Structural Insights into a Novel Class of Aspartate Aminotransferase from Corynebacterium glutamicum

Aspartate aminotransferase from Corynebacterium glutamicum (CgAspAT) is a PLP-dependent enzyme that catalyzes the production of L-aspartate and α-ketoglutarate from L-glutamate and oxaloacetate in L-lysine biosynthesis. In order to understand the molecular mechanism of CgAspAT and compare it with those of other aspartate aminotransferases (AspATs) from the aminotransferase class I, we determined the crystal structure of CgAspAT. CgAspAT functions as a dimer, and the CgAspAT monomer consists of two domains, the core domain and the auxiliary domain. The PLP cofactor is found to be bound to CgAspAT and stabilized through unique residues. In our current structure, a citrate molecule is bound at the active site of one molecule and mimics binding of the glutamate substrate. The residues involved in binding of the PLP cofactor and the glutamate substrate were confirmed by site-directed mutagenesis. Interestingly, compared with other AspATs from aminotransferase subgroup Ia and Ib, CgAspAT exhibited unique binding sites for both cofactor and substrate; moreover, it was found to have unusual structural features in the auxiliary domain. Based on these structural differences, we propose that CgAspAT does not belong to either subgroup Ia or Ib, and can be categorized into a subgroup Ic. The phylogenetic tree and RMSD analysis also indicates that CgAspAT is located in an independent AspAT subgroup.


Introduction
Corynebacterium glutamicum (initially reported as Micrococcus glutamicus) is an aerobic, gram-positive, short and rod-shaped bacterium, and is known to be involved in the fermentation of various amino acids, such as L -lysine and L -glutamate [1]. Over several decades, various C. glutamicum mutants have been isolated, which produce significant amounts of different Lamino acids [2]. In 2003, the genome of C. glutamicum ATCC 13032 was sequenced, providing an abundant source of data on the metabolic pathways used for the biosynthesis of industrially important products [2,3]. During last 10 years, the market size of amino acid for livestock has been continuously increasing. L -methionine, L -lysine, L -threonine, L -tryptophan, and L -valine are essential amino acids manufactured industrially. [2,4,5].
In this study, we report a crystal structure of AspAT from Corynebacterium glutamicum ATCC13032 (CgAspAT) in complex with a PLP and citrate molecule. Compared with AspATs from subgroup Ia and Ib, CgAspAT showed unusual structural features at the auxiliary domain and unique binding sites for both cofactor and substrate. Based on these observations, we propose that CgAspAT can be categorized into a new subgroup, subgroup Ic.

Cloning, expression and purification
The gene coding CgAspAT was amplified through polymerase chain reaction (PCR) using C. glutamicum strain ATCC 13032 chromosomal DNA as the template. The PCR products were subcloned into pET30a (Novagen), and the resulting expression vector pET30a:CgAspAT was transformed into a E. coli BL21(DE3)-T1 R strain, which was grown in 1 L of LB medium containing 100 μM kanamycin at 310 K. At an OD 600 of 0.65, CgAspAT protein expression was induced by adding 1 mM IPTG. After 20 h at 293 K, the cells were harvested through centrifugation at 4,000 g for 20 min at 277 K. The cell pellet was resuspended in buffer A (40 mM Tris-HCl pH 8.0) and disrupted by ultrasonication. The cell debris was removed through centrifugation at 13,500 g for 25 min, and the lysate was applied onto a Ni-NTA agarose column (Qiagen). After washing with buffer B (40 mM Tris-HCl pH 8.0 and 30 mM imidazole), the bound proteins were eluted with buffer C (40mM Tris-HCl pH 8.0 and 300 mM imidazole). Finally, the trace amount of contaminants was removed by size-exclusive chromatography using a Sephacryl S-300 prep-grade column (320 ml, GE Healthcare) equilibrated with buffer A. The eluted protein had a molecular weight of about 90 kDa, indicating a dimeric structure. The protein was concentrated to 50 mg mL -1 using spin column (Amicon Ultra Centrifugal Filter, 30 kDa pore size), and kept at 193 K for further experiments. All purification steps were performed at 277 K.

Crystallization and data collection
Crystallization of the purified CgAspAT protein was initially performed with commercially available sparse-matrix screens including Index, PEG I and II (from Hampton Research), Wizard Classic I and II, Wizard CRYO I and II (from Rigaku Reagents), and Structure Screen I and II (from Molecular Dimensions), using the sitting-drop vapor-diffusion method on the MRC Crystallization plate (Molecular Dimensions) at 295 K. Each experiment consisted of mixing 1.0 μL protein solution (50 mgÁmL −1 , 40 mM Tris-HCl pH 8.0) with 1.0 μL reservoir solution and then equilibrating against 50 μL reservoir solution. CgAspAT crystals were observed in several crystallization screening conditions. After several steps of improvement using the hanging-drop vapor-diffusion methods, crystals of the best quality appeared using 28% polyethylene glycol 3350 and 200 mM Ammonium citrate dibasic as reservoir solution and protein concentration of 50 mgÁmL −1 , were finished out with a loop larger than the crystals and flash-frozen by immersion in liquid nitrogen at 100 K. The data were collected to a resolution of 2.0 Å at beamline 7A at the Pohang Accelerator Laboratory (PAL, Pohang, Republic of Korea) using a Quantum 270 CCD detector (ADSC, USA) ( Table 1). All data was indexed, integrated, and scaled together using the HKL2000 software package [22]. The crystals of CgAspAT belonged to the space group C2. Assuming two CgAspAT molecules in the asymmetric unit, the crystal volume per unit of protein mass was 2.45 Å 3 ÁDa −1 , which means that the solvent content was 49.92% [23].

Structure determination
The structure was determined by molecular replacement using the CCP4 version of MOLREP [24] and the structure of the putative aminotransferase from Corynebacterium diphtheria (PDB code 3D6K) as the search model. Model building was performed manually using the program WinCoot [25], and refinement was performed with CCP4 refmac5 [26] and CNS [27]. The data statistics are summarized in Table 1. The refined CgAspAT model was deposited in protein data bank (PDB code 5IWQ).

Site-directed mutagenesis and activity assay
Site-specific mutations were created using the Quick Change kit (Stratagene), and sequencing was performed to confirm the correct incorporation of the mutations. Mutant proteins were purified in the same manner as the wild type. Because the reaction of AspAT cannot detect directly, the CgAspAT activity assay was performed in duplicate reaction with malate dehydrogenase. At the first reaction, CgAspAT enzymes catalyze L -asparate and α-ketoglutarate to Lglutamate and oxaloacetate, and at the second reaction, oxaloacetate is converted to malate, by malate dehydrogenase using NADH as a cofactor. Oxidation of NADH to NAD was monitored through the decrease of absorbance at 340 nm (extinction coefficient of 6.22 x 10 3 M −1 Ácm −1 ) (S1 Fig). All assays were formed with a reaction mixture of 1 mL total volume, and performed at 303 K. The reaction mixture contained 100 mM Tris-HCl, pH 8.0, 300 mM L -aspartate, 50 mM α-ketoglutarate, 80 μM NADH, and 10 unitÁmL −1 malate dehydrogenase. The reaction was initiated by the addition of enzyme to a final concentration of 1 μgÁmL −1 . The decreased absorbances at 340 nm for 20 sec after the initiation of the reaction were measured.

Results and Discussion
Overall structure of CgAspAT In order to elucidate the molecular mechanism of CgAspAT, we determined the crystal structure of the protein at 2.0 Å. There were two CgAspAT molecules in each asymmetric unit, which form a homodimer of the protein. The atomic structure was in good agreement with the X-ray crystallographic statistics for bond angles, bond lengths, and other geometric parameters ( Table 1). The monomer structure of CgAspAT consists of two domains: a core domain (Leu56-Gly304) and an auxiliary domain (Ala55-Lys34 and Asp305-Asn426). The core domain has 10 α-helices (α4-α13) that surround the 7-stranded (β1-β7) β-sheet located in the middle of the domain (Figs 1B and 2A). The auxiliary domain has 6 α-helices (α1-α3 and α14-α16), with five arranged in a line, and 2 β-strands (β8-β9), which are located between the αhelix line and the core domain (Figs 1B and 2A). Dimerization is mainly mediated through the contact between the core domains of the two subunits with three α-helices (α6, α8, and α12) of one molecule in contact with the corresponding α-helices of the other molecule ( Fig 2B). For dimerization, 63 amino acids are involved in each monomer, and a total of 2472.5 and 2486.2 Å 2 of solvent-accessible surface areas per each monomer are buried, which correspond to 13.9% of the total surface areas of each monomer, respectively. The cofactor-and substratebinding pockets are at the domain interface, which will be described in the following subsections.

PLP binding mode and catalytic residue of CgAspAT
Since we observed the electron density for the PLP cofactor in our structure without adding it in the crystallization process (Fig 3A), we speculate that the PLP cofactor binds strongly to CgAspAT. Moreover, the enzyme exhibited AspAT activity without the addition of PLP in the reaction mixture, supporting the high affinity between the enzyme and the PLP cofactor. The PLP-binding pocket is composed of six loops (α3-α4, β1-α6, β2-α8, β4-α10, β5-α11, and β6-β7). Pyridoxal ring was stabilized by the hydrogen bonds with residues Tyr142, Asn191, Asp220, Tyr223, and Lys259. The nitrogen atom of pyridine ring was hydrogen bonded with Asp220, and the carbonyl group with Asp220 and Tyr223. Tyr223 was also involved in the interaction with the aldehyde-group with the aid of Tyr142 and Lys259 (Fig 3B). Conserved hydrophobic residues, such as Leu105 and Val186, also contribute to the stabilization of pyridoxal ring, among the residues involved in PLP binding, Asn191, Asp220, and Tyr223 were conserved in AspAT enzymes from other organisms (Figs 1B and 3B).
Interestingly, the phosphate moiety was stabilized in a unique mode. Unlike the binding of the phosphate moiety in other AspAT enzymes, where side chains of arginine, threonine, and one or two serine residues are involved, the moiety of CgAspAT was stabilized by the side chains of one tyrosine (Tyr73) and four serine residues (Ser103, Ser104, Ser256, and Ser258) through a hydrogen bond network (Fig 3B). In detail, Tyr73 and Ser258 form hydrogen bonds   with the O1 atom, and Ser103 and Ser256 stabilize the O2 and O3 atoms, respectively. Moreover, Ser104 forms hydrogen bonds with both O3 and O4 atoms (Fig 3B). Near the aldehyde group of the bound PLP, we observed the conserved Lys259 residue. Since this lysine residue is known to function as a catalytic residue in other AspAT enzymes by forming an enzyme aldimine covalent bond, we speculate that CgAspAT may catalyze the enzyme reaction in a way similar to other AspAT enzymes. The residues involved in the cofactor binding were further confirmed by site-directed mutagenesis experiments. Mutants, such as Y73A, Y142A, N191A, D220A, Y223A, and K259A, exhibited almost complete loss of enzyme activity, indicating that these residues are crucial for PLP binding. However, mutations of serine residues (S103A, S104A, and S258A) involved in the stabilization of the phosphate moiety showed~70% AspAT activities compared with the wild type, indicating that these residues, individually, do not influence much the stabilization of the phosphate moiety ( Fig 3C).

Substrate binding mode of CgAspAT
In our current structure, one citrate molecule, a component of the crystallization solution, is observed to be bound (Fig 4A and 4B), and the superposition of our structure with the AspAT from Escherichia coli (EcAspAT) structure in complex with the glutamate substrate (pdb code 1X28) allows us to speculate the substrate binding of CgAspAT. In EcAspAT, the α-carboxyl group of glutamate is stabilized by hydrogen bonds with Asn183 and Arg374 residues. The α-carboxyl group of glutamate might be similarly stabilized in CgAspAT, because residues of Asn191 and Arg394 are located at the corresponding positions of Asn183 and Arg374, respectively ( Fig 4C). However, the binding mode of the γ-carboxyl group of glutamate in CgAspAT seems to be quite different compared with that in EcAspAT. In EcAspAT, four residues such as Trp130, Arg280, Ser284, and Asn285 are involved in stabilization of the γ-carboxyl group of glutamate by direct and water-mediated hydrogen bonds, and none of these residues are conserved in CgAspAT. In CgAspAT, Arg39, Tyr142 and Arg144 are positioned in the vicinity of the γ-carboxyl group of glutamate ( Fig 4C). As observed above, in CgAspAT, a citrate molecule is bound at the position that the glutamate substrate is bound in EcAspAT. Because the citrate molecule is stabilized by hydrogen bonds with the highly positively-charged residues such as Arg39, Lys41, Arg144 and Arg394, binding of the citrate molecule seems to mimic the substrate binding of CgAspAT (Fig 4B and 4C). The involvement of these residues in substrate binding is further confirmed by site-directed mutagenesis experiments. When we replace the Arg39, Tyr142, Arg144, Asn191 and Arg394 residues by alanine, all mutants showed reduced or almost complete loss of enzyme activity, indicating that these unique residues are crucial for the binding of the glutamate substrate in CgAspAT (Fig 3C).

Comparison of CgAspAT with subgroup Ia and Ib
AspAT enzymes belong to the aminotransferase class I. Based on amino acid sequence similarity, class I was further divided into two subgroups, subgroup Ia and Ib. Although amino acid sequence similarities within enzymes in subgroup Ia and Ib are about 40%, that between enzymes in subgroups Ia and Ib is about 15% [12]. Subgroup Ia includes mostly eukaryotic and some bacterial AspAT enzymes, while some extremophiles belong to subgroup Ib. However, PLP binding mode of CgAspAT. The bound PLP cofactor is presented as a stick model with a magenta color. Residues involved in the binding of PLP are shown as stick models and labeled appropriately. The core domain and the auxiliary domain of one molecule are distinguished with orange and light-blue colors, respectively, and the other molecule is with a cyan color. Hydrogen bonds involved in the PLP binding are shown with red-colored dotted lines. (C) Site-directed mutagenesis of CgAspAT. Residues involved in binding of cofactor and substrate are replaced by alanine residues. The relative activities of recombinant mutant proteins were measured and compared with that of wild-type CgAspAT. doi:10.1371/journal.pone.0158402.g003 Crystal Structure of a Novel Class Aspartate Aminotransferase from Corynebacterium glutamicum we observed that CgAspAT has unique structural features compared with AspAT enzymes from subgroups Ia and Ib, and proposed that CgAspAT can be categorized as a new subgroup due to the following reasons.
First, although the overall fold of the core domain in the case of CgAspAT is similar to that of other AspAT enzymes from subgroup Ia and Ib, the auxiliary domain of CgAspAT showed a unique structural conformation. In AspATs from subgroups Ia and Ib, the N-terminal region is positioned in the vicinity of the core domain, constituting part of the substrate binding site. However, in CgAspAT, the N-terminal region is positioned away from the core domain and is not involved in the formation of the substrate binding-site (Fig 5A).
Second, CgAspAT uses unique residues for the stabilization of the PLP cofactor. Although most AspATs share conserved residues for enzyme catalysis and stabilization of the pyridoxal  ring binding, the binding mode of the phosphate moiety was quite different from the other subgroups. In AspATs from subgroup Ia, one each of Thr, Tyr, and Arg and two Ser residues are involved in binding phosphate moiety, while in subgroup Ib only one each of Ser, Tyr and Arg residues. However, CgAspAT uses four Ser and one Tyr residues for the stabilization of the phosphate moiety ( Fig 5B). Moreover, two hydrophobic residues, Leu105 and Val186 that contribute to the stabilization of PLP in CgAspAT, are not conserved in enzymes in subgroup Ia and Ib. In the position of Leu105, Lys and Thr are positioned in subgroup Ia and Ib, respectively. And in the position of Val186, His and Asn are positioned in subgroup Ia and Ib, respectively. These residues are quite conserved within their subgroups, indicating that AspATs from different subgroups have distinctive PLP binding modes (Fig 1B, S2 and S3 Figs).
Finally, CgAspAT also has a unique mode for substrate binding. In AspATs from subgroup Ia, residues, such as two Arg, Asn, and Trp (or Tyr), are involved in substrate binding. AspATs from subgroup Ib stabilize the substrate in a way similar to those from AspATs from subgroup Ia, where Arg, Asn, Tyr, and Trp (or Tyr) are involved. On the other hand, CgAspAT utilizes unique residues, such as one each of Tyr and Lys, dyad and three Arg residues ( Fig 5C). As observed in the residues involved in the PLP binding, these residues are also highly conserved within its subgroup, and we suspect that these unique substrate binding-modes may be one of the decisive structural features of each subgroup (Fig 1B, S2 and S3 Figs). Interestingly, we found three unpublished AspAT structures in the protein data bank, AspATs from Corynebacterium diphtheria (CdAspAT, PDB Code: 3D6K), Mycobacterium tuberculosis (MtAspAT, PDB Code: 5C6U) and Deinococcus geothermalis (DgAspAT, PDB Code: 3EZ1), and these enzymes have structural features similar to CgAspAT (Fig 5D, 5E and 5F), indicating that these enzymes can belong to the category same as CgAspAT.
We then performed a phylogenetic tree analysis of AspATs from various organisms (Fig 6). As expected, AspATs from subgroup Ia and subgroup Ib were located away from each other. However, CgAspAT did not belong to either subgroup Ia or subgroup Ib, confirming that CgAspAT can be categorized as a new subgroup. Moreover, CdAspAT, MtAspAT and DgAs-pAT enzymes show~40% amino acid sequence similarity with CgAspAT and are located close to CgAspAT in the phylogenetic analysis (Fig 6). Here, combined with the unique structural features described above, we propose that CgAspAT, together with CdAspAT, MtAspAT and DgAspAT enzymes, can be categorized as subgroup Ic.

Conclusion
We determined the crystal structure of CgAspAT in complex with the PLP cofactor and we speculated glutamate substrate binding site compare to EcAspAT structure. Many biochemical and structural studies on AspATs have been reported for decades, which categorize these enzymes into two subgroups, subgroup Ia and subgroup Ib. However, CgAspAT shows only 10~15% amino acid sequence similarity with AspATs from subgroup Ia and subgroup Ib and exhibits unique structural features not only in the auxiliary domain but also at the catalytic and substrate binding sites. These amino acid sequence analysis and the structural comparisons reveal that CgAspAT does not belong to either subgroup Ia or subgroup Ib, but rather should be classified to a new category, subgroup Ic. Moreover, the RMSD analysis of the reported AspAT structures supports new classification of enzymes in this family (S1 Table.).