Cloning and functional expression of a soluble form of kynurenine/alpha-aminoadipate aminotransferase from rat kidney.

Several aminotransferases with kynurenine aminotransferase (KAT) activity are able to convert L-kynurenine into kynurenic acid, a putative endogenous modulator of glutamatergic neurotransmission. In the rat, one of the described KAT isoforms has been found to correspond to glutamine transaminase K. In addition, rat kidney alpha-aminoadipate aminotransferase (AadAT) also shows KAT activity. In this report, we describe the isolation of a cDNA clone encoding the soluble form of this aminotransferase isoenzyme from rat (KAT/AadAT). Degenerate oligonucleotides were designed from the amino acid sequences of rat kidney KAT/AadAT tryptic peptides for use as primers for reverse transcription-polymerase chain reaction of rat kidney RNA. The resulting polymerase chain reaction fragment was used to screen a rat kidney cDNA library and to isolate a cDNA clone encoding KAT/AadAT. Analysis of the combined DNA sequences indicated the presence of a single 1275-base pair open reading frame coding for a soluble protein of 425 amino acid residues. KAT/AadAT appears to be structurally homologous to aspartate aminotransferase in the pyridoxal 5'-phosphate binding domain. RNA blot analysis of rat tissues, including brain, revealed a single species of KAT/AadAT mRNA of approximately 2.1 kilobases. HEK-293 cells transfected with the KAT/AadAT cDNA exhibited both KAT and AadAT activities with enzymatic properties similar to those reported for the rat native protein.

Kynurenine aminotransferase (EC 2.6.1.7, KAT 1 ) catalyzes the irreversible transamination of the L-tryptophan metabolite L-kynurenine to form kynurenic acid. Due to an overlapping substrate specificity, multiple forms of pyridoxal 5Ј-phosphate (PLP)-dependent aminotransferases are apparently able to catalyze this reaction (1). Differences probably exist in the various tissues and among different animal species regarding which enzyme form is predominantly responsible for the biosynthesis of kynurenic acid. A soluble form of KAT has recently been identified and cloned from rat (2,3), and it was found that the amino acid sequence of this pyruvate-preferring form of KAT is identical to that reported for rat kidney cysteine S-conjugate ␤-lyase (4), also referred to as glutamine transaminase K (GTK, EC 2.6.1.64) (5). The human form of this aminotransferase with KAT, GTK, and ␤-lyase activity has also been cloned and shown to have 82% amino acid similarity to the rat protein (6).
Due to the postulated role of kynurenic acid as a putative endogenous modulator of glutamatergic neurotransmission (for review, see Ref. 10), particular attention has recently been devoted to the presence of KAT isoenzymes in cerebral tissues. Kynurenic acid is, in fact, an antagonist at the glycine site of N-methyl-D-aspartate receptors, and increased levels of kynurenic acid may exert a neuroprotective action in some pathological conditions (11). Two different aminotransferases able to produce kynurenic acid appear to be present in human brain (12)(13)(14). A cDNA clone from rat brain encoding an aminotransferase with KAT activity has been recently isolated (3), and it has been found that this protein corresponds to the rat kidney ␤-lyase/GTK enzyme cloned by Perry et al. (4). Rat brain also contains a mitochondrial form of this aminotransferase, which differs from cytosolic KAT/GTK exclusively for the presence of an NH 2 -terminal leader peptide, which targets the protein to the mitochondrial matrix (15). Whereas it has been claimed that in rats, a single pyruvate-preferring enzyme with KAT activity may be of physiological relevance in the cerebral synthesis of kynurenic acid (16,17), whether in rat brain other aminotransferase forms are involved in the biosynthesis of kynurenic acid remains to be established. Noteworthy, the presence of AadAT activity has also been described in rat brain (see Ref. 18 and references therein), therefore suggesting that this aminotransferase may represent a second synthetic enzyme for kynurenic acid in the central nervous system of this animal species.
In the present work, we describe the molecular cloning and the functional expression of a soluble aminotransferase from rat kidney displaying both KAT and AadAT activity (KAT/AadAT).

MATERIALS AND METHODS
Enzymatic Activity Determination-KAT activity was assayed as described previously (3). For routine analysis, aliquots of the enzyme preparation were incubated (1 h at 37°C) in the presence of 1 mM * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EMBL Data Bank with accession number(s) Z50144.
AadAT activity was assayed as described in Ref. 19. Briefly, after incubation of the enzyme preparation in the presence of 1.7 mM 2-oxoadipate and 16.7 mM L-glutamate, 2-oxoglutarate formed was measured using the glutamate dehydrogenase assay. Aminotransferase activity toward other L-amino acids was measured essentially as described previously (14).
Kinetic constants were calculated by fitting the experimental data to Michaelis-Menten equation using a computer program (Ultrafit, Biosoft).
Protein content was measured by the Pierce Coomassie Plus protein assay kit.
Rat Kidney KAT/AadAT Digestion, Peptide Mapping, and Amino Acid Sequencing-The purified protein (ϳ100 g) was subjected to SDS-polyacrylamide gel-electrophoresis on a 12.5% mini-gel and subsequently electroblotted onto a polyvinylidene difluoride membrane (Immobilon-P sq , Millipore) (20). After Ponceau S staining, digestion of blotted KAT/AadAT with modified trypsin (Promega, Zü rich, Switzerland) and extraction of the resulting peptides were performed as described previously (20,21). KAT/AadAT tryptic peptides were then separated by reverse-phase HPLC (for details on the chromatographic conditions used, see Ref. 3) and directly subjected to amino acid sequence analysis by means of an Applied Biosystem model 475A protein sequenator (Forster City, CA) with on-line phenylthiohydantoin-derivative detection.
Preparation of Poly(A) ϩ RNA-Total RNA was extracted from different adult rat tissues by the guanidinium isothiocyanate/cesium chloride method (22). Poly(A) ϩ RNA was obtained by two purification cycles on oligo(dT)-cellulose spun columns (Pharmacia). The final PCR product was electrophoresed on a 1% agarose gel. The major band of 1058 base pairs (bp) was cut from the gel, subcloned into the SmaI site of the vector pBC/CMV (23), and sequenced.
From a total of ϳ9 ϫ 10 5 plaques, 10 positive clones were obtained and purified through an additional round of screening at lower plaque density. Bluescript plasmids (pBSK, Stratagene), carrying the cDNA inserts, were then isolated from positive phages via in vivo excision.
DNA Sequencing-All DNA sequences were determined by the method of Sanger et al. (25), as modified for plasmid double-stranded DNA sequencing (Sequenase version 2.0; U. S. Biochemical Corp.). Sequencing from the plasmid vectors, T3 and T7 Bluescript primers, and a set of consecutive inner primers were used. DNA sequence assembly, analysis, and translation were performed using the Gene Jockey 1.3 software package (Biosoft Ltd., Cambridge, United Kingdom).
Expression of Recombinant Rat Kidney KAT/AadAT in HEK-293 Cells-The 1.8-kilobase cDNA insert encoding rat kidney KAT/AadAT was subcloned into the blunt-ended SmaI site of the expression vector pBC/CMV (23), placing transcription of the cDNA under control of the strong immediate early promoter of human cytomegalovirus. Human embryonic kidney fibroblast cells (HEK-293 cell line, ATCC CRL 1573) were transfected with sense and antisense cDNA as described previously (20). 2 days after transfection, the cells were harvested, washed twice with phosphate-buffered saline, and stored at Ϫ80°C until analysis. For activity determination of the recombinant enzyme, the transfected cells were resuspended in 0.5 ml of 15 mM Tris acetate buffer, pH 8.0 (containing 10 mM 2-mercaptoethanol, 0.5 mM phenylmethanesulfonyl fluoride, 50 M PLP), and homogenized in a Polytron homogenizer (Kinematica AG). After centrifugation at 28,000 ϫ g for 20 min at 4°C, aliquots of the supernatant were assayed for enzymatic activity as described above.

Purification of Rat Kidney KAT/AadAT and Internal Peptide
Sequencing-KAT/AadAT was purified from rat kidney following the procedure described in Table I. The purification of the enzyme was monitored by measuring KAT activity in the presence of 2-oxoglutarate, its preferred aminoacceptor. The purified enzyme also displayed AadAT activity with a specific activity of 9.4 mol min Ϫ1 mg protein Ϫ1 . After chromatography on DEAE-Sepharose, this aminotransferase form was well separated from KAT/GTK, a distinct enzyme with KAT activity and preference for pyruvate as cosubstrate (Fig. 1). In accordance with previous reports on rat kidney KAT isoforms (9,26)  the 90 -100-kDa range (as determined by gel filtration on a Pharmacia Superose 12 FPLC column) and was composed of two subunits of equal size. In fact, SDS-polyacrylamide gel electrophoresis analysis of the purified protein under reducing conditions showed a single major band of ϳ45 kDa (not shown).
To obtain information on the amino acid sequence of KAT/ AadAT, the polyvinylidene difluoride-blotted protein was directly submitted to Edman degradation to see if its NH 2 terminus was accessible to sequencing. No sequence information could be obtained, indicating that the NH 2 terminus of the protein was modified either as the result of a posttranslational event or due to protein handling. The amino acid sequences of internal peptides were then obtained after digestion with trypsin and separation of the resulting peptides by reverse-phase HPLC. The sequence of six tryptic peptides (denominated T2a, T2b, T3, T6, T10, and T12, see Figs. 2 and 3) was determined. No significant matches were found in the Swiss-Prot and Protein Identification Resource Protein data banks.
Isolation of Rat Kidney KAT/AadAT cDNA-Using the amino acid sequences of the analyzed tryptic peptides of purified rat kidney KAT/AadAT, we were able to obtain the corresponding cDNA using RT-PCR. Two amino acid sequences from the least degenerate regions of peptides T3 and T10 were selected to construct synthetic degenerate oligonucleotides. Because the relative order of the two tryptic peptides within the protein was not known, degenerate sense and antisense oligonucleotides corresponding to the tryptic peptide sequences were synthesized (Fig. 2, panels A and B). Poly(A) ϩ RNA extracted from rat kidney was reverse-transcribed using each degenerate antisense oligonucleotide as a primer, and the resulting cDNA was used as a PCR template with each possible combination of degenerate sense and antisense oligonucleotides from the two tryptic peptides. After 35 rounds of amplification, one distinct PCR product was detected by agarose gel electrophoresis for the primer combination sense-T3/antisense-T10 (Fig. 2C, lane  2). DNA sequence analysis revealed that the 1058-bp fragment contained an open reading frame coding for a polypeptide of 352 amino acids, starting with the codon of the first amino acid phenylalanine (TTT) of peptide T3 and ending with the second nucleotide of the last glutamine codon (CA) of the T10 sequence. Moreover, this cDNA fragment encoded the tryptic peptides T2a, T2b, and T12, confirming that it was indeed a partial cDNA for rat KAT/AadAT. Molecular Cloning of Rat Kidney KAT/AadAT cDNA-A rat kidney cDNA library (Uni-ZAP TM XR) from Stratagene was screened with the partial cDNA probe of rat kidney KAT/ AadAT. Screening of 9 ϫ 10 5 recombinants revealed 10 positive clones with inserts ranging from 950 to 1828 bp. The four longest clones, designated as rkKAT-8, rkKAT-14, rkKAT-3, and rkKAT-9, were chosen for further characterization. DNA sequence analysis showed that all clones were identical, except for two nucleotide differences. Only one of these nucleotide differences was contained within the coding region, and it did not result in a change in amino acid sequence. The nucleotide and deduced amino acid sequences are shown in Fig. 3  preceded by a 112-nucleotide 5Ј-untranslated region containing two in-frame stop codons at positions 32 and 71, and a 422nucleotide untranslated region in the 3Ј-end. The predicted initiation codon (nucleotide 113) is embedded in the sequence GAGACATG, which does not match perfectly with the consensus sequence CCACCATG frequently found for eukaryotic translation initiation (27). No further ATG was found upstream of this predicted initiation codon, indicating a complete coding sequence.
Comparison of the complete sequence with the EMBL DNA sequence data base using the Genetics Computer Group sequence analysis software (GCG, University of Wisconsin), indicated that the sequence was unique and had not been isolated previously. Analysis of the hydrophilicity plot of the predicted amino acid sequence for KAT/AadAT showed no evidence for membrane-anchoring or -spanning regions, consistent with the soluble nature of the isolated protein. A mitochondrial form of KAT/AadAT has also been described (8,9). However, no structural features resembling those of leader peptides for mitochondrial import (28) were observed in the predicted amino acid sequence of rat KAT/AadAT. Therefore, whether mitochondrial KAT/AadAT is encoded by a different gene or by a splice variant of the isolated form carrying an additional signal-peptide sequence remains to be established. Four potential N-linked glycosylation sites (Asn, Xaa, Ser/Thr) (29) were found in the predicted amino acid sequence of rat kidney KAT/AadAT at Asn residues 2, 57, 101, and 202. No information, however, is presently available indicating whether the native protein is glycosylated.
KAT/AadAT Sequence Comparison with Other Aminotransferases-Binary alignment of the protein sequences of KAT/ AadAT and KAT/GTK (4) using the GAP program contained in GCG showed only 18.4% amino acid identity with 14 gaps located mainly in the NH 2 -and COOH-terminal parts of the two sequences. Considering conservative amino acid substitutions, the two KAT isoforms displayed 46.2% similarity (not shown). Similar degrees of amino acid identity and similarity were observed after comparison with other aminotransferases from different species, including isoenzymes, such as human serine:pyruvate aminotransferase (30), which have been reported to display KAT activity. Sequence homology among aminotransferase isoenzymes is not easily recognizable by standard algorithms for sequence comparison. A multiple sequence alignment of aspartate aminotransferases from various organisms with KAT/AadAT and KAT/GTK showed that several residues in the central region of the sequences are totally conserved, therefore suggesting that the corresponding threedimensional structures might also be conserved. In fact, despite the low degree of amino acid identity, most aminotransferases appear to constitute a group of structurally homologous proteins that originated from a single universal ancestor protein (31). Fig. 4 shows an alignment of KAT/AadAT and KAT/ GTK with E. coli aspartate aminotransferase (AAT), whose three-dimensional structure has been solved in detail. The structure of E. coli AAT consists of three parts, an NH 2 -terminal loose segment and two compact domains, and shows the same fold and active site structure of vertebrate isoenzymes coli AAT for those of the two KAT isoenzymes according to the sequence alignment shown in Fig. 4 and optimized using the Moloc molecular modeling software (Moloc) (33). For both isoenzymes, the scores calculated by the three-dimensional profile method (34) indicated that the model exhibits features of a correctly folded protein. According to the model obtained from the sequence alignment, Lys 263 of KAT/AadAT and Lys 247 of KAT/GTK correspond to the lysine residue (position 246) of the E. coli enzyme, which forms an aldimine linkage with the PLP aldehyde group. In addition, several of the AAT residues involved in the binding of the cofactor (32) appear to be conserved in KAT/AadAT and KAT/GTK (see Fig. 4). Sequence similarities in the NH 2 -and COOH-terminal regions were too low for molecular modeling. This suggests that KAT/AadAT and KAT/GTK should resemble AAT in the large domain comprising the active site but may be different in the domain composed by the NH 2 and COOH termini and possibly involved in the substrate specificity.
Blot Hybridization-Northern blot analysis of poly(A) ϩ RNAs isolated from various rat tissues is shown in Fig. 5. In all tissues tested, a single species of KAT/AadAT mRNA of approximately 2.1 kilobases was detected using a 32 P-labeled nicktranslated cDNA from rkKAT-8, indicating that the in vivo mature KAT/AadAT mRNA may be longer than our cloned cDNA, possibly in the 5Ј-untranslated region. Whereas labeling intensity was highest for mRNA isolated from kidney, a clearly detectable band of equal size was also observed in poly(A) ϩ RNA extracted from total brain and cerebral cortex, indicating the presence of this enzyme form in cerebral tissues. The weaker labeling intensity found in brain in comparison to peripheral tissues is in accordance with the much lower activity of KAT isoenzymes detected in cerebral tissues (see e.g. Ref. 13). The fact that no KAT/AadAT message was detected in the hippocampus might be due to the low mRNA level for this protein in this brain region.
Expression of Recombinant Rat KAT/AadAT in HEK-293 Cells and Characterization of the Enzymatic Activity-To confirm that the isolated cDNA indeed encoded for an aminotransferase with both KAT and AadAT activity, the rat KAT/AadAT cDNA was subcloned into the eukaryotic expression vector pBC/CMV (23). HEK-293 cells were chosen for transfection because they did not exhibit either KAT or AadAT activity. A relatively high expression of both KAT and AadAT activities could be reached after transient transfection of these cells, with most of the enzymatic activity (Ͼ90%) being recovered in the soluble fraction of the cells. No activity was observed after transfection of the cells with the antisense cDNA. Determination of KAT kinetic properties of the recombinant enzyme with 2 mM 2-oxoglutarate as amino acceptor showed K m and V max values for L-kynurenine of 0.95 Ϯ 0.33 mM and 135 Ϯ 31 nmol min Ϫ1 mg protein Ϫ1 , respectively. The enzyme also metabolized L-3-hydroxykynurenine to xanthurenic acid with a similar catalytic efficiency (K m , 1.36 Ϯ 0.16 mM; V max , 166 Ϯ 42 nmol min Ϫ1 mg protein Ϫ1 ). In the presence of 5 mM L-kynurenine as amino donor, the K m of 2-oxoglutarate was 45.5 Ϯ 12.4 M. When the enzyme was assayed for AadAT activity, the K m for L-glutamate was 5.6 Ϯ 2.8 mM, with a V max of 910 Ϯ 210 nmol min Ϫ1 mg protein Ϫ1 . Notably, under our experimental conditions, the catalytic efficiency of KAT/AadAT appeared to be similar for both KAT and AadAT activity, as it can be inferred from the similar V max /K m ratios. The observed kinetic parameters are in accordance with those found for the rat native enzyme (1,8,19).
The KAT activity of the enzyme measured in the presence of various cosubstrates was highest with 2-oxoglutarate and 2-oxoadipate and lowest with pyruvate (Table II). Regarding the specificity of this aminotransferase form toward other Lamino acids, tryptophan, phenylalanine, tyrosine, aspartate and alanine were found to be substrates for KAT/AadAT. However, the specific activities observed in the presence of these amino acids (up to 10 mM) were relatively low, being 15% (tryptophan), 9% (phenylalanine and tyrosine), and 6% (aspartate and alanine) of the activity measured with L-kynurenine. The enzyme did not display significant enzymatic activity toward the other L-amino acids tested (histidine, serine, methionine, glutamine, and leucine). The specificity pattern toward amino acids and cosubstrates of the KAT/AadAT expressed in HEK-293 cells was similar to that observed for the native enzyme purified from rat kidney.
The availability of the cDNA clone for this aminotransferase isoenzyme with KAT activity will contribute to the further clarification of kynurenic acid disposition in peripheral organs as well as in the central nervous system. For instance, in rat FIG. 5. Northern blot hybridization of poly(A) ϩ RNAs from various rat tissues. The blot was hybridized with a 32 P-labeled nicktranslated cDNA insert isolated from rkKAT-8. Numbers on the left indicate kilobases (kb) as determined by RNA size markers (Life Technologies, Inc.). The blot was exposed to x-ray film for 5 days (kidney and liver) or 15 days (brain tissues). brain it appears that at least two distinct aminotransferases with KAT activity are expressed: 1) KAT/AadAT, as described in this paper, and 2) the KAT/GTK form, which was directly isolated from cerebral tissues (3,15). Several factors may determine the role of the various KAT forms in kynurenic acid biosynthesis, such as the different affinity constants of these enzymes for L-kynurenine and the regional and cellular distribution of these enzymes in the different organs. Interestingly, several biochemical characteristics of the two KAT forms identified in rat, including pI value and cosubstrate preference, are similar to those described for the two cerebral KAT forms in humans (12,13), therefore raising the possibility that these human KAT forms may be similar to the identified KAT forms from rat.
Although our attention has focused mainly on the role of KAT/AadAT in L-tryptophan metabolism, its function in L-lysine metabolism also merits further investigation, for instance, in disorders of the lysine metabolic pathway. Interestingly, L-2-aminoadipate, the substrate for AadAT, is a well known astroglial-specific toxin (18). Thus, knowledge of the cerebral disposition of this compound is instrumental for the elucidation of its mechanism of toxicity and possible relevance in pathology.