Kynurenine formamidase: determination of primary structure and modeling-based prediction of tertiary structure and catalytic triad1

Biochimica et Biophysica Acta 1596 (2002) 201^211 www.bba-direct.com

Kynurenine formamidase: determination of primary structure and modeling-based prediction of tertiary structure and catalytic triad1 Michael K. Pabarcus a

a;b

, John E. Casida

a;

Environmental Chemistry and Toxicology Laboratory, Department of Environmental Science, Policy and Management, University of California, Berkeley, CA 94720-3112, USA b Graduate Group in Comparative Biochemistry, University of California, Berkeley, CA 94720-3112, USA Received 9 October 2001; received in revised form 27 December 2001; accepted 25 January 2002

Abstract Kynurenine formamidase (KFase) (EC 3.5.1.9) hydrolyzes N-formyl-L-kynurenine, an obligatory step in the conversion of tryptophan to nicotinic acid. Low KFase activity in chicken embryos, from inhibition by organophosphorus insecticides and their metabolites such as diazoxon, leads to marked developmental abnormalities. While KFase was purportedly isolated previously, the structure and residues important for catalysis and inhibition were not established. KFase was isolated here from mouse liver cytosol by (NH4 )2 SO4 precipitation and three FPLC steps (resulting in 221-fold increase in specific activity for N-formyl-L-kynurenine hydrolysis) followed by conversion to [3 H]diethylphosphoryl-KFase and finally isolation by C4 reverse-phase high-performance liquid chromatography. Determination of tryptic fragment amino acid sequences and cDNA cloning produced a new 305-amino-acid protein sequence. Although an amidase by function, the primary structure of KFase lacks the amidase signature sequence and is more similar to esterases and lipases. Sequence profile analysis indicates KFase is related to the esterase/lipase/thioesterase family containing the conserved active-site serine sequence GXSXG. The K/L-hydrolase fold is suggested for KFase by its primary sequence and predicted secondary conformation. A three-dimensional model based on the structures of homologous carboxylesterase EST2 and brefeldin A esterase implicates Ser162, Asp247 and His279 as the active site triad. ß 2002 Elsevier Science B.V. All rights reserved. Keywords: Amidase; Catalytic triad; Esterase; Homology modeling; Kynurenine formamidase; Serine hydrolase

Abbreviations: AChE, acetylcholinesterase; BFAE, brefeldin A esterase; C4RP, C4 reverse-phase; [3 H]DEP-KFase, [ H]diethylphosphoryl-KFase; EST, expressed sequence tag; EST2, carboxylesterase EST2; FPLC, fast protein liquid chromatography; HPLC, high-performance liquid chromatography; KFase, kynurenine formamidase; MW, molecular weight; NAD(P)þ , nicotinamide adenine dinucleotide (phosphate); OP, organophosphorus; PCR, polymerase chain reaction; PDB, Protein Data Bank; pI, isoelectric point; PSHP, Phenyl Sepharose high-performance; QSFF, Q-Sepharose fast £ow; RACE, rapid ampli¢cation of cDNA ends; RMSD, root mean square deviation; SDX200, Superdex 200; 3D, three-dimensional * Corresponding author. Fax: +1-510-642-6497. E-mail address: [email protected] (J.E. Casida). 1 Portions of this paper were presented at the Fifteenth Symposium of The Protein Society (FASEB), Philadelphia, PA, USA, July 28^ August 1, 2001. 3

0167-4838 / 02 / $ ^ see front matter ß 2002 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 7 - 4 8 3 8 ( 0 2 ) 0 0 2 3 2 - 7

BBAPRO 36608 1-5-02

202

M.K. Pabarcus, J.E. Casida / Biochimica et Biophysica Acta 1596 (2002) 201^211

1. Introduction Kynurenine formamidase (KFase) (EC 3.5.1.9) is a key enzyme in the metabolic sequence of tryptophan, N-formyl-L-kynurenine, L-kynurenine and nicotinic acid to nicotinamide adenine dinucleotide (phosphate) (NAD(P)þ ) cofactors [1]. The importance of KFase in metabolic biochemistry is also evident in toxicological manifestations. Several organophosphorus (OP) insecticides and their metaboites induce severe teratogenic e¡ects in developing chicken embryos associated with inhibition of KFase activity, leading to lowered levels of NAD [2^6]. The potency for a series of OP compounds as mouse liver KFase inhibitors was directly related to the severity of chick teratogenesis induced by these chemicals, indicating the mouse enzyme as a suitable model for study [3]. Diazoxon, the metabolically activated form of diazinon, is a potent KFase inhibitor and avian teratogen [5]. Diazinon inhibits cholinesterase and also produces several abnormalities in mammals [7^10] ranging from skull malformations and extra metacarpal bones in miniature swine [11] to behavioral retardation [12] and increased kynurenic and xanthurenic acid excretion [13] in mice, possibly associated in part with KFase inhibition. Teratogenic e¡ects of OP insecticides are of two types: type I from KFase inhibition creating a block in amino acid metabolism; type II from acetylcholinesterase (AChE) inhibition disrupting nerve function [6]. The structural features of the OP inhibitor and the enzyme target determine the type of teratogenic e¡ect which is induced [5,6]. KFase has not been characterized prior to this investigation relative to primary structure or molecular biology. KFase serves as an amidase in performing its physiological role of hydrolyzing N-formyl-L-kynurenine, but it is also active on other amide substrates and on several esters [14,15]. It might therefore contain the amidase signature sequence or be related to the esterase/lipase/thioesterase family. The catalytic site of KFase is inhibited by diazoxon suggesting that a serine group undergoes diethylphosphorylation. These hypotheses are tested by purifying native KFase from mouse liver cytosol and ¢nally isolating pure [3 H]diethylphosphoryl-KFase ([3 H]DEP-KFase) (Fig. 1). The primary amino acid sequence is deduced

Fig. 1. Puri¢cation of KFase and isolation of [3 H]DEP-KFase from mouse liver cytosol in six steps. 1, (NH4 )2 SO4 precipitation; 2, Q-Sepharose Fast Flow (QSFF) ion exchange fast protein liquid chromatography (FPLC); 3, Phenyl Sepharose highperformance (PSHP) hydrophobic interaction FPLC; 4, Superdex 200 (SDX200) gel ¢ltration FPLC; 5, KFase labeling with [3 H]diazoxon; 6, C4 reverse-phase (C4RP) high-performance liquid chromatography (HPLC).

through isolation and sequencing of the cDNA coding for KFase using gene-speci¢c primers designed from sequences found by analysis of tryptic fragments. Distant sequence similarity and secondary structural threading to proteins with known threedimensional (3D) structures permit the generation of an homology model for KFase. The proposed model tentatively identi¢es residues that may constitute the catalytic triad.

2. Materials and methods 2.1. Chemicals and reagents N-Formyl-L-kynurenine was prepared from L-kynurenine and formic acid/acetic anhydride [16,17] in 85% yield and s 99% purity as a light yellow powder after C18 reverse-phase HPLC. Sources for other chemicals and reagents were as follows: diazoxon and [3 H]diazoxon (58 Ci/mmol) from synthesis in this laboratory [18]; column chromatography resins and gel ¢ltration calibration kit from Amersham Pharmacia Biotech (Piscataway, NJ); mouse liver cDNA and PCR reagents from Clontech (Palo Alto, CA); polymerase chain reaction (PCR) and sequencing primers from Operon Technologies (Alameda, CA); deglycosylation enzymes and reagents from PROzyme (San Leandro, CA).

BBAPRO 36608 1-5-02

M.K. Pabarcus, J.E. Casida / Biochimica et Biophysica Acta 1596 (2002) 201^211

2.2. KFase activity assays Enzyme assays were performed in 96-well £at-bottomed plates in 50 mM sodium phosphate (pH 7.4) bu¡er (200 Wl) to which 5 mM substrate (20 Wl) was added. The same conditions were used for inhibition experiments except for addition of diazoxon in ethanol at a ¢nal concentration of 5% with 90 min incubation at room temperature prior to substrate addition. Activity was monitored for 90 s as the increase of absorbance at 365 nm resulting from liberation of L-kynurenine using a Molecular Devices (Menlo Park, CA) UVmax Kinetic Microplate Reader with SOFTmax v. 2.35 data analysis software. Protein determinations [19] used Bio-Rad (Hercules, CA) Microplate Protein Assay with bovine serum albumin as the standard. 2.3. SDS^PAGE SDS^PAGE was performed in 12% Tris^glycine gels [20] with electrophoresis for 1 h at 150V. Gels were stained with Coomassie Brilliant blue R-250 and de-stained with 50:40:10 water/methanol/acetic acid. Bio-Rad Molecular Weight Markers, low range, were used to calibrate the gels. The molecular weight (MW) contribution from post-translational glycosylation of KFase was determined using the PROzyme Glycopro Deglycosylation kit. Enzymes (PNGase F, Sialidase A and Endo-O-glycosidase) for removal of N- and O-linked sugars were added to unlabeled, C4RP-HPLC puri¢ed KFase and the control glycoprotein, bovine fetuin. Proteins were heat denatured in the presence of SDS and L-mercaptoethanol prior to treatment with enzymes. Samples were analyzed by gel-shift assay. 2.4. Puri¢cation and properties of KFase Frozen mouse liver (204 g) (Pel-Freez Biologicals, Rogers, AZ) was homogenized at 20% w/v in 50 mM sodium phosphate (pH 7.4) and 1 mM EDTA at 4‡C in a Waring blender. Cytosol was prepared by centrifugation, ¢rst at 10 000Ug for 20 min and then at 100 000Ug for 60 min with one wash of the microsomal pellet. The 35^75% (NH4 )2 SO4 precipitate [21] was dissolved in 50 mM sodium phosphate (pH 7.4) and 1 mM EDTA, and desalted into 20 mM Tris^

203

HCl, 1 mM EDTA (pH 6.4) by gel ¢ltration with Sephadex G-25. Further puri¢cation using an Amersham Pharmacia Biotech FPLC system involved three columns monitoring KFase activity. For the ¢rst column the desalted 35^75% (NH4 )2 SO4 fraction was loaded onto a QSFF column (2.6U35 cm) equilibrated in 20 mM Tris^HCl, 1 mM EDTA (pH 6.4) at 5 ml/min. Elution involved a gradient of 0^0.40 M NaCl in the same equilibration bu¡er over 240 min at 5 ml/min. Pooled fractions with KFase activity were precipitated with 80% (NH4 )2 SO4 . For the second FPLC step, a solution of the precipitate in 50 mM sodium phosphate (pH 7.4) was adjusted to 1 M (NH4 )2 SO4 for loading onto a PSHP column (1.6U29 cm) equilibrated in 50 mM sodium phosphate, 1 M (NH4 )2 SO4 (pH 7.4) at 2 ml/min. Elution utilized a gradient of 1.0^0 M (NH4 )2 SO4 in 50 mM sodium phosphate over 200 min at 2 ml/min. The pooled KFase fractions were precipitated with 80% (NH4 )2 SO4 . For the ¢nal FPLC step, the precipitate dissolved in 50 mM sodium phosphate (pH 7.4), 150 mM NaCl and 1 mM EDTA was loaded onto a SDX200 column (2.6U59.5 cm) equilibrated and eluted with the above bu¡er at 1 ml/min. Pooled KFase fractions were concentrated in an Amicon (Beverly, MA) Centricon 10 and stored at 380‡C. The SDX200 gel ¢ltration column was calibrated with standards (blue dextran, catalase, aldolase, bovine serum albumin, ovalbumin, chymotrypsinogen A and ribonuclease A) to determine the apparent MW of eluted KFase activity. The isoelectric point (pI) for KFase (SDX200 preparation) was determined by chromatofocusing on a column of Polybu¡er Exchanger 94 (0.5U20 cm) equilibrated in 25 mM bis-Tris^HCl (pH 6.0) using an in situ pH gradient generated with Amersham Pharmacia Biotech Polybu¡er 74 (diluted 1:8 with water and adjusted to pH 4.0 with HCl) under isocratic conditions at 0.15 ml/min. The apparent pI was measured as the pH at which KFase activity eluted. 2.5. Isolation and identi¢cation of [3 H]DEP-KFase To prepare [3 H]DEP-KFase, protein from the SDX200 fraction (250 Wg) was incubated in 50 mM sodium phosphate (pH 7.4) (50 Wl) for 2 h with [3 H]diazoxon (4 mM, estimated to be 100- to 200fold excess relative to the KFase concentration to

BBAPRO 36608 1-5-02

204

M.K. Pabarcus, J.E. Casida / Biochimica et Biophysica Acta 1596 (2002) 201^211

assure full reaction with the speci¢c labeling site). The labeled protein was isolated using a Vydac (Hesperia, CA) 214TP54 C4 column (4.6 mmU25 cm) with guard cartridge developed with an acetonitrile gradient at 1.5 ml/min on a Waters HPLC system (Millipore Corp., Milford, MA), monitoring continuously at 220 nm and at 0.5-min intervals by liquid scintillation counting. Solvent A was 0.1% tri£uoroacetic acid in water and solvent B was 0.08% tri£uoroacetic acid in acetonitrile. Samples were analyzed on a gradient from 35^55% B over 20 min. The purity of the ¢nal product was examined by SDS^ PAGE and C4RP-HPLC based on protein staining and UV absorbance of radiolabeled protein(s), respectively. To verify the labeled protein as [3 H]DEP-KFase, the SDX200 preparation (5 Wg/ml) was preincubated with unlabeled diazoxon for 120 min, then assayed for KFase activity and available sites for postlabeling by [3 H]diazoxon overnight at room temperature. An excess of [3 H]diazoxon (2 mM) was added to favor phosphorylation by [3 H]diazoxon over the small amount of remaining unlabeled inhibitor present from the initial inhibition assay. Unbound label was removed by passing samples through Bio-Rad 10-DG columns, prior to liquid scintillation counting of the protein fraction. 2.6. Protein sequence Pure [3 H]DEP-KFase (HPLC puri¢ed and diazoxon free) was used to determine the fragment primary structure by N-terminal sequencing of peptides generated from tryptic digestion and C18 reverse-phase HPLC separation in studies carried out at the Protein Structure Facility of the University of California at Davis. Sequence comparisons of peptide fragments with protein and DNA sequence databases were performed using World Wide Web-based tools. Direct protein comparisons used the BLAST tool from the National Center for Biotechnology Information [22] and protein sequence comparison with gaps used the WuBLAST tool from the European Bioinformatics Institute [23]. The tfastxy tool for comparison to translated DNA sequences [24] and InterProScan for protein primary sequence pro¢le analysis [25] were also accessed from the European Bioinformatics Institute. Protein per cent identity

and similarity were calculated using GeneDoc Multiple Sequence Alignment and Shading Utility v. 2.6.001 [26]. 2.7. DNA sequence Multiple DNA sequences obtained from database searches were aligned using GeneDoc as above [26]. PCR^rapid ampli¢cation of cDNA ends (PCR^ RACE) was performed on a MJ Research (Waltham, MA) PTC-200 Peltier Thermal Cycler. Mouse liver Marathon Ready cDNA from Clontech was used along with the Advantage cDNA PCR kit from Clontech to amplify 5P and 3P cDNA fragments using gene speci¢c primers MKP1: ATGGCGTTTCCTTCCCTTTCTGCGGG for the 3P-RACE and MKP2: AGGCCTTCCATCCCACGCGGACA for the 5P-RACE. A library-speci¢c primer, AP1, was used with each gene-speci¢c primer. Thermal cycler conditions were: (1) 94‡C for 30 s; (2) 5 cycles of 94‡C for 5 s, 72‡C for 3 min; (3) 5 cycles of 94‡C for 5 s, 70‡C for 3 min; (4) 25 cycles of 94‡C for 5 s, 65‡C for 30 s, 68‡C for 3 min. PCR reactions were cleaned up with a Qiagen (Valencia, CA) QIAquick PCR puri¢cation kit prior to sequencing. Analysis and documentation of DNA by agarose gel electrophoresis was performed by previously established methods [27]. Automated cycle sequencing was performed by the DNA Sequencing Facility of the University of California at Berkeley. Electropherograms were analyzed and edited with Chromas v. 1.62 (Technelysium, Helensvale, Australia). DNA sequences were analyzed for open reading frames using the ORF ¢nder tool from the National Center for Biotechnology Information World Wide Web site. 2.8. Structural prediction and 3D modeling Secondary structure was predicted by PHDsec [28] and secondary structural threading was performed with TOPITS using the PredictProtein analysis tool from the European Bioinformatics Institute. The atomic coordinates for modeling templates were obtained from the Brookhaven Protein Data Bank (PDB). Structures were visualized and models analyzed using Swiss-PdbViewer v. 3.6b3 with homology modeling performed using Swiss-Model [29].

BBAPRO 36608 1-5-02

M.K. Pabarcus, J.E. Casida / Biochimica et Biophysica Acta 1596 (2002) 201^211

205

2.9. Identi¢cation of catalytic triad The Ser, Asp and His residues likely involved in the catalytic triad of KFase were tentatively identi¢ed through sequence analysis, [3 H]diazoxon labeling and structural alignment of known active site residues in the templates with the homology model.

3. Results 3.1. Puri¢cation and properties of KFase The four-step puri¢cation starting with 19.6 g of mouse liver cytosolic protein (Table 1 and Fig. 2) yielded 12 mg of partially puri¢ed, active KFase with a 221-fold puri¢cation factor for speci¢c activity and a 13% overall recovery of total activity. The 35^ 75% (NH4 )2 SO4 precipitate, with 1.3-fold puri¢cation, contained 80% of the initial activity with some loss attributable to residual supernatant in the unwashed 35% (NH4 )2 SO4 pellet. Ion exchange chromatography on QSFF resulted in a single peak of enzymatic activity between 0.13 and 0.18 M NaCl and 15fold overall puri¢cation. PSHP chromatography utilizing hydrophobic interaction yielded a single active peak at 0.4^0.2 M (NH4 )2 SO4 , with 89-fold cumulative puri¢cation. SDX200 gel ¢ltration also produced a single peak of enzymatic activity with 221-fold overall puri¢cation and an apparent MW of 34 kDa. An apparent pI of 5.4 was observed by chro-

Fig. 2. Puri¢cation of KFase. Samples are designated as in Fig. 1. Fractions (40 Wg each except 10 Wg for C4RP) analyzed by SDS^PAGE (heated at 90‡C for 2 min in Laemmli sample bu¡er and run on 12% Tris^glycine gel). The same mobility was observed for KFase and [3 H]DEP-KFase.

matofocusing for the active peak in the SDX200 fraction. Pure but enzymatically inactive KFase (isolated by C4RP-HPLC as below for [3 H]DEP-KFase but without inhibitor) was used for SDS^PAGE analysis of molecular mass and glycosylation. KFase migrates with an apparent MW of 36 kDa (Fig. 2). There is little or no post-translational glycosylation since treatment with deglycosylation enzymes produced no shift in molecular mass compared to controls (Fig. 3).

Table 1 Puri¢cation of KFase from mouse liver cytosol Puri¢cation step Based on KFase activity Cytosol Ammonium sulfate 35^75% QSFF PSHP SDX200 Based on [3 H]DEP-KFase C4RP-HPLC a b

Total protein (mg)

Total activity (mmol/min/mg)

Speci¢c activity (mmol/min/mg)

19600 12765 578 59 12

5540 4465 2503 1456 747

0.28 0.35 4.3 25 62

a

a

2

No KFase activity. Puri¢cation factor determined from HPLC peak integration.

BBAPRO 36608 1-5-02

Puri¢cation (fold) 1.0 1.3 15 89 221

1206b

Yield (%) 100 80 45 26 13

a

206

M.K. Pabarcus, J.E. Casida / Biochimica et Biophysica Acta 1596 (2002) 201^211

Fig. 3. Possible contribution of glycosylation to the apparent MW of KFase detected by gel shift assay. SDS^PAGE analysis of unlabeled C4RP-HPLC puri¢ed KFase and control protein (bovine fetuin) treated (+) and not treated (3) with deglycosylation enzymes.

3.2. Isolation and identi¢cation of [3 H]DEP-KFase KFase was one of several proteins in the SDX200 preparation based on SDS^PAGE (Fig. 2) and C4RP-HPLC (Fig. 4A). The KFase component was

Fig. that tion data

5. Diazoxon sensitivity of KFase activity compared with for [3 H]diazoxon postlabeling. Linear regression of inhibivalues for KFase activity and [3 H]diazoxon postlabeling gives r = 0.99.

recognized by covalent labeling of a single protein with [3 H]diazoxon at 19.2 min by C4RP-HPLC (Fig. 4B) and at 36 kDa by SDS^PAGE (data not shown). The labeled protein was assigned as [3 H]DEP-KFase (with labeling assumed to be at the catalytic site) by an experiment which showed the same inhibition curve for diazoxon (IC50 45 nM) based on enzyme activity and [3 H]diazoxon postlabeling (r = 0.99) (Fig. 5). [3 H]DEP-KFase was therefore used to monitor further puri¢cation to homogeneity as a single SDS^PAGE band (Fig. 2) and a pure HPLC peak ( s 99%, data not shown). 3.3. Protein sequence

Fig. 4. Analysis of SDX200 KFase preparation and [3 H]DEPKFase under identical conditions by C4RP-HPLC. (A) Protein continuously monitored at 220 nm. (B) [3 H]DEP-KFase detected as cpm for 0.5-min fractions.

Direct sequencing of [3 H]DEP-KFase failed, indicating a blocked (probably acetylated) N-terminus. Four unique peptide sequences obtained by tryptic digestion (Table 2) proved to be su⁄cient for database searching. No signi¢cant similarity to any known protein sequences was found using BLAST and WuBLAST database search tools. However, 100% identity was found for each fragment in ¢ve overlapping and previously unassigned mouse expressed sequence tags (ESTs) identi¢ed using the tfastxy tool comparing protein sequence to translated DNA sequence (Table 2).

BBAPRO 36608 1-5-02

M.K. Pabarcus, J.E. Casida / Biochimica et Biophysica Acta 1596 (2002) 201^211

207

3.4. DNA sequence Cleanup of the overlapping ESTs produced an 892 bp sequence coding for all experimentally generated fragment sequences. Extension of the 5P and 3P ends of KFase cDNA from a mouse liver cDNA library using PCR^RACE resulted in one fragment from each reaction (Fig. 6). The control reaction using the two gene-speci¢c primers also produced one fragment. The size of the 5P-RACE product was between 750 and 1000 bp. Sequencing of the KFase 5P cDNA fragment con¢rmed the previously established sequence. No further 5P sequence data were obtained as all sequencing reactions toward that end failed to produce reliable sequence beyond the previously established 5P end. The 3P-RACE product was approximately 1500 bp in length. The KFase 3P cDNA fragment yielded 576 bp of new sequence. The resulting 1468 bp cDNA sequence (GenBank accession number AF399717) contains an open reading frame from bp 27 to 944. The translated protein of 305 amino acids (Fig. 7) has a predicted MW of 34 kDa and a predicted pI of 5.7. 3.5. Structural prediction and 3D modeling To generate a 3D model for KFase, suitable templates were sought using BLAST and TOPITS. The BLAST results showed some similarity between Table 2 KFase tryptic fragment sequences with corresponding mouse EST GenBank accession numbers Sequencea

Mouse ESTs

(RN)LSSEELEKQYSPSR

AA245789 AA537957 AW611244 AA245789 AA537957 AW611244 AI304009 AA832566 AA537957 AW611244 AA832566

KPEEVVGNFVQIGSQATQK

(R)YPSNEGIYLCGb

(R)HLDVVPAQPVAPACPVLVL a

Amino acids in parentheses are predicted from tryptic cleavage sites but not observed at high con¢dence levels. b Region of unlabeled sequence found in larger 3 H-labeled tryptic fragment. The other tryptic fragments shown were not labeled.

Fig. 6. Ampli¢cation of 5P and 3P cDNA fragments from mouse liver cDNA library using gene-speci¢c primers MKP2 and MKP1 for 5P- and 3P-RACE, respectively. PCR^RACE products analyzed by gel electrophoresis directly after ampli¢cation.

KFase and many esterases and lipases. No mammalian enzyme of known structure was suitable, with prokaryotic templates showing the best similarity to KFase. The search gave 25% similarity for the thermophilic carboxylesterase EST2 from Alicyclobacillus î , PDB entry acidocaldarius (EST2) solved at 2.6 A 1EVQ [30]. Structural threading using TOPITS identi¢ed brefeldin A esterase (BFAE) from Bacillus subî , PDB entry 1JKM [31]. tilis solved at 1.85 A Although several other low scoring matches were identi¢ed, BFAE (with an alignment score of 2.00) was the only serine hydrolase in the group. BFAE is reported as a homolog of human hormone sensitive lipase [31] supporting the use of this prokaryotic template to model the eukaryotic KFase. EST2 and BFAE show 33% identity to each other and share similar 3D structures displaying the K/L-hydrolase fold [32]. The family relationship among the identi¢ed templates and KFase is illustrated by primary sequence elements present in all three proteins, such as the active site serine motif GXSXG and the HGG motif found on the N-terminal side of active site elements [33]. Finally, KFase hydrolyzes ester substrates [14,15] again supporting modeling based on these esterase templates. The 3D homology model of KFase was based on manually adjusted structural alignment between KFase, EST2 and BFAE. Fig. 8A shows the general folding pattern of the KFase model based on these templates. The superimposed backbone traces of î root mean KFase and EST2 displayed a 0.56 A

BBAPRO 36608 1-5-02

208

M.K. Pabarcus, J.E. Casida / Biochimica et Biophysica Acta 1596 (2002) 201^211

Fig. 7. KFase primary sequence, predicted secondary structure and modeling alignment. The primary sequence of KFase determined here is shown relative to the template sequences after manual manipulation of the structural driven alignment of EST2 and BFAE. The predicted (pred) secondary structure of KFase is shown in the line above the modeled (mod) secondary structure. h, K-helix; s, L-sheet. Residues shown in bold represent known components of the template catalytic triads and the modeled components of the KFase catalytic triad (Ser162, Asp247 and His279).

square deviation (RMSD) on 254 Ca atoms using a î cuto¡ distance (data not shown). In a global 2.5 A î, a superposition requiring a cuto¡ distance of 15 A î 2.44 A RMSD is seen on 286 Ca atoms. Good values were also obtained in the superposition with BFAE, î on 198 Ca atoms using a 2.5 A î i.e., RMSD = 1.01 A î cuto¡ distance and RMSD = 3.62 A on 297 Ca atoms î cuto¡ distance. The Ramachandran using a 15 A plot (not shown) indicates that most (92%) residues have P and i angles in the core or allowed regions, with the non-allowed residues residing mainly in loop regions. Of the non-allowed residues, 33% are Gly or Pro. No amino acid clashes were observed and only one residue (Pro235) was found to have a backbone con£ict. Of the 158 non-polar residues in KFase, only 35 (22%) are partially or fully exposed with the remainder in non-accessible protein core regions. The 75 polar residues are located mainly in surface regions with only 25 (33%) partially or fully buried.

3.6. Identi¢cation of catalytic triad Sequence analysis identi¢ed the active site serine motif GHSAG at residues 160^164 (Fig. 7). A KFase tryptic fragment (Table 2) corresponding to residues 150^175 (Fig. 7) was found in sequencing experiments to contain the 3 H-label, presumably at Ser162 of [3 H]DEP-KFase. KFase residues Ser162, Asp247 and His279 were also all seen in the homology model to align with the known catalytic triad components of EST2 and BFAE (Figs. 7 and 8B).

4. Discussion 4.1. Puri¢cation of KFase and identi¢cation of [3 H]DEP-KFase KFase was puri¢ed from mouse liver cytosol by (NH4 )2 SO4 precipitation and three FPLC steps, re-

BBAPRO 36608 1-5-02

M.K. Pabarcus, J.E. Casida / Biochimica et Biophysica Acta 1596 (2002) 201^211

209

Fig. 8. Homology model of KFase and its active site. (A) Schematic representation of structural components. Strands L1^L8 and helices K4, K5, K6, K7, K9 and K11 correspond to the prototypic K/L hydrolase fold. The linear succession of secondary structures in the KFase model is K1-K2-K3-L1-L2-L3-K4-L4-K5-L5-K6-L6-K7-K8-L7-K9-L8-K10-K11. (B) Superposition of catalytic triad elements between KFase (black), EST2 (dark gray) and BFAE (light gray). Note the excellent alignment of backbone traces and the similar positioning of catalytic triad residue side chains. Amino acids are designated relative to KFase sequence.

sulting in a 221-fold increase in speci¢c activity for hydrolysis of N-formyl-L-kynurenine. Each FPLC puri¢cation step gave a single peak of activity indicating one form of active enzyme. Analysis by SDS^ PAGE and C4RP-HPLC showed a mixture that when treated with [3 H]diazoxon produced only a single labeled protein. Diazoxon inhibition of KFase activity correlated with that for [3 H]diazoxon postlabeling establishing that the labeled protein was [3 H]DEP-KFase. Puri¢cation an additional 5.5-fold by C4RP-HPLC gave s 99% pure [3 H]DEP-KFase based both on radioactivity and UV absorbance. This assignment was supported by sequence analysis and comparison of predicted characteristics to experimentally derived properties of KFase described below. 4.2. Protein and DNA sequence [3 H]DEP-KFase gave four tryptic fragment sequences that varied in length from 11 to 19 amino acids. Several of the corresponding mouse ESTs contained multiple fragments with a high degree of iden-

tity to one another over broad ranges that allowed for the extension and clean up of the overlapping regions to produce 892 bp containing, in frame, all the peptide sequences identi¢ed. A possible start codon in bp 27^29 initiated a sequence of 872 bp with no stop codon yielding a protein of 290 amino acids (MW 32 kDa). PCR^RACE experiments to extend the 5P sequence generated a fragment of similar length to the original DNA sequence used to design the gene-speci¢c primers. This indicated that the established fragment contained the majority of the 5P information and that bp 27^29 was potentially the true start codon. The approximately 1500 bp fragment from the 3P-RACE experiment generated 576 bp of new sequence, including a stop codon identi¢ed in nucleotides 942^944 and a poly(A) tail beginning at bp 1443. The unique 305 amino acid sequence coded by this cDNA is the putative full-length primary structure of KFase (MW 34 kDa) (Fig. 7). The open reading frame from 27^944 codes for a protein with theoretical characteristics that match closely the experimentally derived data for KFase. The calculated pI of 5.7 can be compared with the

BBAPRO 36608 1-5-02

210

M.K. Pabarcus, J.E. Casida / Biochimica et Biophysica Acta 1596 (2002) 201^211

value determined by chromatofocusing of 5.4; due to physical interactions in the column the apparent pI by chromatofocusing may only approximate pI values determined by other methods [34]. The theoretical MW (34 kDa) corresponds closely with that observed by gel ¢ltration (34 kDa) and SDS^ PAGE (36 kDa). Glycosylation seems to contribute little or nothing to the apparent MW of KFase, with N-terminal acetylation the only post-translational modi¢cation suggested. KFase is a monomer based on the correspondence of molecular mass by gel permeation chromatography (34 kDa) and denaturing SDS^PAGE (36 kDa), with the slight variation possibly due to the di¡erent processes of size separation. 4.3. Structural prediction and 3D model Full-length KFase shows 20^30% similarity with several hypothetical and known esterases and lipases based on a BLAST search (data not shown). While KFase is an amidase, it does not contain the amidase signature sequence found in many other enzymes hydrolyzing amide bonds [35]. Sequence pro¢le analysis of KFase instead identi¢es residues 87^177 as containing elements related to the esterase/lipase/ thioesterase family active site [25,33]. These elements are the active site serine GXSXG motif and an HGG motif which demonstrate the relationship of KFase to the hormone-sensitive lipase superfamily [33]. Analysis of the primary sequence for predicted secondary structures by PHDsec indicates KFase likely possesses 33% K-helix and 24% L-sheet. A homology model of KFase was generated to gain insight into the 3D structure and active site residues. While the manually adjusted modeling alignment displays low direct identity to KFase (14% for EST2 and 10% for BFAE) the similarity scores are higher (29% and 22%, respectively). Despite the inherently low con¢dence model obtained here, it was nevertheless useful as a predictive tool for possibly identifying potential structures and putative catalytic triad residues. The generated 3D model of KFase showed strong structural correlation with these templates as illustrated by the low RMSD values. Other protein parameters such as the reasonable distribution of polar and nonpolar residues throughout the protein also support

the goodness of the model. In addition, KFase models well as a representative of K/L-hydrolases with strands L1-L8 and helices K4, K5, K6, K7, K9 and K11 representing the core hydrolase domain. 4.4. Identi¢cation of catalytic triad The modeling alignment reveals that Ser162, Asp247 and His279 of KFase match residues of EST2 and BFAE identi¢ed as catalytic triad components [30,31]. The family of serine hydrolases to which EST2, BFAE and KFase belong utilizes the catalytic triad Ser/Asp or Glu/His and their sensitivity to OP compounds is due to phosphorylation of the active site serine. The sequence GXSXG is a conserved motif surrounding the active site serine in esterases and lipases [33,36,37]. This motif is found in KFase at Ser162 with the surrounding sequence GHSAG, which is contained in the larger tryptic moiety identi¢ed as the labeled fragment from digestion of [3 H]DEP-KFase. Ser162 is also identi¢ed in the homology model placing it at the active site with the side chain in the same position as the active serine side chain of the templates. The Asp247 and His279 residues of KFase identi¢ed in the alignment are supported by the 3D model, positioning them in the active site with similar orientations to the templates. Based on the homology model, KFase is proposed to be a serine hydrolase related to the esterase/ lipase family of enzymes possessing the K/L-hydrolase fold with the active site catalytic triad composed of Ser162, Asp247 and His279.

Acknowledgements This work was supported by Grant R01 ES08762 from the National Institute of Environmental Health Sciences (NIEHS), NIH, and its contents are solely the responsibility of the authors and do not represent the o⁄cial views of NIEHS, NIH. It was also supported by the University of California Toxic Substances Research and Teaching Program. We thank our laboratory colleagues Gary Quistad and Susan Sparks for helpful suggestions. We also thank Young M. Lee and the sta¡ of the Protein Structure Facility at the University of California at Davis for protein sequencing.

BBAPRO 36608 1-5-02

M.K. Pabarcus, J.E. Casida / Biochimica et Biophysica Acta 1596 (2002) 201^211

References [1] D. Voet, J.G. Voet, Biochemistry, 2nd Edition, Wiley, New York, 1995. [2] N.H. Proctor, A.D. Moscioni, J.E. Casida, Biochem. Pharmacol. 25 (1976) 757^762. [3] A.D. Moscioni, J.L. Engel, J.E. Casida, Biochem. Pharmacol. 26 (1997) 2251^2258. [4] J. Seifert, J.E. Casida, Biochem. Pharmacol. 27 (1978) 2611^ 2615. [5] M. Eto, J. Seifert, J.L. Engel, J.E. Casida, Toxicol. Appl. Pharmacol. 54 (1980) 20^30. [6] J. Seifert, J.E. Casida, in: D.H. Huston, T.R. Roberts (Eds.), Progress in Pesticide Biochemistry, Vol. 1, Wiley, New York, 1981, pp. 219^246. [7] J.F. Robens, Toxicol. Appl. Pharmacol. 15 (1969) 152^163. [8] J.B. Barnett, J.M. Spyker-Cranmer, D.L. Avery, A.M. Hoberman, J. Environ. Pathol. Toxicol. 4 (1980) 53^63. [9] D.B. Davies, B.J. Holub, Toxicol. Appl. Pharmacol. 54 (1980) 359^367. [10] D.B. Davies, B.J. Holub, Arch. Environ. Contam. Toxicol. 9 (1980) 637^650. [11] F.L. Earl, E. Miller, E.J. Van Loon, in: W.B. Diechmann (Ed.), Pesticides and the Environment: A Continuing Controversy, Intercontinental Medical Book Corp., New York, 1973, pp. 253^266. [12] J.M. Spyker, D.L. Avery, J. Toxicol. Environ. Health 3 (1977) 989^1002. [13] J. Seifert, T. Pewnim, Biochem. Pharmacol. 44 (1992) 2243^ 2250. [14] U. Menge, Hoppe-Seyler’s Z. Physiol. Chem. 360 (1979) 185^196. [15] R.S. Santti, V.K. Hopsu-Havu, Hoppe-Seyler’s Z. Physiol. Chem. 349 (1967) 753^766. [16] V.H. Auerbach, W.E. Knox, in: S.P. Colowick, N.O. Kaplan (Eds.), Methods in Enzymology, Vol. 3, Academic Press, New York, 1957, pp. 620^623. [17] C.E. Dalgliesh, J. Chem. Soc. (1952) 137-141. [18] N. Zhang, H. Morimoto, P.G. Williams, J.E. Casida, J. Label. Compd. Radiopharm. 43 (2000) 1275^1282. [19] M.M. Bradford, Anal. Biochem. 72 (1976) 248^254. [20] U.K. Laemmli, Nature 227 (1970) 680^685.

211

[21] A.A. Green, W.L. Hughes, in: S.P. Colowick, N.O. Kaplan (Eds.), Methods in Enzymology, Vol. 1, Academic Press, New York, 1955, pp. 67^90. [22] S.F. Altschul, T.L. Madden, A.A. Scha«¡er, J. Zhang, Z. Zhang, W. Miller, D.J. Lipman, Nucleic Acids Res. 25 (1997) 3389^3402. [23] W. Gish, BLASTP 2.0MP-WashU http://blast.wustl.edu, 1996^2000. [24] W.R. Pearson, T. Wood, Z. Zhang, W. Miller, Genomics 46 (1997) 24^36. [25] R. Apweiler, T.K. Attwood, A. Bairoch, A. Bateman, E. Birney, M. Biswas, P. Bucher, L. Cerutti, F. Corpet, M.D.R. Croning, R. Durbin, L. Falquet, W. Fleischmann, J. Gouzy, H. Hermjakob, N. Hulo, I. Jonasssen, D. Kahn, A. Kanapin, Y. Karavidopoulou, R. Lopez, B. Marx, N.J. Mulder, T.M. Oinn, M. Pagni, F. Servant, C.J.A. Sirgist, E.M. Zdobnov, Nucleic Acids Res. 29 (2001) 37^40. [26] K.B. Nicholas, H.B.J. Nicholas, Distributed by the author at www.psc.edu/biomed/genedoc, 1997. [27] J. Sambrook, E.F. Fritsch, T. Maniatis, Molecular Cloning : A Laboratory Manual, 2nd Edition, Cold Spring Harbor Laboratory Press, Plainview, NY, 1989. [28] B. Rost, C. Sander, J. Mol. Biol. 232 (1993) 584^599. [29] N. Guex, A. Diemand, M.C. Peitsch, Trends Biochem. Sci. 24 (1999) 364^367. [30] G. De Simone, S. Galdiero, G. Manco, D. Lang, M. Rossi, C. Pedone, J. Mol. Biol. 303 (2000) 761^771. [31] Y. Wei, J.A. Contreras, P. She⁄eld, T. Osterlund, U. Derewenda, R.E. Kneusel, U. Matern, C. Holm, Z.S. Derewenda, Nat. Struct. Biol. 6 (1999) 340^345. [32] D.L. Ollis, E. Cheah, M. Cygler, B. Dijkstra, F. Frolow, S.M. Franken, M. Harel, S.J. Remington, I. Silman, J. Schrag, J.L. Sussman, K.H.G. Verschueren, A. Goldman, Protein Eng. 5 (1992) 197^211. [33] H. Hemila«, T.T. Koivula, I. Palva, Biochim. Biophys. Acta 1210 (1994) 249^253. [34] L.A.Ó. Sluyterman, O. Elgersma, J. Chromatogr. 150 (1978) 17^30. [35] M.P. Patricelli, B.F. Cravatt, J. Biol. Chem. 275 (2000) 19177^19184. [36] S. Brenner, Nature 334 (1988) 528^530. [37] G. Manco, F. Febbraio, E. Adinol¢, M. Rossi, Protein Sci. 8 (1999) 1789^1796.

BBAPRO 36608 1-5-02