Crystal Structure of Human Kynurenine Aminotransferase I*

Received for publication, August 13, 2004, and in revised form, September 7, 2004Published, JBC Papers in Press, September 10, 2004, DOI 10.1074/jbc.M409291200

Franca Rossi‡, Qian Han§, Junsuo Li§, Jianyong Li§, and Menico Rizzi‡¶

From the ‡Dipartimento di Scienze Chimiche, Alimentari, Farmaceutiche, Farmacologiche–Istituto Nazionale di Fisicadella Materia, University of Piemonte Orientale “Amedeo Avogadro,” Via Bovio 6, 28100 Novara, Italy and§Department of Pathobiology, University of Illinois, Urbana, Illinois 61802

The kynurenine pathway has long been regarded asa valuable target for the treatment of several neuro-logical disorders accompanied by unbalanced levels ofmetabolites along the catabolic cascade, kynurenicacid among them. The irreversible transamination ofkynurenine is the sole source of kynurenic acid, and itis catalyzed by different isoforms of the 5�-pyridoxalphosphate-dependent kynurenine aminotransferase(KAT). The KAT-I isozyme has also been reported topossess �-lyase activity toward several sulfur- and se-lenium-conjugated molecules, leading to the proposalof a role of the enzyme in carcinogenesis associatedwith environmental pollutants. We solved the struc-ture of human KAT-I in its 5�-pyridoxal phosphate andpyridoxamine phosphate forms and in complex withthe competing substrate L-Phe. The enzyme active siterevealed a striking crown of aromatic residues deco-rating the ligand binding pocket, which we propose asa major molecular determinant for substrate recogni-tion. Ligand-induced conformational changes affect-ing Tyr101 and the Trp18-bearing �-helix H1 appear toplay a central role in catalysis. Our data reveal a keystructural role of Glu27, providing a molecular basisfor the reported loss of enzymatic activity displayed bythe equivalent Glu 3 Gly mutation in KAT-I of sponta-neously hypertensive rats.

In mammals, the kynurenine pathway is the main route forthe degradation of tryptophan exceeding anabolic needs andrepresents the source for de novo NAD biosynthesis. Severalmetabolites along this pathway, collectively indicated askynurenines, act as potent neuroactive compounds (1) exert-ing their function by either inducing free radicals generation(2) or engaging ionotropic excitatory amino acids receptors inthe central nervous system (3). The amino acid L-kynurenine(L-Kyn)1 is a key metabolite along this pathway, standing at

the central branching point of the metabolic cascade (4).Indeed, L-Kyn undergoes different fates: i) it can be trans-formed to either 3-hydroxy-DL-kynurenine or anthranilic acidand conveyed into the flux of freshly synthesized NAD, or ii)it can be used for the synthesis of kynurenic acid (KA). Amajor role in the central nervous system is ascribed to KA. Infact, it represents the only known endogenous antagonist ofthe excitatory action of excitatory amino acids, showing thehighest affinities for the glycine modulatory site of the N-methyl-D-aspartate subtype of glutamate receptor (5–7) andthe �7-nicotinic acetylcholine receptor (8–10). Its inhibitoryaction underlies its neuroprotective and neurolectic proper-ties; indeed, low endogenous brain KA level profoundly influ-ences vulnerability to excitotoxic attacks (11–14). On theother hand, a KA increase significantly correlates with schiz-ophrenia (15) and cognitive impairment (16–18) suggestingan additional role of KA in the pathophysiology of psychiatricdisorders and mental retardation.

The KA requirement in the central nervous system is satis-fied by the in situ irreversible transamination of L-kynurenine,catalyzed by kynurenine aminotransferases (KATs) (Fig. 1).The catalyzed reaction proceeds first through the transaldimi-nation between the L-Kyn �-amino group and the pyridoxal5�-phosphate (PLP) cofactor following the classical mechanismreported for aminotransferases (19); however, the subsequentcatalytic events leading to the kynurenic acid formation are farfrom fully elucidated. KATs belong to the �-family of PLP-de-pendent enzymes and are considered members of aminotrans-ferases fold type I subfamily (20). So far two KAT isozymes,termed KAT-I and KAT-II, have been identified in rat (21–23)and human (24) brains and have been extensively character-ized. Both isozymes are capable of catalyzing the transamina-tion of L-Kyn and to a lesser extent 3-hydroxy-kynurenine tokynurenic acid and xanthurenic acid, respectively, using manydifferent �-ketoacids as amino group acceptors (25, 26). KAT-I,also known as glutamine transaminase K, (27, 28) is stronglyinhibited by the competing substrates glutamine, tryptophan,and phenylalanine (26). Of functional relevance, KAT-I alsodisplays cysteine conjugate �-lyase activity (29) and has beenshown to activate sulfur- and selenium-conjugated compounds(30, 31). This observation indicates an additional importantrole for KAT-I in the bio-activation of environmental pollutantscontributing to liver- and kidney-associated carcinogenesis(28). Although the kidney and the liver show much greaterKAT activity than the brain, the emphasis of KAT research hasbeen placed on the enzymes in the central nervous system,paralleling the investigation of the pivotal role of KA therein.

We report here the crystal structure of human KAT-I(hKAT-I) at a resolution of 2.0 Å in its PLP and PMP forms aswell as in complex with the competing substrate phenylala-nine. Our results represent the first crystal structure of ahuman KAT to be reported, thus shedding more light into the

* This work was supported by Grant AI 44399 from the NationalInstitutes of Health and Ministero dell’Istruzione, dell’Universita edella Ricera project FIRB2001. The costs of publication of this articlewere defrayed in part by the payment of page charges. This article musttherefore be hereby marked “advertisement” in accordance with 18U.S.C. Section 1734 solely to indicate this fact.

The atomic coordinates and structure factors (codes 1W7L, 1W7N, and1W7M) have been deposited in the Protein Data Bank, Research Col-laboratory for Structural Bioinformatics, Rutgers University, New Bruns-wick, NJ (http://www.rcsb.org/).

¶ To whom correspondence should be addressed: DiSCAFF, Univer-sity of Piemonte Orientale, Via Bovio 6, 28100 Novara, Italy. Tel.:39-0321-375812; Fax: 39-0321-375821; E-mail: rizzi@ipvgen.unipv.it.

1 The abbreviations used are: L-Kyn, L-kynurenine; KA, kynurenicacid; KAT, kynurenine aminotransferase; hKAT-I, human KAT-I; PLP,5�-pyridoxal phosphate; PMP, pyridoxamine phosphate; ttGlnAT,T. thermophilus glutamine:phenylpyruvate aminotransferase; r.m.s.d.,root mean square deviation.

THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 279, No. 48, Issue of November 26, pp. 50214–50220, 2004© 2004 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in U.S.A.

This paper is available on line at http://www.jbc.org50214

at VIV

A, V

irginia Tech, on M

ay 24, 2011w

ww

.jbc.orgD

ownloaded from

catalytic mechanism and substrate specificity. The determina-tion of the three-dimensional structure of hKAT-I may contrib-ute to the rational design of selective inhibitors of high medicalinterest in a number of pathological conditions in humans.

EXPERIMENTAL PROCEDURES

Enzyme Expression and Purification

Construction of Recombinant Transfer Vector for hKAT-I—The openreading frame for hKAT-I (NP_872603) was amplified from human livercDNA (Clontech) using a specific forward (5�-CTCGAGATGGCCAAA-CAGCTGCAG, XhoI site underlined) and reverse (5�-AAGCTTAGAGT-TCCACCTTCCACTT, Hind III site underlined) primers pair. The PCRproduct was inserted into a TOPO TA cloning vector and then subclonedinto the baculovirus transfer vector pBlueBac4.5 (Invitrogen). The re-combinant transfer vector (pBlueBac-HsKAT I) was sequenced forframe verification of the inserted DNA sequence.

Production of Recombinant Baculoviruses for the Expression ofhKAT-I in Insect Cells—pBlueBac-HsKAT I was co-transfected withlinearized Bac-N-BlueTM viral DNA (AcMNPV, Autographa californicamultiple nuclear polyhedrosis virus), in the presence of InsectinPlusTM

insect cell-specific liposomes, to Spodoptera frugiperda (Sf9) insect cells(Invitrogen). The recombinant baculovirus was purified by plaque assayprocedure. Blue putative recombinant plaques were transferred to 12-well microtitre plates and amplified in Sf9 cells. Viral DNA was isolatedfor PCR analysis to determine the purity of recombinant viruses. Hightiter viral stocks for individual recombinant viral strains were gener-ated by amplification in Sf9 cell suspension cultures according to themanufacturer’s instructions.

hKAT-I Expression and Purification—Sf9 cells were cultured at27 °C in an Ultimate InsectTM serum-free medium (Invitrogen) supple-mented with 10 units of heparin/ml (Sigma) in culture spinner flaskswith constant stirring at 80 rev/min. When the cell density reached 2 �106 cells/ml, they were inoculated with the high titer viral stocks of thehKAT-I expressing recombinant baculoviruses at a multiplicity of in-fection of six viral particles per cell.

At day 4 after hKAT-I recombinant baculovirus inoculation, Sf9-infected cells from a 2-liter culture were harvested by centrifugation(800 � g for 15 min at 4 °C). The pellet was dissolved in a lysis buffercontaining 25 mM phosphate buffer, 150 mM NaCl, 0.1 mM PLP, 2 mM

dithiothreitol, 2 mM EDTA, and 1 mM phenylmethylsulfonyl fluoridewith a final pH of 7.4. After a 30-min incubation on ice, cell lysate wascentrifuged at 18,000 � g for 20 min at 4 °C, and the supernatant wascollected and assayed for KAT-I activity. Soluble proteins in superna-tant were precipitated with 65% saturated ammonium sulfate. Proteinprecipitate was re-dissolved in 10% saturated ammonium sulfate andapplied to a phenyl-Sepharose column. A linear gradient from 10–0% ofammonium sulfate in 10 mM sodium phosphate buffer (pH 7.0) was usedfor protein elution. The hKAT-I active fractions were pooled, dialyzed,and separated by DEAE-Sepharose chromatography applying a linearNaCl gradient (0–500 mM) in the same phosphate buffer. The activefractions were pooled again, concentrated, and then separated by aSuperdexTM 200 gel filtration column. The hKAT-I active fractions werefinally collected and concentrated. The purified protein content wasdetermined by a Bio-Rad protein assay kit using bovine serum albuminas the standard. The purity of the enzyme was assessed by SDS-PAGEanalysis. The concentration of the purified hKAT-I stock was adjustedto 20 mg/ml in 20 mM phosphate buffer (pH 7.5), aliquoted, and frozenat �80 °C.

Crystallization, Data Collection, and Structure Determination

Crystallization—Crystals of the PLP form of human hKAT-I havebeen obtained using the vapor diffusion technique in hanging drop. 2 �l

of a protein solution at a concentration of 20 mg/ml were mixed with anequal volume of a reservoir solution containing 4.5 M ammonium for-mate, 0.1 M Tris (pH 7.5) and equilibrated against 500 �l of a reservoirsolution at 4 °C. Yellow crystals grew to a maximum dimension of 0.3 �0.5 � 0.7 mm in about 2 weeks. In the effort to obtain the complex withL-kynurenine, crystals of the PLP form of hKAT-I were soaked over-night at 4 °C in their crystallization solution containing 3 mM L-kynu-renine. A few minutes after soaking, the crystals became colorless,indicating the formation of the PMP form for which the structure isreported in the present study. The crystals of the hKAT-I�Phe complexwere obtained by soaking crystals of the PLP form in their crystalliza-tion solution with the addition of 2.5 mM phenylalanine at 4 °Covernight.

Data Collection—All data collections were performed at 100 K. Crys-tals were directly taken from the crystallization droplet and flash frozenunder a liquid nitrogen stream. Diffraction data for the PLP form werecollected up to 2.0 Å resolution using synchrotron radiation at the ID14EH1 beam line, European Synchrotron Radiation Facility, Grenoble,France. Systematic absences and diffraction symmetry suggested atrigonal P3221 or P3121 space group with the following cell parametersa � b � 146.4 Å and c � 67.5 Å containing one molecule in theasymmetric unit with a corresponding solvent content of 72%. X-raydiffraction data sets for the PMP form of the enzyme and for thehKAT-I�Phe complex were collected up to 2.9 Å and 2.7 Å resolutionrespectively, using in house equipment (RIGAKU RU300 rotating an-ode and RAXIS IV� area detector). Data processing was performedwith the programs of the CCP4 suite (32). Statistics for data collectionsare listed in Table I.

Structure Determination—The structure determination of the PLPform was carried out by means of the molecular replacement technique,using as the search model the structure of Thermus thermophilusaspartate aminotransferase, (34% sequence identity; Protein DataBank code 1GCK) (33). The program AmoRe (34) was employed tocalculate both cross-rotation and translation functions in the 10–4 Åresolution range. The rotation function gave a clear solution, which wassubject to the translation function calculation performed in both theP3221 and P3121 space groups. Only for the former was a clear solutionidentified, unambiguously allowing us to assign P3221 as the correctspace group. The initial model was subjected to iterative cycles ofcrystallographic refinement with the program REFMAC (35) alternatedwith graphic sessions for model building using the program O (36). Arandom sample containing 834 reflections was set apart for the calcu-lation of the free R-factor (37). Solvent molecules were manually addedat positions with density �4 sigma in the Fo � Fc map, considering onlypeaks engaged in at least one hydrogen bond with a protein atom or asolvent atom. The procedure converged to an R-factor and free R-factorof 0.192 and 0.227, respectively, with ideal geometry.

The structures of the PMP form and of the hKAT-I�Phe complex wererefined using (as the starting model) the final coordinates of the PLPform of the enzyme from which all solvent molecules were removed. Inboth cases the crystallographic refinement was carried out by means ofthe program REFMAC alternated with sessions of model building. 564and 722 randomly chosen reflections were excluded from the refine-ment of the PMP and hKAT-I�Phe structures, respectively, and wereused for the free R-factor calculation. For the PMP structure, acareful inspection of the electron density in the enzyme active site hasbeen performed when the R-factor dropped to a value of 0.25 at 2.9 Å.No electron density compatible with either the substrate L-Kyn or theproduct KA was observed, and the enzyme was unambiguously iden-tified as being in the PMP form. In the case of the hKAT-I�Phestructure, the ligand was manually modeled based on both the 2Fo �Fc and Fo � Fc electron density maps only when the R-factor droppedto a value of 0.26 at 2.7 Å resolution. For both the PMP and hKAT-I�Phe structures, solvent molecules were manually added followingthe same criteria adopted for the PLP form. The crystallographicrefinement converged to an R-factor and an R-free of 0.17 and 0.22 forthe PMP form, and an R-factor and R-free of 0.175 and 0.23 for thehKAT�Phe complex. In all the three structures the residues 1–3 and149–151 were not visible in the electron density map. Residuesnumbered 1*–422* refer to the crystallographically related subunitbuilding the hKAT-I functional homodimer. The results of refinementare summarized in Table I.

Deposition—The atomic coordinates of hKAT-I in the PLP and PMPforms and in complex with phenylalanine, have been deposited with theProtein Data Bank (www.rcsb.org) (accession codes 1W7L, 1W7N, and1W7M, respectively).

FIG. 1. Scheme of the two-step reaction catalyzed by kynuren-ine aminotransferases. The first half of the reaction is fullyirreversible.

Human Kynurenine Aminotransferase I Crystal Structure 50215

at VIV

A, V

irginia Tech, on M

ay 24, 2011w

ww

.jbc.orgD

ownloaded from

RESULTS AND DISCUSSION

Overall Quality of the Model—The three-dimensional struc-ture of the recombinant human kynurenine aminotransferase Ihas been solved by means of the molecular replacement tech-nique at a resolution of 2.0 Å. A monomer is present in theasymmetric unit and the functional homodimer could be ob-served in the crystal lattice, resulting from the application ofthe crystallographic 2-fold axis. An excellent electron densitymap allowed the modeling of 415 residues of 422, one PLP, and305 solvent molecules in the PLP form. The final model of thePMP form contains 415 residues, one PMP, and 197 solventmolecules, whereas the hKAT-I�Phe complex consists of 415residues, one PLP molecule, one L-Phe molecule, and 202 sol-vents. The stereochemistry of the models has been assessedwith the program PROCHECK (38). In the case of the PLPform, 90% of the residues were in the most favored regions ofthe Ramachandran plot. Pro123, Pro183, and Pro186 were allrecognized as cis residues. Although Phe278 falls in a disal-lowed region of the Ramachandran plot in all three structuressolved, the excellent electron density allowed us to unambigu-ously assign the observed conformation. All the figures havebeen generated with the programs MOLSCRIPT (39) and BOB-SCRIPT (40).

Overall Structure—The hKAT-I protein architecture revealsthe prototypical fold of aminotransferases subgroup I (19, 41),characterized by an N-terminal arm, a small domain, and alarge domain (Fig. 2). The N-terminal arm consists of a randomcoiled stretch made of residues 4–17. The small domain (resi-dues 18–43 and 302–421) folds in a 5-stranded mainly antipa-rallel �-sheet surrounded by five �-helices. The large domain(residues 44–301) shows a conserved �/� topology, where asharply twisted 7-stranded �-sheet inner core is nested into aconserved array of eight �-helices contributed from both theinterior and the exterior of the molecule.

As observed in other subgroup I aminotransferases, thefunctional unit of hKAT-I consists of a homodimer with sub-units related by a dyad axis, with its two active sites locatedat the domain interface in each subunit, and at the subunitinterface in the dimer (Fig. 3). Helices H2 and H13 partici-pate in dimer stabilization as they are located around the2-fold axis. Moreover, a relevant contribution toward thestability of the hKAT-I dimer is provided by its N-terminalarms. Mutational analysis carried out on pig cytosolic aspar-tate aminotransferase demonstrated that the integrity of theN-terminal region is of crucial importance for the enzymaticactivity (42). In our crystallographic dimer, two strong saltbridges are established between Arg8 and Asp113 (at a dis-

tance of 3.5 Å) and between Arg9 and Glu114 (at a distance of3.1 Å) of the opposite subunits; such interactions, while pro-viding a major contribution to the dimer stability, lock theN-terminal arms in the observed conformation.

The PLP Binding Site—The active site contains one PLPmolecule covalently linked to Lys247 and is hosted in a deepcleft at the domain interface built up by residues from bothsubunits (Fig. 3). Each PLP cofactor sits in a binding pocketdefined by two regions contributed by residues of the largedomains of both subunits (Fig. 4). The bottom of the PLPbinding pocket is entirely defined by residues of the largedomain of the corresponding monomer. With the exception ofLys247 and Gly100, all of these residues stand at or close to theedge of the inner core �-sheet, pointing toward the domaininterface and facing the re-face of the PLP-ring. Distinct arraysof residues form the lateral walls of the PLP binding pocket. Inparticular, the stretch 34–37, together with the side chains ofArg398, Asn181, and Asn185 of the large domain, builds up one ofthe pocket walls. The opposite wall consists of residues Tyr101

and Tyr128 of the same monomer and of residues Phe278*,His279* and Tyr63* of the other subunit. As a consequence, thePLP-pyridine ring can be seen as laterally clamped by a set ofhydrophilic residues on one side and by numerous aromaticresidues on the other one (Fig. 4).

In its PLP form, hKAT-I carries the PLP molecule covalentlybound in the active site by a Schiff-base linkage to the catalyticLys247 (Fig. 4A). This results in the presence of an internalaldimine double bond (C4��N) whose plane forms an angle of123.0° with the pyridine ring of PLP. Several other residuescontact the PLP molecule, participating in its recognition andbinding. The phosphate group of the cofactor is engaged in anumber of interactions with residues building up the strictlyconserved “PLP-phosphate binding cup” featuring PLP-dependent enzymes (43). In particular, its OP1 oxygen makes aset of hydrogen bonds with Ser244 (2.6 Å), Gly100 (with itsbackbone nitrogen atom at 2.8 Å), and with the solvent mole-cule W7 (3.1 Å). The hydroxyl group of Tyr63* makes a hydro-gen bond with the PLP OP2 atom (distance of 2.5 Å). A furtherhydrogen bond involves W34 located at 2.67 Å. Finally, the PLPOP3 atom makes interactions with Lys255 (at 2.7 Å from the NZatom) and with the backbone nitrogen atoms of Gly100 andTyr101 (distances of 3.2 Å and 2.9 Å, respectively). The PLPphenolic oxygen is held in place by its interactions with Tyr216

(at 2.7 Å from its OH atom) and with Asn185 (at a distance of 2.8Å with its NH1 atom). Moreover, the N1 atom of the PLPpyridine moiety forms hydrogen bonds with the carboxylic ox-ygen atoms of Asp213. Several hydrophobic interactions partic-ipate in the further stabilization of PLP. In particular Phe125

and Val215 sandwich the pyridine moiety of the cofactor from itssi-face and re-face, respectively.

Substrate Recognition and Implications for Catalysis—Wehave been unable to produce crystals of the hKAT-I in complexwith the physiological substrate L-Kyn. Attempts at obtainingthe structure of hKAT-I�L-Kyn complex invariably yielded crys-tals of the PMP form (Fig. 4B), emphasizing that our crystalform is catalytically competent. However, to describe the ligandbinding site in hKAT-I, we solved the structure of the Michaeliscomplex with L-phenylalanine, which has been reported to in-hibit hKAT-I by behaving as a very poor substrate and com-peting with L-Kyn for binding to the enzyme active site (26). Inthe hKAT-I�Phe structure (Fig. 4C), L-Phe lies just above thePLP molecule on its si-face. Several residues structurally con-served throughout aminotransferases described so far definethe ligand binding site and contact L-Phe. In particular, Asn185

establishes a hydrogen bond with the ligand �-carboxylatemoiety, which is further stabilized by Arg398. Such a conserved

TABLE IData collection and refinement statistics

Data collection Native PMP L-Phe

Resolution (Å) 2.0 2.9 2.7Observations 306,700 110,645 81,664Unique reflections 54,768 18,243 22,371Rmerge 5.1 9.6 6.8Multiplicity 5.6 6.1 3.7Completeness (%) 99.9 99.3 98.3

RefinementNo. of protein atoms 3321 3321 3321No. of solvent molecules 305 197 202No. of hetero atoms 15 15 27R-work (%) 19.2 17.0 17.5R-free (%) 22.7 22.0 23.0r.m.s.d. bond lengths (Å) 0.010 0.017 0.020r.m.s.d. bond angles (°) 1.17 1.66 1.85Mean B-factor main chain (Å2) 36.7 50.0 55.7Mean B-factor side chain (Å2) 38.7 51.8 57.4Mean B-factor hetero atoms (Å2) 28.6 31.6 55.5Mean B-factor solvent (Å2) 49.6 50.4 53.5

Human Kynurenine Aminotransferase I Crystal Structure50216

at VIV

A, V

irginia Tech, on M

ay 24, 2011w

ww

.jbc.orgD

ownloaded from

bonding scheme at the L-Phe carboxylic side results in thepositioning of the substrate �-amino group just above the PLPC4� reactive center.

A remarkable structural trait of the ligand binding site inhKAT-I consists of a series of aromatic residues, includingTyr63*, His279*, Phe278*, Tyr101, Phe125, and Trp18, that isreminiscent of a crown (Fig. 4A). In the hKAt-I�Phe structure(Fig. 4C), L-Phe is sandwiched between Phe125 and His279*, inwhich the aromatic planes perpendicularly contact the ligandaromatic ring (distances for the closest pair of atoms being: 3.2Å for L-Phe CE2-NE2 His279*, and 3.6 Å for L-Phe CE2–CZPhe125). A further aromatic contact is provided by Phe278*(closest distance of 3.3 Å) oriented with its aromatic ring almostperpendicular to the ligand ring. Notably, Phe278* is the onlyresidue falling in the disallowed region of the Ramachandranplot in hKAT-I structures, suggesting that the observed confor-mation is functional to substrate recognition and catalysis. Amajor question in kynurenine aminotransferases catalysis con-cerns the different substrate specificities of the isozyme. Un-derstanding the molecular determinants resulting in L-Kynrecognition by hKAT-I will prove to be critical for the rationaldesign of selective inhibitors of potential medical interest. Se-quence alignment of subgroup I aminotransferases from differ-ent species revealed that a tyrosine residue peculiarly featureshuman and rodent L-kynurenine aminotransferases (Tyr101 inhKAT-I). In hKAT-I we observed a striking conformationalchange affecting Tyr101. In fact, upon L-Phe binding, the Tyr101

side chain moves 3.5 Å away from its initial position, makingroom for the incoming ligand (Fig. 5). Being displaced from thecrown of aromatic residues that decorates the L-Phe bindingpocket, Tyr101 establishes a new hydrogen bond with Asn275*and becomes in contact with Cys127. Therefore, Tyr101 appearsto assume two alternative conformations: a close conformationin the PLP form and an open conformation in the liganded formof the enzyme. It has to be noted that Tyr101 has been observedin the closed conformation also in the structure of the PMPform of the enzyme (Fig. 4B). We therefore propose that Tyr101,by alternating between an open and a closed conformationalong the catalytic cycle, acts as a gate controlling substrateadmission to the enzyme active site.

In aminotransferases subgroup I, ligand binding inducesconformational changes resulting in the closure of the enzymeactive site (41, 44). Superimposition of the structures of hKAT-Iin its PLP and phenylalanine-bound forms (Fig. 5) reveals ashift of the N-terminal �-helix H1 (residues Pro17-Glu27), whichmoves toward the active site upon substrate binding. As aresult, Trp18 is brought into contact with the bound L-Phe,whose aromatic ring results perpendicularly oriented with re-spect to the Trp18 indole ring at a closest distance of 3.9 Å. Suchan interaction appears to play an important role in ligandrecognition and stabilization, completing the set of aromaticresidues surrounding and contacting the ligand (Fig. 4C). Inthe recently reported structure of T. thermophilus glutamine:phenylpyruvate aminotransferase (ttGlnAT) (45), a bacterial

FIG. 2. Stereo ribbon representa-tion of the h-KAT-I subunit. The smalland large domains are colored in magentaand blue, respectively. The cleft hostingthe enzyme active site can be seen at thedomain interface where the PLP cofactoris shown as a ball-and-stick representa-tion. The N terminus, C terminus, andhelix H1 are indicated.

FIG. 3. Stereo ribbon representa-tion of the functional dimer ofhKAT-I seen along the crystallo-graphic 2-fold axis. The two subunitsare colored differently, and each of thetwo PLP cofactors is drawn as a ball-and-stick representation.

Human Kynurenine Aminotransferase I Crystal Structure 50217

at VIV

A, V

irginia Tech, on M

ay 24, 2011w

ww

.jbc.orgD

ownloaded from

ortholog of human KAT-I (46), an entire subdomain made ofresidues of the small domain, markedly moves to close up theactive site upon ligand binding. However, the observed exten-sion of the conformational change in hKAT-I is less pronouncedthan that in ttGlnAT upon 3-propionic acid binding (45), amolecule isosteric to L-Phe. A likely explanation could be thatthe Trp18 structurally equivalent position in ttGlnAT is occu-pied by the smaller amino acid phenylalanine (Phe15), which tocontact the bound ligand, needs to move longer.

Because of our failure in obtaining the structure of hKAT-I in

complex with the substrate L-Kyn, a crude model of the L-kynurenine Michaelis complex in hKAT-I was built by manu-ally superimposing L-Kyn onto the ligand molecule in thehKAT-I�L-Phe structure, minimizing for sterical collisions.Based on this model, we confirm the central role of the crown ofaromatic residues in stabilizing the anthraniloyl moiety of L-kynurenine, and we suggest Trp18 as an important residue forsubstrate specificity. In fact, in our model a strong hydrogenbond is established between the benzyl amino group of L-kyn-urenine and the indolamine nitrogen atom of Trp18. However,

FIG. 4. Representation of the activesite in human kynurenine amin-otransferase I. The protein residues, thecofactor, and the ligand are depicted asball-and-stick. Portions of the 2Fo � Fcelectron density map, covering the cofac-tor, Lys247, and the bound ligand, areshown countered at 1 sigma level. A, ste-reo view of the active site in the PLP formwith PLP covalently linked to the cata-lytic Lys247. B, stereo view of the activesite in the PMP form. C, stereo view of theactive site in the complex with L-Phe. Thebound L-Phe sits above the PLP cofactoron its si-face.

Human Kynurenine Aminotransferase I Crystal Structure50218

at VIV

A, V

irginia Tech, on M

ay 24, 2011w

ww

.jbc.orgD

ownloaded from

because our structural data clearly indicate a major role of theinduced fit in ligand recognition, the description of the precisemode of binding of the substrate L-Kyn must await for thedetermination of the crystallographic structure of the hKAT-I�L-Kyn complex.

Role of Glu27—Besides of its role as excitatory amino acidsantagonist, in rodents KA is also involved in the control of thecardiovascular function by acting at rostral ventrolateral me-dulla of the central nervous system (47, 48). Spontaneouslyhypertensive rats, the most widely used animal model forstudying genetic hypertension, have higher arterial blood pres-sure with respect to the normotensive Wistar Kyoto controlrats. This defect has been associated with abnormally low KAlevels in the central nervous system-specific districts tuningphysiological blood pressure (49, 50). Sequencing analysis re-vealed the presence of a single point mutation in the KAT-Igene in all of the examined spontaneously hypertensive rats,leading to the substitution of a glutamate by a glycine atposition 61 in the enzyme (51). Remarkably, the resulting mu-tated protein showed a severely reduced enzymatic activity,hinting at the key role of position 61 in catalysis (51). Sequencealignment of hKAT-I and rat KAT revealed the strict conser-vation of a glutamic residue in this position (Glu27 in hKAT-I).

As described above, the conformational change affecting thehelix H1 upon ligand binding is a key event in the catalysis. InhKAT-I, Glu27 represents the C terminus of the helix H1 andappears as a pivotal residue in fixing the helix in the observedconformations. In fact, it makes a strong salt bridge with Lys23

(distance of 3.5 Å), located in the center of the helix and aweaker interaction with His28 (distance of 5.0 Å) located in theloop following the helix H1 (Fig. 6). Based on our structuraldata, we propose that the substitution Glu613 Gly observed inspontaneously hypertensive rats KAT (Glu61 being equivalentto Glu27 in hKAT-I) would affect the conformational changeinduced by substrate binding with dramatic consequences forthe catalysis.

Conclusion—On the basis of our data, we propose that theobserved conformational changes that occur in hKAT-I onligand binding play a key role in catalysis. The shift of the�-helix H1 and the consequent recruitment of Trp18 to theactive site, together with the displacement of Tyr101 (Fig. 6),result in a remodeling of the aromatic crown representing amajor determinant for substrate binding. Whereas Trp18 ap-pears to play a relevant role in substrate specificity, Tyr101,by alternating between its two conformations, controls sub-strate admission. Such structural information will provide asolid foundation for the rational design of small moleculeinhibitors specifically targeting the hKAT-I isozyme, which

will prove useful both as a tool for further investigation of thebiological functions of KATs and as compounds of potentialmedical interest.

Acknowledgments—We thank Dr. Alessandro Coda and Dr. AndreaMattevi (University of Pavia, Italy) for their constant support. We alsothank Dr. Roberto Pellicciari and Dr. Daniele Bellocchi (University ofPerugia, Italy) for helpful discussion. We thank Dr. Ken Hirotsu (Uni-versity of Osaka, Japan) for kindly providing us with the coordinates ofttGlnAT. We thank the European Synchrotron Radiation Facility(Grenoble, France) for data collection at the beam line ID14-EH1.

REFERENCES

1. Stone, T. W., and Darlington, L. G. (2002) Nat. Rev. Drug Discov. 1, 609–6202. Eastman, C. L., and Guilarte, T. R. (1989) Brain Res. 495, 225–2313. Schwarcz, R., and Pellicciari, R. (2002) J. Pharmacol. Exp. Ther. 303, 1–104. Stone, T. W. (2001) Prog. Neurobiol. (Oxf.) 64, 185–2185. Perkins, M. N., and Stone, T. W. (1982) Brain Res. 247, 184–1876. Leeson, P. D., and Iversen, L. L. (1994) J. Med. Chem. 37, 4053–40677. Birch P. J., Grossman, C. J., and Hayes, A. G. (1988) Eur. J. Pharmacol. 154,

85–878. Pereira, E. F., Hilmas, C., Santos, M. D., Alkondon, M., Maelicke, A., and

Albuquerque, E. X. (2002) J. Neurobiol. 53, 479–5009. Hilmas C., Pereira, E. F., Alkondon, M., Rassoulpour, A., Schwarcz, R., and

Albuquerque, E. X. (2001) J. Neurosci. 21, 7463–747310. Alkondon, M., Pereira, E. F., Yu, P., Arruda, E. Z., Almeida, L. E., Guidetti, P.,

Fawcett, W. P., Sapko, M. T., Randall, W. R., Schwarcz, R., Tagle, D. A., andAlbuquerque, E. X. (2004) J. Neurosci. 24, 4635–4648

11. Stone, T. W. (2000) Trends Pharmacol. Sci. 21, 149–154

FIG. 5. Stereo view of part of thehKAT-I active site in the PLP and L-Phe-bound forms after optimal su-perimposition of the two structures.The PLP and L-Phe-bound forms are col-ored in white and green, respectively. Thetwo alternative conformations of Tyr101

can be observed together with the move-ment of helix H1 and Trp18.

FIG. 6. Ribbon representation of the hKAT-I subunit underlin-ing the proposed structural role of Glu27 located at the C termi-nus of helix H1. In rat kynurenine aminotransferase, the indicatedmutation (Glu 3 Gly) results in an hypertensive phenotype. PLP mol-ecule and amino acid residues are drawn as a ball-and-stick represen-tation. The N-terminal arm and the helix H1 are colored in green.

Human Kynurenine Aminotransferase I Crystal Structure 50219

at VIV

A, V

irginia Tech, on M

ay 24, 2011w

ww

.jbc.orgD

ownloaded from

12. Pellicciari, R., Natalini, B., Costantino, G., Mahmoud, M. R., Mattoli, L., andSadeghpour, B. M., Moroni, F., Chiarugi, A., and Carpenedo, R. (1994)J. Med. Chem. 37, 647–655

13. Smith, D. H., Okiyama, K., Thomas, M. J., and McIntosh, T. K. (1993) J. Neu-rosci. 13, 5383–5392

14. Yamamoto, H., Shindo, I., Egawa, B., and Horiguchi, K. (1994) Pediatr. Neurol.10, 9–12

15. Erhardt, S., Schwieler, L., and Engberg, G. (2003) Adv. Exp. Med. Biol. 527,155–165

16. Baran, H., Cairns, N., Lubec, B., and Lubec, G. (1996) Life Sci. 58, 1891–189917. Baran, H., Jellinger, K., and Deecke, L. (1999) J. Neural Transm. 106,

165–18118. Gramsbergen, J. B., Schmidt, W., Turski, W. A., and Schwarcz, R. (1992) Brain

Res. 588, 1–519. Jansonius, J. N. (1998) Curr. Opin. Struct. Biol. 8, 759–76920. Jensen, R. A., and Gu, W. (1996) J. Bacteriol. 178, 2161–217121. Okuno, E., Du, F., Ishikawa, T., Tsujimoto, M., Nakamura, M., Schwarcz, R.,

and Kido, R. (1990) Brain Res. 534, 37–4422. Alberati-Giani, D., Malherbe, P., Kohler, C., Lang, G., Kiefer, V., Lahm, H. W.,

and Cesura, A. M. (1995) J. Neurochem. 64, 1448–145523. Buchli, R., Alberati-Giani, D., Malherbe, P., Kohler, C., Broger, C., and Cesura,

A. M. (1995) J. Biol. Chem. 270, 29330–2933524. Okuno, E., Nakamura, M., and Schwarcz, R. (1991) Brain Res. 542, 307–31225. Guidetti, P., Okuno, E., and Schwarcz, R. (1997) J. Neurosci. Res. 50, 457–46526. Baran, H., Okuno, E., Kido, R., and Schwarcz, R. (1994) J. Neurochem. 62,

730–73827. Cooper, A. J., and Meister, A. (1974) J. Biol. Chem. 249, 2554–256128. Cooper, A. J. L. (2004) Neurochem. Int. 44, 557–57729. Perry, S., Harries, H., Scholfield C., Lock, T., King, L., Gibson, G., and Gold-

farb, P. (1995) FEBS Lett. 360, 277–28030. Commandeur, J. N., Andreadou, I., Rooseboom, M., Out, M., De Leur, L. J.,

Groot, E., and Vermeulen, N. P. (2000) J. Pharmacol. Exp. Ther. 294,753–761

31. Rooseboom, M., Vermeulen, N. P., Durgut, F., and Commandeur, J. N. (2002)Chem. Res. Toxicol. 15, 1610–1618

32. Collaborative Computing Project Number 4 (1994) Acta Crystallogr. Sect. D

Biol. Crystallogr. 50, 760–76333. Ura, H., Nakai, T., Kawaguchi, S., Miyahara, I., Hirotsu, K., and Kuramitsu,

S. (2001) J. Biochem. (Tokyo) 130, 89–9834. Navaza, J. (2001) Acta Crystallogr. Sect. D Biol. Crystallogr. 57, 1367–137235. Murshudov, G. N., Vagin, A. A., and Dodson, E. J. (1997) Acta Crystallogr.

Sect. D Biol. Crystallogr. 53, 240–25536. Jones, T. A., Zou, J. Y., Cowan, S. W., and Kjeldgaard, M. (1991) Acta Crys-

tallogr. Sect. A 47, 110–11937. Brunger, A. T. (1992) Nature 355, 472–47438. Laskowsky, R. A., MacArthur, M. W., Moss, D. S., and Thornton, J. M. (1993)

J. Appl. Crystallogr. 26, 283–29139. Kraulis, P. J. (1991) J. Appl. Crystallogr. 24, 946–95040. Esnouf, R. M. (1997) J. Mol. Graphics 15, 132–13441. Nakai, T., Okada, K., Akutsu, S., Miyahara, I., Kawaguchi S., Kato, R.,

Kuramitsu, S., and Hirotsu, K. (1999) Biochemistry 38, 2413–242442. Fukumoto, Y., Tanase, S., Nagashima, F., Ueda, S., Ikegami, K., and Morino,

Y. (1991) J. Biol. Chem. 266, 4187–419343. Denesyuk, A. I., Denessiouk, K. A., Korpela, T., and Johnson, M. S. (2002) J.

Mol. Biol. 316, 155–17244. Okamoto, A., Higuchi, T., Hirotsu, K., Kuramitsu, S., and Kagamiyama, H.

(1994) J. Biochem. (Tokyo) 116, 95–10745. Goto, M., Omi, R., Miyahara, I., Hosono, A., Mizuguchi, H., Hayashi, H.,

Kagamiyama, H., and Hirotsu, K. (2004) J. Biol. Chem. 279, 16518–1652546. Hosono, A., Mizuguchi, H., Hayashi, H., Goto, M., Miyahara, I., Hirotsu, K.,

and Kagamiyama, H. (2003) J. Biochem. (Tokyo) 134, 843–85147. Dampney, R. A. L. (1994) Physiol. Rev. 74, 323–36448. Colombari, E., Sato, M. A., Cravo, S. L., Bergamaschi C. T., Campos, R. R., and

Lopes, O. U. (2001) Hypertension 38, 549–55449. Kapoor, V., Kapoor, R., and Chalmers, J. (1994) Clin. Exp. Pharmacol. Physiol.

21, 891–89650. Ito, S., Komatsu, K., Tsukamoto, K., and Sved, A. F. (2000) Hypertension 35,

413–41751. Kwok, J. B., Kapoor, R., Gotoda, T., Iwamoto, Y., Iizuka, Y., Yamada, N.,

Isaacs, K. E., Kushwaha, V. V., Church, W. B., Schofield, P. R., and Kapoor,V. (2002) J. Biol. Chem. 277, 35779–35782

Human Kynurenine Aminotransferase I Crystal Structure50220

at VIV

A, V

irginia Tech, on M

ay 24, 2011w

ww

.jbc.orgD

ownloaded from