Crystal structure of a human kynurenine aminotransferase II homologue from Pyrococcus horikoshii OT3 at 2.20 Å resolution



Kynurenine aminotransferase (KAT; EC. is an enzyme that catalyzes the irreversible transamination of L-kynurenine (L-Kyn) to produce kynurenic acid (KA). L-Kyn is a major metabolite in the degradation pathway of tryptophan, and KA acts as an endogenous antagonist of all three ionotropic excitatory amino acid receptors in the central nervous system.1 Enormous attention has recently been paid to structural and functional studies of KAT, because alteration in the endogenous KA level has been suggested to cause a number of brain and neuron diseases.2–8

In mammals, two KAT isozymes, KAT-I and KAT-II, have been identified and characterized.9, 10 KAT-I and KAT-II are also referred to glutamine transaminase K11 and α-aminoadipate aminotransferase,12 respectively. The crystal structures of human KAT-I (hKAT-I),13 and its homologues, such as glutamine-phenylpyruvate aminotransferase14 and aspartate aminotransferase15 from Thermus thermophilus HB8, have been determined. In contrast, neither the crystal structures of KAT-II nor those of its functional homologues have been determined. The atomic coordinates of a KAT-II homologue from Thermotoga maritima MSB8 (Tm-PAT) have been deposited in the Protein Data Bank (PDB) with accession number 1vp4. However, it remains to be determined whether this protein exhibits the KAT activity. KAT-I and KAT-II have been reported to differ in specific activity, substrate specificity, and sensitivity to inhibitors.9, 10, 16 To understand the structural bases for these differences, it is necessary to determine the crystal structures of KAT-II or its functional homologues. It seems difficult to construct a three-dimensional model for KAT-II based on the KAT-I structures because of the poor amino-acid sequence identities between KAT-I and KAT-II (11% for human enzymes).

We have recently overproduced, purified, and characterized a KAT-II homologue from Pyrococcus horikoshii OT3 (phKAT-II).17 Crystallization and preliminary X-ray diffraction analysis of this protein have also been completed. This protein shows an amino-acid sequence identity of 30% to hKAT-II and exhibits the KAT activity in a homodimeric form. In this study, we solved the crystal structure of phKAT-II in order to understand a structural basis for the enzymatic function of KAT-II.

Materials and Methods.

X-ray diffraction data of the native crystal have been obtained previously.17 The structural determination was carried out by means of the molecular replacement technique with MOLREP18 in the CCP4 program suite, using the structure of chain A of a putative aminotransferase from T. maritima termed Tm-PAT (PDB entry 1vp4) as a search model. Model building and refinement of the structure was completed using the programs O19 and CNS.20 The refinement statistics are shown in Table I. Atomic coordinates and structure factors are available from the PDB under accession code 1X0M. The figures were prepared by MOLSCRIPT21 and RASTER3D.22

Table I. Refinement Statistics
  • Values in parentheses correspond to the highest resolution shell (2.34–2.20).

  • a

    Rfree was calculated using 5% of the total reflections chosen randomly and omitted from refinement.

Resolution range (Å)43.42–2.20
Number of reflections used22117 (3435)
R/Rfreea (%)0.212 (0.252)/0.255 (0.341)
Number of protein atoms3241
No. of waters200
RMSD from ideal geometry 
 Bond lengths (Å)0.007
 Bond angles (°)1.2
 Dihedral angles (°)21.8
 Improper angles (°)0.76
Average B factor (Å2) 
 Protein atoms44.7
 Water molecules49.1

The N-terminal amino-acid sequence was determined using a Procise 491 automated protein sequencer (Perkin-Elmer Japan).

Results and Discussion.

The monomer structure model of phKAT-II was refined at a resolution of 2.20 Å to a crystallographic R-factor of 21.2% [Fig. 1(a)]. A single molecule of phKAT-II is present in an asymmetric unit of the crystal, and the functional homodimer could be observed in the crystal lattice. An overview of the dimeric structure model generated according to its crystallographic symmetry is shown in Figure 1(b). The final monomer model of the phKAT-II crystal structure contains 403 of 428 residues (residues 26–428) and 200 water molecules. Electron density for residues 1–25 is not observed, probably due to structural disorder. The N-terminal amino-acid sequence of recombinant phPAT-II purified from E. coli was determined to be MHEDVQLN, indicating that this region is not truncated upon gene expression or protein purification. The refinement statistics are summarized in Table I. The root mean square deviations (RMSDs) from ideal stereochemistry are 0.007 Å for bond lengths, 1.2° for bond angle, 21.8° for dihedral angles, and 0.76° for improper angles. The Ramachandran plot produced by the program PROCHECK23 shows that 90.6% of non-glycine and non-proline residues are in the most favored region, 9.1% in the additionally allowed region, and 0.3% in the generously allowed region.

Figure 1.

(A) Stereo view of the superposition of the phKAT-II and hKAT-I structures. For the phKAT-II structure, an N-terminal arm, a small domain, and a large domain are colored red, green, and blue, respectively. The hKAT-I structure is shown in monochrome. The N- and C-termini of both proteins are indicated. The side chains of Lys269 of phKAT-II and Lys247 of hKAT-I, as well as the structure of PLP covalently linked to Lys247, are also shown. The ϵ-amino groups of these lysine residues and the phosphate groups of PLP are colored blue and red, respectively. (B) Stereo ribbon representation of the functional dimer of phKAT-II seen along the crystallographic 2-fold axis. The subunit (red) represents that in (A) which is rotated by 90° along the y-axis. The N- and C-termini are indicated.

The size of the dimer form of phKAT-II is approximately 75 × 55 × 55 Å, and the monomer form is approximately 55 × 55 × 45 Å. The phKAT-II architecture represents a typical Type 1 fold of aminotransferases and consists of an N-terminal arm (residues 26–44), a small domain (residues 45–69 and 317–428), and a large domain (residues 70–316), as seen in other Type 1 fold aminotansferases.13–15, 24, 25 The monomer model for phKAT-II was sent to the DALI server26 to find proteins with similar structure. A total of 278 proteins were found to have similar structure (with the Z-score of 2.0 or higher). The three proteins with the highest structural similarity are Tm-PAT (1.6 Å RMSD over 357 residues with 45% sequence identity; PDB code: 1vp4), aspartate aminotaransferase from Thermus thermophilus (2.2 Å RMSD over 368 residues with 21% sequence identity; PDB code: 1bjw) and tyrosine aminotransfearse from Trypanosoma cruzi (2.5 Å RMSD over 363 residues with 17% sequence identity; PDB code: 1bw0).

The crystal structure of phKAT-II determined in this study represents the first functional and archaeal KAT-II structure. Superposition of this structure to the hKAT-I structure allowed us to identify the amino-acid residues that are structurally and/or functionally conserved [Fig. 1(a)]. For example, Lys269 of phKAT-II and Lys263 of hKAT-II are identified as the catalytic residue, because the corresponding residue (Lys247) of hKAT-I covalently binds to 5′-pyridoxal phosphate (PLP) by a Shiff-base linkage.13 In addition, the amino-acid residues involved in PLP binding, such as Asn185, Tyr216, Ser244, and Lys255 of hKAT-I, are well conserved as Asn208, Tyr239, Thr265, and Arg276 in phKAT-II, and Asn202, Tyr233, Ser260, and Arg270 in hKAT-II, respectively. In contrast, the amino-acid residues that are involved in substrate binding are only partially conserved. The crystal structure of hKAT-I in complex with L-Kyn remains to be determined. However, according to the crystal structure of hKAT-I in complex with L-Phe, which behaves as a very poor substrate and competes with L-Kyn for binding to the active site of hKAT-I, a series of aromatic residues, including Trp18, Tyr63, Tyr101, Phe125, Phe278, and His279 form a substrate-binding site of hKAT-I.13 Of them, Tyr63 and Phe125 are conserved as Tyr94 and Tyr154 in phKAT-II and Tyr74 and Tyr142 in hKAT-II, respectively. Trp18, Tyr101, and Phe278 are replaced by Val46, Gln130, and Leu299 in phKAT-II, and Thr21, Gln118, and Leu293 in hKAT-II, respectively. His279 is conserved as His294 in hKAT-II, but is replaced by Cys300 in phKAT-II. These differences may account for the differences in substrate and inhibitor specificities of KAT-I and KAT-II. A number of KAT-II homologues, such as α-aminoadipate aminotransferase from T. thermophilus HB27,27 and aromatic aminotransferases from P. furiosus28, 29 and Saccharomyces cerevisiae,30 have been reported to show broad substrate specificity. Therefore, it would be informative to examine whether phKAT-II exhibits broad substrate specificity as well.

When the amino-acid sequence of phKAT-II is compared with that of hKAT-II, most of the residues forming the active site of phKAT-II are well conserved in hKAT-II, suggesting that the conformation of the active site of phKAT-II is highly similar to that of hKAT-II. Therefore, phKAT-II may serve as a useful model for structural and functional studies of hKAT-II.

The P. horikoshii OT3 genome contains various aminotransferase genes in addition to the phKAT-II gene.31 Of them, only the crystal structure of aromatic aminotransferase (Ph-ArAT) has been determined.32 Ph-ArAT shows higher amino-acid sequence identity to hKAT-I than to hKAT-II. However, it remains to be determined whether Ph-ArAT exhibits KAT activity.


The synchrotron radiation experiments were performed at the BL38B1, BL41XU, and BL44XU in the SPring-8 with the approval of the Japan Synchrotron Radiation Research Institute (JASRI) (Proposal No. 2004A0680-NL1-np-P3k, 2004A0682-NL1-np-P3k, C04A44XU-7424-N). We thank Drs. T. Inoue and Y. Kai for their support in X-ray crystallography and helpful discussions.