cloned into pGEX-6P-1 vector by BamHI and XhoI dual digestion and T4 DNA ligase ligation. The ligation product was transferred into E. coli DH5α competent cells for screening of positive recombinant. The recombinant plasmid pGEX-6P-1-PveGS was used as a template for the PCR with the primers (Table S1) to perform site-directed mutagenesis. The PCR products were digested with DpnI to remove the template plasmids, and subsequent experiments were carried out as mentioned above.
Assay of glutamine synthetase activity. The GS activity assay contained 18 mM hydroxylaminehydrochloride, 20 mM L-Glutamine, 1 mM MnCl 2 , 25 mM potassium arsenate, 0.4 mM ADP-Na 2 , and 135 mM imidazole hydrochloride. The reaction was terminated by adding 200 µL 'stop mixture' consisting of 55 g FeCl 3 ▪6H 2 O, 20 g trichloroacetic acid and concentrated HCl (21 mL/L). The absorbance was measured at 540 nm after centrifugation of reaction mixture 36,60 . The L-Glutamic acid γ-monohydroxamate was added into the reaction mixture without enzyme to give a concentration from 0 to 6 mM for quantification by the absorbance at 540 nm. One unit of enzyme activity is defined as the amount of PveGS required convert 1µmol L-Glutamic acid γ-monohydroxamate in one minute in one ml of reaction mixture under optimized conditions.
Effect of temperature and pH. The assay mixture and enzyme were incubated in the temperature gradient from 0 to 60 °C for 5 min and the reaction was initiated by adding L-Glutamine and terminated after 30 min incubation using stop reagent described in previous sections. The thermostability assay was conducted in the temperature gradient from 30 to 70 °C and the samples were taken after every 20 min for 2 h. The residual activity was determined under the optimized conditions. Phosphate-citrate (pH 4.0-8.0) and Glycine-NaOH (pH 8.0-11.0) buffer were used to analyze the effect pH of 4.0 to 8.0 and 8.0 to 11.0, respectively. pH stability was determined by calculating the residual activity left after incubating the enzyme for 24 hours at 4 °C.
The kinetic parameters K m and k cat . The substrates of the PveGS, L-Glutamine, hydroxylaminehydrochloride and ADP-Na 2 were added into the mixture with the concentration of 2-40 mM, 0.05-20 mM and 0.00001-0.01 mM, respectively, to obtain the initial rate of reaction. Michaelis-Menten constant 61 was measured according to Lineweaver-Burkplot, and the value of V max , the molecular mass and the concentration of the purified protein were used to calculate the catalytic constant (k cat ). Graphpad Prism software (Graphpad, San Diego, CA, USA) was also used to calculate the parameters for consistency and accuracy.

Molecular docking.
To search for the active site and investigate the interaction surface of the GS, molecular docking was performed. After obtaining the homologous sequences with a reasonable value, the structure containing the ligand was selected as the target substrate for docking. In present study, a GSIβ glutamine synthetase protein structure from S. typhimurium 41 (PDB ID: 1F1H) was adapted, which containing ligand like ADP, and two important ammonium analogues, Tl + , which were coincidence with ammonium substrate binding site and binding the ammonium group of the glutamate. Based on the homology modelling to WT and S54A, molecular docking analysis was carried out on the instruction of docking tutorial in MOE 2009.