Purification of three aminotransferases from Hydrogenobacter thermophilus TK-6 – novel types of alanine or glycine aminotransferase

Enzymes and catalysis

Authors


M. Ishii, Department of Biotechnology, The University of Tokyo, Yayoi 1-1-1, Bunkyo-ku, Tokyo 113-8657, Japan
Fax: +81 3 5841 5272
Tel: +81 3 5841 5143
E-mail: amishii@mail.ecc.u-tokyo.ac.jp

Abstract

Aminotransferases catalyse synthetic and degradative reactions of amino acids, and serve as a key linkage between central carbon and nitrogen metabolism in most organisms. In this study, three aminotransferases (AT1, AT2 and AT3) were purified and characterized from Hydrogenobacter thermophilus, a hydrogen-oxidizing chemolithoautotrophic bacterium, which has been reported to possess unique features in its carbon and nitrogen anabolism. AT1, AT2 and AT3 exhibited glutamate:oxaloacetate aminotransferase, glutamate:pyruvate aminotransferase and alanine:glyoxylate aminotransferase activities, respectively. In addition, both AT1 and AT2 catalysed a glutamate:glyoxylate aminotransferase reaction. Interestingly, phylogenetic analysis showed that AT2 belongs to aminotransferase family IV, whereas known glutamate:pyruvate aminotransferases and glutamate:glyoxylate aminotransferases are members of family Iγ. In contrast, AT3 was classified into family I, distant from eukaryotic alanine:glyoxylate aminotransferases which belong to family IV. Although Thermococcus litoralis alanine:glyoxylate aminotransferase is the sole known example of family I alanine:glyoxylate aminotransferases, it is indicated that this alanine:glyoxylate aminotransferase and AT3 are derived from distinct lineages within family I, because neither high sequence similarity nor putative substrate-binding residues are shared by these two enzymes. To our knowledge, this study is the first report of the primary structure of bacterial glutamate:glyoxylate aminotransferase and alanine:glyoxylate aminotransferase, and demonstrates the presence of novel types of aminotransferase phylogenetically distinct from known eukaryotic and archaeal isozymes.

Abbreviations
AGT

alanine:glyoxylate aminotransferase

CFE

cell-free extract

GGT

glutamate:glyoxylate aminotransferase

GOT

glutamate:oxaloacetate aminotransferase

GPT

glutamate:pyruvate aminotransferase

2-OG

2-oxoglutarate

PLP

pyridoxal 5′-phosphate

PSOT

phosphoserine:2-oxoglutarate aminotransferase

Introduction

Aminotransferase (EC 2.6.1) catalyses the conversion between amino acids and 2-oxo acids, transferring the amino group of the amino acid onto the 2-oxo acid. This enzyme is widespread, being present in almost all organisms, and plays a key role in the synthesis and degradation of amino acids. As the substrates/products of aminotransferase, namely 2-oxo acids and amino acids, are key metabolites in carbon and nitrogen metabolism, this enzyme can be regarded as a physiologically important linkage within central metabolism. Furthermore, some aminotransferases have been reported to be coupled with further metabolic activities, e.g. enzymes involved in the malate shuttle, porphyrin synthesis [1], maintenance of intracellular redox status [2] or plant photorespiration [3].

A wide variety of substrates for aminotransferases have been reported, including branched-chain amino acids, aromatic amino acids, β-amino acids and their corresponding 2-oxo acids. To categorize diverse aminotransferases, classifications based on the primary structure have been proposed. Such a classification divides aminotransferases into four families, numbered I–IV [4]. Family I is further divided into several subfamilies, such as Iα and Iγ [5]. In this classification system, enzymes belonging to the same family or subfamily share common enzymatic characteristics to some extent.

However, the substrate specificities of aminotransferases are diverse, even within the same family or subfamily; therefore, at present, it is difficult to predict the specificities on the basis of the primary structures only. One reason for this difficulty is that the reaction mechanisms and structures of aminotransferases may be similar to each other, even if they react specifically with different substrates. Moreover, there are only a limited number of aminotransferases whose enzymatic properties and primary sequences have been determined. For these reasons, the function of most putative aminotransferase homologues found in the genome database remains to be ascertained. Some recent studies have revealed properties of several putative aminotransferases by biochemical and enzymatic analyses [6–8], demonstrating the importance of a biochemical approach for the characterization of these enzymes.

Hydrogenobacter thermophilus TK-6 is a thermophilic, hydrogen-oxidizing, obligately chemolithoautotrophic bacterium. The analysis of 16S rRNA sequences has shown that Hydrogenobacter species are located on the deepest branch in the domain Bacteria on the phylogenetic tree, together with other Aquificae species [9]. Reflecting this distinctive phylogenetic position, this bacterium shows many unique characteristics. One such characteristic is its carbon anabolism, where carbon dioxide is fixed via the reductive tricarboxylic acid cycle. Key enzymes in this cycle have been characterized and shown to have novel enzymatic features [10–13]. Furthermore, enzymatically peculiar characteristics have also been found in this bacterium’s nitrogen anabolism [14,15]. Although previous studies have demonstrated that H. thermophilus assimilates nitrogen in the form of ammonium to produce glutamate (Glu), it has not yet been clarified how Glu serves as the nitrogen donor for the synthesis of other nitrogenous compounds.

The study of aminotransferases in this bacterium is of interest, firstly because of the need to characterize biochemically aminotransferases. The importance of this is emphasized by the belief that a novel aminotransferase would be found in this phylogenetically deep-rooted bacterium. Secondly, this study was expected to lead to further elucidation of the metabolism of H. thermophilus. Such elucidation would not be restricted to nitrogen metabolism, but would also include its unique central carbon metabolism. In this study, three aminotransferases were purified and characterized biochemically and presumed to contribute to aspartate (Asp), alanine (Ala) and glycine (Gly) syntheses. Phylogenetic analysis of these enzymes showed a unique combination of substrate specificities and phylogenetic positions, providing novel insights into the aminotransferase classification.

Results

Aminotransferase activities in cell-free extract (CFE)

Given that H. thermophilus operates a distinctive carbon pathway, the reductive tricarboxylic acid cycle, its central carbon metabolism is of interest. Therefore, we focused on amino acids with relatively simple carbon skeletons: Glu, Asp, Ala and Gly. Aminotransferase activities in the CFE were assayed combining Glu, Asp, Ala or Gly as the amino group donor and 2-oxoglutarate (2-OG), oxaloacetate, pyruvate or glyoxylate as the amino group acceptor. Consequently, the following four kinds of activity were detected: 0.96 U·mg−1 glutamate:oxaloacetate aminotransferase (GOT; EC 2.6.1.1), 0.30 U·mg−1 glutamate:pyruvate aminotransferase (GPT; EC 2.6.1.2), 0.30 U·mg−1 glutamate:glyoxylate aminotransferase (GGT; EC 2.6.1.4) and 0.07 U·mg−1 alanine:glyoxylate aminotransferase (AGT; EC 2.6.1.44). Although the GOT reaction was catalysed reversibly, the other reactions proceeded irreversibly as follows:

image
image
image
image

Although GOT is a representative aminotransferase that has been studied extensively in many organisms [16–18], other aminotransferases have been less well studied, especially in bacteria. GPT has been purified and characterized in a few organisms, and only a limited number of GPT sequences have been determined [2,6,19]. GGT and AGT have been subjected to considerably less research. GGT has been purified from a few organisms [20], and only those from Arabidopsis thaliana have been sequenced [3]. AGT has been sequenced and characterized in eukaryotes and archaea [21,22], but not in bacteria. Because of this background, the characterization of these aminotransferase activities was expected to provide new insights into bacterial aminotransferases.

Purification and phylogenetic analysis of aminotransferases

Enzymes that exhibited GOT, GPT, GGT or AGT activity were subjected to purification, and three enzymes (AT1, AT2 and AT3) were purified from H. thermophilus CFE (Table 1). It was shown that GOT, GPT and AGT activities were derived from the single enzymes AT1, AT2 and AT3, respectively (Fig. 1). GGT activity was caused by AT1 and AT2, which exhibited 11 and 60 U·(mg purified protein)−1 of GGT activity, respectively. No other enzymes that exhibited GOT, GGT, GPT or AGT activity were detected throughout the purification, suggesting that the four kinds of activity in CFE were derived from only the three enzymes. Purified AT1, AT2 and AT3 gave single bands of 44, 42 and 45 kDa on SDS/PAGE, respectively (Fig. 2). The N-terminal amino acid sequences of AT1, AT2 and AT3 were determined to be MNLSKRVSHIKPAPT, MYQERLFTPG and MSEEWMFPKVKKL, respectively, and the full-length genes were identified in the H. thermophilus genome (AP011112). The molecular masses of AT1, AT2 and AT3 were calculated from their deduced protein sequences to be 43.7, 41.9 and 45.6 kDa, respectively. These masses were consistent with those calculated from SDS/PAGE.

Table 1.   Purification of AT1, AT2 and AT3 from H. thermophilus.
EnzymeFractionActivity (U)aProtein (mg)Specific activity (U·mg−1)aPurification (fold)Yield (%)
  1. Representing GOT activity (in the direction of Asp synthesis) for AT1, GPT activity for AT2 and AGT activity for AT3.

AT1CFE6366600.961100
Butyl-Toyopearl24513192039
DEAE-Toyopearl741.3596112
MonoQ610.2623924810
AT2CFE2759270.301100
Butyl-Toyopearl65242.7924
DEAE-Toyopearl313.0103511
Hydroxyapatite150.3511715
MonoQ100.13792664
AT3CFE12918110.0711100
Butyl-Toyopearl14710.19311
DEAE-Toyopearl5.43.71.4204
Hydroxyapatite2.10.425.0692
MonoQ2.90.368.01122
Phenyl Superose1.20.063192701
Figure 1.

 Aminotransferase reactions catalysed by AT1, AT2 and AT3. Glyo, glyoxylate; OAA, oxaloacetate; Pyr, pyruvate.

Figure 2.

 SDS/PAGE (13%) of purified AT1, AT2 and AT3. Lane 1, purified AT1; lane 2, purified AT2; lane 3, purified AT3; lane 4, molecular mass markers.

The phylogenetic tree was constructed on the basis of the amino acid sequences (Fig. 3). GOT is known to be divided into two groups in subfamilies Iα and Iγ, and AT1 belongs to aminotransferase subfamily Iγ together with some other GOTs. Unexpectedly, AT2 is classified into family IV together with eukaryotic peroxisomal AGT, whereas other GPTs are members of family I. Interestingly, AT3 was located in family I, unlike eukaryotic AGT. There is only one report of a family I AGT, which was purified from Thermococcus litoralis [22]. The order of divergence of AT3 from enzymes in subfamily Iγ is ambiguous in Fig. 3 because of the low bootstrap values, although more detailed phylogenetic analysis indicated that AT3 is positioned separately from the known members of subfamily Iγ (see below).

Figure 3.

 Phylogenetic tree of aminotransferases on the basis of the amino acid sequences. The numbers at the nodes are bootstrap confidence values expressed as percentages of 1000 bootstrap replicates. The order of the divergence was presumed to be reliable only when the bootstrap values were above 50. The tree was constructed using the neighbor-joining method and showed the same overall topology as that constructed by the maximum likelihood method. Plus signs indicate the activities proven experimentally. The accession numbers of each enzyme are shown in parentheses. Enzymes from the following organisms were used: Arabidopsis thaliana [3,21,26], Bacillus circulans [24], Bacillus sp. YM-2 [17], Corynebacterium glutamicum [6], Escherichia coli [25], Entamoeba histolytica [7], human [35], H. thermophilus, Pyrococcus furiosus [2,36], rat [19,37,38], Saccharomyces cerevisiae [39], Sulfolobus solfataricus [40], T. litoralis [22] and Thermus thermophilus [18].

Enzymatic properties

Gel filtration estimated the molecular mass of AT1 to be 78 kDa, indicating that this enzyme forms a dimer of two identical subunits, as do many known aminotransferases. The molecular masses of AT2 and AT3 were estimated to be 62 and 69 kDa, respectively. These values were 1.5-fold larger than each single peptide mass, indicating that these enzymes are monomers or homodimers. Considering that some thermophilic enzymes have compact folding and their molecular masses are often underestimated by gel filtration [14], AT2 and AT3 might form a homodimer, although it cannot be excluded that they are monomeric.

The effects of pH on the aminotransferase activities of AT1, AT2 and AT3 were tested. AT1 exhibited the highest GOT activities in both directions over a broad pH range, 6.9–7.9 at 70 °C. AT2 and AT3 showed the highest GGT and AGT activities, respectively, at pH 7.9–8.4. These natural or slightly basic optimum pH values are common among known aminotransferases. Some aminotransferases are known to be activated by the addition of pyridoxal 5′-phosphate (PLP), the catalytic cofactor of aminotransferase, to the reaction mixture [2]. The addition of PLP did not affect the activities of AT1, AT2 or AT3, suggesting that PLP binds tightly to these enzymes or extrinsic PLP cannot reactivate the apoenzymes.

AT1 catalyses the GOT reaction reversibly and the GGT reaction only in the direction of Gly synthesis. AT2 catalyses the GPT reaction in the direction of Ala synthesis, and shows only trace activity (< 5% of that in the forward direction) in the reverse direction. This enzyme also irreversibly catalyses the GGT reaction in the direction of Gly synthesis, as well as AT1. Many known GPTs catalyse the GPT reaction reversibly and lack GGT activity. GPTs from A. thaliana share these properties with AT2 [3], although these GPTs belong to subfamily Iγ distant from AT2, which is a member of family IV (Fig. 3). AT3 specifically catalyses the AGT reaction irreversibly in the direction of Gly synthesis. The irreversibility of GGT and AGT is a common feature among known GGTs and AGTs [20,22,23]. Although some eukaryotic AGTs have been reported to exhibit serine:pyruvate aminotransferase activity [21], AT3 did not show this activity, suggesting a high substrate specificity for Ala and glyoxylate compared with these AGTs.

Some members of family IV are known as phosphoserine:2-oxoglutarate aminotransferases (PSOT; EC 2.6.1.52), which catalyse the conversion of phosphoserine and 2-OG to phosphohydroxypyruvate and Glu [7,24,25]. AT2, which belongs to family IV, exhibited PSOT activity at 16 U·mg−1, corresponding to about one-quarter of its GGT activity. It is noteworthy that, although AT2 has a higher similarity to known AGTs than to known PSOTs, it does not have AGT activity but shows PSOT activity (Fig. 3).

Kinetic characterization

The kinetic parameters of AT1, AT2 and AT3 were determined for the reactions that followed typical Michaelis–Menten kinetics (Table 2). AT1 exhibited higher Vmax values in GOT reactions than in the GGT reaction. Km values for Glu, Asp and 2-OG in the GOT reaction were comparable with those of other reported GOTs [16,26]. With regard to GGT activity, both AT1 and AT2 showed Km values as low as those of known GGTs [3,20]. Although the GGT specific activity of AT1 was less than one-fifth of that of AT2, both specific activities were higher than those of reported GGTs (such as 5.71 U·mg−1 from A. thaliana and 3.25 U·mg−1 from Rhodopseudomonas palustris). These data indicate that, not only AT2, but also AT1 has GGT catalytic efficiency comparable with or higher than that of known enzymes. AT2 also showed GPT activity, but its Km value for pyruvate was too high to determine accurately. Further investigations are required to verify the extent to which AT2 contributes to the GPT reaction in vivo. Km values of AT3 were estimated to be equivalent to those of known AGTs.

Table 2.   Kinetic parameters of AT1, AT2 and AT3 (ND, not determined).
EnzymeReactionSubstrateKm (mm)Apparent Vmax (U·mg−1)
  1. The estimate of the Km value for oxaloacetate may be higher than the true value because of the instability of oxaloacetate at the assay temperature.

AT1GOTGlu20 ± 2280 ± 10
Oxaloacetatea0.38 ± 0.05240 ± 10
Asp2.3 ± 0.3110 ± 10
2-OG0.92 ± 0.04110 ± 10
GGTGlu1.5 ± 0.211 ± 0
Glyoxylate4.3 ± 0.813 ± 1
AT2GGTGlu1.2 ± 0.164 ± 2
Glyoxylate6.5 ± 1.870 ± 8
GPTGluNDND
Pyruvate>50>50
PSOTPhosphoserine0.66 ± 0.0717 ± 0
2-OG1.9 ± 0.318 ± 1
AT3AGTAla8.1 ± 0.123 ± 1
Glyoxylate0.90 ± 0.0824 ± 1

All determined Km values, except for that of AT2 for pyruvate, were less than or equivalent to those of known aminotransferases. These results indicate that AT1, AT2 and AT3 are adequately efficient to serve as GOT or GGT, GGT or PSOT, and AGT, respectively.

Discussion

In this study, GOT, GGT, GPT and AGT activities were detected in H. thermophilus, and three aminotransferases were identified. These activities are believed to enable this bacterium to synthesize Asp, Ala and Gly by transferring the amino group of Glu as the nitrogen source. These enzymes were completely purified and characterized and, as such, this report represents, to our knowledge, the first description of the characterization of bacterial GGT and AGT at an enzymatic and gene level.

Comparison of the amino acid sequences with known enzymes showed the phylogenetic position of each aminotransferase. AT2 showed high similarity to eukaryotic AGT in family IV, whereas AT2 possessed GGT, GPT and PSOT activities instead of AGT activity. Most GGTs have been reported to lack GPT activity, with the exception of the GGT from A. thaliana [3]. In addition, GPTs have been identified in several organisms, such as Corynebacterium glutamicum, Pyrococcus furiosus and mammals [2,6,19], and all are classified into subfamily Iγ rather than into family IV. Therefore, it is obvious that AT2 is phylogenetically distinct from known GGTs and GPTs. AT2 also possessed PSOT activity, which is found in some enzymes belonging to family IV. A study of the structure of the Escherichia coli PSOT identified several conserved residues that bind to the substrates [25]. His41, Arg42, His328 and Arg329 in the E. coli PSOT are involved in the interaction with the negatively charged phosphate group of the phosphoserine. These residues are conserved not in AGTs, but are found in all PSOTs (Fig. S1, see Supporting information). Interestingly, AT2 harbours two of these four conserved residues (His29 and Arg30 in AT2). It may be that these partially conserved residues endow AT2 with PSOT activity, which is uncommon among known AGTs of family IV.

AT3 also occupies an unusual phylogenetic position in family I, considering that this enzyme exhibited AGT activity. An AGT belonging to family I has only been found in T. litoralis [22]. This AGT has several characteristics similar to those of AT3, such as comparable specific activity (29 U·mg−1) and strict substrate specificity. However, AT3 seems to be phylogenetically distant from the T. litoralis AGT, because of the low similarity between them: AT3 shows 26% identity to AGT, which is lower than the identity between AT3 and Thermus thermophilus GOT (31%). Furthermore, AT3 lacks several residues that are presumed to affect the substrate specificity of the T. litoralis AGT, e.g. Thr108 in the T. litoralis AGT is supposed to serve to the specificity for Ala [27], but this residue is replaced by Lys105 in AT3 (Fig. S2, see Supporting information). In addition, Leu19, which is located near the substrate in T. litoralis AGT, is replaced by Phe18 in AT3. These phylogenetic and structural differences suggest that AT3 has a substrate recognition mechanism distinct from that presumed in the T. litoralis AGT.

The high similarity between the T. litoralis AGT and kynurenine aminotransferase II [28,29] has noted [27], and they also share similarity with α-aminoadipate aminotransferase [30] and aromatic aminotransferase [31]. These enzymes form a cluster in the phylogenetic tree, but AT3 is clearly located outside of the cluster (Fig. 4). This position also supports the phylogenetic dissimilarity between AT3 and T. litoralis AGT. Instead of these enzymes, AT3-like genes are found in genomes of Aquificales and γ- or δ-proteobacteria (a few of the homologues are depicted in Fig. 4). None of these homologues has been subjected to biochemical studies, and their enzymatic properties and functions are of interest.

Figure 4.

 Phylogenetic tree of AT3, T. litoralis AGT homologues and subfamily Iγ aminotransferases. The numbers at the nodes are bootstrap confidence values expressed as percentages of 1000 bootstrap replicates. The order of the divergence was presumed to be reliable only when the bootstrap values were above 50. The trees were constructed using the neighbor-joining method and showed the same overall topology as the trees constructed by the maximum likelihood method. In addition to the sequences in Fig. 3, those from the following organisms were used: Desulfovibrio vulgaris (YP_010112), Halorhodospira halophila (YP_001001722), human (NP_872603), Hydrogenivirga sp. 128-5-R1-1 (ZP_02176974), Hydrogenobaculum sp. Y04AAS1 (YP_002121232), Nitrococcus mobilis (ZP_01127658), Pyrococcus horikoshii (1X0M_A), Sulfurihydrogenibium sp. YO3AOP1 (YP_001931603) and Thermus thermophilus (BAC76939). AAAAT, α-aminoadipate aminotransferase; KAT-II, kynurenine aminotransferase II.

It has been shown that H. thermophilus has GGT activity and that this activity is derived from two enzymes, AT1 and AT2, with specific activities significantly higher than those of known GGTs. GGT activities derived from AT1 and AT2 in the CFE are calculated to be 0.044 and 0.23 U·mg−1, respectively, from the specific activities and purification factors of each enzyme. These values indicate that most of the GGT activity can be attributed to AT2. Although functional analyses for aminotransferase in vivo are necessary to clarify their physiological roles, it can be speculated that AT2 plays a major role in the GGT reaction to synthesize Gly, and AT1 mainly serves in the GOT reaction.

Although no bacterial GGT gene has been identified, GGT purification has been reported from two species, Rhodopseudomonas palustris and Lactobacillus plantarum [20,23]. AT1 and AT2 homologue genes are found in the genomes of both species (NP_949667 and NP_946142 in R. palustris; NP_785312 and NP_784469 in L. plantarum), and it is possible that the reported GGT activities were derived from these gene products. Further biochemical research is needed to clarify the distribution of these types of homologue with GGT activity.

One of the noteworthy findings in this study is that AT2 and AT3 showed novel substrate specificities from the viewpoint of the well-established aminotransferase classification (Fig. 3), suggesting that the substrate specificity of aminotransferases is broader than previously known. The enzymatic data obtained are expected to be of use in predicting the function of putative aminotransferase homologues that are found in the genome database. It remains unclear whether similar aminotransferases are distributed among a broad range of organisms or whether these enzymes evolved after the divergence from other bacteria early in evolution. Further biochemical study is needed to solve this question. Another intriguing question concerns glyoxylate metabolism in H. thermophilus. Although all three aminotransferases purified in this work use glyoxylate as their substrate, no enzymatic activities for the glyoxylate cycle were detected (not shown), and no genes encoding these enzymes are found in the genome. Glycolate oxidase (EC 1.1.3.15), which catalyses the conversion of glycolate into glyoxylate, may be one of the candidates for physiological glyoxylate synthesis. Several genes in the H. thermophilus genome share similarity with those of glycolate oxidase. However, it remains unclear whether these genes actually encode glycolate oxidase and, furthermore, no genes have been found to explain how glycolate can be synthesized in this bacterium. Moreover, elucidation of an unidentified carbon metabolism is needed to explain glyoxylate and Gly biosyntheses in this bacterium. Studies to clarify these pathways are in progress, and these may elucidate a novel central carbon metabolism in this bacterium.

Materials and methods

Bacterial strain and growth conditions

Hydrogenobacter thermophilus TK-6 (IAM 12695, DSM 6534) was cultivated in an inorganic medium at 70 °C under a gas phase of 75% H2, 10% O2 and 15% CO2, as described previously [32]. Ammonium sulfate in the medium and CO2 in the gas phase were the sole nitrogen and carbon sources, respectively.

Aminotransferase assay

Reaction mixtures contained 50 mm NaPO4 (pH 8.0), 5 mm amino acid, 5 mm 2-oxo acid and the enzyme solution. If necessary, 100 μm PLP was added. For GOT, GGT, GPT, AGT and PSOT assays, substrate concentrations were modified as follows: 100 mm Glu and 10 mm oxaloacetate or 10 mm Asp and 10 mm 2-OG for GOT, 20 mm Glu and 20 mm glyoxylate for GGT, 20 mm Glu and 30 mm pyruvate for GPT, 40 mm Ala and 5 mm glyoxylate for AGT, and 10 mm phosphoserine and 10 mm 2-OG for PSOT. For the AT1 assay, the pH in the reaction mixture was changed to 7.2. The reaction mixtures were incubated at 70 °C, the optimum growth temperature of this bacterium. Aminotransferase activities were determined by measuring the production of the amino acid or the 2-oxo acid.

To measure amino acid production, the reaction mixtures were subjected to phenylthiocarbamyl derivatization, and the derivatized samples were analysed with a reverse-phase column (Inertsil ODS-3, 4.6 mm × 25 cm; GL Science, Tokyo, Japan) to determine the amino acid production [14]. One unit of activity was defined as the activity producing 1 μmol of an amino acid or a 2-oxo acid per minute.

To measure 2-oxo acid production, 150 μL of the reaction mixtures were incubated at 70 °C and the reaction was stopped by the addition of 16 μL of 50% trichloroacetate. Denatured proteins were removed by centrifugation and the supernatants were neutralized with 74 μL of 2 m Tris/HCl (pH 8.0). The concentration of 2-OG was determined in reaction mixtures containing 50 mm NaPO4 (pH 7.2), 0.2 mm NADH, 10 mm NH4Cl and 3 U·mL−1 glutamate dehydrogenase from beef liver (Oriental Yeast, Tokyo, Japan) by measuring the absorbance change at 340 nm. Pyruvate concentration was determined in a reaction buffer containing 1 U·mL−1 lactate dehydrogenase from rabbit muscle (Roche, Basel, Switzerland) instead of NH4Cl and glutamate dehydrogenase.

For the kinetic assay of GOT activity in the direction of Glu synthesis, a coupling method was applied using thermostable malate dehydrogenase from Thermus flavus (Sigma, St Louis, MO, USA). The reaction mixture contained 50 mm NaPO4 (pH 7.2), 10 mm Asp, 5 mm 2-oxoglutarate, 0.2 mm NADH, 1 U·mL−1 malate dehydrogenase and the enzyme solution. The mixture was incubated at 70 °C, and the absorbance was monitored at 340 nm to estimate the decrease in NADH.

Enzyme purification

AT1 was purified from 10 g of wet cells. Active fractions were selected according to GOT and GPT activities. The cells were washed with 20 mm Tris/HCl buffer (pH 8.0) and disrupted by sonication. Cell debris was removed by centrifugation at 100 000 g for 1 h. The supernatant, which was designated CFE, was applied to a DE52 open column (25 mm × 15 cm; Whatman, Brentford, Middlesex, UK) equilibrated with 20 mm Tris/HCl buffer (pH 8.0) containing 1 mm MgCl2. After the elution of bound proteins with buffer containing 1 m NaCl, ammonium sulfate was added to the fractions obtained to 30% saturation, and the samples were applied to a Butyl-Toyopearl column (22 mm × 15 cm; Tosoh, Tokyo, Japan) equilibrated with 20 mm Tris/HCl buffer (pH 8.0) containing 1 mm MgCl2 and ammonium sulfate at 30% saturation. This and subsequent chromatography steps were performed using an ÄKTA purifier system (GE Healthcare, Piscataway, NJ, USA). Proteins were eluted with a gradient of ammonium sulfate from 30% to 0% over 230 mL at a flow rate of 4 mL·min−1. The active fractions were dialysed against 20 mm Tris/HCl buffer (pH 8.0) containing 1 mm MgCl2, and were applied to a DEAE-Toyopearl column (22 mm × 15 cm; Tosoh) equilibrated with 20 mm Tris/HCl buffer (pH 8.0) containing 1 mm MgCl2. Proteins were eluted with a gradient of NaCl from 0 to 1 m over 380 mL at a flow rate of 4 mL·min−1. The active fractions were dialysed against 20 mm Tris/HCl buffer (pH 8.0) containing 1 mm MgCl2, and were applied to a MonoQ HR 5/5 column (bed volume, 1 mL; GE Healthcare) equilibrated with 20 mm Tris/HCl buffer (pH 8.0) containing 1 mm MgCl2. Proteins were eluted with a gradient of NaCl from 0 to 1 m over 40 mL at a flow rate of 0.5 mL·min−1. The active fractions were designated purified AT1, and stored at −80 °C until use.

AT2 was purified from 20 g of wet cells. Active fractions were selected according to GGT and GPT activities. CFE was prepared from the cells and applied to the DE52 column, Butyl-Toyopearl column and DEAE-Toyopearl column, as described above. The active fractions were applied to a CHT Ceramic Hydroxyapatite column (16 mm × 11 cm; Bio-Rad, Hercules, CA, USA) equilibrated with 1 mm KPO4 buffer (pH 7.0). Proteins were eluted with a gradient of KPO4 buffer from 1 to 400 mm over 90 mL at a flow rate of 3 mL·min−1. The active fractions were dialysed against 20 mm Tris/HCl buffer (pH 8.0) containing 1 mm MgCl2, and were applied to the MonoQ column in the same way as AT1. The active fractions were designated purified AT2, and stored at −80 °C until use.

AT3 was purified from 40 g of wet cells. Active fractions were selected according to GGT activity. CFE was prepared from the cells and applied to the DE52 column, Butyl-Toyopearl column, DEAE-Toyopearl column, CHT Ceramic Hydroxyapatite column and MonoQ column in the same way as AT2. Ammonium sulfate was added to the fractions obtained to 30% saturation, and the samples were applied to a Phenyl Superose column (bed volume, 1 mL; GE Healthcare) equilibrated with 20 mm Tris/HCl buffer (pH 8.0) containing 1 mm MgCl2 and ammonium sulfate at 30% saturation. Proteins were eluted with a gradient of ammonium sulfate from 30% to 0% over 15 mL at a flow rate of 0.5 mL·min−1. The active fractions were designated purified AT3, and stored at –80 °C until use.

N-terminal amino acid sequencing

The N-terminal amino acid sequences of purified aminotransferases were determined by Procise 492HT (Applied Biosystems, Foster City, CA, USA) from a blotted membrane [0.2 μm Sequi-Blot poly(vinylidene) difluoride; Bio-Rad].

Protein assay

Protein concentrations were measured using a BCA protein assay kit (Pierce, Rockford, IL, USA). A calibration curve was plotted using bovine serum albumin as a standard protein.

Gel filtration

For the estimation of the molecular mass, gel filtration was performed using a Superose 6 HR 10/30 column (GE Healthcare) or a Shim-pack Diol-300 column (Shimadzu, Kyoto, Japan) equilibrated with 20 mm Tris/HCl (pH 8.0) buffer containing 1 mm MgCl2 and 150 mm NaCl at flow rate of 0.5 or 1 mL·min−1, respectively. Gel Filtration Standard (Bio-Rad) was used as a molecular maker for calibration. Each measurement of standards or samples was performed in triplicate.

Phylogenetic tree construction

Amino acid sequences were aligned using the muscle program [33]. After gap regions had been removed, phylogenetic trees were constructed by the neighbor-joining method or the maximum likelihood method using phylip 3.67 [34].

Nucleotide sequence accession numbers

Nucleotide sequences of AT1, AT2 and AT3 have been deposited in the DDBJ/EMBL/GenBank nucleotide sequence database under accession numbers AB536750, AB536751 and AB536752, respectively.

Acknowledgement

This work was supported by a Grant-in-Aid for JSPS Fellows (20-6284).

Ancillary