Crystal structure of thiamindiphosphate-dependent indolepyruvate decarboxylase from Enterobacter cloacae , an enzyme involved in the biosynthesis of the plant hormone indole-3-acetic acid Anja Schu¨tz 1 , Tatyana Sandalova 2 , Stefano Ricagno 2 , Gerhard Hu¨bner 1 , Stephan Ko¨ nig 1 and Gunter Schneider 2 1 Institute of Biochemistry, Department of Biochemistry and Biotechnology, Martin-Luther-University Halle-Wittenberg, Germany; 2 Division of Molecular Structural Biology, Department of Medical Biochemistry and Biophysics, Karolinska Institutet, Stockholm, Sweden The thiamin diphosphate-dependent enzyme indolepyruvate decarboxylase catalyses the formation of indoleacetaldehyde from indolepyruvate, one step in the indolepyruvate path- way of biosynthesis of the plant hormone indole-3-acetic acid. The crystal structure of this enzyme from Enterobacter cloacae has been determined at 2.65 A ˚ resolution and refined to a crystallographic R-factor of 20.5% (R free 23.6%). The subunit of indolepyruvate decarboxylase contains three domains of open a/b topology, which are similar in structure to that of pyruvate decarboxylase. The tetramer has pseudo 222 symmetry and can be described as a dimer of dimers. It resembles the tetramer of pyruvate decarboxylase from Zymomonas mobilis, but with a relative difference of 20° in the angle between the two dimers. Active site residues are highly conserved in indolepyruvate/pyruvate decarboxylase, suggesting that the interactions with the cofactor thiamin diphosphate and the catalytic mechanisms are very similar. The substrate binding site in indolepyruvate decarboxylase contains a large hydrophobic pocket which can accommo- date the bulky indole moiety of the substrate. In pyruvate decarboxylases this pocket is smaller in size and allows dis- crimination of larger vs. smaller substrates. In most pyruvate decarboxylases, restriction of cavity size is due to replace- ment of residues at three positions by large, hydrophobic amino acids such as tyrosine or tryptophan. Keywords: crystal structure; protein crystallography; pyru- vate decarboxylase; substrate specificity; thiamin diphos- phate. Plant hormones play central roles in the regulation of plant growth and development. The first plant hormone to be described was indole-3-acetic acid (IAA), which is synthe- sized by plants [1,2] and plant-associated bacteria [3,4]. Several pathways for the synthesis of IAA in these organisms have been described, and most of them start from L -tryptophan as precursor. One of the tryptophan- dependent biosynthetic routes to IAA is the indolepyruvic acid (IPA) pathway. This pathway starts from L -trypto- phan, and consists of three steps: (a) the conversion of tryptophan to indole-3-pyruvic acid; (b) the formation of indole-3-acetaldehyde; and (c) the production of IAA (Fig. 1). The first step of the pathway is catalysed by L -tryptophan aminotransferase, a pyridoxal-5-phosphate- dependent enzyme [5]. The intermediate, IPA, is decarboxy- lated by the action of indolepyruvate decarboxylase (IPDC) [6] and the resulting indole-3-acetaldehyde is oxidized by an aldehyde oxidase to IAA [7]. Genes encoding IPDC from several microorganisms have been cloned and characterized. These organisms include Enterobacter cloacae [8], Pantoea agglomerans [9], Klebsiella aerogenes [10], Azospirillum brasilense [11,12] and Azospirillum lipoferum [13]. The IPDC genes code for polypeptides of about 550 amino acids in length, corres- ponding to a molecular mass of  60 kDa per subunit. The enzyme from E. cloacae, which has been characterized biochemically to some extent, has a molecular mass of 240 kDa, suggesting a tetrameric structure in solution [6]. The enzyme is dependent on Mg 2+ and thiamin diphos- phate as cofactors and has a high affinity for the substrate, indolepyruvate (K M ¼ 20 l M ; 1 [13a]). The amino acid sequences of IPDC show homology to pyruvate decarb- oxylases (PDC) with, for instance, 40% identity between IPDC from E. cloacae and PDC from Klyveromyces lactis, 38% identity to PDC from Saccharomyces cerevisiae (ScPDC) and 32% identity to PDC from Zymomonas mobilis (ZmPDC) [8]. Correspondence to G. Schneider, Department of Medical Biochemistry and Biophysics, Tomtebodava ¨ gen 6, Karolinska Institutet, S-171 77 Stockholm, Sweden. Fax: +46 8327626, Tel.: +46 87287675, E-mail: gunter@alfa.mbb.ki.se Abbreviations: IPDC, indolepyruvate decarboxylase; PDC, pyruvate decarboxylase; EcIPDC, indole-pyruvate decarboxylase from Ente- robacter cloacae; ZmPDC, PDC from Zymomonas mobilis; ScPDC, PDC from Saccharomyces cerevisiae;IAA,indole-3-aceticacid; IPA, indolepyruvic acid; ThDP, thiamin diphosphate. Note: To facilitate comparison, we are using the nomenclature defined by Muller et al. (1993) [48] to identify the various domains in ThDP-dependent enzymes. (Received 19 January 2003, revised 28 March 2003, accepted 2 April 2003) Eur. J. Biochem. 270, 2312–2321 (2003) Ó FEBS 2003 doi:10.1046/j.1432-1033.2003.03601.x This study reports the three-dimensional structure of IPDC from E. cloacae, determined to 2.65 A ˚ resolution by protein crystallography. The fold of the subunit is similar to that of ScPDC [14] and ZmPDC [15]. However, the packing of the two dimers in the tetramer is different from that of the PDCs of known structure, best described as a 20° rotation of one dimer towards the other when compared to the Z. mobilis enzyme. The active site shows a substantially larger substrate binding pocket in IPDC in order to accommodate the bulky indole moiety of the substrate. Materials and methods Protein production and purification The Escherichia coli strain JM109 harbouring the plasmid pIP362 (kindly provided by J. Koga, Meiji Seika Kaisha Ltd, Japan) was used for expression. The plasmid contains theIPDCgenefromE. cloacae inserted into the high production vector pUC19. A 6-L culture of Escherichia coli strain JM109 was grown in a medium containing 2% (w/v) bactotryptone, 1% (w/v) yeast extract, 0.5% (w/v) sodium chloride, 0.1 m M thiamine, 0.1 m M magnesium sulphate, 0.01% (w/v) ampicillin, and 0.15 M potassium phosphate pH 6.5 for 24 h at 30 °C. Expression of the IPDC gene was induced by addition of 1 m M isopropyl thio-b- D -galacto- side. Cells were harvested by centrifugation, quickly frozen in liquid nitrogen and stored at )80 °C. About 25 g of cells were suspended in 40 mL 0.1 M potassium phosphate pH 6.5, containing 10 m M thiamin diphosphate (ThDP), 10 m M magnesium sulphate, 1 m M EDTA, 5 m M dithio- threitol 2 , and disrupted in a French Press at 1200 bar (Gaulin, APV Homogeniser GmbH, Lu ¨ beck, Germany). The mixture was centrifuged at 70 000 g for 10 min and the pellet was discarded. Nucleic acids were precipitated by incubation with 0.1% (w/v) streptomycin sulphate for 45 min at 8 °C. A 15–30% (w/v) ammonium sulphate fractionation was performed at a protein concentration of 20 mgÆmL )1 . After centrifugation at 30 000 g for 5 min, the precipitate was dissolved in 10 mL 50 m M Mes/NaOH pH 6.5, containing 10 m M magnesium sulphate, 0.15 M ammonium sulphate and 1 m M dithiothreitol. The solution was applied to a Sephacryl S200 H column (5 · 95 cm; Amersham Biosciences) and eluted with the same buffer at 1mLÆmin )1 . The IPDC-containing fractions were pooled and concentrated by precipitation with ammonium sulphate (0.5 mgÆmL )1 ). After centrifugation the precipitate was dissolved in 20 m M Mes/NaOH pH 6.5, 1 m M dithiothrei- tol and this solution was desalted on a Hiprep column (2.6 · 10 cm; Amersham Biosciences) and applied to a Source 15Q column (2.6 · 7 cm; Amersham Biosciences). Elution was performed using a linear gradient of 120 mL 0–25% of 20 m M Mes/NaOH pH 6.5, 1 m M dithiothreitol, 0.25 M ammonium sulphate. The fractions with the highest catalytic activity and homogeneity were pooled, and after addition of 0.2 M ammonium sulphate quickly frozen in liquidnitrogen,andstoredat)80 °C. Crystallization The purified enzyme was concentrated to  4mgÆmL )1 . Simultaneously, the buffer was changed to 20 m M Mes/ NaOH pH 6.5 and 1 m M dithiothreitol. IPDC was crystal- lized by the hanging drop vapour diffusion method. Crystals were grown at 20 °C using poly(ethylene glycol) 2000 monomethylether as precipitating agent. Drops contained equal volumes (2 lL) of reservoir solution [0.1 M sodium citrate pH 5.0, 8–12% (w/v) poly(ethylene glycol)] and IPDC (4 mgÆmL )1 in 20 m M Mes/NaOH pH 6.5, 1 m M dithiothreitol, 5 m M ThDP, 5 m M magnesium sulphate). Before setting the drops, IPDC was incubated with the cofactors at room temperature for 30 min. The best crystals were obtained at 9–10% (w/v) poly(ethylene glycol). Within 3–4 days bundles of needles appeared. Streak seeding was then used to improve the crystal size. After transfer of seeds to fresh drops, single crystals appeared within 1 day and grew to a maximum size of 0.6 · 0.4 · 0.2 mm in 3 days. Data collection X-ray data were collected at cryo-conditions with a ADSC Quantum-4 CCD detector on beam line ID29 (ESRF, Grenoble, France). The crystals were soaked in crystallization buffer supplemented with 20% glycerol Fig. 1. The indole-3-pyruvic acid pathway for the biosynthesis of the plant hormone indole-3-acetic acid in Entero bacter cloa cae. 1, L -tryptophan aminotransferase, 2, indolepyruvate decarboxylase, 3, indoleacetaldehyde oxidase. Ó FEBS 2003 3D structure of indolepyruvate decarboxylase (Eur. J. Biochem. 270) 2313 before flash freezing directly in the nitrogen stream. The diffraction data was collected at wavelength 0.979 A ˚ and processed with MOSFLM 3 [16]. The CCP4 suite of programs [17] was used for scaling and reduction of the data. The space group and cell dimensions were determined using the auto-indexing option of MOSFLM and by the analysis of pseudo-precession images [18]. Structure determination The structure was solved by molecular replacement using the program package AMORE [19]. The self-rotation function and the estimated solvent content [20] indicate that the asym- metric unit contains four subunits, arranged as a tetramer. The structure of a dimer of ZmPDC was used as a search model, as calculation of low resolution models of E. cloacae PDC (EcPDC) from small angle X-ray solution scattering data [21] had indicated that the quaternary structure of IPDC is more similar to that of ZmPDC than ScPDC. A poly serine model of ZmPDC without cofactors and solvent atoms was used as search model. The best solution had a correlation coefficient of 0.235 after rigid body refinement. This solution was fixed, and the search for the second dimer gave a solution with a correlation coefficient of 0.32 and an R-factor of 50.1% after rigid body refinement. Model building and crystallographic refinement Refinement of the model was performed with CNS 4 [22]. To monitor progress 5% of each data set was set aside for calculation of R free [23]. Initial improvement of the model was achieved by rigid body refinement, first with the dimers, and subsequently with the subunits as independent rigid bodies. As the asymmetric unit contains one tetramer, tight noncrystallographic symmetry restraints (Wa ¼ 300 kcalÆmol )1 ÆA ˚ )2 ) 5 were imposed on the crystal- lographically independent monomers throughout the refinement procedure. Bulk solvent correction was used with default CNS parameters. Manual rebuilding of the model was performed using the program O [24] based on sigma-weighted 2F o ) F c and F o ) F c electron density maps. The parts of the polypeptide chain which differ most from the search model due to insertions/deletions in the amino acid sequence were modelled based on omit electron density maps [22]. The coenzyme ThDP was excluded from the search model and the correctness of the solution was confirmed by electron density for ThDP and Mg 2+ appearing at the expected positions (Fig. 2). The model was further refined by simulated annealing and isotropic B-factor refinement. Water molecules were modelled using the automatic water picking option in CNS. All water molecules were checked for hydrogen bonds with protein atoms. The final R-values and other refinement statistics are given in Table 1. The X-ray data and the atomic coordinates have been deposited at the Protein Data Bank, accession number 1ovm. Structure analysis Structure comparisons were carried out using the pro- grams TOP [25] and O [24] using default parameters. Sequence alignments were performed with MULTALIN Fig. 2. Stereoview of the ThDP binding site in IPDC. The initial, unrefined 2Fo-Fc map, showing the electron density for the bound magnesium ion and ThDP is contoured at 1 r. The refined protein model is superposed. The magnesium ion is shown by a green sphere and red spheres represent bound solvent molecules. Table 1. Data collection and refinement statistics. Space group P2 1 2 1 2 Cell dimensions (A ˚ ) 132.2, 151.6, 107.6 Resolution (A ˚ ) 2.65 Completeness (%) 99.9 (99.9) a Total number of reflections 315 465 Unique reflections 63 426 I/r 7.8 (2.0) R sym (%) 8.7 (36.9) R (%) 20.5 R free (%) 23.6 Number of protein atoms 16404 Number of solvent molecules 347 Root mean square bond lengths 0.007 Root mean square bond angles 1.388 B-factors (A ˚ 2 ) Overall 32.2 ThDP 26.0 Solvent 23.4 Ramachandran plot Percentage of nonglycine residues in: Favourable regions 87.9 Additionally allowed regions 12.1 a Numbers in parentheses are for the highest resolution shell. 2314 A. Schu ¨ tz et al. (Eur. J. Biochem. 270) Ó FEBS 2003 (http://prodes.toulouse.inra.fr/multalin) 6 . The analysis of protein interfaces was done using the Protein–Protein interaction server (http://www.biochem.ucl.ac.uk/bsm/PP/ server/). Figures were created with MOLSCRIPT [26], BOBSCRIPT [27] and RASTER 3 D [28]. Results Purification of IPDC The procedure comprises four steps: streptomycin sulphate treatment, ammonium sulphate precipitation, gel filtration, and anion exchange chromatography. The resulting enzyme is the homogenous apo enzyme, free of cofactors. A molecular mass of 60 000 Da per subunit resulted from SDS/PAGE, which corresponded to the value calculated from the nucleotide sequence of the structural gene. The identity of the purified enzyme was confirmed by N-terminal amino acid sequence analysis (Met-Arg-Thr-Pro-Tyr-Cys- Val-Ala). Structure determination The crystals of holo-IPDC belong to the space group P2 1 2 1 2 with unit cell dimensions a ¼ 132.2 A ˚ ,b¼ 151.6 A ˚ ,c¼ 107.6 A ˚ and one tetramer in the asymmetric unit, corres- ponding to a solvent content of 45%. The structure of IPDC was solved by molecular replacement using a dimer of ZmPDC as an initial search model and refined to final R free / R-values of 23.6%/20.5%. The stereochemistry of the model is as expected for this resolution (Table 1). In general, the electron density for the polypeptide chain is well defined. However, there is no continuous electron density for the long loop connecting the middle and the C-terminal domains (residues 342–355) and these residues were not included in the model (Fig. 4, 7 lowercase). The analysis of the model during refinement showed that almost all residues obey noncrystallographic symmetry, except the C termini and the side chains of residues His227, Asp278, Arg367, Ile379, and Arg394. After superposition of the subunits the rmsd between all corresponding C a atomsis0.13A ˚ for two monomers in the dimer, and 0.17 A ˚ for two dimers. The final model includes residues 3–341, and 356–551 of the protein, four magnesium ions, four molecules of ThDP and citrate, and 347 water molecules. The crystallographic refinement statistics are presented in Table 1. Overall structure of IPDC IPDC is a homo-tetramer with overall dimensions of 92 · 94 · 116 A ˚ . Each monomer consists of three domains with an open a/b class topology: the N-terminal PYR 1 domain (residues 3–180), which binds the pyrimidine part of ThDP; the middle domain (residues 181–340); and the C-terminal PP domain (residues 356–551), which binds the diphosphate moiety of the cofactor (Fig. 3). The PYR and PP domains contain a six-stranded parallel b-sheet flanked by a number of helices, whereas the middle domain contains a six-stranded mixed b-sheet (four strands are parallel, two antiparallel), with several helices packing against the sheet. The secondary structure elements of IPDC are shown in Fig. 4, together with the aligned amino acid sequences of IPDC and ZmPDC. The topology of the IPDC monomer is similar to that of ScPDC and ZmPDC with some variations in the length and orientation of helices. The superposition of the IPDC monomer on the subunit of ScPDC and ZmPDC results in rmsd of 1.24 A ˚ for 470 out of 563 Ca atoms and 1.48 A ˚ for 496 out of 568 Ca atoms, respectively. All insertion/deletions are short, they occur in the loop regions and do not effect the overall structure. The loop connecting the middle and PP domain is five residues longer in ZmPDC and most residues of this loop are invisible in the structure of IPDC. None of the insertions/deletions occur near the active site, however, some of them are at the dimerization/ tetramerization interfaces (Fig. 4). Two monomers interact tightly to form the dimer. The accessible surface area buried in the monomor–monomer interface is 3590 A ˚ 2 (17% of the whole accessible surface area). The interface is mostly nonpolar (65% of residues), but it also contains 26 hydrogen bonds and two salt bridges. All three domains of the monomer participate in the dimer interactions (Fig. 4, residues marked ÔdÕ), with most residues at the interface coming from the PYR and PP domains. This is in agreement with the average mobility of the domains in the crystal; the PYR domain has the lowest average B-factor, 23 A ˚ 2 (comparable to the B-factor of bound ThDP), whereas the middle domain has the highest overall B-factor, 37 A ˚ 2 . One-hundred and two residues make up the mono- mer–monomer interface; 57 of these residues are conserved between IPDC and ZmPDC, and 15 residues are invariant in all IPDC/PDC sequences (Fig. 4). The IPDC dimer interface is with 3414 A ˚ 2 comparable to that of pyruvamide-activated ScPDC [29]. In ZmPDC, the interaction area is larger (4387 A ˚ 2 ),whereasitissmallerin Fig. 3. Fold of the subunit of IPDC from Enterobacter cloacae. The PYR domain is shown in blue, the middle domain in green and the PP domain in red. The secondary structure elements are labelled as defined in Fig. 4. The cofactor ThDP and the magnesium ion are included as ball-and-stick models. The broken line indicates the disordered loop comprising residues 342–355. Ó FEBS 2003 3D structure of indolepyruvate decarboxylase (Eur. J. Biochem. 270) 2315 nonactivated ScPDC (2892 A ˚ 2 ) [14]. The number of hydrogen bonds is also far fewer than in ZmPDC (26 vs. 66). In part, this is due to the shorter C-terminal region in IPDC, because the last five residues of ZmPDC are responsible for a dimer interface area of 400 A ˚ 2 . Another difference of about 400 A ˚ 2 in the dimer interface can be accounted for by a deletion of five residues in the IPDC amino acid sequence after helix a21 (Fig. 4). In ZmPDC, residues 496–504, which are inserted at this position, participate in the monomer–monomer interface. In addition to these two deletions in the IPDC sequence, there are several amino acid substitutions resulting in a reduced number of hydrogen bonds in the IPDC dimer interface, for instance Ser74 fi Gly, Asn102 fi Gly, Asn104 fi Ala, Fig. 4. Structural alignment of EcIPDC and ZmPDC sequences. Sequences were denoted as IPDC if biochemical and/or genetic data support such an activity of the enzyme. Residues conserved in six known/putative IPDCs (Enterobacter cloacae, Pseudomonas putida, Pantoea agglomerans, Azospirillum brasiliense, Azospirillum lipoferum and Klebsiella aerogenes) are shown in red in the EcIPDC sequence (DCIP_ENTCL). Residues conserved in 12 PDC sequences (DCP1_MAIZE, DCP1_ORYSA, DCP1_PEA, DCP2_TOBAC, DCP2_ORYSA, DCPY_ZYMMO, DCP1_YEAST, DCP2_YEAST, DCP3_YEAST, DCPY_KLULA, DCPY_KLUMA, DCPY_HANUV) are also shown in red in the ZmPDC sequence (DCPY_ZYMMO). Conservative amino acid replacements are shown in blue. Residues common to ZmPDC and IPDC are shown in bold. a-Helices are displayed as rectangles, b-strands as arrows. ÔdÕ indicates residues in the dimer interface and ÔtÕ residues in tetramer interface. Residues lining the active site cavity are underlined. Residues binding ThDP are highlighted with a blue background, and those involved in substrate binding by yellow. Amino acids of EcIPDC invisible in the electron density map are shown in lowercase. 2316 A. Schu ¨ tz et al. (Eur. J. Biochem. 270) Ó FEBS 2003 Gln411 fi Leu, Lys485 fi Ala, Asn486 fi Leu (first resi- due is that of ZmPDC) (Fig. 4). Two dimers form a tetramer (Fig. 5), as seen in many other ThDP-dependent enzymes. However, the dimer– dimer interface is smaller than the monomer–monomer interface within the dimer. Only 2030 A ˚ 2 (9.5%) of the dimer accessible surface area is buried in IPDC upon tetramer formation. That corresponds to 44 interacting residues, which are marked by ÔtÕ in Fig. 4. Ten of them are conserved between IPDC and ZmPDC, but none are invariant in the whole IPDC/PDC family. The majority of residues contributing to these interfaces is located in the PYR and middle domains. The interface contains 10 hydrogen bonds in IPDC. The dimer–dimer interface in IPDC is smaller than the corresponding interface in ZmPDC (4400 A ˚ 2 ). It is significantly larger than in non- activated ScPDC (1344 A ˚ 2 ), and comparable to pyruv- amide-activated ScPDC (1920 A ˚ 2 )[15]. The tetramer of IPDC differs significantly from other tetrameric ThDP in the packing of the dimers within the tetramer. The pseudo 222 symmetry is preserved, and the molecule can be best described as a Ôdimer of dimersÕ. The closest relative is ZmPDC, where the second dimer is rotated by about 20° when tetramers of IPDC and ZmPDC are compared (Fig. 5). It is noteworthy that the relative orientation of the dimers in the tetramer is different in all of the tetrameric PDCs of known three-dimensional structure. Binding of the cofactors ThDP and Mg 2+ The homo-tetrameric IPDC binds four molecules of the cofactors ThDP and Mg 2+ . The ThDP binding sites are located in narrow clefts at the interfaces formed by the PYR domains from one subunit and the PP domains of the other subunit within the dimer. ThDP adopts the V-conformation [30,31] and is completely buried in the cofactor binding cleft. Several hydrogen bonds that are responsible for binding and proper orientation of the aminopyrimidine ring, are conserved in all ThDP-dependent enzymes. One of these, the hydrogen bond between the N1¢ atom of the pyrimidine ring of ThDP and the side chain of an invariant glutamate residue of the neighbouring subunit (Glu52), is essential for catalysis [32–35]. The C2 carbon atom of the thiazolium ring points into the active site cavity and is accessible for external ligands. The diphosphate moiety of ThDP is bound exclusively to the PP domain of the subunit through Fig. 5. Quaternary structure of IPDC. Upper panel: stereo view of the quaternary structure of IPDC from Enterobacter cloacae. The four subunits of the tetramer are shown in different colours. The cofactor molecules are included as ball-and-stick models. Lower panel: different packing of the dimers in the tetramer of EcIPDC and ZmPDC. After superposition of one dimer in the tetramer of EcIPDC and ZmPDC (green), the difference in the orientation of the second dimer (IDPC, blue, ZmPDC, red) in the two enzymes is clearly evident. 10 Ó FEBS 2003 3D structure of indolepyruvate decarboxylase (Eur. J. Biochem. 270) 2317 hydrogen bonds and a bridging magnesium ion. The magnesium ion is octahedrally coordinated to oxygen atoms from the diphosphate group of ThDP, the side chains of Asp435 and Asn462, the main chain oxygen atom of Gly464, and a water molecule. All these interactions are highly conserved among ThDP-dependent enzymes. Substrate binding site and catalytic residues The active site cavity in IPDC extends from the thiazolium ring of the cofactor to the surface of the protein. The entrance of the active site cleft is covered by the C-terminal helix and this part of the polypeptide chain must move in order to allow entry of the substrate. This structural feature was also found in ZmPDC [15], and it could be shown that the kinetic properties of ZmPDC variants, truncated at the C-terminal helix, are consistent with a role of this helix in closure of the active site [36]. A model of the a-carbanion/enamine intermediate of the substrate indole-3-pyruvate with ThDP in the active site of IPDC was built based on the three-dimensional structure of the corresponding intermediate in transketolase [37] and the model derived for ScPDC [38] (Fig. 6). In the immediate vicinity of the modelled a-carbanion/enamine, there are a number of invariant amino acids, Asp29, His115, His116, and Glu468, which are conserved in all PDCs. Site-directed mutagenesis has confirmed the essentiality of these residues for catalysis in ScPDC [39,40] and ZmPDC [41–43], 8 respectively. These studies, together with structural data from crystallography [14,15,29] and modelling [38] have provided considerable insights into the role of these residues in PDC. As all amino acids, which were suggested to participate in catalytic steps of PDC are conserved in IPDC, the enzymatic mechanism seems to be very similar, if not identical, in the two enzymes. A significant difference in the active site between PDC and IPDC appears to be Gln383, which is replaced by Thr in most PDCs (Fig. 4). In the structure of holo-IPDC the side chain of Gln383 points away from the active site cavity and cannot interact with bound substrate and/or reaction intermediates. However, only side chain movement would be sufficient to allow interactions of this residue with bound substrate, suggesting that Gln383 might be involved in substrate binding and, possibly, specific recognition of indole-3-pyruvate. Substrate recognition The indole moiety of the modelled intermediate is bound in a large hydrophobic pocket, lined by residues from three helices, Ala387, Phe388 (helix a16), Val467, Ile471 (helix a20), Leu538, Leu542, and Leu546 (a23), and is completely buried in the protein. Three of these residues (Phe388, Val467 and Ile471) are either invariant or have conservative substitutions in all PDC/IPDC sequences (Fig. 4). The assignment of this hydrophobic pocket as part of the substrate binding site is further supported by mutational studies of ZmPDC, because residue substitutions at the positions corresponding to 467 and 471 in IPDC influence substrate binding and specificity [43]. The volume of the active site cavity is larger in IPDC (130 A ˚ 3 )thaninZmPDC(85A ˚ 3 ), where it is partially filled with bulky amino acids, Tyr290, Trp387, and Trp542 (IPDC sequence numbering). These large aromatic side chains effectively restrict the size of the pocket and prevent binding of larger substrates (Fig. 6). The structural model is thus consistent with the finding that ZmPDC does not recognize indolepyruvate as a substrate [13a] 9 .Aminoacid sequence comparisons of residues lining this substrate recognition pocket reveal identical residues at these posi- tions also in all plant PDCs. A change in substrate specificity from pyruvate to indolepyruvate thus involves at least substitution of three residues in the substrate binding pocket. In all IPDC sequences, residue 290 is replaced by threonine, position 387 by alanine or leucine, and position 542 by residues which are smaller than tryptophan, resulting in a larger cavity size. Restriction of the cavity size thus seems to be a major cause of discrimination against large substrates in PDCs. Yeast PDCs do not follow this substitution pattern as the basis of discrimination towards large aromatic substrates. Consequently, ScPDC is, in contrast with ZmPDC, able to decarboxylate indole-pyruvate (Schu ¨ tz et al. unpublished data). While yeast PDCs show a similar substitution at position 290 (Thr fi Phe) as ZmPDC, there are no replacements at position 387 by amino acids with a large hydrophobic side chain. Furthermore, there are significant structural differences between ZmPDC and IPDC on the one hand, and ScPDC on the other hand involving the C-terminal part of the polypeptide chain (Fig. 7). In the latter, differences in the conformation of the loop between Fig. 6. Stereo picture of the model of the a-carbanion/enamine intermediate (light grey) in the active site of EcIPDC. The three resi- dues, which restrict the size of the substrate binding cavity in ZmPDC (Tyr290, Trp387 and Trp542), are shown in red. 2318 A. Schu ¨ tz et al. (Eur. J. Biochem. 270) Ó FEBS 2003 strands b11 and b12 prevent the C-terminal helix from approaching the other subunit in the dimer sufficiently to shield the active site, as it does in ZmPDC and IPDC. There is therefore no residue in ScPDC which is structurally equivalent to 542 in ZmPDC and IPDC. These differences result in a larger volume of the active site cavity in ScPDC, allowing accommodation of larger substrates such as indole-3-pyruvate. Substrate activation IPDC follows Michaelis–Menten kinetics (Schu ¨ tz et al. unpublished data). In this regard, the enzyme is similar to ZmPDC that in contrast with all other PDCs investigated so far is not subject to substrate activation [44]. Several models to account for substrate activation in ScPDC have been proposed [45–47], involving Cys221 as the site where the substrate activation cascade is triggered. More recently, an additional pathway for signal transduction between active sites in ScPDC has been suggested, based on a detailed kinetic study [40]. An alternative model is based on the structure of ScPDC with bound activator pyruvamide, which revealed a disorder–order transition of two active site loops (residues 104–113 and 290–304), and which appears to be a key event in the activation process [29]. These conformational transitions are accompanied by large-scalechangesintherelativeorientationofthedimers in the tetramer. In the three-dimensional structure of ZmPDC,theactivesiteloopsarewellorderedand observed in a conformation suitable for catalysis to occur [15]. The much tighter packing of the subunits in the ZmPDC tetramer, leading to more extensive interactions in the dimer–dimer interface compared to ScPDC most likely excludes such large-scale conformational changes during catalysis, and these structural features explain the lack of substrateactivationinZmPDC.InIPDC,theassemblyof the subunits in the tetramer resembles that of ZmPDC rather than ScPDC. As in ZmPDC, the active site loops are folded in a conformation poised for catalysis even in the absence of substrate or other activators. The structure of IPDC supports the conclusion that the substrate activation observed in most PDC species may be linked to the packing of the subunits in the tetramer. Enzyme species with a rather loose packing such as ScPDC maintain the possibility of conformational changes during catalysis, and thus allow for cooperativity, whereas in ZmPDC and IPDC with tighter and more extensive dimer– dimer interfaces large-scale conformational changes would be energetically too costly and are not used for control of enzyme activity. Acknowledgements We thank J. Koga for providing a plasmid producing Enterobacter cloacae indolepyruvate decarboxylase, and K P. Ru ¨ cknagel (Max- Planck-Society, Research Unit ÔEnzymology of protein foldingÕ,Halle/ Saale, Germany) for the amino acid sequence analysis. We acknow- ledge access to synchrotron radiation at beamline ID29, ESRF, Grenoble. A.S. acknowledges travel support by the Deutscher Akademischer Austauschdienst (DAAD). This work was supported by the Science Research Council, Sweden and the Graduiertenkolleg of Sachsen-Anhalt. References 1. Bartel, B. (1997) Auxin biosynthesis. Ann. Rev. Plant Physiol. Plant Mol. Biol. 48, 51–66. 2. Normanly, J. & Bartel, B. (1999) Redundancy as a way of life – IAA metabolism. Curr. Opin. Plant Biol. 2, 207–213. 3. Kawaguchi, M. & Syono, K. (1996) The excessive production of indole-3-acetic acid and its significance in studies of the bio- synthesis of this regulator of plant growth and development. Plant Cell Physiol. 37, 1043–1048. 4. Patten, C.L. & Glick, B.R. (1996) Bacterial biosynthesis of indole- 3-acetic acid. Can. J. Microbiol. 42, 207–220. 5. Koga, J., Syono, K., Ichikawa, T. & Adachi, T. (1994) Involve- ment of 1-tryptophan aminotransferase in indole-3-acetic acid biosynthesis of Enterobacter cloacae. Biochim. Biophsy. Acta 1209, 241–247. 6. Koga, J., Adachi, T. & Hidaka, H. (1992) Purification and char- acterization of indolepyruvate decarboxylase. J. Biol. Chem. 267, 15823–15828. 7. Sekimoto, H., Seo, M., Kawakami, N., Komano, T., Desloire, S., Liotenberg, S., Marion-Poll, A., Caboche, M., Kamiya, Y. & Koshiba, T. (1998) Molecular cloning and characterization of aldehyde oxidases in Arabidopsis thaliana. Plant Cell Physiol. 39, 433–442. 8. Koga, J., Adachi, T. & Hidaka, H. (1991) Molecular cloning of the gene for indolepyruvate decarboxylase from Enterobacter cloacae. Mol. Gen. Genet. 226, 10–16. 9. Brandl, M. & Lindow, S.E. (1996) Cloning and characterization of a locus encoding anindolepyruvate decarboxylase involved in Fig. 7. Comparison of the substrate binding pockets in IPDC (red) and activated ScPDC (green). In the latter, the loop between strands b11 and b12 (shown in blue) prevents the C-terminal helix a23 from approaching the active site, and thus leads to a more open substrate binding pocket in ScPDC. A model of the a-carbanion/enamine intermediate is included. The blue sphere indicates the posi- tion of the Mg 2+ ion. Ó FEBS 2003 3D structure of indolepyruvate decarboxylase (Eur. J. Biochem. 270) 2319 indole-3-acetic acid synthesis in Erwiniaherbicola. Appl. Environ. Microbiol. 62, 4121–4128. 10. Zimmer, W., Hundeshagen, B. & Niederau, E. (1994) Demon- stration of the indolepyruvate decarboxylase gene in different auxin-producing species of the Enterobacteriaceae. Can. J. Microbiol. 40, 1072–1076. 11. Costacurta, A., Keijers, V. & Vanderleyden, J. (1994) Molecular cloning and sequence analysis of an Azospirillum brasilense indole- 3-pyruvate decarboxylase gene. Mol. Gen. Genet. 243, 463–472. 12.Carreno-Lopez,R.,Campos-Reales,N.,Elmerich,C.&Baca, B.E. (2000) Physiological evidence for differently regulated tryp- tophan-dependent pathways for indole-3-acetic acid synthesis in Azospirillum brasilense. Mol. Gen. Genet. 264, 521–530. 13. Yagi, K., Chujo, T., Nojiri, H., Omori, T., Nishiyama, M. & Yamane, H. (2001) Evidence for the presence of DNA-binding proteins involved in regulation of the gene expression of indole-3- pyruvic acid decarboxylase, a key enzyme in indole-3-acetic acid biosynthesis in Azospirillum lipoferum FS. Biosci. Biotechnol. Biochem. 65, 1265–1269. 13a. Schu ¨ tz, A., Golbik, R., Tittmann, K., Svergun, D.I., Koch, M.H.J., Hu ¨ bner,G.&Ko ¨ nig, S. (2003) Studies on structure– function relationships of indolepyruvate decarboxylase from Enterobacter cloacae, a key enzyme of the indole acetic acid path- way. Eur. J. Biochem. 270, 2322–2331. 14. Arjunan, P., Umland, T., Dyda, F., Swaminathan, S., Furey, W., Sax,M.,Farrenkopf,B.,Gao,Y.,Zhang,D.&Jordan,F.(1996) Crystal structure of the thiamine diphosphate-dependent enzyme pyruvate decarboxylase from the yeast Saccharomyces cerevisiae at 2.3 A ˚ resolution. J. Mol. Biol. 256, 590–600. 15. Dobritzsch, D., Ko ¨ nig, S., Schneider, G. & Lu, G. (1998) High resolution crystal structure of pyruvate decarboxylase from Zymomonas mobilis. J. Biol. Chem. 273, 20196–20204. 16. Leslie, A.G.W. (1992) in Joint CCP4 and ESF-EACMB News- letter on Protein Crystallography, no. 26. Daresbury Laboratory, Warrington, UK. 17. Collaborative Computational Project, Number, 4. (1994) The CCP.4 Suite: Programs for Protein Crystallography. Acta Crys- tallogr. D50, 760–763. 18. Lu, G. (1999) PATTERN: a precession simulation programme for displaying reciprocal-space diffraction data. J. Appl. Crystallogr. 32, 375–376. 19. Navaza, J. (1994) AMoRe: an automated package for molecular replacement. Acta Crystallogr. A50, 157–163. 20. Matthews, B.W. (1968) Solvent content of protein crystals. J. Mol. Biol. 33, 491–497. 21. Ko ¨ nig, S., Schu ¨ tz, A., Svergun, D.I. & Koch, M.H.J. (2000) First SAXS measurements on indolepyruvate decarboxylase. Hasylab Annual Report 295–296. 22. Bru ¨ nger, A.T., Adams, P.D., Clore, G.M., DeLano, W.L., Gros, P., Grosse-Kunstleve, R.W., Jiang, J S., Kuszewski, J., Nilges, M., Pannu, N.S., Read, R.J., Rice, L.M., Simonson, T. & Warren, G.L. (1998) Crystallography & NMR system: a new software suite for macromolecular structure determination. Acta Crystallogr. D54, 905–921. 23. Brunger, A.T. (1993) Assessment of phase accuracy by cross vali- dation: the free R value. Methods and applications. Acta Crys- tallogr. D49, 24–36. 24. Jones, T.A., Zou, J.Y., Cowan, S.W. & Kjeldgaard, M. (1991) Improved methods for building protein models in electron density maps and the location of errors in these models. Acta Crystallogr. A47, 110–119. 25. Lu, G. (2000) TOP: a new method for protein structure compari- sons and similarity searches. J. Appl. Crystallogr. 33, 176–183. 26. Kraulis, P. (1991) Molscript: a program to produce both detailed and schematic plots of protein structures. J. Appl. Crystallogr. 26, 282–291. 27. Esnouf, R.M. (1997) An extensively modified version of MolScript that includes generally enhanced colouring capabilities. J. Mol. Graph. Model 15 (132–134), 112–113. 28. Merritt, E.A. & Bacon, D.J. (1997) Raster3D: Photorealistic Molecular Graphics. Methods Enzymol. 277, 505–524. 29. Lu, G., Dobritzsch, D., Baumann, S., Schneider, G. & Ko ¨ nig, S. (2000) The structural basis of substrate activation in yeast pyru- vate decarboxylase: a crystallographic and kinetic study. Eur. J. Biochem. 267, 861–868. 30. Schellenberger, A. (1967) Structure and mechanism of action of the active center of yeast pyruvate decarboxylase. Angew. Chem. 6, 1024–1035. 31. Shin, W., Pletcher, J., Blank, G. & Sax, M. (1977) Ring stacking interactions between thiamin and planar molecules as seen in the crystal structure of a thiamin picrolonatedihydrate complex. J. Am. Chem. Soc. 99, 3491–3499. 32. Wikner, C., Meshalkina, L., Nilsson, U., Nikkola, M., Lindqvist, Y., Sundstro ¨ m, M. & Schneider, G. (1994) Analysis of an invariant cofactor–protein interaction in thiamin diphosphate- dependent enzymes by site-directed mutagenesis. Glutamic acid 418 in transketolase is essential for catalysis. J. Biol. Chem. 269, 32144–32150. 33. Candy, J.M., Koga, J., Nixon, P.F. & Duggleby, R.G. (1996) The role of residues glutamate-50 and phenylalanine-496 in Zymomonas mobilis pyruvate decarboxylase. Biochem. J. 315, 745–751. 34. Killenberg-Jabs, M., Ko ¨ nig, S., Eberhardt, I., Hohmann, S. & Hu ¨ bner, G. (1997) Role of Glu51 for cofactor binding and cata- lytic activity in pyruvate decarboxylase from yeast studied by site- directed mutagenesis. Biochemistry 36, 1900–1905. 35. Kern, D., Kern, G., Neef, H., Tittmann, K., Killenberg-Jabs, M., Schneider, G. & Hu ¨ bner, G. (1997) How thiamine diphosphate is activated in enzymes. Science 275, 67–70. 36. Chang, A.K., Nixon, P.F. & Duggleby, R.G. (2000) Effects of deletions at the carboxyl terminus of Zymomonas mobilis pyruvate decarboxylase on the kinetic properties and substrate specificity. Biochemistry 39, 9430–9437. 37. Fiedler, E., Thorell, S., Sandalova, T., Golbik, R., Ko ¨ nig, S. & Schneider, G. (2002) Snapshot of a key intermediate in enzymatic thiamin catalysis: Crystal structure of the a-carbanion of (a,b-dihydroxyethyl) -thiamin diphosphate in the active site of transketolase from Saccharomyces cerevisiae. Proc.NatlAcad.Sci. USA 99, 591–595. 38. Lobell, M. & Crout, D.H.G. (1996) Pyruvate decarboxylase: a molecular modeling study of pyruvate decarboxylation and acy- loin formation. J. Am. Chem. Soc. 118, 1867–1873. 39.Liu,M.,Sergienko,E.A.,Guo,F.,Wang,J.,Tittmann,K., Hu ¨ bner, G., Furey, W. & Jordan, F. (2001) Catalytic acid-base groups in yeast pyruvate decarboxylase. 1. Site-directed muta- genesis and steady-state kinetic studies on the enzyme with the D28A, H114F, H115F, and E477Q substitutions. Biochemistry 40, 7355–7368. 40. Sergienko, E.A. & Jordan, F. (2001) Catalytic acid-base groups in yeast pyruvate decarboxylase.3. a steady state kinetic model con- sistent with the behavior of both wild-type and variant enzymes at all relevant pH values. Biochemistry 40, 7382–7403. 41. Schenk, G., Leeper, F.J., England, R., Nixon, P.F. & Duggleby, R. (1997) The role of His113 and His114 in pyruvate decarboxy- lase from Zymomonas mobilis. Eur. J. Biochem. 248, 63–71. 42. Chang, A.K., Nixon, P.F. & Duggleby, R.G. (1999) Aspartate-27 and glutamate-473 are involved in catalysis of Zymomonas mobilis pyruvate decarboxylase. Biochem. J. 339, 255–260. 43. Pohl,M.,Siegert,P.,Mesch,K.,Bruhn,H.&Grotzinger,J.(1998) Active site mutants of pyruvate decarboxylase from Zymomonas mobilis—a site-directed mutagenesis study of L112, I472, I476, E473, and N482. Eur. J. Biochem. 257, 538–546. 2320 A. Schu ¨ tz et al. (Eur. J. Biochem. 270) Ó FEBS 2003 44. Bringer-Meyer, S., Schimz, K.L. & Sahm, H. (1986) Pyruvate decarboxylase from Zymomonas mobilis. Isolation and partial characterization. Arch. Microbiol. 146, 105–110. 45. Hu ¨ bner, G., Ko ¨ nig, S. & Schellenberger, A. (1988) The functional role of thiol groups of pyruvate decarboxylase from brewer’s yeast. Biomed Biochim. Acta 47, 9–18. 46. Baburina, I., Gao, Y., Hu, Z., Jordan, F., Hohmann, S. & Furey, W. (1994) Substrate activation of brewers’ yeast pyruvate decarboxylase is abolished by mutation of cysteine 221 to serine. Biochemistry 33, 5630–5635. 47. Baburina,I.,Li,H.,Bennion,B.,Furey,W.&Jordan,F.(1998) Interdomain information transfer during substrate activation of yeast pyruvate decarboxylase: the interaction between cysteine 221 and histidine 92. Biochemistry 37, 1235–1244. 48. Muller, Y.A., Lindqvist, Y., Furey, W., Schulz, G.E., Jordan, F. & Schneider, G. (1993) A thiamin diphosphate binding fold revealed by comparision of the crystal structures of transketolase, pyruvate oxidase and pyruvate decarboxylase. Structure 1, 95–103. Ó FEBS 2003 3D structure of indolepyruvate decarboxylase (Eur. J. Biochem. 270) 2321 . Crystal structure of thiamindiphosphate-dependent indolepyruvate decarboxylase from Enterobacter cloacae , an enzyme involved in the biosynthesis of the. catalyses the formation of indoleacetaldehyde from indolepyruvate, one step in the indolepyruvate path- way of biosynthesis of the plant hormone indole-3-acetic acid.